1 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
7 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
8 class pass_optimize_bswap, bswap_replace,
9 pass_optimize_bswap::execute): Moved to ...
10 * gimple-ssa-store-merging.c: ... this file.
11 Include optabs-tree.h.
12 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
13 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
14 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
15 anonymous namespace, remove static keywords.
16 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
17 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
19 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
22 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
24 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
26 * vec.h (debug_helper): New function.
27 (DEFINE_DEBUG_VEC): New macro.
28 * hash-set.h (debug_helper): New function.
29 (DEFINE_DEBUG_HASH_SET): New macro.
30 * cfg.c (debug_slim (edge)): New function.
31 Call DEFINE_DEBUG_VEC for edges.
32 Call DEFINE_DEBUG_HASH_SET for edges.
33 * cfghooks.c (debug_slim (basic_block)): New function.
34 Call DEFINE_DEBUG_VEC for basic blocks.
35 Call DEFINE_DEBUG_HASH_SET for basic blocks.
36 * print-tree.c (debug_slim): New function to handle trees.
37 Call DEFINE_DEBUG_VEC for trees.
38 Call DEFINE_DEBUG_HASH_SET for trees.
39 (debug (vec<tree, va_gc>) &): Remove.
40 (debug (<vec<tree, va_gc>) *): Remove.
41 * print-rtl.c (debug_slim): New function to handle const_rtx.
42 Call DEFINE_DEBUG_VEC for rtx_def.
43 Call DEFINE_DEBUG_VEC for rtx_insn.
44 Call DEFINE_DEBUG_HASH_SET for rtx_def.
45 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
46 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
47 (debug (vec<rtx_insn *> *ptr): Remove.
48 (debug_insn_vector): Remove.
49 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
51 2017-11-20 Tom de Vries <tom@codesourcery.com>
53 PR rtl-optimization/82020
54 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
55 IF_THEN_ELSE condition.
57 2017-11-19 Jeff Law <law@redhat.com>
59 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
60 of degenerates resulting from ignoring an edge.
62 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
65 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
67 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
70 * profile-count.c (profile_count::to_sreal_scale): Fix return value
71 for uninitialied counts.
73 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
76 * tree-inline.c (estimate_num_insns): Set to 1 at least.
78 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
81 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
84 2017-11-19 Tom de Vries <tom@codesourcery.com>
86 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
89 2017-11-19 Tom de Vries <tom@codesourcery.com>
91 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
94 2017-11-19 Tom de Vries <tom@codesourcery.com>
96 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
99 2017-11-19 Tom de Vries <tom@codesourcery.com>
101 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
103 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
105 2017-11-19 Tom de Vries <tom@codesourcery.com>
107 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
109 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
110 semicolon after MCORE_EXPORT_NAME call.
112 2017-11-19 Tom de Vries <tom@codesourcery.com>
115 * vmsdbgout.c (vmsdbgout_early_finish): New function.
116 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
118 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
120 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
121 (cgraph_edge::clone): Cleanup updating of profile.
122 * ipa-cp.c (update_profiling_info): Likewise.
123 * ipa-inline-transform.c (inline_transform): Likewise.
124 * ipa-inline.c (inline_small_functions): Add missing space to dump.
125 * ipa-split.c (execute_split_functions): Do not split when function
127 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
128 * profile-count.c (profile_count::dump): Add global0.
129 (profile_count::to_cgraph_frequency): Do not ICE when entry is
131 (profile_count::to_sreal_scale): Likewise.
132 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
133 (profile_count::combine_with_ipa_count): New function.
134 * profile-count.h (profile_guessed_global0adjusted): New.
135 (profile_count::adjusted_zero): New.
136 (profile_count::global0adjusted): New.
137 (profile_count::combine_with_ipa_count): New.
138 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
139 correct profile of return block of split functions.
140 (copy_cfg_body): Remove unused profile_count.
141 (copy_body): Likewise.
142 (expand_call_inline): Update.
143 (tree_function_versioning): Update.
145 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
147 * hash-set.h (hash_set::empty): New.
148 * tree-ssa-threadbackward.h: Delete.
149 * tree-ssa-threadbackward.c (class thread_jumps): New.
150 Move max_threaded_paths into class.
151 (fsm_find_thread_path): Remove arguments that are now in class.
152 (profitable_jump_thread_path): Rename to...
153 (thread_jumps::profitable_jump_thread_path): ...this.
154 (convert_and_register_jump_thread_path): Rename to...
155 (thread_jumps::convert_and_register_current_path): ...this.
156 (check_subpath_and_update_thread_path): Rename to...
157 (thread_jumps::check_subpath_and_update_thread_path): ...this.
158 (register_jump_thread_path_if_profitable): Rename to...
159 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
160 (handle_phi): Rename to...
161 (thread_jumps::handle_phi): ...this.
162 (handle_assignment): Rename to...
163 (thread_jumps::handle_assignment): ...this.
164 (fsm_find_control_statement_thread_paths): Rename to...
165 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
166 (find_jump_threads_backwards): Rename to...
167 (thread_jumps::find_jump_threads_backwards): ...this.
168 Initialize path local data.
169 (pass_thread_jumps::execute): Call find_jump_threads_backwards
170 from within thread_jumps class.
171 (pass_early_thread_jumps::execute): Same.
173 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
175 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
177 2017-11-17 Steve Ellcey <sellcey@cavium.com>
179 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
180 to canonical location.
182 2017-11-17 Steve Ellcey <sellcey@cavium.com>
185 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
187 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
189 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
191 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
192 rcrt1.o%s/grcrt1.o%s for -static-pie.
194 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
196 * i386.c (ix86_multiplication_cost, ix86_division_cost,
197 ix86_shift_rotate_cost): Break out from ...
198 (ix86_rtx_costs): ... here.
199 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
202 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
204 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
205 when entry block was promoted unlikely.
206 (estimate_bb_frequencies): Increase frequency scale.
207 * profile-count.h (profile_count): Export precision info.
209 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
211 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
212 disturbin profile of entry block.
214 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
216 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
219 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
221 * ipa-cp.c (update_profiling_info): Handle conversion to local
223 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
225 2017-11-17 Jeff Law <law@redhat.com>
227 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
228 from evrp_dom_walker class. Various methods moved into new class.
229 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
230 (evrp_range_analyzer::enter): New method.
231 (evrp_range_analyzer::leave): New method.
232 (evrp_dom_walker): Remove delegators no longer needed by this class.
233 Replace vr_values data member with evrp_range_analyzer
235 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
236 method extracted from evrp_dom_walker::before_dom_children.
237 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
238 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
240 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
241 Add private copy constructor and move assignment operators.
242 Privatize methods and class data where trivially possible.
243 (evrp_dom_walker::cleanup): New function, extracted from
244 execute_early_vrp. Simplify access to class data.
246 * vr-values.h (get_output_for_vrp): Prototype.
247 * vr-values.c (get_output_for_vrp): New function extracted from
248 vrp_visit_assignment_or_call and extract_range_from_stmt.
249 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
251 2017-11-17 Luis Machado <luis.machado@linaro.org>
253 * config/aarch64/aarch64.c
254 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
255 (qdf24xx_tunings) <autoprefetcher_model>: Set to
256 tune_params::AUTOPREFETCHER_WEAK.
258 2017-11-17 Tamar Christina <tamar.christina@arm.com>
261 * config/arm/arm.c (arm_valid_target_attribute_rec):
262 Parse "arch=" and "+<ext>".
263 (arm_valid_target_attribute_tree): Re-init global options.
264 (arm_option_override): Make non-static.
265 (arm_options_perform_arch_sanity_checks): Make errors fatal.
266 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
267 (__ARM_FEATURE_CRC32): Support undef.
268 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
269 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
271 2017-11-17 David Malcolm <dmalcolm@redhat.com>
273 * gdbinit.in (break-on-diagnostic): New command.
275 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
277 * config/i386/i386.c (ix86_expand_epilogue): Change simple
278 return to indirect jump for EH return if control-flow
279 protection is enabled. Change explicit 'false' argument in
280 pro_epilogue_adjust_stack with a value of
282 * config/i386/i386.md (simple_return_indirect_internal):
283 Remove SImode restriction to support 64-bit.
285 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
287 * combine.c (added_notes_insn): New.
288 (try_combine): Handle added_notes_insn like added_links_insn.
289 Rewrite return value code.
290 (distribute_notes): Set added_notes_insn to the earliest insn we added
293 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
295 PR rtl-optimization/82621
296 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
297 dest of one of those SETs is unused.
299 2017-11-17 Richard Biener <rguenther@suse.de>
302 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
303 * tree-pretty-print.c (dump_generic_node): Handle
304 annot_expr_parallel_kind.
305 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
306 * gimplify.c (gimple_boolify): Likewise.
308 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
310 * config.gcc (extra_headers): Add cet.h for x86 targets.
311 * config/i386/cet.h: New file.
312 * doc/install.texi: Add --enable-cet/--disable-cet.
314 2017-11-17 Richard Biener <rguenther@suse.de>
316 PR tree-optimization/83017
317 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
318 (gen_parallel_loop): Properly count iterations.
319 (parallelize_loops): Handle loop->can_be_parallel independent
320 of flag_loop_parallelize_all. Make static profitability test match
322 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
323 * invoke.texi (parloops-min-per-thread): Document.
325 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
327 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
328 upstreaming review comments.
330 2017-11-17 Tamar Christina <tamar.christina@arm.com>
332 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
333 with fast unaligned access.
334 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
336 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
338 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
339 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
341 2017-11-17 Richard Biener <rguenther@suse.de>
343 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
344 folding of references.
346 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
349 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
350 of replacing call to strncmp with corresponding call to strcmp when
353 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
355 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
356 option prefer-avx256 for skylake-avx512 configuration.
357 * config/i386/i386.c (ix86_option_override_internal): Ditto.
358 (get_builtin_code_for_version): Ditto.
360 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
361 Monk Chiang <sh.chiang04@gmail.com>
363 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
364 (FIXED_REGISTERS): Reserve more register numbers.
365 (CALL_USED_REGISTERS): Likewise.
366 (REG_ALLOC_ORDER): Likewise.
367 (REG_CLASS_CONTENTS): Likewise.
368 (REGISTER_NAMES): Likewise.
370 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
371 Kito Cheng <kito.cheng@gmail.com>
373 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
375 * config/nds32/iterators.md: Add vector mode iterators and attributes.
377 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
379 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
380 parameter list for vec_splats.
382 2017-11-16 Joseph Myers <joseph@codesourcery.com>
384 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
386 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
389 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
392 * doc/install.texi: Document incompatibility of Perl >=5.6.26
393 with the required version of automake 1.11.6.
395 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
397 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
399 (DU_C2_3_power9): Correct reservation combinations.
400 (FP_DIV_power9, VEC_DIV_power9): New.
401 (power9-alu): Split out rotate/shift...
402 (power9-rot): ...to here, correct dispatch resource.
403 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
405 (power9-fp): Correct latency.
406 (power9-sdiv): Add div/sqrt resource.
407 (power9-ddiv): Correct latency, add div/sqrt resource.
408 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
409 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
411 (power9-qpdiv, power9-qpmul): Adjust resource usage.
413 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
415 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
416 switch statement mapping KF built-ins to TF built-ins if we don't
417 have the proper ISA 3.0 assembler support.
419 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
421 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
422 (lower_emutls_function_body): Do not compute it.
424 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
426 * ipa-split.c (split_bb_info): Turn time to sreal.
427 (split_point): Likewise.
428 (dump_split_point): Likewise.
429 (fine_split_points): Likewise.
430 (execute_split_functions): Only zero split_bbs; turn time to sreals.
432 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
434 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
437 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
439 * predict.c (combine_predictions_for_bb): Preserve zero predicted
441 (expensive_function_p): Remove useless assert.
442 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
444 2017-11-16 Martin Sebor <msebor@redhat.com>
446 PR tree-optimization/82588
447 PR tree-optimization/82583
448 * tree-vrp.c (check_array_ref): Handle flexible array members,
449 string literals, and inner indices.
450 (search_for_addr_array): Add detail to diagnostics.
452 2017-11-16 Nathan Sidwell <nathan@acm.org>
456 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
457 (SET_DECL_ASSEMBLER_NAME): Forward to
458 overwrite_decl_assembler_name.
459 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
460 (overwrite_decl_assembler_name): Declare.
461 * tree.c (overwrite_decl_assembler_name): New.
462 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
463 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
464 (LANG_HOOKS_INITIALIZER): Add it.
465 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
466 * langhooks.c (lhd_set_decl_assembler_name): Use
467 SET_DECL_ASSEMBLER_NAME.
468 (lhd_overwrite_decl_assembler_name): Default implementation.
470 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
471 Jackson Woodruff <jackson.woodruff@arm.com>
473 PR tree-optimization/71026
474 * match.pd: Canonicalize constant multiplies in division.
476 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
478 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
479 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
480 Remove OPT_fomit_frame_pointer entry.
481 * common/config/arc/arc-common.c: Likewise.
482 * common/config/arm/arm-common.c: Likewise.
483 * common/config/avr/avr-common.c: Likewise.
484 * common/config/c6x/c6x-common.c: Likewise.
485 * common/config/cr16/cr16-common.c: Likewise.
486 * common/config/cris/cris-common.c: Likewise.
487 * common/config/epiphany/epiphany-common.c: Likewise.
488 * common/config/fr30/fr30-common.c: Likewise.
489 * common/config/frv/frv-common.c: Likewise.
490 * common/config/ia64/ia64-common.c: Likewise.
491 * common/config/iq2000/iq2000-common.c: Likewise.
492 * common/config/lm32/lm32-common.c: Likewise.
493 * common/config/m32r/m32r-common.c: Likewise.
494 * common/config/mcore/mcore-common.c: Likewise.
495 * common/config/microblaze/microblaze-common.c: Likewise.
496 * common/config/mips/mips-common.c: Likewise.
497 * common/config/mmix/mmix-common.c: Likewise.
498 * common/config/mn10300/mn10300-common.c: Likewise.
499 * common/config/nios2/nios2-common.c: Likewise.
500 * common/config/pa/pa-common.c: Likewise.
501 * common/config/pdp11/pdp11-common.c: Likewise.
502 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
503 * common/config/riscv/riscv-common.c: Likewise.
504 * common/config/rs6000/rs6000-common.c: Likewise.
505 * common/config/rx/rx-common.c: Likewise.
506 * common/config/s390/s390-common.c: Likewise.
507 * common/config/sh/sh-common.c: Likewise.
508 * common/config/sparc/sparc-common.c: Likewise.
509 * common/config/tilegx/tilegx-common.c: Likewise.
510 * common/config/tilepro/tilepro-common.c: Likewise.
511 * common/config/v850/v850-common.c: Likewise.
512 * common/config/visium/visium-common.c: Likewise.
513 * common/config/xstormy16/xstormy16-common.c: Likewise.
514 * common/config/xtensa/xtensa-common.c: Likewise.
515 * invoke.texi (-fomit-frame-pointer): Update documentation.
517 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
519 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
521 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
523 * cfg.c (scale_bbs_frequencies_int,
524 cale_bbs_frequencies_gcov_type): Remove.
525 * cfg.h (scale_bbs_frequencies_int,
526 cale_bbs_frequencies_gcov_type): Remove.
528 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
530 * tree-ssa-loop-manip.c
531 (scale_dominated_blocks_in_loop): Update to profile counts.
532 (tree_transform_and_unroll_loop): Likewise.
534 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
536 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
537 scale_bbs_frequencies_int.
539 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
541 * final.c (compute_alignments): Use counts rather than frequencies.
543 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
545 * cfgloopanal.c: Include sreal.h
546 (average_num_loop_insns): Use counts and sreal for accounting.
548 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
550 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
553 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
554 Kito Cheng <kito.cheng@gmail.com>
556 * config/nds32/constraints.md: Provide more constraints.
557 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
558 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
559 support constraints usage.
561 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
563 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
565 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
566 Kito Cheng <kito.cheng@gmail.com>
568 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
569 * config/nds32/nds32.opt: Refine the layout.
570 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
571 TARGET_EXT_STRING): Support new options.
572 * config/nds32/nds32.h: Likewise.
573 * config/nds32/nds32.md: Likewise.
574 * config/nds32/nds32-predicates.c: Likewise.
575 * config/nds32/constraints.md: Likewise.
576 * common/config/nds32/nds32-common.c: Likewise.
578 2017-11-16 Julia Koval <julia.koval@intel.com>
581 * config/i386/gfniintrin.h: Add sse check.
582 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
584 2017-11-16 Julia Koval <julia.koval@intel.com>
586 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
587 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
588 (ix86_handle_option): Handle -mavx512vbmi2.
589 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
590 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
591 * config/i386/i386-c.c (__AVX512VBMI2__): New.
592 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
593 (ix86_valid_target_attribute_inner_p): Ditto.
594 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
595 * config/i386/i386.opt (mavx512vbmi2): New option.
596 * doc/invoke.texi: Add new option.
598 [2017-11-16 Julia Koval <julia.koval@intel.com>
600 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
601 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
602 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
603 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
604 _mm512_gf2p8mul_epi8): New intrinsics.
605 * config/i386/i386-builtin-types.def
606 (V64QI_FTYPE_V64QI_V64QI): New type.
607 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
608 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
609 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
610 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
611 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
612 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
614 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
616 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
617 explicitly as a stream of bytes.
619 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
621 * config/rs6000/altivec.h (vec_xst_be): New #define.
622 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
623 and externalize from *altivec_vperm_<mode>_internal.
624 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
626 (XL_BE_V8HI): Likewise.
627 (XL_BE_V4SI): Likewise.
628 (XL_BE_V4SI): Likewise.
629 (XL_BE_V2DI): Likewise.
630 (XL_BE_V4SF): Likewise.
631 (XL_BE_V2DF): Likewise.
632 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
633 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
634 all array entries with these keys: VSX_BUILTIN_VEC_XL,
635 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
636 VSX_BUILTIN_VEC_XST_BE.
637 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
638 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
640 (altivec_init_builtins): Replace conditional calls to def_builtin
641 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
642 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
643 with unconditional calls. Remove calls to def_builtin for
644 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
645 __builtin_vec_xst_be.
646 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
647 to define_expand, and add alternate RTL generation for P8.
648 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
650 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
651 add alternate RTL generation for P8.
652 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
653 vsx_ld_elemrev_v16qi.
654 (vsx_st_elemrev_v8hi): Convert define_insn
655 to define_expand, and add alternate RTL generation for P8.
656 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
658 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
659 add alternate RTL generation for P8.
660 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
661 vsx_st_elemrev_v16qi.
663 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
666 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
667 TARGET_AVX512ER check.
668 (ix86_option_override_internal): Set MASK_VZEROUPPER if
669 neither -mzeroupper nor -mno-zeroupper is used and
670 TARGET_EMIT_VZEROUPPER is set.
671 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
672 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
674 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
677 folding of vector compares.
678 (fold_build_vec_cmp): New helper function.
679 (fold_compare_helper): New helper function.
680 (builtin_function_type): Add compare builtins to the list of functions
681 having unsigned arguments. Cosmetic updates to comment indentation.
682 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
683 the not+eq combination.
685 2017-11-15 Bin Cheng <bin.cheng@arm.com>
687 PR tree-optimization/82726
688 PR tree-optimization/70754
689 * tree-predcom.c (order_drefs_by_pos): New function.
690 (combine_chains): Move code setting has_max_use_after to...
691 (try_combine_chains): ...here. New parameter. Sort combined chains
692 according to position information.
693 (tree_predictive_commoning_loop): Update call to above function.
694 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
696 2017-11-15 Bin Cheng <bin.cheng@arm.com>
698 PR tree-optimization/82726
700 2017-01-23 Bin Cheng <bin.cheng@arm.com>
702 PR tree-optimization/70754
703 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
704 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
705 combined stmt before it if not NULL.
706 (combine_chains): Process refs reversely and compute dominance point
710 2017-02-23 Bin Cheng <bin.cheng@arm.com>
712 PR tree-optimization/79663
713 * tree-predcom.c (combine_chains): Process refs in reverse order
714 only for ZERO length chains, and add explaining comment.
716 2017-11-15 Tamar Christina <tamar.christina@arm.com>
718 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
719 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
720 * doc/invoke.texi (ARM Options): Add armv8.3-a.
722 2017-11-15 Tamar Christina <tamar.christina@arm.com>
724 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
726 2017-11-15 Martin Liska <mliska@suse.cz>
728 * tree-cfg.c (pass_warn_function_return::execute):
729 Compare warn_return_type for greater than zero.
731 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
735 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
736 to return true on Xeon and not on Xeon Phi.
737 (ix86_check_avx256_register): Changed to ...
738 (ix86_check_avx_upper_register): ... this. Add extra check for
739 VALID_AVX512F_REG_OR_XI_MODE.
740 (ix86_avx_u128_mode_needed): Changed
741 ix86_check_avx256_register to ix86_check_avx_upper_register.
742 (ix86_check_avx256_stores): Changed to ...
743 (ix86_check_avx_upper_stores): ... this. Changed
744 ix86_check_avx256_register to ix86_check_avx_upper_register.
745 (ix86_avx_u128_mode_after): Changed
746 avx_reg256_found to avx_upper_reg_found. Changed
747 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
748 (ix86_avx_u128_mode_entry): Changed
749 ix86_check_avx256_register to ix86_check_avx_upper_register.
750 (ix86_avx_u128_mode_exit): Ditto.
751 * config/i386/i386.h: (host_detect_local_cpu): New define.
753 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
755 * config/arm/xgene1.md (xgene1): Split into automatons
756 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
757 (xgene1_f_load): Adjust reservations and/or types.
758 (xgene1_f_store): Likewise.
759 (xgene1_load_pair): Likewise.
760 (xgene1_store_pair): Likewise.
761 (xgene1_fp_load1): Likewise.
762 (xgene1_load1): Likewise.
763 (xgene1_store1): Likewise.
764 (xgene1_move): Likewise.
765 (xgene1_alu): Likewise.
766 (xgene1_simd): Likewise.
767 (xgene1_bfm): Likewise.
768 (xgene1_neon_load1): Likewise.
769 (xgene1_neon_store1): Likewise.
770 (xgene1_neon_logic): Likewise.
771 (xgene1_neon_st1): Likewise.
772 (xgene1_neon_ld1r): Likewise.
773 (xgene1_alu_cond): Added.
774 (xgene1_shift_reg): Likwise.
775 (xgene1_bfx): Likewise.
776 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
778 2017-11-15 Jakub Jelinek <jakub@redhat.com>
781 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
783 (can_widen_mult_without_libcall): New function.
784 (expand_mul_overflow): If only checking unsigned mul overflow,
785 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
786 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
787 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
788 (expand_DIVMOD): Formatting fix.
789 * expmed.h (expand_mult): Add NO_LIBCALL argument.
790 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
791 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
793 PR tree-optimization/82977
794 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
795 constructed temporary to strlen_to_stridx.put.
797 2017-11-15 Martin Liska <mliska@suse.cz>
799 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
800 * configure: Regenerate.
802 2017-11-15 Martin Liska <mliska@suse.cz>
805 * config/sh/sh-mem.cc: Use proper probability for
808 2017-11-14 Jeff Law <law@redhat.com>
810 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
811 the red zone for stack_clash_protection_final_dynamic_probe targets
812 when the total dynamic stack size is zero bytes.
814 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
815 blocks is post order.
817 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
819 * dumpfile.h (TDF_COMPARE_DEBUG): New.
820 * final.c (rest_of_clean_state): Set it for the
821 -fcompare-debug dump.
822 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
823 class when TDF_COMPARE_DEBUG is set.
825 * dwarf2out.c (gen_producer_string): Discard
828 2017-11-15 Joseph Myers <joseph@codesourcery.com>
831 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
832 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
833 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
834 (__TGMATH_REAL_2_3): Remove macros.
835 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
836 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
838 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
840 (remquo): Define using __TGMATH_REAL_3.
842 2017-11-14 Jeff Law <law@redhat.com>
844 * vr-values.c: New file with contents extracted from tree-vrp.c.
845 * Makefile.in (OBJS): Add vr-values.o
846 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
847 (set_value_range, set_and_canonicalize_value_range): Likewise.
848 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
849 (value_range_constant_singleton, symbolic_range_p): Likewise.
850 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
851 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
852 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
853 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
854 (find_case_label_range, find_case_label_index): Likewise.
855 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
856 (range_int_cst_singleton_p, value_inside_range): Likewise.
857 (get_single_symbol): Likewise.
858 (switch_update): Move structure definition here.
859 (to_remove_edges, to_update_switch_stmts): Provide externs.
860 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
861 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
862 (vrp_val_is_min, set_value_range): Likewise.
863 (set_and_canonicalize_value_range, copy_value_range): Likewise.
864 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
865 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
866 (range_is_nonnull, range_int_cst_p): Likewwise.
867 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
868 (get_single_symbol, operand_less_p): Likewise
869 (compare_values_warnv, compare_values): Likewise.
870 (value_inside_range, value_range_constant_singleton): Likewise.
871 (zero_nonzero_bitgs_from_vr): Likewise.
872 (extract_range_from_binary_expr_1): Likewise.
873 (overflow_comparison_p): Likewise.
874 (to_remove_edges, to_update_switch_stmts): Likewise.
875 (find_case_label-index, find_case_label_range): Likewise.
876 (switch_update, set_value_range_to_nonnegative): Remove.
877 (set_value_range_to_truthvalue): Likewise.
878 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
879 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
880 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
881 (find_case_label_ranges, test_for_singularity): Likewise.
882 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
883 (x_vr_values): Move to its remaining use site.
885 2017-11-10 Jeff Law <law@redhat.com>
887 * vr-values.h (VR_INITIALIZER): Move #define here.
888 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
889 * Makefile.in (OBJS): Add tree-evrp.o
890 * tree-vrp.h (assert_info): Move structure definition here.
891 (set_value_range_to_varying): Prototype.
892 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
893 (infer_value_range, register_edge_assert_for): Likewise.
894 (stmt_interesting_for_vrp): Likewise.
895 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
896 (set_value_range_to_varying): No longer static.
897 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
898 (infer_value_range, register_edge_assert_for): Likewise.
900 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
902 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
903 generate the XXBRD instruction.
905 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
906 (rs6000_builtin_type_compatible): Treat _Float128 and long double
907 as being compatible if -mabi=ieeelongdouble.
908 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
909 to setup float128 built-ins with hardware support.
910 (BU_FLOAT128_HW_2): Likewise.
911 (BU_FLOAT128_HW_3): Likewise.
912 (BU_FLOAT128_HW_VSX_1): Likewise.
913 (BU_FLOAT128_HW_VSX_2): Likewise.
914 (scalar_extract_expq): Change float128 built-in functions to
915 accommodate having both KFmode and TFmode functions. Use the
916 KFmode variant as the default.
917 (scalar_extract_sigq): Likewise.
918 (scalar_test_neg_qp): Likewise.
919 (scalar_insert_exp_q): Likewise.
920 (scalar_insert_exp_qp): Likewise.
921 (scalar_test_data_class_qp): Likewise.
922 (sqrtf128_round_to_odd): Delete processing the round to odd
923 built-in functions as special built-in functions, and define them
924 as float128 built-ins. Use the KFmode variant as the default.
925 (truncf128_round_to_odd): Likewise.
926 (addf128_round_to_odd): Likewise.
927 (subf128_round_to_odd): Likewise.
928 (mulf128_round_to_odd): Likewise.
929 (divf128_round_to_odd): Likewise.
930 (fmaf128_round_to_odd): Likewise.
931 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
932 support for KFmode and TFmode xststdcqp calls.
933 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
934 point, switch the built-in handlers for the get/set float128
935 exponent, get float128 mantissa, float128 test built-ins, and the
936 float128 round to odd built-in functions. Eliminate creating the
937 float128 round to odd built-in functions as special built-ins.
938 (rs6000_init_builtins): Eliminate special creation of the float128
939 round to odd built-in functions.
940 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
941 function insns to support both TFmode and KFmode variants.
942 (xsxsigqp_<mode>): Likewise.
943 (xsiexpqpf_<mode>): Likewise.
944 (xsiexpqp_<mode>): Likewise.
945 (xststdcqp_<mode>): Likewise.
946 (xststdcnegqp_<mode>): Likewise.
947 (xststdcqp_<mode>): Likewise.
949 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
951 * tree-ssa-threadupdate.c (compute_path_counts): Remove
952 unused path_in_freq_ptr parameter.
953 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
955 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
957 * ipa-inline.c (edge_badness): Dump sreal frequency.
958 (compute_inlined_call_time): Match natural implementaiton ...
959 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
960 forgotten division by CGRAPH_FREQ_BASE.
962 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
964 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
965 Solaris 11. Update comment.
966 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
968 * config/sol2.h (STARTFILE_SPEC): Likewise.
969 * configure: Regenerate.
971 2017-11-14 Carl Love <cel@us.ibm.com>
973 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
974 le_ and be_ prefixes to swap* variables. Remove
975 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
978 2017-11-14 Jason Merrill <jason@redhat.com>
980 Support GTY((cache)) on hash_map.
981 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
982 (ggc_cache_remove): Override it instead of ggc_mx.
983 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
984 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
985 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
986 (simple_cache_map_traits): Override maybe_mx.
987 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
988 (hash_map): Friend gt_cleare_cache.
989 (gt_cleare_cache): New.
990 * tree.h (tree_cache_traits): New hash_map traits class.
991 (tree_cache_map): New typedef.
993 2017-11-14 Richard Biener <rguenther@suse.de>
995 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
996 paramter and handling.
997 (cleanup_control_flow_bb): Likewise.
998 (cleanup_control_flow_pre): New helper performing a DFS walk
999 to call cleanup_control_flow_bb in PRE order.
1000 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1001 via cleanup_control_flow_pre.
1003 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1005 * config/aarch64/aarch64-simd.md
1006 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1007 (*aarch64_simd_bsl<mode>_alt): Likewise.
1008 (aarch64_simd_bsldi_internal): New.
1009 (aarch64_simd_bsldi_alt): Likewise.
1011 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1013 * tracer.c (better_p): Do not compare frequencies.
1014 * reg-stack.c (better_edge): Likewise.
1015 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1018 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1020 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1021 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1022 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1023 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1024 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1025 * predict.c (maybe_hot_frequency_p): Inline to ...
1026 (maybe_hot_count_p): ... here; rewrite to counts.
1027 (counts_to_freqs): Rename to ...
1028 (update_max_bb_count): ... this one.
1029 (expensive_function_p): Use counts.
1030 (estimate_bb_frequencies): Update.
1031 (rebuild_frequencies): Update.
1032 * predict.h (counts_to_freqs): Rename to ...
1033 (update_max_bb_count): ... this one.
1034 * profile.c (compute_branch_probabilities): Add debug info
1035 * tree-inline.c (expand_call_inline): Update debug info.
1036 (optimize_inline_calls): Use update_max_bb_count..
1037 (tree_function_versioning): Use update_max_bb_count..
1038 * value-prof.c (gimple_value_profile_transformations):
1039 Do not use update_max_bb_count.
1041 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1043 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1044 always use frequencies.
1046 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1048 * bb-reorder.c: Remove frequencies from comments.
1049 (better_edge_p): Use profile counts.
1050 (find_traces): Dump profile counts.
1051 (rotate_loop): Use profile counts.
1052 (find_traces_1_round): Likewise.
1053 (connect_better_edge_p): Use counts instead of probabilities for
1055 (copy_bb_p): Drop early check for non-0 frequency.
1056 (sanitize_hot_paths): Update comments.
1058 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1060 * ipa-split.c (struct split_point): Add count.
1061 (consider_split): Do not compute incoming frequency; compute incoming
1062 count and store it to split_point.
1063 (split_function): Set count of the call to split part correctly.
1065 2017-11-13 Carl Love <cel@us.ibm.com>
1067 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1069 2017-11-13 Tom Tromey <tom@tromey.com>
1071 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1073 2017-11-13 Carl Love <cel@us.ibm.com>
1075 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1076 Add support for builtins:
1077 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1078 vector {un,}signed {char,int,short},
1079 vector {un,}signed {char,int,short}) arguments.
1080 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1081 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1082 Add BU_P9V_AV_2 expansions for the builtins.
1083 * config/rs6000/altivec.h (vec_first_match_index,
1084 vec_first_mismatch_index, vec_first_match_or_eos_index,
1085 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1086 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1087 new extern declaration.
1088 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1089 * config/rs6000/vsx.md (first_match_index_<mode>,
1090 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1091 first_mismatch_or_eos_index_<mode>): Add define expand.
1092 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1093 * doc/extend.texi: Update the built-in documenation file for the new
1096 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1098 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1099 into the min/max operations for _Float<N> and _Float<N>X types.
1101 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1104 * dwarf2out.c (do_eh_frame): New static variable.
1105 (dwarf2out_begin_prologue): Set it.
1106 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1108 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1110 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1112 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1114 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1117 * debug.h (dwarf2out_do_eh_frame): Declare.
1118 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1119 (dwarf2out_do_frame): Use it.
1120 (dwarf2out_do_cfi_asm): Likewise.
1121 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1122 (dwarf2out_assembly_start): Likewise.
1123 (dwarf2out_begin_prologue): Fix comment.
1124 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1125 if the target needs either debug or unwind DWARF2 info.
1126 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1127 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1128 -ftrapping-math, -ftrapv and -fwrapv.
1130 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1132 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1133 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1134 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1135 (estimate_edge_size_and_time): Likewise.
1136 (ipa_merge_fn_summary_after_inlining): Likewise.
1137 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1138 (compute_uninlined_call_time): Use sreal_frequency.
1139 (compute_inlined_call_time): Likewise.
1140 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1141 * profile-count.c: Include sreal.h.
1142 (profile_count::to_sreal_scale): New.
1143 * profile-count.h: Forward declare sreal.
1144 (profile_count::to_sreal_scale): Declare.
1146 2017-11-13 Nathan Sidwell <nathan@acm.org>
1148 * diagnostic.c (maybe_line_and_column): New.
1149 (diagnostic_get_location_text): Use it.
1150 (diagnostic_report_current_module): Likewise.
1151 (test_diagnostic_get_location_text): Add tests.
1153 2017-11-13 Luis Machado <luis.machado@linaro.org>
1155 * doc/md.texi (Specifying processor pipeline description): Fix
1156 incorrect latency for the div instruction example.
1158 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1160 PR tree-optimization/78821
1161 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1162 that bit_not_p is the same.
1163 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1164 (split_group): Count precisely bit_not_p bits in each statement.
1165 (invert_op): New function.
1166 (imm_store_chain_info::output_merged_store): Use invert_op to
1167 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1168 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1170 2017-11-13 Martin Liska <mliska@suse.cz>
1172 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1173 (struct source_info): Likewise.
1174 (add_branch_counts): Likewise.
1175 (add_line_counts): Likewise.
1176 (function_summary): Likewise.
1177 (output_intermediate_line): Likewise.
1178 (generate_results): Likewise.
1180 2017-11-13 Martin Liska <mliska@suse.cz>
1182 * gcov.c (struct block_info): Remove typedef for block_t.
1183 (struct line_info): Likewise.
1184 (line_info::has_block): Likewise.
1185 (EXIT_BLOCK): Likewise.
1186 (unblock): Likewise.
1187 (circuit): Likewise.
1188 (get_cycles_count): Likewise.
1189 (process_file): Likewise.
1190 (read_graph_file): Likewise.
1191 (solve_flow_graph): Likewise.
1192 (find_exception_blocks): Likewise.
1193 (add_line_counts): Likewise.
1194 (accumulate_line_info): Likewise.
1195 (output_line_details): Likewise.
1197 2017-11-13 Martin Liska <mliska@suse.cz>
1199 * gcov.c (struct arc_info): Remove typedef for arc_t.
1200 (struct line_info): Likewise.
1201 (add_branch_counts): Likewise.
1202 (output_branch_count): Likewise.
1203 (function_info::~function_info): Likewise.
1204 (circuit): Likewise.
1205 (output_intermediate_line): Likewise.
1206 (read_graph_file): Likewise.
1207 (solve_flow_graph): Likewise.
1208 (find_exception_blocks): Likewise.
1209 (add_line_counts): Likewise.
1210 (accumulate_line_info): Likewise.
1211 (output_line_details): Likewise.
1212 (output_function_details): Likewise.
1214 2017-11-13 Martin Liska <mliska@suse.cz>
1216 * gcov.c (struct function_info): Remove typedef for function_t.
1217 (struct source_info): Likewise.
1218 (source_info::get_functions_at_location): Likewise.
1219 (solve_flow_graph): Likewise.
1220 (find_exception_blocks): Likewise.
1221 (add_line_counts): Likewise.
1222 (output_intermediate_file): Likewise.
1223 (process_file): Likewise.
1224 (generate_results): Likewise.
1225 (release_structures): Likewise.
1226 (read_graph_file): Likewise.
1227 (read_count_file): Likewise.
1228 (accumulate_line_counts): Likewise.
1229 (output_lines): Likewise.
1231 2017-11-13 Martin Liska <mliska@suse.cz>
1233 * gcov.c (function_info::function_info): Remove num_counts
1234 and add vector<gcov_type>.
1235 (function_info::~function_info): Use the vector.
1236 (process_file): Likewise.
1237 (read_graph_file): Likewise.
1238 (read_count_file): Likewise.
1239 (solve_flow_graph): Likewise.
1241 2017-11-13 Martin Liska <mliska@suse.cz>
1243 * gcov.c (function_info::is_artificial): New function.
1244 (process_file): Erase all artificial early.
1245 (generate_results): Skip as all artificial are already
1248 2017-11-13 Martin Liska <mliska@suse.cz>
1250 * gcov.c (read_graph_file): Store to global vector of functions.
1251 (read_count_file): Iterate the vector.
1252 (process_file): Likewise.
1253 (generate_results): Likewise.
1254 (release_structures): Likewise.
1256 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1258 PR tree-optimization/82954
1259 * gimple-ssa-store-merging.c
1260 (imm_store_chain_info::coalesce_immediate_stores): If
1261 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1263 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1265 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1266 Upddate call to ENDIAN_LANE_N.
1267 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1268 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1269 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1270 and use aarch64_endian_lane_rtx.
1271 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1273 2017-11-12 Tom de Vries <tom@codesourcery.com>
1275 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1277 2017-11-12 Tom de Vries <tom@codesourcery.com>
1279 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1281 2017-11-12 Tom de Vries <tom@codesourcery.com>
1283 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1285 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1286 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1287 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1289 2017-11-11 Martin Sebor <msebor@redhat.com>
1292 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1295 * prefic.c (translate_name): Replace strncpy with memcpy to
1296 avoid -Wstringop-truncation.
1298 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1300 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1302 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1304 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1306 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1308 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1311 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1313 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1314 enable generating XXBRH if the value is in a vector register.
1315 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1316 value is in a vector register.
1317 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1318 register to register bswap64's instead of doing the GPR sequence
1319 used on previous machines.
1320 (bswapdi2_xxbrd): New insn.
1321 (bswapdi2_reg): Disallow on ISA 3.0.
1322 (register to register bswap64 splitter): Do not split the insn on
1323 ISA 3.0 systems that use XXBRD.
1325 2017-11-10 Martin Sebor <msebor@redhat.com>
1328 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1330 (find_subframework_file): Same.
1332 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1334 * auto-profile.c (afdo_indirect_call): Drop frequency.
1335 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1336 (cgraph_node::create_edge): Drop frequency argument.
1337 (cgraph_node::create_indirect_edge): Drop frequency argument.
1338 (cgraph_edge::make_speculative): Drop frequency arguments.
1339 (cgraph_edge::resolve_speculation): Do not update frequencies
1340 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1341 (cgraph_node::dump): Check consistency in IPA mode.
1342 (cgraph_edge::maybe_hot_p): Use IPA counter.
1343 (cgraph_edge::verify_count_and_frequency): Rename to ...
1344 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1345 (cgraph_node::verify_node): Update.
1346 * cgraph.h (struct cgraph_edge): Drop frequency.
1347 (cgraph_edge::frequency): New function.
1348 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1350 (cgraph_edge::rebuild_edges): Likewise.
1351 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1352 (duplicate_thunk_for_node): Do not pass frequency.
1353 (cgraph_node::create_clone): Scale only counts.
1354 (cgraph_node::create_virtual_clone): Do not pass frequency.
1355 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1356 (cgraph_node::create_version_clone): Do not pass frequency.
1357 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1358 (cgraph_node::expand_thunk): Do not pass frequency.
1359 (cgraph_node::create_wrapper): Do not pass frequency.
1360 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1362 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1363 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1364 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1365 (gather_caller_stats): Use frequency function.
1366 (ipcp_cloning_candidate_p): Use frequency function.
1367 (ipcp_propagate_stage): Use frequency function.
1368 (get_info_about_necessary_edges): Use frequency function.
1369 (update_profiling_info): Update only IPA profile.
1370 (update_specialized_profile): Use frequency functoin.
1371 (perhaps_add_new_callers): Update only IPA profile.
1372 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1373 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1374 (dump_ipa_call_summary): Use frequency function.
1375 (estimate_edge_size_and_time): Use frequency function.
1376 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1377 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1378 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1379 (update_noncloned_counts): ... ths one; scale counts only.
1380 (clone_inlined_nodes): Do not scale frequency.
1381 (inline_call): Do not pass frequency.
1382 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1383 (compute_inlined_call_time): Use IPA profile.
1384 (want_inline_small_function_p): Use IPA profile.
1385 (want_inline_self_recursive_call_p): Use IPA profile.
1386 (edge_badness): Use IPA profile.
1387 (lookup_recursive_calls): Use IPA profile.
1388 (recursive_inlining): Do not pass frequency.
1389 (resolve_noninline_speculation): Do not update frequency.
1390 (inline_small_functions): Collect max of IPA profile.
1391 (dump_overall_stats): Dump IPA porfile.
1392 (dump_inline_stats): Dump IPA porfile.
1393 (ipa_inline): Collect IPA stats.
1394 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1395 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1396 (ipa_propagate_frequency): Use frequency function.
1397 (ipa_profile): Cleanup.
1398 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1399 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1400 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1401 (input_node): Do not stream frequency.
1402 (input_edge): Do not stream frequency.
1403 (merge_profile_summaries): Scale only IPA profiles.
1404 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1405 * predict.c (drop_profile): Do not recompute frequency.
1406 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1407 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1408 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1409 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1410 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1411 * tree-inline.c (copy_bb): Do not scale frequency.
1412 (expand_call_inline): Do not scale frequency.
1413 (tree_function_versioning): Do not scale frequency.
1414 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1416 2017-11-10 Julia Koval <julia.koval@intel.com>
1418 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1419 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1420 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1421 (_mm256_maskz_gf2p8affine_epi64_epi8)
1422 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1423 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1424 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1425 (__builtin_ia32_vgf2p8affineqb_v32qi)
1426 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1427 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1429 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1433 (arm_option_override): Refactor.
1434 (arm_option_reconfigure_globals): New.
1435 (arm_options_perform_arch_sanity_checks): New.
1436 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1438 (arm_options_perform_arch_sanity_checks): Likewise
1440 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1442 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1443 (power9-qpmul): New.
1444 * rs6000/rs6000.md ("type" attr): Add qmul.
1445 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1446 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1447 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1449 2017-11-10 Martin Sebor <msebor@redhat.com>
1452 * builtins.c (compute_objsize): Handle arrays that
1453 compute_builtin_object_size likes to fail for. Make extern.
1454 * builtins.h (compute_objsize): Declare.
1455 (check_strncpy_sizes): New function.
1456 (expand_builtin_strncpy): Call check_strncpy_sizes.
1457 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1458 -Wstringop-truncation.
1459 (gimple_fold_builtin_strncat): Same.
1460 * gimple.c (gimple_build_call_from_tree): Set call location.
1461 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1462 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1463 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1464 (handle_builtin_strlen): Use strlen_to_stridx.
1465 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1467 Use strlen_to_stridx.
1468 (pass_strlen::execute): Release strlen_to_stridx.
1469 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1470 (-Wstringop-truncation): Document new option.
1472 2017-11-10 Martin Liska <mliska@suse.cz>
1474 PR gcov-profile/82702
1475 * gcov.c (main): Handle intermediate files in a different
1477 (get_gcov_intermediate_filename): New function.
1478 (output_gcov_file): Remove support of intermediate files.
1479 (generate_results): Allocate intermediate file.
1480 (release_structures): Clean-up properly fn_end.
1481 (output_intermediate_file): Start iterating with line 1.
1483 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1485 PR tree-optimization/82929
1486 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1487 ops_swapped_p non-static data member.
1488 (store_immediate_info::store_immediate_info): Clear it.
1489 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1490 ops set ops_swapped_p.
1491 (count_multiple_uses): Handle ops_swapped_p.
1493 2017-11-10 Martin Liska <mliska@suse.cz>
1495 * coverage.c (coverage_init): Stream information about
1496 support of has_unexecuted_blocks.
1497 * doc/gcov.texi: Document that.
1498 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1499 * gcov.c (read_graph_file): Likewise.
1500 (output_line_beginning): Fix a small issue with
1503 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1505 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1506 reference of trivial component.
1508 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1511 * gimple-ssa-store-merging.c
1512 (pass_store_merging::terminate_all_aliasing_chains): For
1513 gimple_store_p stmts also call refs_output_dependent_p.
1515 PR rtl-optimization/82913
1516 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1519 2017-11-09 Jeff Law <law@redhat.com>
1521 * vr-values.h: New file with vr_values class.
1522 * tree-vrp.c: Include vr-values.h
1523 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1524 data objects into the vr_values class.
1525 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1526 (get_value_range): Make it a member function within vr_values class.
1527 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1528 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1529 (op_with_constant_singleton_value_range): Likewise.
1530 (extract_range_for_var_from_comparison_expr): Likewise.
1531 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1532 (extract_range_from_binary_expr): Likewise.
1533 (extract_range_from_unary_expr): Likewise.
1534 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1535 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1536 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1537 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1538 (compare_name_with_value, compare_names): Likewise.
1539 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1540 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1541 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1542 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1543 (extract_range_from_phi_node): Likewise.
1544 (simplify_truth_ops_using_ranges): Likewise.
1545 (simplify_div_or_mod_using_ranges): Likewise.
1546 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1547 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1548 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1549 (simplify_float_conversion_using_ranges): Likewise.
1550 (simplify_internal_call_using_ranges): Likewise.
1551 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1552 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1553 poitner via x_vr_values for calls into gimple folder.
1554 (vrp_initialize_lattice): Make this the vr_values ctor.
1555 (vrp_free_lattice): Make this the vr_values dtor.
1556 (set_vr_value): New function.
1557 (class vrp_prop): Add vr_values data member. Add various member
1558 functions as well as member functions that delegate to vr_values.
1559 (check_array_ref): Make a member function within vrp_prop class.
1560 (search_for_addr_array, vrp_initialize): Likewise.
1561 (vrp_finalize): Likewise. Revamp to avoid direct access to
1562 vr_value, values_propagated, etc.
1563 (check_array_bounds): Extract vrp_prop class instance pointer from
1564 walk info structure. Use it to call member functions.
1565 (check_all_array_refs): Make a member function within vrp_prop class.
1566 Smuggle class instance pointer via walk info structure.
1567 (x_vr_values): New local static.
1568 (vrp_valueize): Use x_vr_values to get class instance.
1569 (vr_valueize_1): Likewise.
1570 (class vrp_folder): Add vr_values data member. Add various member
1571 functions as well as member functions that delegate to vr_values.
1572 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1573 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1574 class instance from vr_values. Use it to call member functions.
1575 (vrp_dom_walker): Add vr_values data member.
1576 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1577 instance via x_vr_values.
1578 (identify_jump_threads): Accept vr_values as argument. Store
1579 it into the walker structure.
1580 (evrp_dom_walker): Add vr_values class data member. Add various
1582 (evrp_dom_walker::try_find_new_range): Use vr_values data
1583 member to access the memory allocator.
1584 (evrp_dom_walker::before_dom_children): Store vr_values class
1585 instance into the vrp_folder class.
1586 (evrp_dom_walker::push_value_range): Rework to avoid direct
1587 access to num_vr_values and vr_value.
1588 (evrp_dom_walker::pop_value_range): Likewise.
1589 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1590 Use vr_values to get to dump_all_value_ranges member function.
1591 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1592 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1593 Pass vr_values class instance down to identify_jump_threads.
1594 Remove call to vrp_free_lattice.
1595 (debug_all_value_ranges): Remove.
1597 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1598 (vrp_folder): Likewise.
1600 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1601 Get it from the existing bitmap instead.
1602 (vrp_intersect_ranges_1): Likewise.
1604 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1606 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1608 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1609 set bit_not_p to it.
1610 (imm_store_chain_info::coalesce_immediate_stores): Break group
1611 if bit_not_p is different.
1612 (count_multiple_uses, split_group,
1613 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1614 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1615 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1616 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1617 to store_immediate_info ctor.
1619 2017-11-09 Jim Wilson <jimw@sifive.com>
1621 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1622 (scan_prog_file): Likewise.
1624 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1626 * bb-reorder.c (max_entry_frequency): Remove.
1627 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1628 connect_traces, push_to_next_round_p): Remove prototypes.
1629 (find_traces_1_round): Use counts only.
1630 (push_to_next_round_p): Likewise.
1631 (find_traces): Likewise.
1632 (rotate_loop): Likewise.
1633 (find_traces_1_round): Likewise.
1634 (connect_traces): Likewise.
1635 (edge_order): Likewise.
1637 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1639 * config/arm/arm.c (output_return_instruction): Add comments to
1640 indicate requirement for cmse_nonsecure_entry return to account
1641 for the size of clearing instruction output here.
1642 (thumb_exit): Likewise.
1643 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1644 return in hardfloat mode.
1646 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1648 * config/rs6000/rs6000.c (machine_function): Add a bool,
1649 "toc_is_wrapped_separately".
1650 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1651 if it wasn't explicitly set or unset, we are optimizing for speed, and
1652 doing separate shrink-wrapping.
1653 (rs6000_get_separate_components): Enable the TOC component if
1654 saving the TOC register in the prologue.
1655 (rs6000_components_for_bb): Handle the TOC component.
1656 (rs6000_emit_prologue_components): Store the TOC register where needed.
1657 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1658 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1660 2017-11-09 Martin Jambor <mjambor@suse.cz>
1662 * ipa-param-manipulation.c: New file.
1663 * ipa-param-manipulation.h: Likewise.
1664 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1665 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1666 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1667 (ipa_parm_adjustment): Likewise.
1668 (ipa_parm_adjustment_vec): Likewise.
1669 (ipa_get_vector_of_formal_parms): Moved declaration to
1670 ipa-param-manipulation.h.
1671 (ipa_get_vector_of_formal_parm_types): Likewise.
1672 (ipa_modify_formal_parameters): Likewise.
1673 (ipa_modify_call_arguments): Likewise.
1674 (ipa_combine_adjustments): Likewise.
1675 (ipa_dump_param_adjustments): Likewise.
1676 (ipa_modify_expr): Likewise.
1677 (ipa_get_adjustment_candidate): Likewise.
1678 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1679 ipa-param-manipulation.c.
1680 (ipa_get_vector_of_formal_parm_types): Likewise.
1681 (ipa_modify_formal_parameters): Likewise.
1682 (ipa_modify_call_arguments): Likewise.
1683 (ipa_modify_expr): Likewise.
1684 (get_ssa_base_param): Likewise.
1685 (ipa_get_adjustment_candidate): Likewise.
1686 (index_in_adjustments_multiple_times_p): Likewise.
1687 (ipa_combine_adjustments): Likewise.
1688 (ipa_dump_param_adjustments): Likewise.
1689 * tree-sra.c: Also include ipa-param-manipulation.h
1690 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1693 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1694 Alan Hayward <alan.hayward@arm.com>
1695 David Sherwood <david.sherwood@arm.com>
1697 * doc/sourcebuild.texi (vect_masked_store): Document.
1699 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1700 Alan Hayward <alan.hayward@arm.com>
1701 David Sherwood <david.sherwood@arm.com>
1703 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1705 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1706 Alan Hayward <alan.hayward@arm.com>
1707 David Sherwood <david.sherwood@arm.com>
1709 * doc/sourcebuild.texi (vect_variable_length): Document.
1711 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1712 Alan Hayward <alan.hayward@arm.com>
1713 David Sherwood <david.sherwood@arm.com>
1715 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
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 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1723 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1724 Alan Hayward <alan.hayward@arm.com>
1725 David Sherwood <david.sherwood@arm.com>
1727 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1728 previously undocumented selectors.
1729 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1731 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1733 * doc/rtl.texi (const_vector): Say that elements can be
1734 const_wide_ints too.
1735 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1736 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1737 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1738 * optabs.c (expand_vector_broadcast): Likewise.
1740 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1741 Alan Hayward <alan.hayward@arm.com>
1742 David Sherwood <david.sherwood@arm.com>
1744 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1745 scaled index even if the unscaled address was invalid.
1746 Don't increase the complexity of using a scale in that case.
1748 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1749 Alan Hayward <alan.hayward@arm.com>
1750 David Sherwood <david.sherwood@arm.com>
1752 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1753 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1754 UNITS_PER_WORD bytes.
1755 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1756 based on REGMODE_NATURAL_SIZE of the inner mode.
1757 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1758 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1759 * expmed.c (lowpart_bit_field_p): Divide the value up into
1760 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1761 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1762 whether something is likely to occupy more than one register.
1764 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1767 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1768 profile_count::adjust_for_ipa_scaling.
1769 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1770 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1772 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1774 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1776 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1777 (split_group): Add total_orig and total_new arguments, estimate the
1778 number of statements related to the store group without store merging
1779 and with store merging.
1780 (imm_store_chain_info::output_merged_store): Adjust split_group
1781 callers, punt if estimated number of statements with store merging
1782 is not smaller than estimated number of statements without it.
1784 (handled_load): Remove has_single_use checks.
1785 (pass_store_merging::process_store): Likewise.
1787 2017-11-09 Richard Biener <rguenther@suse.de>
1789 PR tree-optimization/82902
1790 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1792 2017-11-09 Martin Liska <mliska@suse.cz>
1795 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1798 2017-11-09 Martin Liska <mliska@suse.cz>
1800 PR tree-optimization/82669
1801 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1803 2017-11-09 Martin Liska <mliska@suse.cz>
1805 PR gcov-profile/48463
1806 * coverage.c (coverage_begin_function): Output also end locus
1807 of a function and information whether the function is
1809 * gcov-dump.c (tag_function): Parse and print the information.
1810 * gcov.c (INCLUDE_MAP): Add include.
1811 (INCLUDE_SET): Likewise.
1812 (struct line_info): Move earlier in the source file because
1813 of vector<line_info> in function_info structure.
1814 (line_info::line_info): Likewise.
1815 (line_info::has_block): Likewise.
1816 (struct source_info): Add new member index.
1817 (source_info::get_functions_at_location): New function.
1818 (function_info::group_line_p): New function.
1819 (output_intermediate_line): New function.
1820 (output_intermediate_file): Use the mentioned function.
1821 (struct function_start): New.
1822 (struct function_start_pair_hash): Likewise.
1823 (process_file): Add code that identifies group functions.
1824 Assign lines either to global or function scope.
1825 (generate_results): Skip artificial functions.
1826 (find_source): Assign index for each source file.
1827 (read_graph_file): Read new flag artificial and end_line.
1828 (add_line_counts): Assign it either to global of function scope.
1829 (accumulate_line_counts): Isolate core of the function to
1830 accumulate_line_info and call it for both function and global
1832 (accumulate_line_info): New function.
1833 (output_line_beginning): Fix GNU coding style.
1834 (print_source_line): New function.
1835 (output_line_details): Likewise.
1836 (output_function_details): Likewise.
1837 (output_lines): Iterate both source (global) scope and function
1839 (struct function_line_start_cmp): New class.
1840 * doc/gcov.texi: Reflect changes in documentation.
1842 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1845 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1846 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1847 and similarly for not instead of neg.
1849 2017-11-08 Andi Kleen <ak@linux.intel.com>
1851 * config/i386/i386.opt: Add -mforce-indirect-call.
1852 * config/i386/predicates.md: Check for flag_force_indirect_call.
1853 * doc/invoke.texi: Document -mforce-indirect-call
1855 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1857 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1859 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1860 config/riscv/riscv.c (predict.h): New include.
1861 (riscv_slow_unaligned_access_p): No longer static.
1862 (riscv_block_move_straight): Add require.
1863 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1865 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1868 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1869 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1870 nonimmediate_operand predicate for operand 1 instead of
1873 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1875 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1877 * config/aarch64/constraints.md (Uml): New constraint.
1878 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1881 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1883 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1884 of two vec_duplicates into a vec_concat.
1886 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1888 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1889 Simplify vec_merge of vec_duplicate and vec_concat.
1890 * config/aarch64/constraints.md (Utq): New constraint.
1891 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1894 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1896 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1897 Simplify vec_merge of vec_duplicate and const_vector.
1898 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1900 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1901 mode iterator. Update predicate on operand 1 to
1902 handle non-const_vec constants. Delete constraints.
1903 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1905 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1907 PR tree-optimization/78821
1908 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1910 (store_operand_info::store_operand_info): Initialize it to false.
1911 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1912 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1913 store in the group, and if chain_info is non-NULL, to ignore altogether
1915 (compatible_load_p): Fail if bit_not_p does not match.
1916 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1917 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1918 (pass_store_merging::process_store): Adjust
1919 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1920 call terminate_all_aliasing_chains newly when adding a store into
1921 a chain with non-NULL chain_info.
1923 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1925 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1927 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1929 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1931 (aarch64_layout_frame): Initialise emit_frame_chain.
1932 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1933 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1935 2017-11-08 Martin Liska <mliska@suse.cz>
1937 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1938 of gimple_call_internal_p.
1940 2017-11-07 Tom de Vries <tom@codesourcery.com>
1942 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1944 2017-11-07 Tom de Vries <tom@codesourcery.com>
1946 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1949 2017-11-08 Martin Liska <mliska@suse.cz>
1952 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1954 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1956 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1958 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1961 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1962 (*cmp<mode>_ccz_1): New insn with $k alternative.
1965 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1966 RTX_COMM_COMPARE as commutative as well.
1967 (ix86_binary_operator_ok): Formatting fix.
1968 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1969 *<code><mode>3<mask_name><round_saeonly_name>,
1970 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1971 *<s>mul<mode>3_highpart<mask_name>,
1972 *vec_widen_umult_even_v16si<mask_name>,
1973 *vec_widen_umult_even_v8si<mask_name>,
1974 *vec_widen_umult_even_v4si<mask_name>,
1975 *vec_widen_smult_even_v16si<mask_name>,
1976 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1977 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1978 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1979 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1980 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1981 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1982 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1983 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1984 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1985 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1986 ix86_binary_operator_ok. Formatting fixes.
1987 (*<plusminus_insn><mode>3<mask_name><round_name>,
1988 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1991 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1993 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1994 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1995 eq and ne if TARGET_ISEL.
1996 (cmp): New code_iterator.
1997 (UNS, UNSU_, UNSIK): New code_attrs.
1998 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1999 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2001 ("eq<mode>3"): ... this.
2002 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2004 ("ne<mode>3"): ... this.
2006 2017-11-07 Julia Koval <julia.koval@intel.com>
2009 * common/config/i386/i386-common.c
2010 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2011 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2012 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2013 * config/i386/i386.opt: Ditto.
2014 * config/i386/i386.c (ix86_target_string): Ditto.
2015 (ix86_option_override_internal): Ditto.
2016 (ix86_init_mpx_builtins): Move MPX to args2.
2017 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2018 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2019 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2020 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2021 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2022 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2023 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2025 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2028 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2029 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2030 (zero-extendsidi peephole2): Remove peephole.
2032 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2035 * stor-layout.c: Include attribs.h.
2036 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2037 explicit lookup of "aligned" attribute.
2039 2017-11-07 Andrew Waterman <andrew@sifive.com>
2041 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2042 (riscv_expand_block_move): Likewise.
2043 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2045 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2046 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2047 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2048 (riscv_adjust_block_mem): Likewise.
2049 (riscv_block_move_loop): Likewise.
2050 (riscv_expand_block_move): Likewise.
2051 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2053 2017-11-07 Michael Clark <michaeljclark@mac.com>
2055 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2056 (MUSL_DYNAMIC_LINKER): Likewise.
2058 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2060 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2061 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2064 2017-11-07 Richard Biener <rguenther@suse.de>
2066 * match.pd: Fix build.
2068 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2069 Jackson Woodruff <jackson.woodruff@arm.com>
2071 PR tree-optimization/71026
2072 * match.pd: Canonicalize negate in division.
2074 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2077 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2079 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2081 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2082 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2084 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2086 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2087 non-scalar integral types.
2088 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2089 (-(A-B), -(~A)): New transformations.
2091 2017-11-07 Tom de Vries <tom@codesourcery.com>
2093 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2094 semicolon after "do {} while (0)".
2095 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2096 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2097 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2098 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2099 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2101 2017-11-07 Tom de Vries <tom@codesourcery.com>
2103 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2104 after "do {} while (0)".
2105 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2106 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2107 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2108 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2109 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2111 2017-11-07 Tom de Vries <tom@codesourcery.com>
2114 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2116 (arm_rtx_costs_internal): Add missing semicolon after
2117 HANDLE_NARROW_SHIFT_ARITH call.
2119 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2122 disable isel if it was not set explicitly.
2124 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2126 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2128 (add_type_attribute) likewise.
2130 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2132 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2133 of struct ix86_frame.
2134 (ix86_initial_elimination_offset): Likewise.
2135 (ix86_expand_split_stack_prologue): Likewise.
2137 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2139 * tree-vrp.h (enum value_range_type): Update stale comment.
2141 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2142 Alan Hayward <alan.hayward@arm.com>
2143 David Sherwood <david.sherwood@arm.com>
2145 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2146 (aarch64_expand_vec_perm_const): Take the number of units too.
2147 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2148 (aarch64_expand_vec_perm_const): Likewise.
2149 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2150 (vec_perm<mode>): Update accordingly.
2152 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2153 Alan Hayward <alan.hayward@arm.com>
2154 David Sherwood <david.sherwood@arm.com>
2156 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2157 Take the number of units too.
2158 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2159 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2160 but check for a vector mode before rather than after the call.
2161 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2162 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2163 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2164 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2165 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2166 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2167 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2168 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2169 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2170 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2171 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2172 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2173 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2174 (aarch64_sqdmull2_n<mode>): Update accordingly.
2176 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2177 Alan Hayward <alan.hayward@arm.com>
2178 David Sherwood <david.sherwood@arm.com>
2180 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2181 the number of units too.
2182 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2183 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2184 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2185 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2186 (vec_store_lanesxi<mode>): Update accordingly.
2188 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2189 Alan Hayward <alan.hayward@arm.com>
2190 David Sherwood <david.sherwood@arm.com>
2192 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2193 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2194 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2195 of units rather than the mode.
2196 * config/aarch64/iterators.md (nunits): New mode attribute.
2197 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2198 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2199 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2200 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2201 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2202 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2203 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2204 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2205 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2206 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2207 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2208 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2209 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2210 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2211 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2212 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2213 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2214 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2215 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2216 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2217 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2218 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2219 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2220 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2221 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2222 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2223 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2224 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2225 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2226 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2227 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2228 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2229 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2230 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2231 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2232 (aarch64_simd_vec_setv2di): Likewise.
2234 2017-11-06 Carl Love <cel@us.ibm.com>
2236 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2238 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2239 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2241 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2243 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2244 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2245 Add power 8 macro expansions.
2246 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2247 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2248 power 8 instructions. (VSX_XXBR): Add iterator.
2250 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2252 * config/arm/arm.md (predicable_short_it): Change default to "no",
2253 improve documentation, remove uses that are identical to the default.
2254 (enabled_for_depr_it): Rename to enabled_for_short_it.
2255 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2256 Remove default uses.
2257 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2258 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2259 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2260 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2262 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2265 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2266 float128 helper macros, which are no longer used after deleting
2267 the old 'q' built-in functions, and moving the round to odd
2268 built-in functions to being special built-in functions.
2269 (BU_FLOAT128_2): Likewise.
2270 (BU_FLOAT128_1_HW): Likewise.
2271 (BU_FLOAT128_2_HW): Likewise.
2272 (BU_FLOAT128_3_HW): Likewise.
2273 (FABSQ): Delete old 'q' built-in functions.
2274 (COPYSIGNQ): Likewise.
2275 (SQRTF128_ODD): Move round to odd built-in functions to be
2276 special built-in functions, so that we can handle
2277 -mabi=ieeelongdouble.
2278 (TRUNCF128_ODD): Likewise.
2279 (ADDF128_ODD): Likewise.
2280 (SUBF128_ODD): Likewise.
2281 (MULF128_ODD): Likewise.
2282 (DIVF128_ODD): Likewise.
2283 (FMAF128_ODD): Likewise.
2284 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2285 built-in names to 'f128'.
2286 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2287 old 'q' built-in functions, as the machine independent code for
2288 'f128' built-in functions handles this.
2289 (rs6000_expand_builtin): Add expansion for float128 round to odd
2290 functions, keying off on -mabi=ieeelongdouble of whether to use
2291 the KFmode or TFmode variant.
2292 (rs6000_init_builtins): Initialize the _Float128 round to odd
2294 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2295 _Float128 'q' built-in functions are now mapped into the new
2296 'f128' built-in functions.
2298 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2300 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2301 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2303 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2306 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2307 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2308 * toplev.c: Include "ipa-fnsummary.h".
2309 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2311 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2313 PR tree-optimization/82838
2314 * gimple-ssa-store-merging.c
2315 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2316 on a separate gimple_seq which is then appended to seq.
2318 2017-11-06 Jeff Law <law@redhat.com>
2321 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2322 (get_probe_interval): New functions.
2323 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2324 (ix86_adjust_stack_and_probe): Likewise.
2325 (output_adjust_stack_and_probe): Likewise.
2326 (ix86_emit_probe_stack_range): Likewise.
2327 (ix86_expand_prologue): Likewise.
2329 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2331 PR tree-optimization/82816
2332 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2333 if the modes of the two types are the same.
2334 (convert_plusminus_to_widen): Likewise.
2336 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2338 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2340 (usadv16qi): New define_expand.
2341 (usadv8hi): New define_expand.
2343 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2346 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2348 (inline_transform): Likewise.
2349 * predict.c (counts_to_freqs): Remove useless conditional.
2350 * profile-count.h (profile_count::apply_scale): Move sanity check.
2351 * tree-inline.c (copy_bb): Always scale.
2352 (copy_cfg_body): Likewise.
2354 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2357 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2359 (*cmp_ite0): Add enabled_for_depr_it attribute.
2360 (*cmp_ite1): Likewise.
2362 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2364 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2367 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2369 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2370 return the value by pointer.
2371 (extract_range_from_multiplicative_op_1): Update accordingly.
2372 Return as soon as an operation fails.
2374 2017-11-05 Tom de Vries <tom@codesourcery.com>
2377 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2378 (DEF_SANITIZER_BUILTIN): ... here.
2379 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2380 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2382 2017-11-05 Tom de Vries <tom@codesourcery.com>
2385 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2387 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2388 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2389 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2391 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2392 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2393 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2395 2017-11-05 Tom de Vries <tom@codesourcery.com>
2398 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2401 2017-11-04 Michael Clark <michaeljclark@mac.com>
2403 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2404 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2406 (*addsi3_extended): Likewise.
2407 (*addsi3_extended2): Likewise.
2408 (<optab>si3): Likewise.
2409 (<optab>di3): Likewise.
2410 (<optab><mode>3): Likewise.
2411 (<*optabe>si3_internal): Likewise.
2412 (zero_extendqi<SUPERQI:mode>2): Likewise.
2413 (*add<mode>hi3): Likewise.
2414 (*xor<mode>hi3): Likewise.
2415 (<optab>di3): Likewise.
2416 (*<optab>si3_extend): Likewise.
2417 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2418 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2419 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2421 2017-11-04 Andrew Waterman <andrew@sifive.com>
2423 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2424 TARGET_STRICT_ALIGN based upon -mtune argument.
2426 2017-11-04 Andrew Waterman <andrew@sifive.com>
2428 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2430 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2432 config/i386/i386.c (choose_basereg): Use optional scratch
2433 register and add assertion.
2434 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2435 needed, and don't allocate stack.
2436 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2437 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2438 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2440 2017-11-03 Jeff Law <law@redhat.com>
2442 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2443 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2444 to probe at the start of a noreturn function.
2446 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2448 PR tree-optimization/78821
2449 * gimple-ssa-store-merging.c: Update the file comment.
2450 (MAX_STORE_ALIAS_CHECKS): Define.
2451 (struct store_operand_info): New type.
2452 (store_operand_info::store_operand_info): New constructor.
2453 (struct store_immediate_info): Add rhs_code and ops data members.
2454 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2455 and op1r arguments to the ctor, initialize corresponding data members.
2456 (struct merged_store_group): Add load_align_base and load_align
2458 (merged_store_group::merged_store_group): Initialize them.
2459 (merged_store_group::do_merge): Update them.
2460 (merged_store_group::apply_stores): Pick the constant for
2461 encode_tree_to_bitpos from one of the two operands, or skip
2462 encode_tree_to_bitpos if neither operand is a constant.
2463 (class pass_store_merging): Add process_store method decl. Remove
2464 bool argument from terminate_all_aliasing_chains method decl.
2465 (pass_store_merging::terminate_all_aliasing_chains): Remove
2466 var_offset_p argument and corresponding handling.
2467 (stmts_may_clobber_ref_p): New function.
2468 (compatible_load_p): New function.
2469 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2470 if there is overlap and rhs_code is not INTEGER_CST. For
2471 non-overlapping stores terminate group if rhs is not mergeable.
2472 (get_alias_type_for_stmts): Change first argument from
2473 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2474 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2475 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2477 (get_location_for_stmts): Change first argument from
2478 auto_vec<gimple *> & to vec<gimple *> &.
2479 (struct split_store): Remove orig_stmts data member, add orig_stores.
2480 (split_store::split_store): Create orig_stores rather than orig_stmts.
2481 (find_constituent_stmts): Renamed to ...
2482 (find_constituent_stores): ... this. Change second argument from
2483 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2484 to info structures rather than the statements.
2485 (split_group): Rename ALLOW_UNALIGNED argument to
2486 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2487 it. Adjust find_constituent_stores caller.
2488 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2489 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2490 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2491 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2492 (mem_valid_for_store_merging): New function.
2493 (handled_load): New function.
2494 (pass_store_merging::process_store): New method.
2495 (pass_store_merging::execute): Use process_store method. Adjust
2496 terminate_all_aliasing_chains caller.
2498 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2500 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2501 Return true for more constants, symbols and label references.
2502 (aarch64_valid_floating_const): Remove unused function.
2504 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2507 * config/aarch64/aarch64.c (aarch64_layout_frame):
2508 Undo forcing of LR at bottom of frame.
2510 2017-11-03 Jeff Law <law@redhat.com>
2513 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2514 for int_registers_saved.
2516 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2517 extracted from tree-ssa-dom.c.
2518 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2519 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2520 (record_equivalences_from_incoming_edge): Add additional argument
2521 to single_pred_edge_ignoring_loop_edges call.
2522 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2523 (uncprop_dom_walker::before_dom_children): Add additional argument
2524 to single_pred_edge_ignoring_loop_edges call.
2525 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2526 single_pred_edge_ignoring_loop_edges rather than open coding.
2527 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2529 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2531 * match.pd (-(-A)): Rewrite.
2533 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2535 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2536 (rs6000_emit_int_cmove): New declaration.
2537 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2538 (rs6000_emit_sISEL): Delete.
2539 (rs6000_emit_int_cmove): Make non-static.
2540 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2541 instead of rs6000_emit_sISEL.
2543 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2545 * asan.c (create_cond_insert_point): Maintain profile.
2546 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2548 * basic-block.h (struct basic_block_def): Remove frequency.
2549 (EDGE_FREQUENCY): Use to_frequency
2550 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2552 (find_traces): Update to use to_frequency.
2553 (find_traces_1_round): Likewise; use only IPA counts.
2554 (bb_to_key): Likewise.
2555 (connect_traces): Use IPA counts only.
2556 (copy_bb_p): Update to use to_frequency.
2557 (fix_up_crossing_landing_pad): Likewise.
2558 (sanitize_hot_paths): Likewise.
2559 * bt-load.c (basic_block_freq): Likewise.
2560 * cfg.c (init_flow): Set count_max to uninitialized.
2561 (check_bb_profile): Remove frequencies; check counts.
2562 (dump_bb_info): Do not dump frequencies.
2563 (update_bb_profile_for_threading): Update counts only.
2564 (scale_bbs_frequencies_int): Likewise.
2565 (MAX_SAFE_MULTIPLIER): Remove.
2566 (scale_bbs_frequencies_gcov_type): Update counts only.
2567 (scale_bbs_frequencies_profile_count): Update counts only.
2568 (scale_bbs_frequencies): Update counts only.
2569 * cfg.h (struct control_flow_graph): Add count-max.
2570 (update_bb_profile_for_threading): Update prototype.
2571 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2572 (find_many_sub_basic_blocks): Likewise.
2573 * cfgcleanup.c (try_forward_edges): Likewise.
2574 (try_crossjump_to_edge): Likewise.
2575 * cfgexpand.c (expand_gimple_cond): Likewise.
2576 (expand_gimple_tailcall): Likewise.
2577 (construct_init_block): Likewise.
2578 (construct_exit_block): Likewise.
2579 * cfghooks.c (verify_flow_info): Check consistency of counts.
2580 (dump_bb_for_graph): Do not dump frequencies.
2581 (split_block_1): Do not update frequencies.
2582 (split_edge): Do not update frequencies.
2583 (make_forwarder_block): Do not update frequencies.
2584 (duplicate_block): Do not update frequencies.
2585 (account_profile_record): Do not update frequencies.
2586 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2587 for global heuristics.
2588 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2589 (expected_loop_iterations_unbounded): Use counts only.
2590 * cfgloopmanip.c (scale_loop_profile): Simplify.
2591 (create_empty_loop_on_edge): Simplify
2593 (duplicate_loop_to_header_edge): Simplify
2594 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2595 (update_br_prob_note): Take care of removing note when profile
2597 (relink_block_chain): Do not dump frequency.
2598 (rtl_account_profile_record): Use to_frequency.
2599 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2600 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2601 (cgraph_update_edges_for_call_stmt_node): Likewise.
2602 (cgraph_edge::verify_count_and_frequency): Update.
2603 (cgraph_node::verify_node): Temporarily disable frequency verification.
2604 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2605 to_cgraph_frequency.
2606 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2607 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2609 (cgraph_node::expand_thunk): Update profile.
2610 * except.c (dw2_build_landing_pads): Do not update frequency.
2611 * final.c (compute_alignments): Use to_frequency.
2612 (dump_basic_block_info): Do not dump frequency.
2613 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2614 (dump_gimple_bb_header): Do not dump frequency.
2615 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2617 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2618 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2619 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2620 (init_before_recovery): Do not update frequency.
2621 (sched_create_recovery_edges): Do not update frequency.
2622 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2623 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2624 (ipa_cp_c_finalize): Set max_count to uninitialized.
2625 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2626 (param_change_prob): Use counts.
2627 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2629 * ipa-split.c (consider_split): Use to_frequency.
2630 (split_function): Use to_frequency.
2631 * ira-build.c (loop_compare_func): Likewise.
2632 (mark_loops_for_removal): Likewise.
2633 (mark_all_loops_for_removal): Likewise.
2634 * loop-doloop.c (doloop_modify): Do not update frequency.
2635 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2637 * lto-streamer-in.c (input_function): Update count_max.
2638 * omp-expand.c (expand_omp_taskreg): Update count_max.
2639 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2640 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2641 (maybe_hot_count_p): Use ipa counts only.
2642 (maybe_hot_bb_p): Simplify.
2643 (maybe_hot_edge_p): Simplify.
2644 (probably_never_executed): Do not take frequency argument.
2645 (probably_never_executed_bb_p): Do not pass frequency.
2646 (probably_never_executed_edge_p): Likewise.
2647 (combine_predictions_for_bb): Check that profile is nonzero.
2648 (propagate_freq): Do not set frequency.
2649 (drop_profile): Simplify.
2650 (counts_to_freqs): Simplify.
2651 (expensive_function_p): Use to_frequency.
2652 (propagate_unlikely_bbs_forward): Simplify.
2653 (determine_unlikely_bbs): Simplify.
2654 (estimate_bb_frequencies): Add hack to silence graphite issues.
2655 (compute_function_frequency): Use ipa counts.
2656 (pass_profile::execute): Update.
2657 (rebuild_frequencies): Use counts only.
2658 (force_edge_cold): Use counts only.
2659 * profile-count.c (profile_count::dump): Dump new count types.
2660 (profile_count::differs_from_p): Check compatiblity.
2661 (profile_count::to_frequency): New function.
2662 (profile_count::to_cgraph_frequency): New function.
2663 * profile-count.h (struct function): Declare.
2664 (enum profile_quality): Add profile_guessed_local and
2665 profile_guessed_global0.
2666 (class profile_proability): Decrease number of bits to 29;
2667 update from_reg_br_prob_note and to_reg_br_prob_note.
2668 (class profile_count: Update comment; decrease number of bits
2669 to 61. Check compatibility.
2670 (profile_count::compatible_p): New private member function.
2671 (profile_count::ipa_p): New member function.
2672 (profile_count::operator<): Handle global zero correctly.
2673 (profile_count::operator>): Handle global zero correctly.
2674 (profile_count::operator<=): Handle global zero correctly.
2675 (profile_count::operator>=): Handle global zero correctly.
2676 (profile_count::nonzero_p): New member function.
2677 (profile_count::force_nonzero): New member function.
2678 (profile_count::max): New member function.
2679 (profile_count::apply_scale): Handle IPA scalling.
2680 (profile_count::guessed_local): New member function.
2681 (profile_count::global0): New member function.
2682 (profile_count::ipa): New member function.
2683 (profile_count::to_frequency): Declare.
2684 (profile_count::to_cgraph_frequency): Declare.
2685 * profile.c (OVERLAP_BASE): Delete.
2686 (compute_frequency_overlap): Delete.
2687 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2688 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2689 * sched-ebb.c (rank): Use counts only.
2690 * shrink-wrap.c (handle_simple_exit): Use counts only.
2691 (try_shrink_wrapping): Use counts only.
2692 (place_prologue_for_one_component): Use counts only.
2693 * tracer.c (find_best_predecessor): Use to_frequency.
2694 (find_trace): Use to_frequency.
2695 (tail_duplicate): Use to_frequency.
2696 * trans-mem.c (expand_transaction): Do not update frequency.
2697 * tree-call-cdce.c: Do not update frequency.
2698 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2699 (gimple_merge_blocks): Likewise.
2700 (gimple_split_edge): Likewise.
2701 (gimple_duplicate_sese_region): Likewise.
2702 (gimple_duplicate_sese_tail): Likewise.
2703 (move_sese_region_to_fn): Likewise.
2704 (gimple_account_profile_record): Likewise.
2705 (insert_cond_bb): Likewise.
2706 * tree-complex.c (expand_complex_div_wide): Likewise.
2707 * tree-eh.c (lower_resx): Update profile.
2708 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2710 (initialize_cfun): Do not initialize frequencies
2711 (freqs_to_counts): Delete.
2712 (copy_cfg_body): Ignore count parameter.
2713 (copy_body): Update.
2714 (expand_call_inline): Update count_max.
2715 (optimize_inline_calls): Update count_max.
2716 (tree_function_versioning): Update count_max.
2717 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2718 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2720 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2721 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2722 (try_peel_loop): Likewise.
2723 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2725 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2726 (tree_transform_and_unroll_loop): Do not use frequencies
2727 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2728 Use reliable prediction only.
2729 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2730 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2731 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2732 probability scaling.
2733 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2734 not update frequency
2735 (any_remaining_duplicated_blocks): Likewise.
2736 (update_profile): Likewise.
2737 (estimated_freqs_path): Delete.
2738 (freqs_to_counts_path): Delete.
2739 (clear_counts_path): Delete.
2740 (ssa_fix_duplicate_block_edges): Likewise.
2741 (duplicate_thread_path): Likewise.
2742 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2743 * tree-tailcall.c (decrease_profile): Do not update frequency.
2744 (eliminate_tail_call): Likewise.
2745 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2746 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2747 (optimize_mask_stores): Likewise.
2748 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2749 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2750 (ubsan_expand_ptr_ifn): Update profile.
2751 * value-prof.c (gimple_ic): Simplify.
2752 * value-prof.h (gimple_ic): Update prototype.
2753 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2754 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2756 (want_inline_self_recursive_call_p): Likewise.
2757 (resolve_noninline_speculation): Only cummulate defined counts.
2758 (inline_small_functions): Use nonzero_p.
2759 (ipa_inline): Do not access freed node.
2761 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2763 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2764 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2766 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2768 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2769 non-legitimate address.
2771 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2773 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2774 (*lt0_<mode>di, *lt0_<mode>si): New.
2776 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2778 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2779 TARGET_PAIRED_FLOAT.
2781 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2782 Jim Wilson <jim.wilson@linaro.org>
2784 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2785 * config/aarch64/aarch64-tune.md: Regenerated.
2786 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2787 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2789 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2791 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2792 (arc_expand_prologue): Restore blink for millicode.
2793 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2795 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2798 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2799 gen_vec_duplicate after forcing the scalar into a register.
2801 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2803 * combine (try_combine): Print the insns input to try_combine to the
2806 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2809 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2810 Remove second argument from aarch64_process_target_attr call.
2811 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2813 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2815 (aarch64_handle_attr_arch): Remove second argument.
2816 (aarch64_handle_attr_cpu): Ditto.
2817 (aarch64_handle_attr_tune): Ditto.
2818 (aarch64_handle_attr_isa_flags): Ditto.
2819 (aarch64_process_one_target_attr): Ditto.
2820 (aarch64_process_target_attr): Ditto.
2821 (aarch64_option_valid_attribute_p): Remove second argument.
2822 on aarch64_process_target_attr call.
2824 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2826 * diagnostic.c: Include "selftest-diagnostic.h".
2827 (selftest::assert_location_text): New function.
2828 (selftest::test_diagnostic_get_location_text): New function.
2829 (selftest::diagnostic_c_tests): Call it.
2831 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2833 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2834 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2835 (class selftest::test_diagnostic_context): Move to...
2836 * selftest-diagnostic.c: New file.
2837 * selftest-diagnostic.h: New file.
2839 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2841 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2842 offset range for FT32B.
2843 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2844 * config/ft32/ft32.md: Add TARGET_NOPM.
2845 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2846 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2848 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2850 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2852 2017-11-02 Jeff Law <law@redhat.com>
2854 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2855 virtual keyword on FINAL OVERRIDE members.
2857 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2858 virtuals together. Add virtual destructor.
2859 (substitute_and_fold_engine): Similarly.
2861 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2863 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2865 2017-11-02 Richard Biener <rguenther@suse.de>
2867 PR tree-optimization/82795
2868 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2870 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2872 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2874 (gcc_GAS_CHECK_FEATURE): Remove.
2875 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2876 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2877 * configure: Regenerate.
2879 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2881 * config/arc/arc.c (hwloop_optimize): Account for empty
2884 2017-11-02 Richard Biener <rguenther@suse.de>
2887 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2888 Truncate ARRAY_REF index and element size.
2890 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2892 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2894 2017-11-01 Jeff Law <law@redhat.com>
2896 * tree-ssa-ccp.c (ccp_folder): New class derived from
2897 substitute_and_fold_engine.
2898 (ccp_folder::get_value): New member function.
2899 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2900 (ccp_fold_stmt): Remove prototype.
2901 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2902 * tree-ssa-copy.c (copy_folder): New class derived from
2903 substitute_and_fold_engine.
2904 (copy_folder::get_value): Renamed from get_value.
2905 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2906 * tree-vrp.c (vrp_folder): New class derived from
2907 substitute_and_fold_engine.
2908 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2909 (vrp_folder::get_value): New member function.
2910 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2911 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2912 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2913 provide a class interface to folder/substitute routines.
2914 (ssa_prop_fold_stmt_fn): Remove typedef.
2915 (ssa_prop_get_value_fn): Likewise.
2916 (subsitute_and_fold): Remove prototype.
2917 (replace_uses_in): Likewise.
2918 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2919 Renamed from replace_uses_in. Call the virtual member function
2920 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2921 (substitute_and_fold_dom_walker): Remove initialization of
2922 data member entries for calbacks. Add substitute_and_fold_engine
2923 member and initialize it.
2924 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2925 member functions for get_value, replace_phi_args_in c
2926 replace_uses_in, and fold_stmt calls.
2927 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2928 substitute_and_fold. Remove assert. Update ctor call.
2930 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2931 (ssa_prop_visit_phi_fn): Likewise.
2932 (class ssa_propagation_engine): New class to provide an interface
2934 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2936 (ssa_prop_visit_phi): Likewise.
2937 (ssa_propagation_engine::simulate_stmt): Moved into class.
2938 Call visit_phi/visit_stmt from the class rather than via
2939 file scoped static variables.
2940 (ssa_propagation_engine::simulate_block): Moved into class.
2941 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2942 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2943 set file scoped statics for the visit_stmt/visit_phi callbacks.
2944 * tree-complex.c (complex_propagate): New class derived from
2945 ssa_propagation_engine.
2946 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2947 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2948 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2950 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2951 ssa_propagation_engine.
2952 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2953 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2954 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2955 * tree-ssa-copy.c (copy_prop): New class derived from
2956 ssa_propagation_engine.
2957 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2958 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2959 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2960 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2961 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2962 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2963 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2965 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2967 PR rtl-optimization/82778
2968 PR rtl-optimization/82597
2969 * compare-elim.c (struct comparison): Add in_a_setter field.
2970 (find_comparison_dom_walker::before_dom_children): Remove killed
2971 bitmap and df_simulate_find_defs call, instead walk the defs.
2972 Compute last_setter and initialize in_a_setter. Merge definitions
2973 with first initialization for a few variables.
2974 (try_validate_parallel): Use insn_invalid_p instead of
2975 recog_memoized. Return insn rather than just the pattern.
2976 (try_merge_compare): Fix up comment. Don't uselessly test if
2977 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2979 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2982 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2983 Alan Hayward <alan.hayward@arm.com>
2984 David Sherwood <david.sherwood@arm.com>
2986 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2987 aarch64_hard_regno_nregs to get the number of registers
2990 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2991 Alan Hayward <alan.hayward@arm.com>
2992 David Sherwood <david.sherwood@arm.com>
2994 * config/aarch64/constraints.md (Upl): Rename to...
2996 * config/aarch64/aarch64.md
2997 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3000 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3001 Alan Hayward <alan.hayward@arm.com>
3002 David Sherwood <david.sherwood@arm.com>
3004 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3005 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3008 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3009 Alan Hayward <alan.hayward@arm.com>
3010 David Sherwood <david.sherwood@arm.com>
3012 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3013 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3014 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3016 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3017 of a single element.
3018 * config/aarch64/iterators.md: Add a comment above the permute
3019 unspecs to say that they are generated directly by
3020 aarch64_expand_vec_perm_const.
3021 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3023 2017-11-01 Nathan Sidwell <nathan@acm.org>
3025 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3027 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3029 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3030 and medany code models, and describe what they do.
3032 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3034 Revert accidental duplicate:
3036 * combine.c (can_change_dest_mode): Reject changes in
3037 REGMODE_NATURAL_SIZE.
3039 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3041 PR rtl-optimization/64682
3042 PR rtl-optimization/69567
3043 PR rtl-optimization/69737
3044 PR rtl-optimization/82683
3045 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3046 register mentioned in the note, drop the note, unless it came from I3,
3047 in which case it should go to I3 again.
3049 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3051 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3052 and return false if not.
3053 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3055 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3057 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3058 range is known to be empty.
3060 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3061 Alan Hayward <alan.hayward@arm.com>
3062 David Sherwood <david.sherwood@arm.com>
3064 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3065 and CONST_VECTOR_NUNITS instead of computing the number of units from
3066 the byte sizes of the vector and element.
3067 (simplify_binary_operation_1): Likewise.
3068 (simplify_const_binary_operation): Likewise.
3069 (simplify_ternary_operation): Likewise.
3071 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3072 Alan Hayward <alan.hayward@arm.com>
3073 David Sherwood <david.sherwood@arm.com>
3075 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3076 (int_mem_offset): ...this new function.
3077 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3078 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3079 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3082 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3083 Alan Hayward <alan.hayward@arm.com>
3084 David Sherwood <david.sherwood@arm.com>
3086 * lower-subreg.c (interesting_mode_p): New function.
3087 (compute_costs, find_decomposable_subregs, decompose_register)
3088 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3089 (resolve_clobber, dump_choices): Use it.
3091 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3092 Alan Hayward <alan.hayward@arm.com>
3093 David Sherwood <david.sherwood@arm.com>
3095 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3097 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3098 Alan Hayward <alan.hayward@arm.com>
3099 David Sherwood <david.sherwood@arm.com>
3101 * alias.c (find_base_value, find_base_term): Only process integer
3102 truncations. Check the precision rather than the size.
3104 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3105 Alan Hayward <alan.hayward@arm.com>
3106 David Sherwood <david.sherwood@arm.com>
3108 * machmode.h (is_narrower_int_mode): New function
3109 * optabs.c (expand_float, expand_fix): Use it.
3110 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3112 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3113 Alan Hayward <alan.hayward@arm.com>
3114 David Sherwood <david.sherwood@arm.com>
3116 * rtl.h (narrower_subreg_mode): New function.
3117 * ira-color.c (update_costs_from_allocno): Use it.
3119 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3120 Alan Hayward <alan.hayward@arm.com>
3121 David Sherwood <david.sherwood@arm.com>
3123 * optabs-query.h (convert_optab_p): New function, split out from...
3124 (convert_optab_handler): ...here.
3125 (widening_optab_handler): Delete.
3126 (find_widening_optab_handler): Remove permit_non_widening parameter.
3127 (find_widening_optab_handler_and_mode): Likewise. Provide an
3128 override that operates on mode class wrappers.
3129 * optabs-query.c (widening_optab_handler): Delete.
3130 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3131 parameter. Assert that the two modes are the same class and that
3132 the "from" mode is narrower than the "to" mode. Use
3133 convert_optab_handler instead of widening_optab_handler.
3134 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3135 instead of widening_optab_handler.
3136 * expr.c (expand_expr_real_2): Update calls to
3137 find_widening_optab_handler.
3138 * optabs.c (expand_widen_pattern_expr): Likewise.
3139 (expand_binop_directly): Take the insn_code as a parameter.
3140 (expand_binop): Only call find_widening_optab_handler for
3141 conversion optabs; use optab_handler otherwise. Update calls
3142 to find_widening_optab_handler and expand_binop_directly.
3143 Use convert_optab_handler instead of widening_optab_handler.
3144 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3145 find_widening_optab_handler and use scalar_mode rather than
3147 (convert_plusminus_to_widen): Likewise.
3149 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3150 Alan Hayward <alan.hayward@arm.com>
3151 David Sherwood <david.sherwood@arm.com>
3153 * machmode.h (fixed_size_mode): New class.
3154 * rtl.h (get_pool_mode): Return fixed_size_mode.
3155 * gengtype.c (main): Add fixed_size_mode.
3156 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3157 (get_raw_arg_mode): Likewise.
3158 * doc/tm.texi: Regenerate.
3159 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3160 * targhooks.c (default_get_reg_raw_mode): Likewise.
3161 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3162 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3163 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3164 (msp430_get_raw_result_mode): Likewise.
3165 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3166 * dbxout.c (dbxout_parms): Require fixed-size modes.
3167 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3168 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3169 * omp-low.c (lower_oacc_reductions): Likewise.
3170 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3171 (simplify_subreg): Update accordingly.
3172 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3173 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3174 that aren't fixed-size.
3175 (get_pool_mode): Return a fixed_size_mode.
3176 (output_constant_pool_2): Take a fixed_size_mode.
3178 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3179 Alan Hayward <alan.hayward@arm.com>
3180 David Sherwood <david.sherwood@arm.com>
3182 * doc/rtl.texi (vec_series): Document.
3183 (const): Say that the operand can be a vec_series.
3184 * rtl.def (VEC_SERIES): New rtx code.
3185 * rtl.h (const_vec_series_p_1): Declare.
3186 (const_vec_series_p): New function.
3187 * emit-rtl.h (gen_const_vec_series): Declare.
3188 (gen_vec_series): Likewise.
3189 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3190 (gen_vec_series): Likewise.
3191 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3192 * simplify-rtx.c (simplify_unary_operation): Handle negations
3194 (simplify_binary_operation_series): New function.
3195 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3196 (test_vector_ops_series): New function.
3197 (test_vector_ops): Call it.
3198 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3199 gen_const_vec_series.
3200 (altivec_lvsr): Likewise.
3201 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3203 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3204 Alan Hayward <alan.hayward@arm.com>
3205 David Sherwood <david.sherwood@arm.com>
3207 * doc/rtl.texi (const): Update description of address constants.
3208 Say that vector constants are allowed too.
3209 * common.md (E, F): Use CONSTANT_P instead of checking for
3211 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3212 checking for CONST_VECTOR.
3213 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3215 * expr.c (expand_expr_real_2): Check for vector modes instead
3216 of checking for CONST_VECTOR.
3217 * rtl.h (const_vec_p): New function.
3218 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3219 (unwrap_const_vec_duplicate): Handle them here too.
3221 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3222 David Malcolm <dmalcolm@redhat.com>
3223 Alan Hayward <alan.hayward@arm.com>
3224 David Sherwood <david.sherwood@arm.com>
3226 * rtl.h (vec_duplicate_p): New function.
3227 * selftest-rtl.c (assert_rtx_eq_at): New function.
3228 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3229 (assert_rtx_eq_at): Declare.
3230 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3231 * selftest-run-tests.c (selftest::run_tests): Call it.
3232 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3233 (simplify_unary_operation_1): Recursively handle vector duplicates.
3234 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3236 (simplify_subreg): Handle subregs of vector duplicates.
3237 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3238 (selftest::simplify_rtx_c_tests): New functions.
3240 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3241 Alan Hayward <alan.hayward@arm.com>
3242 David Sherwood <david.sherwood@arm.com>
3244 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3245 (gen_vec_duplicate): Likewise.
3246 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3248 (gen_const_vector): ...here.
3249 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3250 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3251 whose elements are all equal.
3252 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3253 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3254 (simplify_relational_operation): Likewise.
3255 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3257 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3258 (aarch64_expand_vector_init): Likewise.
3259 * config/arm/arm.c (neon_vdup_constant): Likewise.
3260 (neon_expand_vector_init): Likewise.
3261 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3262 (arm_block_set_unaligned_vect): Likewise.
3263 (arm_block_set_aligned_vect): Likewise.
3264 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3265 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3266 (expand_vec_perm_even_odd_pack): Likewise.
3267 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3268 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3269 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3270 gen_const_vec_duplicate.
3271 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3272 * config/mips/mips.c (mips_gen_const_int_vector): Use
3273 gen_const_vec_duplicate.
3274 (mips_expand_vector_init): Use CONST0_RTX.
3275 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3276 (define_split): Use gen_const_vec_duplicate.
3277 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3278 (define_split): Use gen_const_vec_duplicate.
3279 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3280 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3281 * config/spu/spu.c (spu_const): Likewise.
3283 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3284 Alan Hayward <alan.hayward@arm.com>
3285 David Sherwood <david.sherwood@arm.com>
3287 * combine.c (can_change_dest_mode): Reject changes in
3288 REGMODE_NATURAL_SIZE.
3290 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3292 * configure.ac (--enable-libssp): New.
3293 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3294 trying to determine target-specific default. Adjust indentation.
3295 * configure: Regenerated.
3296 * doc/install.texi (Configuration): Expand --disable-libssp
3299 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3301 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3304 2017-10-31 Martin Jambor <mjambor@suse.cz>
3307 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3309 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3311 * auto-profile.c (autofdo_source_profile::read): Use
3312 UNKNOWN_LOCATION rather than 0.
3313 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3314 (warning_at): ...this overload.
3315 (warning_at_rich_loc_n): Rename to...
3316 (warning_n): ...this overload.
3317 (error_at_rich_loc): Rename to...
3318 (error_at): ...this overload.
3319 (pedwarn_at_rich_loc): Rename to...
3320 (pedwarn): ...this overload.
3321 (permerror_at_rich_loc): Rename to...
3322 (permerror): ...this overload.
3323 (inform_at_rich_loc): Rename to...
3324 (inform): ...this overload.
3325 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3326 (diagnostic_n_impl_richloc): Rename to...
3327 (diagnostic_n_impl): ...this rich_location *-based decl.
3328 (inform_at_rich_loc): Rename to...
3329 (inform): ...this, and add an assertion.
3330 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3331 (warning_at_rich_loc): Rename to...
3332 (warning_at): ...this, and add an assertion.
3333 (warning_at_rich_loc_n): Rename to...
3334 (warning_n): ...this, and add an assertion.
3335 (warning_n): Update location_t-based implementation for removal of
3336 location_t-based diagnostic_n_impl.
3337 (pedwarn_at_rich_loc): Rename to...
3338 (pedwarn): ...this, and add an assertion.
3339 (permerror_at_rich_loc): Rename to...
3340 (permerror): ...this, and add an assertion.
3341 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3342 (error_at_rich_loc): Rename to...
3343 (error_at): ...this, and add an assertion.
3344 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3345 (driver::do_spec_on_infiles): Likewise.
3346 * substring-locations.c (format_warning_va): Update for renaming
3347 of inform_at_rich_loc.
3349 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3351 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3352 _Float<N>X built-in functions so that the variant without the
3353 "__builtin_" prefix is only enabled for the GNU C and Objective C
3354 languages when they are in non-strict ANSI/ISO mode.
3355 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3356 * target.def (floatn_builtin_p): Add a target hook to control
3357 whether _Float<N> and _Float<N>X built-in functions without the
3358 "__builtin_" prefix are enabled, and return true for C and
3359 Objective C in the default hook. Include langhooks.h in
3361 * targhooks.h (default_floatn_builtin_p): Likewise.
3362 * targhooks.c (default_floatn_builtin_p): Likewise.
3363 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3364 floatn_builtin_p target hook.
3365 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3367 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3368 Eric Botcazou <ebotcazou@adacore.com>
3370 PR rtl-optimization/81803
3371 * lra-constraints.c (curr_insn_transform): Also reload the whole
3372 register for a strict subreg no wider than a word if this is for
3373 a WORD_REGISTER_OPERATIONS target.
3375 2017-10-31 Jason Merrill <jason@redhat.com>
3377 * gdbinit.in: Skip over inlines from timevar.h.
3379 2017-10-31 Martin Liska <mliska@suse.cz>
3381 * doc/gcov.texi: Document new option.
3382 * gcov.c (print_usage): Likewise print it.
3383 (process_args): Support the argument.
3384 (format_count): New function.
3385 (format_gcov): Use the function.
3387 2017-10-31 Martin Liska <mliska@suse.cz>
3389 * gcov.c (struct name_map): do not use typedef.
3390 Define operator== and operator<.
3391 (name_search): Remove.
3392 (name_sort): Remove.
3393 (main): Do not allocate names.
3394 (process_file): Add vertical space.
3395 (generate_results): Use std::find.
3396 (release_structures): Do not release memory.
3397 (find_source): Use std::find.
3399 2017-10-31 Martin Liska <mliska@suse.cz>
3401 * gcov.c (struct line_info): Remove it's typedef.
3402 (line_info::line_info): Add proper ctor.
3403 (line_info::has_block): Do not use a typedef.
3404 (struct source_info): Do not use typedef.
3405 (circuit): Likewise.
3406 (get_cycles_count): Likewise.
3407 (output_intermediate_file): Iterate via vector iterator.
3408 (add_line_counts): Use std::vector methods.
3409 (accumulate_line_counts): Likewise.
3410 (output_lines): Likewise.
3412 2017-10-31 Martin Liska <mliska@suse.cz>
3414 * gcov.c (struct source_info): Remove typedef.
3415 (source_info::source_info): Add proper ctor.
3416 (accumulate_line_counts): Use struct, not it's typedef.
3417 (output_gcov_file): Likewise.
3418 (output_lines): Likewise.
3419 (main): Do not allocate an array.
3420 (output_intermediate_file): Use size of vector container.
3421 (process_file): Resize the vector.
3422 (generate_results): Do not preallocate, use newly added vector
3424 (release_structures): Do not release sources.
3425 (find_source): Use vector methods.
3426 (add_line_counts): Do not use typedef.
3428 2017-10-31 Martin Liska <mliska@suse.cz>
3430 * doc/gcov.texi: Document that.
3431 * gcov.c (add_line_counts): Mark lines with a non-executed
3433 (output_line_beginning): Handle such lines.
3434 (output_lines): Pass new argument.
3435 (output_intermediate_file): Print it in intermediate format.
3437 2017-10-31 Martin Liska <mliska@suse.cz>
3439 * color-macros.h: New file.
3440 * diagnostic-color.c: Factor out color related to macros to
3442 * doc/gcov.texi: Document -k option.
3443 * gcov.c (INCLUDE_STRING): Include string.h.
3444 (print_usage): Add -k option.
3445 (process_args): Parse it.
3446 (pad_count_string): New function.
3447 (output_line_beginning): Likewise.
3448 (DEFAULT_LINE_START): New macro.
3449 (output_lines): Support color output.
3451 2017-10-31 Martin Liska <mliska@suse.cz>
3453 PR gcov-profile/82633
3454 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3455 their interaction with GCOV infrastructure.
3456 * configure.ac: Add -fkeep-{inline,static}-functions to
3458 * configure: Regenerate.
3460 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3463 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3465 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3468 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3469 into a register if it does not fit into an immediate offset field.
3471 2017-10-31 Olivier Hainque <hainque@adacore.com>
3473 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3475 2017-10-31 Julia Koval <julia.koval@intel.com>
3477 * config.gcc: Add gfniintrin.h.
3478 * config/i386/gfniintrin.h: New.
3479 * config/i386/i386-builtin-types.def
3480 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3481 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3482 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3483 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3484 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3485 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3486 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3487 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3488 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3489 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3490 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3491 * config/i386/immintrin.h: Include gfniintrin.h.
3492 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3494 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3496 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3498 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3500 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3501 (arm_ashldi3_1bit): Remove pattern.
3502 (ashrdi3): Remove shift by 1 expansion.
3503 (arm_ashrdi3_1bit): Remove pattern.
3504 (lshrdi3): Remove shift by 1 expansion.
3505 (arm_lshrdi3_1bit): Remove pattern.
3506 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3507 cost of ashldi3 by 1.
3508 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3509 (<shift>di3_neon): Likewise.
3511 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3513 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3514 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3515 and (*aarch64_simd_mov<VQ:mode>).
3516 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3517 pattern alternative.
3518 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3519 attributes to match pattern alternative.
3521 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3523 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3524 * config/rs6000/emmintrin.h: New file.
3525 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3527 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3529 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3530 * (movdi_vfp_cortexa8): Remove pattern.
3532 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3534 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3536 (Specific, alpha*-dec-osf5.1): Remove.
3537 (Specific, mips-sgi-irix5): Remove.
3538 (Specific, mips-sgi-irix6): Remove.
3540 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3543 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3544 arguments to clear_bit_region_be.
3546 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3548 * gimplify.c: Include memmodel.h.
3550 2017-10-30 Martin Jambor <mjambor@suse.cz>
3552 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3553 condition whether loop should be executed at all.
3555 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3557 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3558 gimple folding of vec_madd() intrinsics.
3559 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3560 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3561 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3563 2017-10-30 Richard Biener <rguenther@suse.de>
3565 PR tree-optimization/82762
3567 2017-10-23 Richard Biener <rguenther@suse.de>
3569 PR tree-optimization/82129
3571 2017-08-01 Richard Biener <rguenther@suse.de>
3573 PR tree-optimization/81181
3574 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3575 (compute_antic): ... end of iteration here.
3577 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3579 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3580 -std=iso9899:2017 and -std=gnu17.
3581 * doc/standards.texi (C Language): Document C17 support.
3582 * doc/cpp.texi (Overview): Mention -std=c17.
3583 (Standard Predefined Macros): Document C11 and C17 values of
3584 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3585 * doc/extend.texi (Inline): Do not list individual options for
3586 standards newer than C99.
3587 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3589 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3592 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3594 * asan.c (asan_finish_file): Align asan globals array by shadow
3597 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3600 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3601 (struct store_immediate_info): Add bitregion_start and bitregion_end
3603 (store_immediate_info::store_immediate_info): Add brs and bre
3604 arguments and initialize bitregion_{start,end} from those.
3605 (struct merged_store_group): Add bitregion_start, bitregion_end,
3606 align_base and mask fields. Drop unnecessary struct keyword from
3607 struct store_immediate_info. Add do_merge method.
3608 (clear_bit_region_be): Use memset instead of loop storing zeros.
3609 (merged_store_group::do_merge): New method.
3610 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3611 stores as long as the surrounding bitregions have no gaps.
3612 (merged_store_group::merge_overlapping): Use do_merge.
3613 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3614 is byte aligned, rather than requiring that start and width are
3615 byte aligned. Drop unnecessary struct keyword from
3616 struct store_immediate_info. Allocate and populate also mask array.
3617 Make start of the arrays relative to bitregion_start rather than
3618 start and size them according to bitregion_{end,start} difference.
3619 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3620 struct store_immediate_info.
3621 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3622 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3623 struct keyword from struct store_immediate_info.
3624 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3625 between stores as long as the surrounding bitregions have no gaps.
3627 (struct split_store): Add orig non-static data member.
3628 (split_store::split_store): Initialize orig to false.
3629 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3630 if there is exactly a single original stmt. Change stmts argument
3631 to pointer from reference, if NULL, don't push anything to it. Add
3632 first argument, use it to optimize skipping over orig stmts that
3633 are known to be before bitpos already. Simplify.
3634 (split_group): Return unsigned int count how many stores are or
3635 would be needed rather than a bool. Add allow_unaligned argument.
3636 Change split_stores argument from reference to pointer, if NULL,
3637 only do a dry run computing how many stores would be produced.
3638 Rewritten algorithm to use both alignment and misalign if
3639 !allow_unaligned and handle bitfield stores with gaps.
3640 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3641 from bitregion_start instead of start. Compute allow_unaligned
3642 here, if true, do 2 split_group dry runs to compute which one
3643 produces fewer stores and prefer aligned if equal. Punt if
3644 new count is bigger or equal than original before emitting any
3645 statements, rather than during that. Remove no longer needed
3646 new_ssa_names tracking. Replace num_stmts with
3647 split_stores.length (). Use 32-bit stack allocated entries
3648 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3649 if possible. Handle bitfields with gaps.
3650 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3651 Compute bitregion_{start,end} for the stores and construct
3652 store_immediate_info with that. Formatting fixes.
3654 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3657 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3658 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3660 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3662 * gimplify.c: Include tm_p.h.
3664 * common.opt (gcoff): Re-add as ignored option.
3665 (gcoff1, gcoff2, gcoff3): Likewise.
3667 * Makefile.in (OBJS): Delete sdbout.o.
3668 (GTFILES): Delete $(srcdir)/sdbout.c.
3669 * debug.h: Delete sdb_debug_hooks.
3670 * final.c: Delete sdbout.h include.
3671 (final_scan_insn): Delete SDB_DEBUG check.
3672 (rest_of_clean_state): Likewise.
3673 * output.h: Delete sdb_begin_function_line.
3676 * toplev.c: Delete sdbout.h include.
3677 (process_options): Delete SDB_DEBUG check.
3678 * tree-core.h (tree_type_common): Delete pointer field of
3680 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3681 TYPE_SYMTAB_POINTER.
3682 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3683 (TYPE_SYMTAB_IS_POINTER): Delete.
3684 (TYPE_SYMTAB_IS_DIE): Renumber.
3685 * xcoffout.c: Refer to former sdbout.c file.
3686 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3688 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3689 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3690 (Debugging Options): Delete -gcoff.
3691 (-gstabs): Delete SDB reference.
3693 (-gcoff@var{level}): Delete.
3694 * doc/passes.texi (Debugging information output): Delete SDB and
3695 sdbout.c references.
3696 * doc/tm.texi: Regenerate.
3697 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3698 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3700 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3701 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3702 (SDB_DEBUGGING_INFO): Delete.
3703 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3704 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3705 * target.def (output_source_filename): Delete COFF reference.
3707 * common.opt (gcoff): Delete.
3708 (gxcoff+): Update Negative chain.
3709 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3711 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3712 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3713 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3715 (expand_function_start): Change sdb reference to past tense.
3716 (expand_function_end): Change sdb reference to past tense.
3717 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3718 * opts.c (debug_type_names): Delete coff entry.
3719 (common_handle_option): Delete OPT_gcoff case.
3720 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3722 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3723 * config/cris/cris.h: Delete SDB reference in comment.
3724 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3725 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3726 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3727 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3729 (ix86_expand_prologue): Likewise.
3730 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3731 * config/ia64/ia64.h: Likewise.
3732 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3733 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3735 * config/mmix/mmix.h: Likewise.
3736 * config/nds32/nds32.c: Likewise.
3737 * config/stormy/storym16.h: Likewise.
3738 * config/visium/visium.h: Likewise.
3739 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3741 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3743 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3744 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3745 FRAME_POINTER_REGNUM point at high end of local var area.
3747 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3749 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3750 Move comment around. Do not reset best_edge for a copiable
3751 destination if the copy would cause a partition change.
3752 (better_edge_p): Remove redundant check.
3754 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3756 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3758 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3760 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3761 for math functions that have _Float<N> and _Float<N>X variants.
3762 (mathfn_built_in_2): Add support for math functions that have
3763 _Float<N> and _Float<N>X variants.
3764 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3765 (expand_builtin_mathfn_ternary): Add support for fma with
3766 _Float<N> and _Float<N>X variants.
3767 (expand_builtin): Likewise.
3768 (fold_builtin_3): Likewise.
3769 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3770 create math function _Float<N> and _Float<N>X variants as external
3772 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3773 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3774 the __builtin_ prefix and if not strict ansi, without the prefix.
3775 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3776 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3777 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3778 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3779 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3780 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3781 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3782 function signatures for fma _Float<N> and _Float<N>X variants.
3783 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3784 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3785 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3786 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3787 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3788 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3789 * gencfn-macros.c (print_case_cfn): Add support for math functions
3790 that have _Float<N> and _Float<N>X variants.
3791 (print_define_operator_list): Likewise.
3792 (fltfn_suffixes): Likewise.
3794 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3795 for math functions that have _Float<N> and _Float<N>X variants.
3796 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3797 and _Float<N>X variants.
3798 (COPYSIGN): Likewise.
3801 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3802 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3804 (integer_valued_read_call_p): Likewise.
3805 * fold-const-call.c (fold_const_call_ss): Likewise.
3806 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3807 _Float<N> and _Float<N>X variants.
3808 (fold_const_call_ssss): Add support for fma _Float<N> and
3809 _Float<N>X variants.
3810 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3811 support for copysign and fma _Float<N> and _Float<N>X variants.
3812 (backprop::process_builtin_call_use): Likewise.
3813 * tree-call-cdce.c (can_test_argument_range); Add support for
3814 sqrt _Float<N> and _Float<N>X variants.
3815 (edom_only_function): Likewise.
3816 (get_no_error_domain): Likewise.
3817 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3818 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3819 copysign _Float<N> and _Float<N>X variants.
3820 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3821 handled by machine independent code.
3822 (FMAF128): Likewise.
3823 * doc/cpp.texi (Common Predefined Macros): Document defining
3824 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3825 fma _Float<N> and _Float<N>X variants.
3827 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3830 * config/i386/i386-modes.def (CCFPU): Remove definition.
3831 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3832 (ix86_cc_modes_compatible): Ditto.
3833 (ix86_expand_carry_flag_compare): Ditto.
3834 (ix86_expand_int_movcc): Ditto.
3835 (ix86_expand_int_addcc): Ditto.
3836 (ix86_reverse_condition): Ditto.
3837 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3838 Return true/false for unordered/ordered fp comparisons.
3839 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3840 (ix86_prepare_fp_compare_args): Update for rename.
3841 (ix86_expand_fp_compare): Update for rename. Generate unordered
3842 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3843 (ix86_expand_sse_compare_and_jump): Ditto.
3844 * config/i386/predicates.md (fcmov_comparison_operator):
3845 Remove CCFPU mode handling.
3846 (ix86_comparison_operator): Ditto.
3847 (ix86_carry_flag_operator): Ditto.
3848 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3849 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3850 (*cmpu<mode>_cc_i387): Ditto.
3851 (FPCMP): Remove mode iterator.
3852 (unord): Remove mode attribute.
3853 (unord_subst): New define_subst transformation
3854 (unord): New define_subst attribute.
3856 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3857 (*cmpi<unord>xf_i387): Ditto.
3858 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3859 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3860 using unord_subst transformation.
3861 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3862 (round_saeonly): Also handle CCFP mode.
3863 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3864 Remove UNSPEC_SAHF unspec handling.
3866 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3868 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3870 2017-10-27 Jeff Law <law@redhat.com>
3872 * gimple-ssa-sprintf.c: Include domwalk.h.
3873 (class sprintf_dom_walker): New class, derived from dom_walker.
3874 (sprintf_dom_walker::before_dom_children): New function.
3875 (struct call_info): Moved into sprintf_dom_walker class
3876 (compute_formath_length, handle_gimple_call): Likewise.
3877 (sprintf_length::execute): Call the dominator walker rather
3878 than walking the statements.
3880 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3881 gimple statement locations.
3882 (check_array_bounds): Corresponding changes. Get the statement's
3883 location directly from wi->stmt.
3885 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3888 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3890 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3892 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3893 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3895 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3898 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3899 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3900 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3901 maybe_get_pool_constant.
3902 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3905 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3907 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3908 2.26 caveat. Update gas and gld versions.
3909 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3912 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3914 * cgraph.h (set_malloc_flag): Declare.
3915 * cgraph.c (set_malloc_flag_1): New function.
3916 (set_malloc_flag): Likewise.
3917 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3918 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3920 (read_ipa_call_summary): Add support for reading is_return_callee.
3921 (write_ipa_call_summary): Stream is_return_callee.
3922 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3923 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3924 ipa-prop.h, ipa-fnsummary.h.
3925 (pure_const_names): Change to static.
3926 (malloc_state_e): Define.
3927 (malloc_state_names): Define.
3928 (funct_state_d): Add field malloc_state.
3929 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3930 (check_retval_uses): New function.
3931 (malloc_candidate_p): Likewise.
3932 (analyze_function): Add support for malloc attribute.
3933 (pure_const_write_summary): Stream malloc_state.
3934 (pure_const_read_summary): Add support for reading malloc_state.
3935 (dump_malloc_lattice): New function.
3936 (propagate_malloc): New function.
3937 (warn_function_malloc): New function.
3938 (ipa_pure_const::execute): Call propagate_malloc and
3939 ipa_free_fn_summary.
3940 (pass_local_pure_const::execute): Add support for malloc attribute.
3941 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3942 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3944 2017-10-27 Martin Liska <mliska@suse.cz>
3946 PR gcov-profile/82457
3947 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3948 for fork-like functions to be properly instrumented.
3950 2017-10-27 Richard Biener <rguenther@suse.de>
3953 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3954 info when we redirected EH.
3956 2017-10-26 Michael Collison <michael.collison@arm.com>
3958 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3960 (<optab>_trunchf<GPI:mode>2: New pattern.
3961 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3962 * config/aarch64/iterators.md (wv): New mode attribute.
3963 (vf, VF): New mode attributes.
3964 (vgp, VGP): New mode attributes.
3965 (s): Update attribute with SImode and DImode prefixes.
3967 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3969 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3970 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3971 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3972 (nios2_option_overide): Initialize it. Don't allow R0-relative
3973 addressing with PIC.
3974 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3975 (nios2_symbolic_constant_p): Likewise.
3976 (nios2_legitimate_address_p): Likewise.
3977 (nios2_r0rel_section_name_p): New.
3978 (nios2_symbol_ref_in_r0rel_data_p): New.
3979 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3980 (r0rel_constant_p): New.
3981 (nios2_print_operand_address): Handle r0rel_constant_p.
3982 (nios2_cdx_narrow_form_p): Likewise.
3983 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3984 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3985 (Nios II Options): Document -mr0rel-sec.
3987 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3989 * config/nios2/nios2.c: Include xregex.h.
3990 (nios2_gprel_sec_regex): New.
3991 (nios2_option_overide): Initialize it. Don't allow GP-relative
3992 addressing with PIC.
3993 (nios2_small_section_name_p): Check for regex match.
3994 * config/nios2/nios2.opt (mgprel-sec=): New option.
3995 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3996 (Nios II Options): Document -mgprel-sec.
3998 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4000 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4002 2017-10-26 Tom de Vries <tom@codesourcery.com>
4004 PR tree-optimization/82707
4005 * gimple.c (gimple_copy): Fix unsharing of
4006 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4008 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4010 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4011 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4012 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4013 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4014 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4015 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4016 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4017 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4018 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4019 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4020 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4021 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4022 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4023 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4024 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4025 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4026 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4028 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4030 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4032 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4033 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4034 warning to rs6000.c. Remove the Undocumented flag, since it has
4036 (-mabi=ibmlongdouble): Likewise.
4037 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4038 already set, set the default format for long double.
4039 (rs6000_debug_reg_global): Print whether long double is IBM or
4041 (rs6000_option_override_internal): Rework setting long double
4042 format. Only warn if the user is changing the long double default
4043 and they did not use -Wno-psabi.
4044 * doc/invoke.texi (PowerPC options): Update the documentation for
4045 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4047 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4048 Alan Hayward <alan.hayward@arm.com>
4049 David Sherwood <david.sherwood@arm.com>
4051 * rtl.h (wider_subreg_mode): New function.
4052 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4053 rather than an unsigned int *.
4054 * ira-color.c (regno_max_ref_width): Replace with...
4055 (regno_max_ref_mode): ...this new variable.
4056 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4057 Use wider_subreg_mode.
4058 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4059 rather than an unsigned int *.
4060 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4061 (process_alt_operands): Likewise.
4062 (invariant_p): Likewise.
4063 * lra-spills.c (assign_mem_slot): Likewise.
4064 (add_pseudo_to_slot): Likewise.
4065 * lra.c (collect_non_operand_hard_regs): Likewise.
4066 (add_regs_to_insn_regno_info): Likewise.
4067 * reload1.c (regno_max_ref_width): Replace with...
4068 (regno_max_ref_mode): ...this new variable.
4069 (reload): Update accordingly. Update call to
4070 ira_sort_regnos_for_alter_reg.
4071 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4072 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4073 (scan_paradoxical_subregs): Likewise.
4075 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4077 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4078 (aarch64_frame): Add emit_frame_chain boolean.
4079 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4080 Move eh_return case to aarch64_layout_frame.
4081 (aarch64_layout_frame): Initialize emit_frame_chain.
4082 (aarch64_expand_prologue): Use emit_frame_chain.
4084 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4086 * config/aarch64/aarch64.c (aarch64_layout_frame):
4087 Ensure LR is always stored at the bottom of the callee-saves.
4088 Remove rarely used frame layout which saves callee-saves at top of
4089 frame, so the store of LR can be used as a valid probe in all cases.
4091 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4093 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4094 Improve unaligned TImode/TFmode base/offset split.
4096 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4097 Alan Hayward <alan.hayward@arm.com>
4098 David Sherwood <david.sherwood@arm.com>
4100 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4101 * combine.c (find_single_use_1): Likewise.
4102 (expand_field_assignment): Likewise.
4103 (move_deaths): Likewise.
4104 * lra-constraints.c (simplify_operand_subreg): Likewise.
4105 (curr_insn_transform): Likewise.
4106 * lra.c (collect_non_operand_hard_regs): Likewise.
4107 (add_regs_to_insn_regno_info): Likewise.
4108 * rtlanal.c (reg_referenced_p): Likewise.
4109 (covers_regno_no_parallel_p): Likewise.
4111 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4113 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4114 Don't print any bits outside the precision of the value.
4115 * wide-int.cc (test_printing): Add some new tests.
4117 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4119 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4120 supports -xbrace_comment option.
4121 * configure: Regenerate.
4122 * config.in: Regenerate.
4123 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4124 (ASM_CPU_SPEC): Use it.
4126 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4128 * target.def (static_rtx_alignment): New hook.
4129 * targhooks.h (default_static_rtx_alignment): Declare.
4130 * targhooks.c (default_static_rtx_alignment): New function.
4131 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4132 * doc/tm.texi: Regenerate.
4133 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4134 instead of targetm.constant_alignment. Remove call to
4136 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4137 (cris_preferred_mininum_alignment): New function, split out from...
4138 (cris_constant_alignment): ...here.
4139 (cris_static_rtx_alignment): New function.
4140 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4142 (ix86_constant_alignment): ...here.
4143 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4144 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4145 (mmix_static_rtx_alignment): New function.
4146 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4147 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4149 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4152 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4153 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4155 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4157 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4158 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4159 mark as a sign-extending load.
4160 (local_pic_load_u): Define.
4162 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4165 * fold-const.c (operand_equal_for_comparison_p): Also return true
4166 if ARG0 is a simple variant of ARG1 with narrower precision.
4167 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4169 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4171 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4173 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4174 scatter_static, scatter_per_elt.
4175 * x86-tune-costs.h: Add new cost entries.
4177 2017-10-25 Richard Biener <rguenther@suse.de>
4179 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4180 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4181 class pass_fre): Move to ...
4182 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4183 class pass_fre): ... here and adjust for statistics.
4185 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4188 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4189 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4190 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4191 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4194 2017-10-25 Richard Biener <rguenther@suse.de>
4196 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4197 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4198 eliminate_push_avail, eliminate_insert): Move inside...
4199 (class eliminate_dom_walker): ... this class in preparation
4201 (fini_eliminate): Remove by merging with ...
4202 (eliminate): ... this function. Adjust for class changes.
4203 (pass_pre::execute): Remove fini_eliminate call.
4204 (pass_fre::execute): Likewise.
4206 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4209 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4210 (VPERMI2, VPERMI2I): New mode iterators.
4211 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4212 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4214 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4215 mode iterator. Remove 3 old define_insn patterns.
4216 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4217 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4218 VPERMI2 mode iterator, remove the other two expanders.
4219 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4220 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4221 instructions, remove the other two patterns.
4222 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4223 mode iterator, remove the other two patterns.
4224 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4225 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4226 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4227 and adjust argument order accordingly.
4228 (ix86_expand_vec_perm): Adjust caller.
4229 (expand_vec_perm_1): Likewise.
4230 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4231 (expand_vec_perm_vpermt2_vpshub2): ... this.
4232 (ix86_expand_vec_perm_const_1): Adjust caller.
4233 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4236 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4237 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4238 (vec_shl_<mode>): Remove unused expander.
4239 (avx512bw_<shift_insn><mode>3): New define_insn.
4240 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4241 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4243 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4246 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4249 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4251 PR rtl-optimization/82396
4252 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4253 (autopref_multipass_init): Simplify initialization.
4254 (autopref_rank_data): Simplify sort order.
4255 * gcc/sched-int.h (autopref_multipass_data_): Remove
4256 multi_mem_insn_p, min_offset and max_offset.
4258 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4261 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4262 Check special value of flag_omit_frame_pointer.
4263 (aarch64_can_eliminate): Likewise.
4264 (aarch64_override_options_after_change_1): Simplify handling of
4265 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4267 2017-10-24 Richard Biener <rguenther@suse.de>
4269 PR tree-optimization/82697
4270 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4271 zero for conditional load and unconditional store.
4273 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4275 * doc/install.texi: Document bootstrap-cet.
4277 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4280 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4281 ENDBR instruction at function entrance if function is only
4284 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4287 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4288 patterns to better describe from which operation the CF is computed.
4289 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4290 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4291 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4292 is 0, use _0 suffixed expanders instead of emitting a comparison
4295 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4297 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4298 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4300 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4303 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4304 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4305 * loop-iv.c (iv_get_reaching_def): Likewise.
4306 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4307 variable is promoted and the partition contains undefined values.
4309 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4311 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4313 (nios2_address_cost): Define.
4314 (nios2_legitimize_address): Recognize (exp + constant) directly.
4315 (TARGET_ADDRESS_COST): Define.
4317 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4319 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4320 (nios2_symbolic_memory_operand_p): Declare.
4321 (nios2_split_large_constant): Declare.
4322 (nios2_split_symbolic_memory_operand): Declare.
4323 * config/nios2/nios2.c: Adjust includes.
4324 (nios2_symbolic_constant_allowed): New.
4325 (nios2_symbolic_constant_p): New.
4326 (nios2_plus_symbolic_constant_p): New.
4327 (nios2_valid_addr_expr_p): Recognize addresses involving
4329 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4330 (nios2_symbolic_memory_operand_p): New.
4331 (nios2_large_constant_p): New.
4332 (nios2_split_large_constant): New.
4333 (nios2_split_plus_large_constant): New.
4334 (nios2_split_symbolic_memory_operand): New.
4335 (nios2_legitimize_address): Code refactoring. Handle addresses
4336 involving symbolic constants.
4337 (nios2_emit_move_sequence): Likewise.
4338 (nios2_print_operand): Improve error output.
4339 (nios2_print_operand_address): Handle LO_SUM.
4340 (nios2_cdx_narrow_form_p): Likewise.
4341 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4342 operands involving symbolic constants.
4343 (movhi_internal, movsi_internal): Likewise.
4344 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4345 (extendhisi2, extendqi<mode>2): Likewise.
4347 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4349 * tree-pass.h (PROP_rtl_split_insns): Define.
4350 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4352 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4354 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4356 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4359 * target.def (const_not_ok_for_debug_p): Default to
4360 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4361 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4362 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4363 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4364 which targetm.const_not_ok_for_debug_p returned true.
4365 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4367 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4369 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4370 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4371 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4373 (ix86_const_not_ok_for_debug_p): New function.
4374 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4375 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4377 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4380 * system.h: Conditionally include "unique-ptr.h" if
4381 INCLUDE_UNIQUE_PTR is defined.
4382 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4383 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4385 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4387 * config/rl78/rl78.md: New define_expand "subdi3".
4389 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4392 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4393 DF_REF_INSN if DF_REF_INSN_INFO is false.
4395 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4397 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4398 xmm_move instead of sse_move.
4399 (sse_store_index): New function.
4400 (ix86_register_move_cost): Be more sensible about mismatch stall;
4401 model AVX moves correctly; make difference between sse->integer and
4403 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4404 moves; make difference between SSE and AVX.
4405 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4406 and zmm_move. Increase size of sse load and store tables;
4407 add unaligned load and store tables; add ssemmx_to_integer.
4408 * x86-tune-costs.h: Update all entries according to real
4409 move latencies from Agner Fog's manual and chip documentation.
4411 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4414 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4415 * config/i386/constraints.md (Wf): New constraint.
4416 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4417 (cmp<dwi>_doubleword): Removed.
4418 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4419 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4420 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4421 expand with cmp<dwi>_doubleword. For LTU and GEU use
4422 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4424 * common.opt (gcolumn-info): Enable by default.
4425 * doc/invoke.texi (gcolumn-info): Document new default.
4427 2017-10-23 Richard Biener <rguenther@suse.de>
4429 PR tree-optimization/82672
4430 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4431 Fold the stmt if we propagated into it.
4433 2017-10-23 Richard Biener <rguenther@suse.de>
4435 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4436 (bitmap_remove_expr_from_set): ... this. All callers call this
4437 for non-constant values.
4438 (bitmap_set_subtract): Rename to...
4439 (bitmap_set_subtract_expressions): ... this. Adjust and
4441 (bitmap_set_contains_value): Remove superfluous check.
4442 (bitmap_set_replace_value): Inline into single caller ...
4443 (bitmap_value_replace_in_set): ... here and simplify.
4444 (dependent_clean): Merge into ...
4445 (clean): ... this using an overload. Adjust.
4446 (prune_clobbered_mems): Adjust.
4447 (compute_antic_aux): Likewise.
4448 (compute_partial_antic_aux): Likewise.
4450 2017-10-23 Richard Biener <rguenther@suse.de>
4452 PR tree-optimization/82129
4454 2017-08-01 Richard Biener <rguenther@suse.de>
4456 PR tree-optimization/81181
4457 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4458 (compute_antic): ... end of iteration here.
4460 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4462 * target.def (starting_frame_offset): New hook.
4463 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4464 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4465 * doc/tm.texi.in: Regenerate.
4466 * hooks.h (hook_hwi_void_0): Declare.
4467 * hooks.c (hook_hwi_void_0): New function.
4468 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4469 STARTING_FRAME_OFFSET.
4470 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4471 * reload1.c (reload): Likewise.
4472 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4473 instead of STARTING_FRAME_OFFSET.
4474 * function.c (try_fit_stack_local): Likewise.
4475 (assign_stack_local_1): Likewise
4476 (instantiate_virtual_regs): Likewise.
4477 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4478 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4479 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4480 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4481 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4482 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4483 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4484 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4485 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4486 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4487 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4488 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4489 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4490 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4491 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4492 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4493 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4494 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4495 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4496 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4497 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4498 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4499 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4500 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4501 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4502 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4503 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4504 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4505 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4506 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4507 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4508 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4509 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4510 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4511 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4512 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4513 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4514 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4515 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4516 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4517 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4518 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4519 return a HOST_WIDE_INT.
4520 (avr_builtin_setjmp_frame_value): Use it instead of
4521 STARTING_FRAME_OFFSET.
4522 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4523 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4524 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4526 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4527 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4528 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4529 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4530 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4531 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4532 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4533 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4534 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4535 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4536 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4537 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4539 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4540 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4541 * config/mips/mips.c (mips_compute_frame_info): Refer to
4542 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4543 (mips_starting_frame_offset): New function.
4544 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4545 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4546 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4547 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4548 and return a HOST_WIDE_INT.
4549 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4550 (mmix_initial_elimination_offset): Refer to
4551 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4552 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4553 * config/pa/pa.c (pa_starting_frame_offset): New function.
4554 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4555 (pa_expand_prologue): Likewise.
4556 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4557 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4558 !FRAME_GROWS_DOWNWARD handling to...
4559 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4560 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4561 !FRAME_GROWS_DOWNWARD handling to...
4562 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4563 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4564 !FRAME_GROWS_DOWNWARD handling to...
4565 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4566 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4568 (rs6000_starting_frame_offset): New function.
4569 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4570 !FRAME_GROWS_DOWNWARD handling to...
4571 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4572 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4573 !FRAME_GROWS_DOWNWARD handling to...
4574 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4575 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4576 !FRAME_GROWS_DOWNWARD handling to...
4577 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4578 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4579 (rs6000_starting_frame_offset): New function.
4580 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4581 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4582 * config/vax/vax.c (vax_starting_frame_offset): New function.
4583 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4584 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4585 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4586 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4587 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4588 * system.h (STARTING_FRAME_OFFSET): Poison.
4590 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4592 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4593 SCALAR_TYPE_MODE instead of TYPE_MODE.
4595 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4596 Alan Hayward <alan.hayward@arm.com>
4597 David Sherwood <david.sherwood@arm.com>
4599 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4601 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4602 Alan Hayward <alan.hayward@arm.com>
4603 David Sherwood <david.sherwood@arm.com>
4605 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4607 2017-10-23 Richard Biener <rguenther@suse.de>
4609 PR tree-optimization/82129
4610 * tree-ssa-pre.c (bitmap_set_and): Remove.
4611 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4612 canonicalizing expressions in the set to those with lowest
4613 ID rather than taking that from the first edge.
4615 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4617 * combine.c (rtx_equal_for_field_assignment_p): Use
4618 byte_lowpart_offset.
4620 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4621 Alan Hayward <alan.hayward@arm.com>
4622 David Sherwood <david.sherwood@arm.com>
4624 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4625 to a SUBREG_PROMOTED_VAR.
4627 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4628 Alan Hayward <alan.hayward@arm.com>
4629 David Sherwood <david.sherwood@arm.com>
4631 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4632 (expand_debug_source_expr): Likewise.
4633 * combine.c (combine_simplify_rtx): Likewise.
4634 * cse.c (fold_rtx): Likewise.
4635 * optabs.c (expand_float): Likewise.
4636 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4637 (simplify_binary_operation_1): Likewise.
4639 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4640 Alan Hayward <alan.hayward@arm.com>
4641 David Sherwood <david.sherwood@arm.com>
4643 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4644 (record_promoted_value): Likewise.
4645 * expr.c (expand_expr_real_2): Likewise.
4646 * ree.c (update_reg_equal_equiv_notes): Likewise.
4647 (combine_set_extension): Likewise.
4648 * rtlanal.c (low_bitmask_len): Likewise.
4649 * simplify-rtx.c (neg_const_int): Likewise.
4650 (simplify_binary_operation_1): Likewise.
4652 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4653 Alan Hayward <alan.hayward@arm.com>
4654 David Sherwood <david.sherwood@arm.com>
4656 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4657 * regcprop.c (maybe_mode_change): Likewise.
4658 * reload1.c (alter_reg): Likewise.
4660 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4662 * inchash.h (inchash::hash::add_wide_int): New function.
4663 * lto-streamer-out.c (hash_tree): Use it.
4665 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4667 * inchash.h (inchash::hash::add_wide_int): Rename to...
4668 (inchash::hash::add_hwi): ...this.
4669 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4670 (polymorphic_call_target_hasher::hash): Likewise.
4671 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4672 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4673 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4674 * lto-streamer-out.c (hash_tree): Likewise.
4675 * optc-save-gen.awk: Likewise.
4676 * tree.c (add_expr): Likewise.
4678 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4681 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4682 for ordered inequality comparisons even with TARGET_IEEE_FP.
4684 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4687 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4688 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4689 Expand with cmp<dwi>_doubleword.
4691 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4693 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4694 List CET intrinsics.
4695 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4696 specific to -fcf-protection option.
4698 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4700 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4701 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4702 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4703 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4704 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4705 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4706 (extra_objs): Add cet.o for Linux/x86 targets.
4707 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4708 * config/i386/cet.c: New file.
4709 * config/i386/cetintrin.h: Likewise.
4710 * config/i386/t-cet: Likewise.
4711 * config/i386/cpuid.h (bit_SHSTK): New.
4712 (bit_IBT): Likewise.
4713 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4714 pass IBT and SHSTK bits.
4715 * config/i386/i386-builtin-types.def
4716 (VOID_FTYPE_UNSIGNED_PVOID): New.
4717 (VOID_FTYPE_UINT64_PVOID): Likewise.
4718 * config/i386/i386-builtin.def: Add CET intrinsics.
4719 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4720 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4721 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4722 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4724 * config/i386/i386.c (rest_of_insert_endbranch): New.
4725 (pass_data_insert_endbranch): Likewise.
4726 (pass_insert_endbranch): Likewise.
4727 (make_pass_insert_endbranch): Likewise.
4728 (ix86_notrack_prefixed_insn_p): Likewise.
4729 (ix86_target_string): Add -mibt, -mshstk flags.
4730 (ix86_option_override_internal): Add flag_cf_protection
4732 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4733 (ix86_print_operand): Add 'notrack' prefix output.
4734 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4735 (ix86_expand_builtin): Expand CET intrinsics.
4736 (x86_output_mi_thunk): Add 'endbranch' instruction.
4737 * config/i386/i386.h (TARGET_IBT): New.
4738 (TARGET_IBT_P): Likewise.
4739 (TARGET_SHSTK): Likewise.
4740 (TARGET_SHSTK_P): Likewise.
4741 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4742 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4743 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4744 (builtin_setjmp_setup): New pattern.
4745 (builtin_longjmp): Likewise.
4746 (rdssp<mode>): Likewise.
4747 (incssp<mode>): Likewise.
4748 (saveprevssp): Likewise.
4749 (rstorssp): Likewise.
4750 (wrss<mode>): Likewise.
4751 (wruss<mode>): Likewise.
4752 (setssbsy): Likewise.
4753 (clrssbsy): Likewise.
4754 (nop_endbr): Likewise.
4755 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4757 * config/i386/immintrin.h: Include <cetintrin.h>.
4758 * config/i386/linux-common.h
4759 (file_end_indicate_exec_stack_and_cet): New prototype.
4760 (TARGET_ASM_FILE_END): New.
4762 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4764 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4765 latencies instead of having separate table; make difference between
4766 integer and float costs.
4767 * i386.h (processor_costs): Remove scalar_stmt_cost,
4768 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4769 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4771 * x86-tune-costs.h: Remove entries which has been removed in
4772 procesor_costs from all tables; make cond_taken_branch_cost
4773 and cond_not_taken_branch_cost COST_N_INSNS based.
4775 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4777 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4779 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4781 * config/i386/i386.md (isa): Remove fma_avx512f.
4782 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4783 <avx512>_fmadd_<mode>_mask3<round_name>,
4784 <avx512>_fmsub_<mode>_mask<round_name>,
4785 <avx512>_fmsub_<mode>_mask3<round_name>,
4786 <avx512>_fnmadd_<mode>_mask<round_name>,
4787 <avx512>_fnmadd_<mode>_mask3<round_name>,
4788 <avx512>_fnmsub_<mode>_mask<round_name>,
4789 <avx512>_fnmsub_<mode>_mask3<round_name>,
4790 <avx512>_fmaddsub_<mode>_mask<round_name>,
4791 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4792 <avx512>_fmsubadd_<mode>_mask<round_name>,
4793 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4794 (*vec_widen_umult_even_v16si<mask_name>,
4795 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4796 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4798 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4800 * extend.texi: Add 'nocf_check' documentation.
4801 * gimple.texi: Add second parameter to
4802 gimple_build_call_from_tree.
4803 * invoke.texi: Add -fcf-protection documentation.
4804 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4806 2017-10-20 Richard Biener <rguenther@suse.de>
4808 PR tree-optimization/82473
4809 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4810 the largest input type.
4812 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4814 * c-attribs.c (handle_nocf_check_attribute): New function.
4815 (c_common_attribute_table): Add 'nocf_check' handling.
4816 * gimple-parser.c: Add second argument NULL to
4817 gimple_build_call_from_tree.
4818 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4819 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4821 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4822 * common.opt: Add fcf-protection flag.
4823 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4824 * flag-types.h: Add enum cf_protection_level.
4825 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4826 Add 'nocf_check' attribute propagation to gimple call.
4827 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4828 (gimple_build_call_from_tree): Update prototype.
4829 (gimple_call_nocf_check_p): New function.
4830 (gimple_call_set_nocf_check): Likewise.
4831 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4832 * ipa-icf.c: Add nocf_check attribute in statement hash.
4833 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4834 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4835 * toplev.c (process_options): Add flag_cf_protection handling.
4837 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4839 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4841 2017-10-20 Richard Biener <rguenther@suse.de>
4843 PR tree-optimization/82603
4844 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4845 remove false predicated stores.
4847 2017-10-20 Richard Biener <rguenther@suse.de>
4849 * graphite-isl-ast-to-gimple.c
4850 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4851 Remove return value and simplify, dump copied stmt after lhs
4853 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4854 Reduce dump verbosity.
4855 (gsi_insert_earliest): Likewise.
4856 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4857 * graphite.c (print_global_statistics): Adjust dumping.
4858 (print_graphite_scop_statistics): Likewise.
4859 (print_graphite_statistics): Do not dump loops here.
4860 (graphite_transform_loops): But here.
4862 2017-10-20 Nicolas Roche <roche@adacore.com>
4864 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4865 * configure: Regenerate.
4867 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4870 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4871 functions when optimizing replace GIMPLE_RETURN stmts with
4872 calls to __builtin_unreachable ().
4875 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4876 for -fsanitize=thread link of executables.
4877 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4878 link of executables.
4881 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4883 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4884 define_insns for logical vector shifts to use VI248_AVX512BW
4885 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4886 condition, useless isa and prefix attributes. Change the first
4887 2 of these define_insns to ...
4888 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4889 define_insn for avx512vl.
4890 (<shift_insn><mode>3): ... and this, new define_insn without
4891 masking for non-avx512vl.
4894 * config/i386/sse.md (*andnot<mode>3,
4895 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4896 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4897 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4898 not applied use empty suffix even for TARGET_AVX512VL.
4899 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4900 is applied, supply evex,evex or evex,evex,evex instead of just
4903 2017-10-20 Julia Koval <julia.koval@intel.com>
4905 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4906 (OPTION_MASK_ISA_GFNI_UNSET): New.
4907 (ix86_handle_option): Handle OPT_mgfni.
4908 * config/i386/cpuid.h (bit_GFNI): New.
4909 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4910 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4911 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4912 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4913 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4914 * config/i386/i386.opt: Add mgfni.
4916 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4918 * config/msp430/msp430.c (msp430_option_override): Disable
4919 -fdelete-null-pointer-checks.
4920 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4922 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4924 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4925 of x87 and SSE instructions.
4927 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4929 * asan.c (create_cond_insert_point): Do not update edge count.
4930 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4931 (afdo_propagate_circuit): Likewise.
4932 (afdo_calculate_branch_prob): Likewise.
4933 (afdo_annotate_cfg): Likewise.
4934 * basic-block.h (struct edge_def): Remove count.
4935 (edge_def::count): New accessor.
4936 * bb-reorder.c (rotate_loop): Update.
4937 (find_traces_1_round): Update.
4938 (connect_traces): Update.
4939 (sanitize_hot_paths): Update.
4940 * cfg.c (unchecked_make_edge): Update.
4941 (make_single_succ_edge): Update.
4942 (check_bb_profile): Update.
4943 (dump_edge_info): Update.
4944 (update_bb_profile_for_threading): Update.
4945 (scale_bbs_frequencies_int): Update.
4946 (scale_bbs_frequencies_gcov_type): Update.
4947 (scale_bbs_frequencies_profile_count): Update.
4948 (scale_bbs_frequencies): Update.
4949 * cfganal.c (connect_infinite_loops_to_exit): Update.
4950 * cfgbuild.c (compute_outgoing_frequencies): Update.
4951 (find_many_sub_basic_blocks): Update.
4952 * cfgcleanup.c (try_forward_edges): Update.
4953 (try_crossjump_to_edge): Update
4954 * cfgexpand.c (expand_gimple_cond): Update
4955 (expand_gimple_tailcall): Update
4956 (construct_exit_block): Update
4957 * cfghooks.c (verify_flow_info): Update
4958 (redirect_edge_succ_nodup): Update
4959 (split_edge): Update
4960 (make_forwarder_block): Update
4961 (duplicate_block): Update
4962 (account_profile_record): Update
4963 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4964 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4965 * cfgloopmanip.c (scale_loop_profile): Update.
4967 (lv_adjust_loop_entry_edge): Update.
4968 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4969 (force_nonfallthru_and_redirect): Update.
4970 (purge_dead_edges): Update.
4971 (rtl_flow_call_edges_add): Update.
4972 * cgraphunit.c (init_lowered_empty_function): Update.
4973 (cgraph_node::expand_thunk): Update.
4974 * gimple-pretty-print.c (dump_probability): Update.
4975 (dump_edge_probability): Update.
4976 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4977 * haifa-sched.c (sched_create_recovery_edges): Update.
4978 * hsa-gen.c (convert_switch_statements): Update.
4979 * ifcvt.c (dead_or_predicable): Update.
4980 * ipa-inline-transform.c (inline_transform): Update.
4981 * ipa-split.c (split_function): Update.
4982 * ipa-utils.c (ipa_merge_profiles): Update.
4983 * loop-doloop.c (add_test): Update.
4984 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4985 * lto-streamer-in.c (input_cfg): Update.
4986 (input_function): Update.
4987 * lto-streamer-out.c (output_cfg): Update.
4988 * modulo-sched.c (sms_schedule): Update.
4989 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4990 * predict.c (maybe_hot_edge_p): Update.
4991 (unlikely_executed_edge_p): Update.
4992 (probably_never_executed_edge_p): Update.
4993 (dump_prediction): Update.
4994 (drop_profile): Update.
4995 (propagate_unlikely_bbs_forward): Update.
4996 (determine_unlikely_bbs): Update.
4997 (force_edge_cold): Update.
4998 * profile.c (compute_branch_probabilities): Update.
4999 * reg-stack.c (better_edge): Update.
5000 * shrink-wrap.c (handle_simple_exit): Update.
5001 * tracer.c (better_p): Update.
5002 * trans-mem.c (expand_transaction): Update.
5003 (split_bb_make_tm_edge): Update.
5004 * tree-call-cdce.c: Update.
5005 * tree-cfg.c (gimple_find_sub_bbs): Update.
5006 (gimple_split_edge): Update.
5007 (gimple_duplicate_sese_region): Update.
5008 (gimple_duplicate_sese_tail): Update.
5009 (gimple_flow_call_edges_add): Update.
5010 (insert_cond_bb): Update.
5011 (execute_fixup_cfg): Update.
5012 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5013 * tree-complex.c (expand_complex_div_wide): Update.
5014 * tree-eh.c (lower_resx): Update.
5015 (unsplit_eh): Update.
5016 (cleanup_empty_eh_move_lp): Update.
5017 * tree-inline.c (copy_edges_for_bb): Update.
5018 (freqs_to_counts): Update.
5019 (copy_cfg_body): Update.
5020 * tree-ssa-dce.c (remove_dead_stmt): Update.
5021 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5022 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5023 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5024 (unloop_loops): Update.
5025 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5026 * tree-ssa-loop-split.c (connect_loops): Update.
5027 (split_loop): Update.
5028 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5029 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5030 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5031 * tree-ssa-reassoc.c (branch_fixup): Update.
5032 * tree-ssa-tail-merge.c (replace_block_by): Update.
5033 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5034 (compute_path_counts): Update.
5035 (update_profile): Update.
5036 (recompute_probabilities): Update.
5037 (update_joiner_offpath_counts): Update.
5038 (estimated_freqs_path): Update.
5039 (freqs_to_counts_path): Update.
5040 (clear_counts_path): Update.
5041 (ssa_fix_duplicate_block_edges): Update.
5042 (duplicate_thread_path): Update.
5043 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5044 (case_bit_test_cmp): Update.
5045 (collect_switch_conv_info): Update.
5046 (gen_inbound_check): Update.
5047 (do_jump_if_equal): Update.
5048 (emit_cmp_and_jump_insns): Update.
5049 * tree-tailcall.c (decrease_profile): Update.
5050 (eliminate_tail_call): Update.
5051 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5052 (vect_do_peeling): Update.
5053 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5054 * ubsan.c (ubsan_expand_null_ifn): Update.
5055 (ubsan_expand_ptr_ifn): Update.
5056 * value-prof.c (gimple_divmod_fixed_value): Update.
5057 (gimple_mod_pow2): Update.
5058 (gimple_mod_subtract): Update.
5059 (gimple_ic): Update.
5060 (gimple_stringop_fixed_value): Update.
5062 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5065 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5067 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5069 PR rtl-optimization/82395
5070 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5071 based on non_spilled_static_chain_regno_p.
5073 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5075 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5076 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5077 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5079 2017-10-19 Martin Sebor <msebor@redhat.com>
5081 PR tree-optimization/82596
5082 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5084 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5086 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5087 (get_mem_refs_of_builtin_call): Likewise.
5088 * builtins.c (expand_builtin_apply): Adjust call to
5089 allocate_dynamic_stack_space.
5090 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5091 the third argument to allocate_dynamic_stack_space, otherwise -1.
5092 (expand_builtin): Deal with all alloca variants.
5093 (is_inexpensive_builtin): Likewise.
5094 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5095 * calls.c (special_function_p): Deal with all alloca variants.
5096 (initialize_argument_information): Adjust call to
5097 allocate_dynamic_stack_space.
5098 (expand_call): Likewise.
5099 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5100 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5101 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5102 use it for the stack usage computation.
5103 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5104 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5105 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5106 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5107 (in_loop_p): Remove first argument and useless check.
5108 (pass_walloca::execute): Remove useless test and adjust call to above.
5109 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5110 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5111 (gimplify_call_expr): Deal with all alloca variants.
5112 * hsa-gen.c (gen_hsa_alloca): Likewise.
5113 (gen_hsa_insns_for_call): Likewise.
5114 * ipa-pure-const.c (special_builtin_state): Likewise.
5115 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5116 * tree-object-size.c (alloc_object_size): Likewise.
5117 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5118 (call_may_clobber_ref_p_1): Likewise.
5119 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5120 (ccp_fold_stmt): Likewise.
5121 (optimize_stack_restore): Likewise.
5122 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5123 (mark_all_reaching_defs_necessary_1): Likewise.
5124 (propagate_necessity): Likewise.
5125 (eliminate_unnecessary_stmts): Likewise.
5126 * tree.c (build_common_builtin_nodes): Build
5127 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5128 (build_alloca_call_expr): New function.
5129 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5130 (CASE_BUILT_IN_ALLOCA): Likewise.
5131 (build_alloca_call_expr): Declare.
5132 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5134 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5137 * dwarf2out.c (new_die_raw): New static inline function.
5138 (new_die): Use it to create the DIE.
5139 (add_AT_external_die_ref): Likewise.
5140 (clone_die): Likewise.
5141 (clone_as_declaration): Likewise.
5142 (dwarf2out_vms_debug_main_pointer): Likewise.
5143 (base_type_die): Likewise. Remove early return for corner cases.
5144 Do not call add_pubtype on the DIE here.
5145 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5146 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5147 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5148 native order exists for base types, attach the DIE manually and call
5149 add_pubtype on it. Do not equate a reverse order DIE to the type.
5151 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5153 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5154 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5156 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5158 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5160 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5161 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5162 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5163 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5164 * opts.c (sanitizer_opts): Add builtin.
5165 * ubsan.c (instrument_builtin): New function.
5166 (pass_ubsan::execute): Call it.
5167 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5168 * doc/invoke.texi: Document -fsanitize=builtin.
5170 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5171 builtins, store max (log2 (align), 0) into uchar field instead of
5172 align into uptr field.
5173 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5174 store uchar 0 field instead of uptr 0 field.
5175 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5176 instead of passing one address of struct with 2 locations pass
5177 two addresses of structs with 1 location each.
5178 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5179 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5180 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5181 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5182 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5183 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5184 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5185 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5187 2017-10-19 Martin Liska <mliska@suse.cz>
5190 * file-find.c (remove_prefix): Remove.
5191 * file-find.h (remove_prefix): Likewise.
5192 * gcc-ar.c: Remove smartness of lookup.
5194 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5196 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5197 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5198 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5200 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5203 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5204 (setcc + and to xor + setcc): New peephole2.
5206 2017-10-19 Tom de Vries <tom@codesourcery.com>
5208 * doc/sourcebuild.texi (Test Directives, Variants of
5209 dg-require-support): Add dg-require-stack-size.
5211 2017-10-19 Martin Liska <mliska@suse.cz>
5214 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5215 that have a large alignment.
5216 (gimplify_target_expr): Likewise.
5218 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5220 PR rtl-optimization/82602
5221 * ira.c (rtx_moveable_p): Return false for volatile asm.
5223 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5226 * config/i386/i386-modes.def (CCGZ): New CC mode.
5227 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5228 * config/i386/predicates.md (ix86_comparison_operator):
5230 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5231 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5232 with double-word subtraction.
5233 (put_condition_code): Handle CCGZmode.
5235 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5237 * wide-int.cc (debug (const wide_int &)): New.
5238 (debug (const wide_int *)): New.
5239 (debug (const widest_int &)): New.
5240 (debug (const widest_int *)): New.
5242 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5245 * lra-constraints.c (curr_insn_transform): Use non-input operand
5246 instead of output one for matched reload.
5248 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5250 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5251 (tree-ssa-loop-ivopts.h): New header file.
5252 (struct builtin_info): New fields.
5253 (classify_builtin_1): Compute and record base and offset parts for
5254 memset builtin partition by calling strip_offset.
5255 (offset_cmp, fuse_memset_builtins): New functions.
5256 (finalize_partitions): Fuse adjacent memset partitions by calling
5258 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5259 Expose the interface.
5260 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5262 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5264 PR tree-optimization/82574
5265 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5266 that data reference must be executed exactly once per iteration
5267 against the outermost loop in nest.
5268 (classify_partition): Update call to above function.
5270 2017-10-18 Richard Biener <rguenther@suse.de>
5272 PR tree-optimization/82591
5273 * graphite.c (graphite_transform_loops): Move code gen message
5275 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5276 Here. Handle scop_to_isl_ast failing.
5277 (scop_to_isl_ast): Limit the number of ISL operations.
5279 2017-10-18 Richard Biener <rguenther@suse.de>
5281 * graphite-isl-ast-to-gimple.c
5282 (translate_isl_ast_to_gimple::set_rename): Simplify.
5283 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5284 (graphite_copy_stmts_from_block): ... here.
5285 (copy_bb_and_scalar_dependences): Simplify.
5286 (add_parameters_to_ivs_params): Canonicalize.
5287 (generate_entry_out_of_ssa_copies): Simplify.
5288 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5289 by passing in ISL dimension.
5290 (parameter_index_in_region_1): Rename to ...
5291 (parameter_index_in_region): ... this.
5292 (extract_affine): Adjust assert, pass down parameter index.
5293 (add_param_constraints): Use range-info when available.
5294 (build_scop_context): Adjust.
5295 * sese.c (new_sese_info): Adjust.
5296 (free_sese_info): Likewise.
5297 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5298 Remove unused typedefs.
5299 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5301 2017-10-18 Martin Liska <mliska@suse.cz>
5303 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5305 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5307 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5308 (s390_sched_init): Do not reset s390_sched_state if we entered the
5309 current basic block via a fallthru edge and all others are unlikely.
5311 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5313 * config/s390/s390.c (NUM_SIDES): New variable.
5314 (LONGRUNNING_THRESHOLD): New variable.
5315 (LATENCY_FACTOR): New variable.
5316 (s390_sched_score): Decrease score for long-running instructions on
5318 (s390_sched_variable_issue): Perform bookkeeping for long-running
5321 2017-10-18 Richard Biener <rguenther@suse.de>
5323 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5324 Simplify with removal of the parameter rename map.
5325 (set_rename): Likewise.
5326 (should_copy_to_new_region): Likewise.
5327 (graphite_copy_stmts_from_block): Likewise.
5328 (copy_bb_and_scalar_dependences): Remove initialization of
5329 unused copied_bb_map.
5331 (copy_internal_parameters): Likewise.
5332 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5333 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5334 Use INTEGRAL_TYPE_P.
5335 (parameter_index_in_region_1): Rename to ...
5336 (assign_parameter_index_in_region): ... this. Assert we have
5337 a parameter we handle.
5338 (scan_tree_for_params): Adjust.
5339 * sese.h (parameter_rename_map_t): Remove.
5340 (struct sese_info_t): Remove unused parameter_rename_map and
5341 copied_bb_map members.
5342 * sese.c (new_sese_info): Adjust.
5343 (free_sese_info): Likewise.
5345 2017-10-18 Martin Liska <mliska@suse.cz>
5348 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5349 on an abnormal edge.
5351 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5353 * doc/invoke.texi (ffunction-sections and fdata-sections):
5356 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5358 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5359 the use statement can throw internally.
5361 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5363 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5364 any RTX present on the RHS of a SET.
5365 * compare-elim.c (try_eliminate_compare): Restore comment.
5367 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5369 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5370 may be also called on tcc_type nodes.
5371 * langhooks.c (lhd_tree_size): Likewise.
5373 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5375 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5376 format_warning_at_substring.
5377 (maybe_warn): Convert source_range * param to a location_t. Pass
5378 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5379 (format_directive): Remove code to extract source_ranges and
5380 source_range * in favor of just a location_t.
5381 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5383 * substring-locations.c (format_warning_va): Convert
5384 source_range * param to a location_t.
5385 (format_warning_at_substring): Likewise.
5386 * substring-locations.h (format_warning_va): Likewise.
5387 (format_warning_at_substring): Likewise.
5389 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5391 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5393 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5394 and scatter/gather ops.
5396 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5397 vec_gather_load and vec_scatter_store.
5398 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5399 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5400 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5401 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5402 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5403 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5405 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5407 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5408 Detect FTST instruction and handle its register pops. Only pop
5409 second operand if can_pop_second_op is true.
5410 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5411 set can_pop_second_op to false in the compare_for_stack_reg call.
5413 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5414 output_fp_compare for stack register operands.
5415 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5416 instructions here. Do not emit stack register pops here. Assert
5417 that FCOMPP pops next to top stack register. Rewrite function.
5419 2017-10-17 Nathan Sidwell <nathan@acm.org>
5422 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5423 use DECL_ASSEMBLER_NAME_RAW.
5426 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5429 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5430 Wilco Dijkstra <wilco.dijkstra@arm.com>
5432 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5433 converstion to Pmode from the buf_addr.
5435 2017-10-17 Richard Biener <rguenther@suse.de>
5437 * graphite-dependences.c (scop_get_reads_and_writes): Change
5438 output parameters to references.
5440 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5442 PR 71026/tree-optimization
5443 * fold-const.c (distribute_real_division): Removed.
5444 (fold_binary_loc): Remove calls to distribute_real_divison.
5446 2017-10-17 Richard Biener <rguenther@suse.de>
5448 * graphite-scop-detection.c
5449 (scop_detection::stmt_has_simple_data_refs_p): Always use
5450 the full nest as region.
5451 (try_generate_gimple_bb): Likewise.
5452 * sese.c (scalar_evolution_in_region): Simplify now that
5453 SCEV can handle instantiation in regions.
5454 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5455 in the non-loop part of a function if requested.
5457 2017-10-17 Richard Biener <rguenther@suse.de>
5459 PR tree-optimization/82563
5460 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5462 (graphite_regenerate_ast_isl): Call it.
5463 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5465 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5467 PR tree-optimization/82549
5468 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5469 Formatting fixes. Instead of calling make_bit_field_ref with negative
5472 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5474 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5475 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5476 _mm_maskz_reduce_ss): New.
5477 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5478 __builtin_ia32_reducess_mask): Ditto..
5479 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5480 * config/i386/sse.md (reduces<mode>): Renamed to ...
5481 (reduces<mode><mask_scalar_name>): ... this.
5482 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5484 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5485 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5487 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5489 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5490 * selftest-run-tests.c (selftest::run_tests): Call
5491 selftest::unique_ptr_tests_cc_tests.
5492 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5493 * unique-ptr-tests.cc: New file.
5495 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5498 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5500 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5501 (make_hard_regno_born, make_hard_regno_dead): Update
5502 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5504 2017-10-16 Jeff Law <law@redhat.com>
5506 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5508 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5510 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5512 2017-10-16 Olivier Hainque <hainque@adacore.com>
5514 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5515 with_cpu if we were configured for an e500v2 target cpu name.
5517 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5519 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5520 * doc/invoke.texi: Document +nodsp as a valid extension for
5523 2017-10-16 Martin Liska <mliska@suse.cz>
5525 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5526 (test_set_range): Likewise.
5527 (test_range_functions): Rename to ...
5528 (test_bit_in_range): ... this.
5529 (sbitmap_c_tests): Add new test.
5531 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5533 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5535 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5536 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5538 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5540 * config/aarch64/aarch64-builtins.c
5541 (aarch64_types_quadopu_lane_qualifiers): New.
5542 (TYPES_QUADOPU_LANE): New.
5543 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5544 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5545 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5546 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5547 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5548 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5549 (Vdottype, DOTPROD): New.
5550 (sur): Add SDOT and UDOT.
5552 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5554 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5555 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5556 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5558 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5559 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5560 Enable TARGET_DOTPROD.
5561 (cortex-a75.cortex-a55): Likewise.
5562 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5564 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5566 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5567 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5569 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5571 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5572 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5573 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5574 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5575 * config/arm/types.md (neon_dot, neon_dot_q): New.
5576 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5578 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5580 * config/arm/arm.h (TARGET_DOTPROD): New.
5581 * config/arm/arm.c (arm_arch_dotprod): New.
5582 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5583 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5584 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5585 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5586 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5587 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5588 * doc/invoke.texi (armv8.2-a): Document dotprod
5590 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5592 * i386.c (ix86_vec_cost): New function.
5593 (ix86_rtx_costs): Handle vector operations better.
5594 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5595 * x86-tune-costs.h: Add new costs to all tables.
5597 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5599 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5601 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5602 divsd, sqrtss and sqrtsd
5603 * x86-tune-costs.h: Add new entries to all costs.
5604 (znver1_cost): Fix to match real instruction latencies.
5606 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5607 Michael Collison <michael.collison@arm.com>
5609 * compare-elim.c: Include emit-rtl.h.
5610 (can_merge_compare_into_arith): New function.
5611 (try_validate_parallel): Likewise.
5612 (try_merge_compare): Likewise.
5613 (try_eliminate_compare): Call the above when no previous clobber
5615 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5618 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5622 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5623 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5627 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5628 to be any operand_equal_p operands. For & (B - 1) require
5629 B to be power of 2. Recognize
5630 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5632 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5635 * optabs.c (expand_memory_blockage): Fix call of
5636 targetm.have_memory_blockage.
5638 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5641 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5642 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5643 objects to extra_objs instead of overwriting it.
5645 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5647 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5648 Use any_fp_register_operand as operand[3] predicate. Simplify
5649 equality test for operands[2] and operands[4] memory location.
5650 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5651 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5652 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5653 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5654 any_fp_register_operand as operand[1] predicate. Simplify
5655 equality test for operands[0] and operands[3] memory location.
5656 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5657 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5658 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5660 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5662 * target-insns.def: Add memory_blockage.
5663 * optabs.c (expand_memory_blockage): New function.
5664 (expand_asm_memory_barrier): Rename ...
5665 (expand_asm_memory_blockage): ... to this.
5666 (expand_mem_thread_fence): Call expand_memory_blockage
5667 instead of expand_asm_memory_barrier.
5668 (expand_mem_singnal_fence): Ditto.
5669 (expand_atomic_load): Ditto.
5670 (expand_atomic_store): Ditto.
5671 * doc/md.texi (Standard Pattern Names For Generation):
5672 Document memory_blockage instruction pattern.
5674 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5676 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5677 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5678 * config/rl78/rl78.md: New define_expand "adddi3".
5680 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5682 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5685 2017-10-13 Jeff Law <law@redhat.com>
5687 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5689 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5692 * internal-fn.c (expand_mul_overflow): If both operands have
5693 the same highpart of -1 or 0 and the topmost bit of lowpart
5694 is different, overflow is if res <= 0 rather than res < 0.
5696 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5698 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5699 TARGET_P9_VECTOR code for unaligned_load case.
5701 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5703 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5705 2017-10-13 Nathan Sidwell <nathan@acm.org>
5707 * tree-core.h (tree_contains_struct): Make bool.
5708 * tree.c (tree_contains_struct): Likewise.
5709 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5710 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5711 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5712 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5713 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5715 2017-10-13 Richard Biener <rguenther@suse.de>
5717 * graphite-isl-ast-to-gimple.c
5718 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5719 parameters and dominance check.
5720 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5721 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5722 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5723 Do not update SSA form here or do intermediate IL verification.
5724 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5725 (graphite_initialize): Remove check on the number of loops in
5726 the function and inline into graphite_transform_loops.
5727 (graphite_finalize): Inline into graphite_transform_loops.
5728 (graphite_transform_loops): Perform SSA update and IL verification
5730 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5732 2017-10-13 Richard Biener <rguenther@suse.de>
5734 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5735 graphite_expression_type_precision): Avoid global constructor
5737 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5738 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5739 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5740 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5741 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5742 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5744 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5747 * config/i386/i386.h (ix86_red_zone_size): New.
5748 * config/i386/i386.md (push peephole2s): Replace
5749 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5751 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5752 Alan Hayward <alan.hayward@arm.com>
5753 David Sherwood <david.sherwood@arm.com>
5755 * combine.c (can_change_dest_mode): Reject changes in
5756 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5763 (expand_debug_source_expr): Likewise.
5764 * combine.c (combine_simplify_rtx): Likewise.
5765 * cse.c (fold_rtx): Likewise.
5766 * fwprop.c (canonicalize_address): Likewise.
5767 * targhooks.c (default_shift_truncation_mask): 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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5774 (widened_mode): Likewise.
5775 (expand_unop): Likewise.
5776 * ree.c (transform_ifelse): Likewise.
5777 (merge_def_and_ext): Likewise.
5778 (combine_reaching_defs): Likewise.
5779 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5781 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5782 Alan Hayward <alan.hayward@arm.com>
5783 David Sherwood <david.sherwood@arm.com>
5785 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5786 * combine.c (gen_lowpart_for_combine): Likewise.
5787 * dwarf2out.c (rtl_for_decl_location): Likewise.
5788 * final.c (alter_subreg): Likewise.
5789 * rtlhooks.c (gen_lowpart_general): Likewise.
5790 (gen_lowpart_if_possible): Likewise.
5792 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5793 Alan Hayward <alan.hayward@arm.com>
5794 David Sherwood <david.sherwood@arm.com>
5796 * calls.c (expand_call): Use subreg_lowpart_offset.
5797 * cse.c (cse_insn): Likewise.
5798 * regcprop.c (copy_value): Likewise.
5799 (copyprop_hardreg_forward_1): Likewise.
5801 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5804 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5805 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5806 =Q constraints to +Q and into insn condition add check
5807 that operands[0] and operands[1] are equal.
5808 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5809 =Q constraints to +Q and into insn condition add check
5810 that operands[0] is equal to either operands[1] or operands[2].
5813 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5814 instead of handling MINUS_EXPR twice (once for each argument),
5815 canonicalize operand order and handle just once, use rtype where
5816 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5819 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5820 any values of __C while still being pattern recognizable as a simple
5823 2017-10-13 Richard Biener <rguenther@suse.de>
5825 PR tree-optimization/82451
5827 2017-10-02 Richard Biener <rguenther@suse.de>
5829 PR tree-optimization/82355
5830 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5831 a mapping for the enclosing loop but avoid generating one for
5833 (copy_bb_and_scalar_dependences): Remove premature codegen
5834 error on PHIs in blocks duplicated into multiple places.
5835 * graphite-scop-detection.c
5836 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5837 in the region use it as loop and nest to analyze the DR in.
5838 (try_generate_gimple_bb): Likewise.
5839 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5840 (add_loop_constraints): For blocks in a loop not in the region
5841 create a dimension with a single iteration.
5842 * sese.h (gbb_loop_at_index): Remove assert.
5844 * cfgloop.c (loop_preheader_edge): For the loop tree root
5845 return the single successor of the entry block.
5846 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5847 Reset the SCEV hashtable and niters.
5848 * graphite-scop-detection.c
5849 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5850 assert that we only have POLYNOMIAL_CHREC that vary in loops
5851 contained in the region.
5852 (scop_detection::graphite_can_represent_expr): Adjust.
5853 (scop_detection::stmt_has_simple_data_refs_p): For loops
5854 not in the region set loop to NULL. The nest is now the
5855 entry edge to the region.
5856 (try_generate_gimple_bb): Likewise.
5857 * sese.c (scalar_evolution_in_region): Adjust for
5858 instantiate_scev change.
5859 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5860 Make nest parameter the edge into the region.
5861 (create_data_ref): Likewise.
5862 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5863 entry edge into a region and adjust instantiate_scev calls.
5864 (create_data_ref): Likewise.
5865 (graphite_find_data_references_in_stmt): Likewise.
5866 (find_data_references_in_stmt): Pass the loop preheader edge
5867 from the nest argument.
5868 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5869 parameter the edge into the region.
5870 (instantiate_parameters): Use the loop preheader edge as entry.
5871 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5873 (get_instantiated_value_entry): Make instantiate_below parameter
5874 the edge into the region.
5875 (instantiate_scev_name): Likewise. Adjust dominance checks,
5876 when we cannot use loop-based instantiation instantiate by
5877 walking use-def chains.
5878 (instantiate_scev_poly): Adjust.
5879 (instantiate_scev_binary): Likewise.
5880 (instantiate_scev_convert): Likewise.
5881 (instantiate_scev_not): Likewise.
5882 (instantiate_array_ref): Remove.
5883 (instantiate_scev_3): Likewise.
5884 (instantiate_scev_2): Likewise.
5885 (instantiate_scev_1): Likewise.
5886 (instantiate_scev_r): Do not blindly handle N-operand trees.
5887 Do not instantiate array-refs. Handle all constants and invariants.
5888 (instantiate_scev): Make instantiate_below parameter
5889 the edge into the region.
5890 (resolve_mixers): Use the loop preheader edge for the region
5891 parameter to instantiate_scev_r.
5892 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5894 2017-10-13 Richard Biener <rguenther@suse.de>
5896 PR tree-optimization/82525
5897 * graphite-isl-ast-to-gimple.c
5898 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5900 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5901 Fail code generation when we cannot represent the isl integer.
5902 (binary_op_to_tree): Elide modulo operations that are no-ops
5903 in the type we code generate. Remove now superfluous code
5906 2017-10-13 Richard Biener <rguenther@suse.de>
5908 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5909 (scop_detection::harmful_loop_in_region): Remove premature
5910 IV type restriction.
5911 (scop_detection::graphite_can_represent_scev): We can handle
5912 pointer IVs just fine.
5914 2017-10-13 Alan Modra <amodra@gmail.com>
5916 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5917 "Clobbers and Scratch Registers". Add paragraph on
5918 alternative to clobbers for scratch registers and OpenBLAS
5921 2017-10-13 Alan Modra <amodra@gmail.com>
5923 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5924 example of a memory input for a string of known length. Move
5925 commentary out of table. Add a number of new examples
5926 covering array memory inputs.
5928 2017-10-12 Martin Liska <mliska@suse.cz>
5930 PR tree-optimization/82493
5931 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5932 (test_range_functions): New function.
5933 (sbitmap_c_tests): Likewise.
5934 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5935 * selftest.h (sbitmap_c_tests): New function.
5937 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5939 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5941 * config/rs6000/amo.h: Fix spacing issue.
5943 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5946 * config/i386/i386.md (*ashl<mode>3_mask_1,
5947 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5948 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5951 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5953 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5954 (profile_probability): Set max_probability
5955 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5957 * profile-count.c (profile_probability::differs_from_p): Do not
5958 rely on max_probaiblity == 10000
5960 2017-10-12 Jeff Law <law@redhat.com>
5962 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5965 2017-10-12 Martin Sebor <msebor@redhat.com>
5969 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5970 (handle_alias_pairs): Call it.
5971 * common.opt (-Wattribute-alias): New option.
5972 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5973 * doc/invoke.texi (-Wattribute-alias): Document.
5975 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5978 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5980 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5982 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5983 (make_hard_regno_born, make_hard_regno_dead): Update
5984 bb_killed_pseudos and bb_gen_pseudos.
5986 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5988 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5990 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5992 * config/alpha/alpha.c (alpha_split_conditional_move):
5993 Use std::swap instead of manually swapping.
5994 (alpha_stdarg_optimize_hook): Ditto.
5995 (alpha_canonicalize_comparison): Ditto.
5997 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5999 * tree-loop-distribution.c (struct builtin_info): New struct.
6000 (struct partition): Refactor fields into struct builtin_info.
6001 (partition_free): Free struct builtin_info.
6002 (build_size_arg_loc, build_addr_arg_loc): Delete.
6003 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6004 information from struct builtin_info.
6005 (find_single_drs): New function refactored from classify_partition.
6006 Also moved builtin validity checks to this function.
6007 (compute_access_range, alloc_builtin): New functions.
6008 (classify_builtin_st, classify_builtin_ldst): New functions.
6009 (classify_partition): Refactor code into functions find_single_drs,
6010 classify_builtin_st and classify_builtin_ldst.
6011 (distribute_loop): Don't do runtime alias check when distributing
6013 (find_seed_stmts_for_distribution): New function.
6014 (pass_loop_distribution::execute): Refactor code finding seed
6015 stmts into above function. Support distribution for the innermost
6016 two-level loop nest. Adjust dump information.
6018 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6020 * tree-loop-distribution.c: Adjust the general comment.
6021 (NUM_PARTITION_THRESHOLD): New macro.
6022 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6023 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6024 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6025 in call to build_partition_graph.
6026 (finalize_partitions): New parameter. Make loop distribution more
6027 conservative by fusing more partitions.
6028 (distribute_loop): Don't do runtime alias check in case of loop nest
6030 (find_seed_stmts_for_distribution): New function.
6031 (prepare_perfect_loop_nest): New function.
6032 (pass_loop_distribution::execute): Refactor code finding seed stmts
6033 and loop nest into above functions. Support loop nest distribution.
6034 Adjust dump information accordingly.
6036 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6038 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6039 and set PTYPE_SEQUENTIAL for merged partition.
6041 2017-10-12 Richard Biener <rguenther@suse.de>
6043 PR tree-optimization/69728
6045 2017-09-19 Richard Biener <rguenther@suse.de>
6047 PR tree-optimization/69728
6048 * graphite-sese-to-poly.c (schedule_error): New global.
6049 (add_loop_schedule): Handle empty domain by failing the
6051 (build_original_schedule): Handle schedule_error.
6053 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6054 domain by returning an unchanged schedule.
6056 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6058 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6059 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6061 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6063 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6066 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6069 * expr.c (store_field): Don't optimize away bitsize == 0 store
6070 from CALL_EXPR with addressable return type.
6072 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6074 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6075 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6076 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6077 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6078 TARGET_ISEL instead of TARGET_ISEL<sel>.
6080 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6082 * config/rs6000/rs6000.c
6083 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6085 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6087 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6089 (reg_or_cint_operand): Fix comment.
6090 (reg_or_zero_operand): New predicate.
6091 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6092 * config/rs6000/rs6000.c (output_isel): Delete.
6093 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6094 instead of reg_or_cint_operand. Output instruction directly (not via
6096 (isel_unsigned_<mode>): Ditto.
6097 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6098 gpc_reg_operand. Add an instruction alternative for this. Output
6099 instruction directly.
6100 (*isel_reversed_unsigned_<mode>): Ditto.
6102 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6104 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6105 (TARGET_CANONICALIZE_COMPARISON): Define.
6107 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6110 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6111 Check whether the dest is REG before adding REG_EQUIV note.
6113 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6116 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6118 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6119 (make_hard_regno_born, make_hard_regno_dead): Update
6120 bb_killed_pseudos and bb_gen_pseudos.
6122 2017-10-11 Nathan Sidwell <nathan@acm.org>
6124 * incpath.h (enum incpath_kind): Name enum, prefix values.
6125 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6126 * incpath.c (heads, tails): Use INC_MAX.
6127 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6128 (merge_include_chains, split_quote_chain,
6129 register_include_chains): Update incpath_kind names.
6130 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6131 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6133 (add_framework_path, darwin_register_objc_includes): Likewise.
6134 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6136 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6138 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6139 Do not use float_operator operator predicate.
6140 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6141 * config/i386/predicates.md (float_operator): Remove predicate.
6143 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6145 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6146 (*jccxf_i387): Ditto.
6147 (*jcc<mode>_i387): Ditto.
6148 (*jccu<mode>_i387): Ditto.
6149 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6150 (*jcc_*_i387 splitters): Remove.
6151 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6152 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6153 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6156 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6158 * profile-count.h (slow_safe_scale_64bit): New function.
6159 (safe_scale_64bit): New inline.
6160 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6161 * profile-count.c: Include wide-int.h
6162 (slow_safe_scale_64bit): New.
6164 2017-10-11 Nathan Sidwell <nathan@acm.org>
6166 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6167 HAS_DECL_ASSEMBLER_NAME_P.
6168 * gimple-expr.c (gimple_decl_printable_name: Check
6169 HAS_DECL_ASSEMBLER_NAME_P too.
6170 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6171 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6172 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6173 * passes.c (rest_of_decl_compilation): Check
6174 HAS_DECL_ASSEMBLER_NAME_P too.
6175 * recog.c (verify_changes): Likewise.
6176 * tree-pretty-print.c (dump_decl_name): Likewise.
6177 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6179 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6180 (SET_DECL_ASSEMBLER_NAME): Use it.
6181 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6182 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6183 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6185 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6187 * config.gcc (i386, x86_64): Add extra objects.
6188 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6189 (ix86_min_insn_size): Declare.
6190 (ix86_issue_rate): Declare.
6191 (ix86_adjust_cost): Declare.
6192 (ia32_multipass_dfa_lookahead): Declare.
6193 (ix86_macro_fusion_p): Declare.
6194 (ix86_macro_fusion_pair_p): Declare.
6195 (ix86_bd_has_dispatch): Declare.
6196 (ix86_bd_do_dispatch): Declare.
6197 (ix86_core2i7_init_hooks): Declare.
6198 (ix86_atom_sched_reorder): Declare.
6199 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6200 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6201 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6202 (rip_relative_addr_p): Rename to ...
6203 (ix86_rip_relative_addr_p): ... this one; export.
6204 (memory_address_length): Update.
6205 (ix86_issue_rate): Move to x86-tune-sched.c.
6206 (ix86_flags_dependent): Move to x86-tune-sched.c.
6207 (ix86_agi_dependent): Move to x86-tune-sched.c.
6208 (exact_dependency_1): Move to x86-tune-sched.c.
6209 (exact_store_load_dependency): Move to x86-tune-sched.c.
6210 (ix86_adjust_cost): Move to x86-tune-sched.c.
6211 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6212 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6213 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6214 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6215 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6216 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6217 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6218 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6219 (min_insn_size): Rename to ...
6220 (ix86_min_insn_size): ... this one; export.
6221 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6222 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6223 (core2i7_first_cycle_multipass_backtrack): Move to
6224 x86-tune-sched-core.c.
6225 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6226 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6227 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6228 (ix86_avoid_jump_mispredicts): Update.
6229 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6230 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6231 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6232 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6233 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6234 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6235 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6236 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6237 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6238 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6239 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6240 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6241 (BIG): Move to ix86-tune-sched-bd.c.
6242 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6243 (enum insn_path): Move to ix86-tune-sched-bd.c.
6244 (get_mem_group): Move to ix86-tune-sched-bd.c.
6245 (is_cmp): Move to ix86-tune-sched-bd.c.
6246 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6247 (is_branch): Move to ix86-tune-sched-bd.c.
6248 (is_prefetch): Move to ix86-tune-sched-bd.c.
6249 (init_window): Move to ix86-tune-sched-bd.c.
6250 (allocate_window): Move to ix86-tune-sched-bd.c.
6251 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6252 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6253 (process_end_window): Move to ix86-tune-sched-bd.c.
6254 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6255 (find_constant): Move to ix86-tune-sched-bd.c.
6256 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6257 (has_immediate): Move to ix86-tune-sched-bd.c.
6258 (get_insn_path): Move to ix86-tune-sched-bd.c.
6259 (get_insn_group): Move to ix86-tune-sched-bd.c.
6260 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6261 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6262 (add_insn_window): Move to ix86-tune-sched-bd.c.
6263 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6264 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6265 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6266 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6267 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6268 (do_dispatch): Move to ix86-tune-sched-bd.c.
6269 (has_dispatch): Move to ix86-tune-sched-bd.c.
6270 * config/i386/t-i386: Add new object files.
6271 * config/i386/x86-tune-costs.h: New file.
6272 * config/i386/x86-tune-sched-atom.c: New file.
6273 * config/i386/x86-tune-sched-bd.c: New file.
6274 * config/i386/x86-tune-sched-core.c: New file.
6275 * config/i386/x86-tune-sched.c: New file.
6277 2017-10-11 Liu Hao <lh_mouse@126.com>
6279 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6280 the generic version below instead.
6281 (should_colorize): Recognize Windows consoles as terminals
6283 * pretty-print.c [__MINGW32__] (write_all): New function.
6284 [__MINGW32__] (find_esc_head): Likewise.
6285 [__MINGW32__] (find_esc_terminator): Likewise.
6286 [__MINGW32__] (eat_esc_sequence): Likewise.
6287 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6289 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6292 2017-10-11 Richard Biener <rguenther@suse.de>
6294 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6295 Properly call analyze_scalar_evolution with the loop of the stmt.
6297 2017-10-11 Richard Biener <rguenther@suse.de>
6299 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6300 * tree-core.h (tree_base): Add chrec_var union member.
6301 * tree.h (CHREC_VAR): Remove.
6302 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6303 * tree-chrec.h (build_polynomial_chrec): Adjust.
6304 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6305 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6307 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6309 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6310 * match.pd: ... here.
6311 ((T) X == (T) Y): Relax condition.
6313 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6315 PR tree-optimization/82472
6316 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6318 (break_alias_scc_partitions): Update postorder number.
6320 2017-10-11 Martin Liska <mliska@suse.cz>
6323 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6325 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6327 2017-10-11 Martin Liska <mliska@suse.cz>
6329 * print-rtl.c (print_insn): Move declaration of idbuf
6330 to same scope as name.
6332 2017-10-11 Martin Liska <mliska@suse.cz>
6337 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6340 2017-10-11 Martin Liska <mliska@suse.cz>
6343 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6346 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6348 * config.gcc (default_gnu_indirect_function): Default to yes for
6349 arm*-*-linux* with glibc.
6351 2017-10-11 Richard Biener <rguenther@suse.de>
6353 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6354 default-defs and types we do not want to analyze.
6355 (interpret_loop_phi): Replace unreachable code with an assert.
6356 (compute_scalar_evolution_in_loop): Remove and inline ...
6357 (analyze_scalar_evolution_1): ... here, replacing condition with
6358 what makes the intent clearer. Remove handling of cases
6359 get_scalar_evolution now handles.
6361 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6363 PR rtl-optimization/81434
6364 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6365 comment for main loop. In sched_group_found if, also add checks for
6366 pass and min_cost_group.
6368 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6370 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6371 (rs6000_insn_cost): New function.
6372 * config/rs6000/rs6000.md (cost): New attribute.
6374 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6375 H.J. Lu <hongjiu.lu@intel.com>
6379 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6380 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6381 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6382 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6383 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6384 OPTION_MASK_ISA_MMX and some other bit set require both
6385 mmx and the ISAs without the mmx bit.
6386 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6387 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6388 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6389 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6390 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6391 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6392 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6393 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6394 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6395 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6396 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6397 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6398 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6399 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6400 Add OPTION_MASK_ISA_MMX.
6402 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6404 * config.gcc (armv7*-*-freebsd*): New target.
6405 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6407 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6409 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6410 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6413 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6415 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6416 a template parameter.
6417 (WIDE_INT_REF_FOR): Update accordingly.
6418 * tree.h (wi::int_traits <const_tree>): Delete.
6419 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6420 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6421 (wi::tree_to_wide_ref): New typedef.
6422 (wi::to_wide): New function.
6423 * calls.c (get_size_range): Use wi::to_wide when operating on
6425 * cgraph.c (cgraph_node::create_thunk): Likewise.
6426 * config/i386/i386.c (ix86_data_alignment): Likewise.
6427 (ix86_local_alignment): Likewise.
6428 * dbxout.c (stabstr_O): Likewise.
6429 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6430 * expr.c (const_vector_from_tree): Likewise.
6431 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6432 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6433 (fold_negate_expr_1, int_const_binop_1, const_binop)
6434 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6435 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6436 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6437 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6438 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6439 (fold_not_const, round_up_loc): Likewise.
6440 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6441 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6442 (alloca_call_type): Likewise.
6443 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6444 * godump.c (go_output_typedef): Likewise.
6445 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6446 * internal-fn.c (get_min_precision): Likewise.
6447 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6448 * ipa-polymorphic-call.c
6449 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6450 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6451 (ipa_modify_call_arguments): Likewise.
6452 * match.pd: Likewise.
6453 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6454 * print-tree.c (print_node_brief, print_node): Likewise.
6455 * stmt.c (expand_case): Likewise.
6456 * stor-layout.c (layout_type): Likewise.
6457 * tree-affine.c (tree_to_aff_combination): Likewise.
6458 * tree-cfg.c (group_case_labels_stmt): Likewise.
6459 * tree-data-ref.c (dr_analyze_indices): Likewise.
6460 (prune_runtime_alias_test_list): Likewise.
6461 * tree-dump.c (dequeue_and_dump): Likewise.
6462 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6463 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6464 * tree-pretty-print.c (dump_generic_node): Likewise.
6465 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6466 (simple_iv_with_niters): Likewise.
6467 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6468 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6469 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6470 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6471 (refine_value_range_using_guard, number_of_iterations_ne_max)
6472 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6473 (get_cst_init_from_scev, record_nonwrapping_iv)
6474 (scev_var_range_cant_overflow): Likewise.
6475 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6476 * tree-ssa-pre.c (compute_avail): Likewise.
6477 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6478 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6479 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6480 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6481 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6482 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6483 (dump_case_nodes, try_switch_expansion): Likewise.
6484 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6485 (vect_do_peeling): Likewise.
6486 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6487 * tree-vect-stmts.c (vectorizable_load): Likewise.
6488 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6489 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6490 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6491 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6492 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6493 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6494 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6495 (evrp_dom_walker::before_dom_children): Likewise.
6496 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6497 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6498 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6499 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6500 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6501 (get_range_pos_neg): Likewise.
6502 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6503 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6504 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6505 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6506 * config/avr/avr.c (avr_fold_builtin): Likewise.
6507 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6508 * config/msp430/msp430.c (msp430_attr): Likewise.
6509 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6510 * config/powerpcspe/powerpcspe-c.c
6511 (altivec_resolve_overloaded_builtin): Likewise.
6512 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6513 (rs6000_expand_ternop_builtin): Likewise.
6514 * config/rs6000/rs6000-c.c
6515 (altivec_resolve_overloaded_builtin): Likewise.
6516 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6517 (rs6000_expand_ternop_builtin): Likewise.
6518 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6520 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6522 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6523 when copying loop nest with only one inner loop.
6525 2017-10-10 Richard Biener <rguenther@suse.de>
6527 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6528 blocks if SCEV is active.
6529 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6531 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6532 (scev_initialize): Assert we are not yet initialized.
6534 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6536 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6537 inner loop's exit stmt by making it always exit the loop, otherwise
6538 we would generate an infinite empty loop.
6540 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6542 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6543 renaming variables in new preheader if it's deleted.
6545 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6547 * tree-loop-distribution.c (struct partition): Remove unused field
6548 loops of the structure.
6549 (partition_alloc, partition_free): Ditto.
6550 (build_rdg_partition_for_vertex): Ditto.
6552 2017-10-09 Jeff Law <law@redhat.com>
6554 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6555 return type to match prototype and documentation.
6557 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6559 * config/rs6000/rs6000.c (processor_costs): Move to ...
6560 * config/rs6000/rs6000.h: ... here.
6561 (rs6000_cost): Declare.
6563 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6565 * except.c (setjmp_fn): New global variable.
6566 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6567 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6568 if DONT_USE_BUILTIN_SETJMP is defined.
6570 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6572 * target.def (insn_cost): New hook.
6573 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6574 * doc/tm.texi: Regenerate.
6575 * rtlanal.c (insn_cost): Use the new hook.
6577 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6579 * combine.c (combine_validate_cost): Compute the new insn_cost,
6580 not just pattern_cost.
6581 (try_combine): Adjust comment.
6583 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6585 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6587 * combine.c (uid_insn_cost): Adjust comment.
6588 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6590 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6591 * dse.c (find_shift_sequence): Ditto.
6592 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6593 (bb_valid_for_noce_process_p): Use pattern_cost.
6594 * rtl.h (insn_rtx_cost): Delete.
6595 (pattern_cost): New prototype.
6596 (insn_cost): New prototype.
6597 * rtlanal.c (insn_rtx_cost): Rename to...
6598 (pattern_cost): ... this.
6601 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6603 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6604 (*jcc<mode>_0_r_i387): Ditto.
6605 (*jccxf_r_i387): Ditto.
6606 (*jcc<mode>_r_i387): Ditto.
6607 (*jccu<mode>_r_i387): Ditto.
6608 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6609 (*jcc): Rename from *jcc_1.
6611 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6613 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6614 deferred rescans after the lvx/stvx recombination pre-pass.
6616 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6618 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6619 memory operation instruction support.
6620 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6621 (rs6000-ibm-aix[789]*): Likewise.
6622 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6623 Document new functions.
6625 2017-10-09 Richard Biener <rguenther@suse.de>
6627 PR tree-optimization/82397
6628 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6629 equality only for semantically equal trees.
6631 2017-10-09 Richard Biener <rguenther@suse.de>
6633 PR tree-optimization/82449
6634 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6635 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6636 allow constant addresses.
6637 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6640 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6642 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6645 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6648 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6651 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6654 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6656 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6659 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6660 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6662 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6664 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6665 (WI_BINARY_PREDICATE_RESULT): Likewise.
6666 (wi::binary_traits::operator_result): New type.
6667 (wi::binary_traits::predicate_result): Likewise.
6668 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6669 (generic_wide_int::operator==, generic_wide_int::operator!=)
6670 (generic_wide_int::operator&, generic_wide_int::and_not)
6671 (generic_wide_int::operator|, generic_wide_int::or_not)
6672 (generic_wide_int::operator^, generic_wide_int::operator+
6673 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6675 (operator~, unary operator-, operator==, operator!=, operator&)
6676 (operator|, operator^, operator+, binary operator-, operator*): New
6678 * expr.c (get_inner_reference): Use wi::bit_and_not.
6679 * fold-const.c (fold_binary_loc): Likewise.
6680 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6681 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6682 (bit_value_binop): Likewise.
6683 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6684 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6685 (extract_range_from_binary_expr_1): Likewise.
6686 (masked_increment): Likewise.
6687 (simplify_bit_ops_using_ranges): Likewise.
6689 2017-10-09 Martin Jambor <mjambor@suse.cz>
6692 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6693 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6694 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6696 (hsa_fixup_mov_insn_type): New function.
6697 (hsa_op_with_type::get_in_type): Use it.
6698 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6699 immediates in an assert.
6700 (hsa_op_with_type::extend_int_to_32bit): New method.
6701 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6702 types. Convert to dest type if necessary.
6703 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6704 (reg_for_gimple_ssa): Pass false as min32int to
6705 hsa_type_for_scalar_tree_type.
6706 (gen_hsa_addr): Fixup type when creating addresable temporary.
6707 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6708 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6709 necessary. Call hsa_fixup_mov_insn_type.
6710 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6711 extend operands and convert to dest type if necessary.
6712 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6713 to dest type if necessary.
6714 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6715 if conversion nt necessary and size matches.
6716 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6717 to dest type if necessary.
6718 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6719 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6721 (gen_hsa_clrsb): Likewise.
6722 (gen_hsa_ffs): Likewise.
6723 (gen_hsa_divmod): Extend operands and convert to dest type if
6725 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6727 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6729 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6730 Remove empty default arguments. Use a brace block as output
6732 (conditional return): Ditto.
6734 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6735 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6737 (group_ending_nop): Ditto.
6738 (doloop_end): Ditto.
6739 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6740 (splitters for those): Ditto.
6742 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6744 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6745 a conditional jump (and the compare for it) so that pc_rtx is the
6747 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6748 for the deleted and renamed ctr<mode>_internal[234] patterns.
6749 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6750 Delete second conditional return pattern.
6751 (ctr<mode>_internal2): Delete this second bdnz pattern.
6752 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6753 (ctr<mode>_internal4): Delete this second bdz pattern.
6755 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6757 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6758 (always_initialized_rtx_for_ssa_name_p): New predicate.
6759 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6760 (finish_out_of_ssa): Free new field of SA.
6761 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6762 * tree-ssa-coalesce.c: Include tree-ssa.h.
6763 (get_parm_default_def_partitions): Remove extern keyword.
6764 (get_undefined_value_partitions): New function.
6765 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6766 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6768 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6770 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6772 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6774 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6776 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6777 for targets that preffer 128bit.
6779 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6781 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6783 2017-10-08 Olivier Hainque <hainque@adacore.com>
6785 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6786 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6788 (thumb_set_return_address): Likewise.
6790 2017-10-08 Olivier Hainque <hainque@adacore.com>
6792 * common/config/arm/arm-common.c (arm_except_unwind_info):
6793 Handle DWARF2_UNWIND_INFO.
6795 2017-10-07 Michael Collison <michael.collison@arm.com>
6797 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6800 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6802 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6803 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6804 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6805 defined, force the creation of a new block for a dispatch label.
6807 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6809 * invoke.texi (Wsuggest-attribute=cold): Document.
6810 * common.opt (Wsuggest-attribute=cold): New
6811 * ipa-pure-const.c (warn_function_cold): New function.
6812 * predict.c (compute_function_frequency): Use it.
6813 * predict.h (warn_function_cold): Declare.
6815 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6817 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6820 2017-10-06 Martin Liska <mliska@suse.cz>
6822 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6823 keyword for member functions.
6824 (struct sanopt_tree_couple): New struct.
6825 (struct sanopt_tree_couple_hash): New function.
6826 (struct sanopt_ctx): Add new hash_map.
6827 (has_dominating_ubsan_ptr_check): New function.
6828 (record_ubsan_ptr_check_stmt): Likewise.
6829 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6830 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6831 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6833 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6836 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6837 aarch64_simd_valid_immediate on CONST_VECTORs.
6838 (aarch64_reg_or_bic_imm): Likewise.
6840 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6842 PR rtl-optimization/82396
6843 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6845 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6847 * graphite-dependences.c (scop_get_reads): Move code to...
6848 (scop_get_must_writes): Move code to...
6849 (scop_get_may_writes): Move code to...
6850 (scop_get_reads_and_writes): ... here.
6851 (scop_get_dependences): Call scop_get_reads_and_writes.
6853 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6855 PR tree-optimization/82434
6856 * fold-const.h (can_native_encode_type_p,
6857 can_native_encode_string_p): Remove.
6858 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6859 don't encode anything, just return what would be otherwise returned.
6860 (native_encode_fixed, native_encode_complex, native_encode_vector):
6862 (native_encode_string): Likewise. Inline by hand
6863 can_native_encode_string_p.
6864 (can_native_encode_type_p): Remove.
6865 (can_native_encode_string_p): Remove.
6866 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6867 STRING_CSTs using can_native_encode_string_p, test all
6868 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6869 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6870 argument from native_encode_expr.
6871 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6872 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6875 2017-10-06 Richard Biener <rguenther@suse.de>
6877 PR tree-optimization/82397
6878 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6879 operand_equal_p but rely on data_ref_compare_tree for detecting
6881 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6882 to match up with dr_group_sort_cmp.
6884 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6887 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6889 * config/s390/s390-builtin-types.def: Regenerate.
6891 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6894 * config/s390/s390-builtin-types.def: Regenerate.
6895 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6896 Change flag from B_VXE to B_VX.
6897 (s390_vec_min_dbl): Remove B_VXE flag.
6899 2017-10-06 Richard Biener <rguenther@suse.de>
6901 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6902 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6903 translate_isl_ast_to_gimple::is_valid_rename,
6904 translate_isl_ast_to_gimple::get_rename,
6905 translate_isl_ast_to_gimple::get_def_bb_for_const,
6906 translate_isl_ast_to_gimple::get_new_name,
6907 translate_isl_ast_to_gimple::collect_all_ssa_names,
6908 translate_isl_ast_to_gimple::copy_loop_phi_args,
6909 translate_isl_ast_to_gimple::collect_all_ssa_names,
6910 translate_isl_ast_to_gimple::copy_loop_phi_args,
6911 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6912 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6913 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6914 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6915 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6916 translate_isl_ast_to_gimple::copy_cond_phi_args,
6917 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6918 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6919 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6920 translate_isl_ast_to_gimple::rename_uses,
6921 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6922 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6923 (set_rename_for_each_def): Likewise.
6924 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6925 here. Handle rewriting SCEV analyzable uses here.
6926 (copy_bb_and_scalar_dependences): Generate code for PHI
6928 (graphite_regenerate_ast_isl): Adjust.
6929 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6930 (add_write, add_read): New functions.
6931 (build_cross_bb_scalars_def): Use it and simplify.
6932 (build_cross_bb_scalars_use): Likewise.
6933 (graphite_find_cross_bb_scalar_vars): Inline into...
6934 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6935 simulating out-of-SSA. Compute liveout and add dependencies.
6936 (build_scops): Force an empty entry block.
6937 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6939 (sese_build_liveouts): Declare.
6940 (sese_trivially_empty_bb_p): Likewise.
6941 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6942 compute liveout and debug_liveout.
6943 (sese_bad_liveouts_use): Remove.
6944 (sese_reset_debug_liveouts_bb): Likewise.
6945 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6946 (sese_build_liveouts): Build liveout and debug_liveout and store
6948 (new_sese_info): Adjust.
6949 (free_sese_info): Likewise.
6950 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6951 do not build liveout here.
6952 (move_sese_in_condition): Adjust region entry.
6953 (scev_analyzable_p): Match up with chrec_apply requirements.
6954 (sese_trivially_empty_bb_p): New.
6955 * tree-into-ssa.c (get_reaching_def): Properly support generating
6956 default-defs for incremental rewrite of anonymous names.
6958 2017-10-06 Richard Biener <rguenther@suse.de>
6960 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6961 precision do not perform modulo reduction.
6963 2017-10-06 Richard Biener <rguenther@suse.de>
6965 PR tree-optimization/82436
6966 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6967 conservatively choose the vectorization factor when checking
6968 whether we can perform the required load permutation.
6969 (vect_transform_slp_perm_load): Assert when we may not fail.
6971 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6973 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6974 message for incompatible -msdata=* and -mcall-* options.
6976 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6978 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6979 rate for post-reload scheduling.
6981 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6983 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6985 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6987 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6988 to improve monte carlo in scimark.
6990 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6992 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6993 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6994 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6995 pentium4_cost, nocona_cost): Set reassociation width to 1.
6996 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6997 width to 2 for fp operations and 1 otherwise.
6998 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7000 (atom_cost): Set reassociation width to 2.
7001 (slm_cost, generic_cost): Set fp reassociation width
7002 to 2 and 1 otherwise.
7003 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7004 (core_cost): Set fp reassociation width to 4 and vector to 2.
7005 (ix86_reassociation_width): Rewrite using cost table; special case
7006 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7007 and TARGET_AVX128_OPTIMAL.
7008 * config/i386/i386.h (processor_costs): Add
7009 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7010 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7011 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7012 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7013 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7014 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7016 2017-10-05 Nathan Sidwell <nathan@acm.org>
7018 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7020 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7022 * config/arm/arm.c (arm_test_fpu_data): New.
7023 (arm_run_selftests): Call arm_test_fpu_data.
7025 2017-10-04 Nathan Sidwell <nathan@acm.org>
7027 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7029 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7031 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7033 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7034 check type for aarch64_simd_valid_immediate.
7035 (aarch64_output_simd_mov_immediate): Update prototype.
7036 (aarch64_simd_valid_immediate): Update prototype.
7037 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7038 support for ORR-immediate.
7039 (and<mode>3): modified pattern to add support for BIC-immediate.
7040 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7041 now checks for valid immediate for BIC and ORR based on new enum
7043 (aarch64_output_simd_mov_immediate): Function now used to output
7044 BIC/ORR imm as well based on new enum argument.
7045 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7047 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7048 (aarch64_reg_or_bic_imm): Likewise.
7050 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7052 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7053 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7055 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7059 PR rtl-optimization/82396
7060 * haifa-sched.c (autopref_multipass_init): Simplify
7062 (autopref_rank_data): Simplify sort order.
7063 * sched-int.h (autopref_multipass_data_): Remove
7064 multi_mem_insn_p, min_offset and max_offset.
7066 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7068 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7070 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7072 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7075 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7077 * doc/sourcebuild.texi: Document vect_long_mult.
7079 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7081 PR tree-optimization/82413
7082 * fold-const.c (build_range_check): Use widest_int when comparing
7083 the maximum ETYPE value with HIGH.
7085 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7087 PR rtl-optimization/82396
7088 * haifa-sched.c (autopref_multipass_init): Simplify
7090 (autopref_rank_data): Simplify sort order.
7091 * sched-int.h (autopref_multipass_data_): Remove
7092 multi_mem_insn_p, min_offset and max_offset.
7094 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7096 PR tree-optimization/82381
7097 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7098 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7101 PR tree-optimization/82374
7102 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7103 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7104 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7105 current_function_decl to the new decl.
7107 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7109 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7110 helper macro for IEEE float128 hardware built-in functions.
7111 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7113 (TRUNCF128_ODD): Likewise.
7114 (ADDF128_ODD): Likewise.
7115 (SUBF128_ODD): Likewise.
7116 (MULF128_ODD): Likewise.
7117 (DIVF128_ODD): Likewise.
7118 (FMAF128_ODD): Likewise.
7119 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7120 UNSPEC_TRUNC_ROUND_TO_ODD.
7121 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7122 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7123 floating point round to odd instructions.
7124 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7125 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7126 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7127 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7128 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7129 (trunc<mode>sf2_hw): Change the truncate with round to odd
7130 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7131 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7132 to odd hardware instructions.
7133 (sub<mode>3_odd): Likewise.
7134 (mul<mode>3_odd): Likewise.
7135 (div<mode>3_odd): Likewise.
7136 (sqrt<mode>2_odd): Likewise.
7137 (fma<mode>4_odd): Likewise.
7138 (fms<mode>4_odd): Likewise.
7139 (nfma<mode>4_odd): Likewise.
7140 (nfms<mode>4_odd): Likewise.
7141 (trunc<mode>df2_odd): Change the truncate with round to odd
7142 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7144 * doc/extend.texi (PowerPC built-in functions): Update documentation
7145 for existing IEEE float128-bit built-in functions. Add built-in
7146 functions that generate the IEEE 128-bit floating point round to
7149 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7151 PR rtl-optimization/77729
7152 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7153 to (X&(C1&~C2))|C2 transformations.
7155 2017-10-03 Martin Jambor <mjambor@suse.cz>
7157 PR tree-optimization/82363
7158 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7159 mismatch, mark lacc written regardless of racc.
7161 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7163 PR tree-optimization/82381
7164 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7165 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7166 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7167 fallthrough into reassoc_stmt_dominates_stmt_p.
7170 * combine.c (combine_instructions): Don't combine in unreachable
7173 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7176 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7177 function to not use the have_cpu variable. Do not set cpu_index,
7178 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7180 (rs6000_valid_attribute_p): Remove duplicate initializations of
7181 old_optimize and func_optimize.
7182 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7183 (rs6000_activate_target_options): Make global.
7184 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7187 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7189 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7190 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7191 Set *poffset to 0 and *psize and *pmax_size to -1 if
7192 *poffset + *psize overflows in HOST_WIDE_INT.
7194 PR tree-optimization/82387
7195 PR tree-optimization/82388
7196 PR tree-optimization/82389
7197 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7198 instead of live_bytes non-NULL.
7200 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7203 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7206 2017-10-02 Richard Biener <rguenther@suse.de>
7208 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7209 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7210 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7212 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7214 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7215 requested precision matches the type's.
7216 * calls.c (alloc_max_size): Calculate the new candidate size as
7217 a widest_int and use wi::to_widest when comparing it with the
7218 current candidate size.
7219 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7220 zero rather than integer_zero_node.
7221 * match.pd: Check for a no-op conversion before using wi::add
7222 rather than after. Use tree_to_uhwi when summing small shift
7223 counts into an unsigned int.
7225 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7226 Alan Hayward <alan.hayward@arm.com>
7227 David Sherwood <david.sherwood@arm.com>
7230 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7231 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7232 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7233 POINTER_AND_FP_REGS.
7235 2017-10-02 Richard Biener <rguenther@suse.de>
7237 PR tree-optimization/82355
7238 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7239 a mapping for the enclosing loop but avoid generating one for
7241 (copy_bb_and_scalar_dependences): Remove premature codegen
7242 error on PHIs in blocks duplicated into multiple places.
7243 * graphite-scop-detection.c
7244 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7245 in the region use it as loop and nest to analyze the DR in.
7246 (try_generate_gimple_bb): Likewise.
7247 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7248 (add_loop_constraints): For blocks in a loop not in the region
7249 create a dimension with a single iteration.
7250 * sese.h (gbb_loop_at_index): Remove assert.
7252 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7254 * omp-expand.c (adjust_context_scope): New function.
7255 (expand_parallel_call): Call adjust_context_scope.
7257 2017-10-01 Jeff Law <law@redhat.com>
7259 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7260 dom_opt_dom_walker class with direct access to private members.
7261 Add comments. Call test_for_singularity.
7262 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7263 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7264 m_dummy_cond anymore.
7265 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7267 (pass_dominator:execute): Build the dummy_cond here and pass it
7268 to the dom_opt_dom_walker ctor.
7269 (test_for_singularity): New function.
7271 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7272 Maya Rashish <coypu@sdf.org>
7274 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7275 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7276 (alpha*-*-netbsd*) Use nbsd_tm_file.
7277 (arm*-*-netbsdelf*) Likewise.
7278 (i[34567]86-*-netbsdelf*) Likewise.
7279 (x86_64-*-netbsd*) Likewise.
7280 (mips*-*-netbsd*) Likewise.
7281 (powerpc-*-netbsd*) Likewise.
7282 (sh*-*-netbsd*) Likewise.
7283 (sparc-*-netbsdelf*) Likewise.
7284 (sparc64-*-netbsd*) Likewise.
7285 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7287 (vax-*-netbsdelf*) Likewise.
7288 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7289 (UINT_FAST8_TYPE) Likewise.
7290 (INT_FAST16_TYPE) Check CHAR_FAST16.
7291 (UINT_FAST16_TYPE) Likewise.
7293 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7296 * config/i386/i386.md
7297 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7298 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7299 *divmodsi4_zext_2): New define_insn_and_split.
7300 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7301 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7302 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7303 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7304 New define_insn_and_split.
7305 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7306 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7307 operands[1] having DImode when mode is SImode.
7309 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7310 always SImode for DIV and MOD in REG_EQUAL notes.
7312 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7315 * match.pd: Fix handling of NaNs in pattern.
7317 2017-09-29 Jeff Law <law@redhat.com>
7319 * sbitmap.c (bitmap_bit_in_range_p): New function.
7320 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7321 * tree-ssa-dse.c (live_bytes_read): New function.
7322 (dse_classify_store): Ignore reads of dead bytes.
7324 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7325 typos and whitespace errors.
7326 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7327 * config/s390/s390.c (s390_emit_prologue): Likewise.
7329 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7332 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7333 with a symbol for LRA.
7335 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7337 PR rtl-optimization/82338
7338 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7340 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7342 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7343 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7344 (qsort_chk): Declare.
7345 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7346 (qsort_chk): New function.
7348 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7350 PR tree-optimization/82337
7351 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7352 phi definition if the PHI result appears in an abnormal PHI.
7353 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7354 basis appears in an abnormal PHI.
7356 2017-09-29 Richard Biener <rguenther@suse.de>
7358 * graphite-isl-ast-to-gimple.c
7359 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7360 (binary_op_to_tree): Use it.
7361 (get_rename_from_scev): Likewise.
7362 (copy_loop_phi_nodes): Likewise.
7363 (copy_bb_and_scalar_dependences): Likewise.
7364 (translate_pending_phi_nodes): Likewise.
7366 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7369 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7370 for movabsq $(i32 << shift), r64.
7372 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7374 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7375 index when encoding %esp as %rsp to avoid 0x67 prefix.
7377 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7379 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7380 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7382 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7384 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7385 Extensions with more than 16 double VFP registers.
7386 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7387 to_clear_mask and all code related to it. Replace the remaining
7388 entry by a sbitmap and adapt code accordingly.
7390 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7392 * brig-builtins.def: Change pure attributes to const.
7394 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7396 * config.gcc (default_gnu_indirect_function): Default to yes for
7397 sparc*-*-linux* with glibc.
7399 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7401 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7402 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7403 when creating .init_array and .fini_array sections with priority
7406 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7409 * config/aarch64/aarch64.c
7410 (aarch64_builtin_support_vector_misalignment): Always return false
7411 when misalignment is unknown.
7413 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7415 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7416 this function to return false if the definition used by the swap
7417 instruction is artificial, or if the memory address from which the
7418 constant value is loaded is not represented by a base address held
7419 in a register or if the base address register is a frame or stack
7420 pointer. Additionally, return false if the base address of the
7421 loaded constant is a SYMBOL_REF but is not considered to be a
7423 (replace_swapped_load_constant): New function.
7424 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7425 loaded constant vector with a load of a swapped constant vector.
7427 2017-09-27 Carl Love <cel@us.ibm.com>
7429 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7430 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7431 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7434 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7436 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7437 first, always call autopref_rank_data otherwise.
7439 2017-09-27 Richard Biener <rguenther@suse.de>
7441 * graphite-scop-detection.c (find_scop_parameters): Move
7442 loop bound handling ...
7443 (gather_bbs::before_dom_children): ... here, avoiding the need
7444 to build scop_info->loop_nest.
7445 (record_loop_in_sese): Remove.
7446 * sese.h (sese_info_t::loop_nest): Remove.
7447 * sese.c (new_sese_info): Do not allocate loop_nest.
7448 (free_sese_info): Do not free loop_nest.
7450 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7453 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7454 lhs from calls if the lhs has addressable type.
7456 2017-09-27 Richard Biener <rguenther@suse.de>
7458 * graphite.h (scop::max_alias_set): New member.
7459 * graphite-scop-detection.c: Remove references to non-existing
7460 --param in comments.
7461 (build_alias_sets): Record the maximum alias set used for drs.
7462 (build_scops): Support zero as unlimited for
7463 --param graphite-max-arrays-per-scop.
7464 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7466 (build_poly_sr_1): ... here. Compute alias set based on the
7467 maximum alias set used for drs rather than
7468 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7470 2017-09-27 Richard Biener <rguenther@suse.de>
7472 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7473 --param loop-block-tile-size=0 to disable tiling.
7475 2017-09-27 Richard Biener <rguenther@suse.de>
7477 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7478 (graphite-max-nb-scop-params): Document special value zero.
7479 * domwalk.h (dom_walker::STOP): New symbolical constant.
7480 (dom_walker::dom_walker): Add optional parameter for bb to
7482 (dom_walker::~dom_walker): Declare.
7483 (dom_walker::before_dom_children): Document STOP return value.
7484 (dom_walker::m_user_bb_to_rpo): New member.
7485 (dom_walker::m_bb_to_rpo): Likewise.
7486 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7487 mapping here if not provided by the user.
7488 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7489 provided by the user.
7490 (dom_walker::STOP): Define.
7491 (dom_walker::walk): Do not compute bb to RPO mapping here.
7492 Support STOP return value from before_dom_children to stop
7494 * graphite-optimize-isl.c (optimize_isl): If the schedule
7495 is the same still generate code if -fgraphite-identity
7496 or -floop-parallelize-all are given.
7497 * graphite-scop-detection.c: Include cfganal.h.
7498 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7500 (gather_bbs::before_dom_children): Return STOP for BBs
7502 (build_scops): Compute bb to RPO mapping and pass it to
7503 the domwalk. Treat --param graphite-max-nb-scop-params=0
7504 as not limiting the number of params.
7505 * graphite.c (graphite_initialize): Remove limit on the
7506 number of basic-blocks in a function.
7507 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7508 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7509 default value of 10.
7511 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7513 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7514 Adjust code to eliminate needing to do the shift right 32-bits
7515 operation after XSCVDPSPN.
7517 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7519 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7520 ((X / Y) != 0 -> X >= Y): Likewise.
7522 2017-09-26 Carl Love <cel@us.ibm.com>
7524 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7525 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7526 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7527 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7528 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7529 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7530 definitions and overloading.
7531 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7532 statement for P9V_BUILTIN_XST_LEN_R.
7533 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7534 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7535 define_expand and define_insn for the instructions and builtins.
7536 * doc/extend.texi: Update the built-in documentation file for the new
7538 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7539 define_insn for the instructions
7541 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7544 * gcc/config/netbsd-protos.h: New file.
7545 * gcc/config/netbsd.c: New file.
7546 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7547 * gcc/config/t-netbsd: New file.
7548 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7549 (tmake_file) Add t-netbsd.
7550 (extra_objs) Add netbsd.o.
7552 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7556 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7558 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7560 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7561 sign extension from a vector register to a GPR by doing a 32-bit
7562 direct move and then an EXTSW.
7563 (extendsi<mode>2 splitter): Likewise.
7564 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7565 right or vector extract after doing XSCVDPSPN. Use
7566 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7568 (movdi_from_sf_zero_ext): Likewise.
7569 (reload_gpr_from_vsxsf): Likewise.
7570 (p8_mfvsrd_4_disf): Delete, no longer used.
7571 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7572 then needing to move the SFmode to a GPR to use the XSCVDPSP
7573 instruction instead of FRSP and XSCVDPSPN.
7574 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7575 it is adjacent to the other XSCVSPDP insns.
7576 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7577 SFmode to be in traditional Altivec registers.
7578 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7579 (vsx_xscvspdpn): Likewise.
7580 (vsx_xscvspdpn_scalar): Likewise.
7582 2017-09-26 Martin Jambor <mjambor@suse.cz>
7584 * tree-sra.c (compare_access_positions): Put integral types first,
7585 stabilize sorting of integral types, remove conditions putting
7586 non-full-precision integers last.
7587 (sort_and_splice_var_accesses): Disable scalarization if a
7588 non-integert would be represented by a non-full-precision integer.
7590 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7592 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7593 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7594 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7595 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7596 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7597 conditionals inside the function instead of around it. Call
7598 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7599 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7601 2017-09-26 Richard Biener <rguenther@suse.de>
7603 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7605 (scop_detection::build_scop_breadth): ... this. Removed.
7606 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7607 (scop_detection::harmful_stmt_in_bb): Likewise.
7608 (scop_detection::graphite_can_represent_stmt): Likewise.
7609 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7610 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7611 (scop_detection::can_represent_loop_1): ... this. Removed.
7612 (scop_detection::harmful_loop_in_region): Simplify after inlining
7613 the above and remove more quadraticness.
7614 (build_scops): Adjust.
7615 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7618 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7621 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7622 REGNO (base) == SP_REG if base is a REG.
7625 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7626 if it is different SSA_NAME.
7627 (optimize_range_tests_cmp_bitwise): New function.
7628 (optimize_range_tests): Call it.
7630 2017-09-26 Richard Biener <rguenther@suse.de>
7632 PR tree-optimization/82321
7633 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7634 for the def being inside the loop.
7636 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7638 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7640 * config/s390/s390-builtins.def: Fix constraint on op4.
7642 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7644 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7645 independent expanders.
7646 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7647 ("vec_ordered", "vec_unordered"): New expanders.
7649 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7651 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7654 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7656 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7657 vec_unpacks_lo_v16qi.
7658 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7660 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7662 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7663 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7664 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7666 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7668 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7670 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7672 ("*vec_slb<mode>"): New insn pattern.
7673 ("vec_shr_<mode>"): New expander.
7674 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7675 and force the shift count operand to V16QImode.
7676 ("vec_srb<mode>"): Set shift count mode to V16QI.
7678 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7680 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7681 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7682 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7684 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7687 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7689 2017-09-26 Richard Biener <rguenther@suse.de>
7691 PR tree-optimization/82320
7692 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7695 2017-09-25 Jeff Law <law@redhat.com>
7697 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7698 prototype for new argument.
7699 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7700 mostly extracted from rs6000_emit_allocate_stack.
7701 (rs6000_emit_probe_stack_range_stack_clash): New function.
7702 (rs6000_emit_allocate_stack): Call
7703 rs6000_emit_probe_stack_range_stack_clash as needed.
7704 (rs6000_emit_probe_stack_range): Add additional argument
7705 to call to gen_probe_stack_range{si,di}.
7706 (output_probe_stack_range): New.
7707 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7708 (output_probe_stack_range_stack_clash): New.
7709 (rs6000_emit_prologue): Emit notes into dump file as requested.
7710 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7711 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7712 Add additional operand and pass it to output_probe_stack_range.
7714 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7716 PR tree-optimization/82163
7717 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7718 (checking_verify_loop_closed_ssa): New parameter.
7719 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7720 (check_loop_closed_ssa_stmt): Delete.
7721 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7722 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7723 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7726 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7728 * brig-builtins.def: Treat HSAIL barrier builtins as
7729 setjmp/longjump style functions.
7731 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7733 * target.def (constant_alignment): New hook.
7734 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7735 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7736 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7737 * doc/tm.texi: Regenerate.
7738 * targhooks.h (default_constant_alignment): Declare.
7739 (constant_alignment_word_strings): Likewise.
7740 * targhooks.c (default_constant_alignment): New function.
7741 (constant_alignment_word_strings): Likewise.
7742 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7743 instead of CONSTANT_ALIGNMENT.
7744 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7745 (force_const_mem): Likewise.
7746 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7747 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7748 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7749 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7750 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7752 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7753 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7754 constant_alignment_word_strings.
7755 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7756 (CONSTANT_ALIGNMENT): Likewise.
7757 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7758 (arm_constant_alignment): New function.
7759 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7760 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7761 constant_alignment_word_strings.
7762 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7763 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7764 constant_alignment_word_strings.
7765 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7766 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7767 (cris_constant_alignment): New function.
7768 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7769 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7770 (epiphany_constant_alignment): New function.
7771 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7772 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7773 constant_alignment_word_strings.
7774 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7775 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7776 constant_alignment_word_strings.
7777 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7778 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7779 constant_alignment_word_strings.
7780 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7781 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7782 * config/i386/i386.c (ix86_constant_alignment): Make static.
7783 Use the same interface as the target hook.
7784 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7785 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7786 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7787 constant_alignment_word_strings.
7788 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7789 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7790 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7791 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7792 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7793 constant_alignment_word_strings.
7794 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7795 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7796 constant_alignment_word_strings.
7797 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7798 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7799 constant_alignment_word_strings.
7800 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7801 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7803 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7804 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7805 * config/mips/mips.c (mips_constant_alignment): New function.
7806 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7807 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7808 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7809 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7810 (mmix_constant_alignment): Make static. Use the same interface
7812 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7813 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7814 constant_alignment_word_strings.
7815 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7816 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7817 constant_alignment_word_strings.
7818 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7819 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7820 constant_alignment_word_strings.
7821 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7822 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7823 (rs6000_constant_alignment): New function.
7824 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7825 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7826 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7827 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7828 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7829 (rs6000_constant_alignment): New function.
7830 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7831 * config/s390/s390.c (s390_constant_alignment): New function.
7832 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7833 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7834 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7835 constant_alignment_word_strings.
7836 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7837 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7838 (sparc_constant_alignment): New function.
7839 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7840 * config/spu/spu.c (spu_constant_alignment): New function.
7841 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7842 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7843 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7844 constant_alignment_word_strings.
7845 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7846 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7847 constant_alignment_word_strings.
7848 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7849 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7850 constant_alignment_word_strings.
7851 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7852 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7853 (visium_constant_alignment): New function.
7854 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7855 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7856 (xtensa_constant_alignment): New function.
7857 * system.h (CONSTANT_ALIGNMENT): Poison.
7859 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7861 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7862 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7863 (rs6000_builtin_valid_without_lhs): New helper function.
7864 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7865 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7867 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7869 * target.h (vec_perm_indices): Use unsigned short rather than
7871 (auto_vec_perm_indices): Likewise.
7872 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7873 Use unsigned int rather than unsigned char.
7874 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7876 2017-09-25 Richard Biener <rguenther@suse.de>
7878 * cfgloop.h (sort_sibling_loops): Declare.
7879 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7880 (sort_sibling_loops): New function sorting the sibling loop list
7882 * graphite.c (graphite_transform_loops): Sort sibling loops.
7884 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7886 * target.def (vec_perm_const_ok): Change sel parameter to
7888 * optabs-query.c (can_vec_perm_p): Update accordingly.
7889 * doc/tm.texi: Regenerate.
7890 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7891 auto_vec_perm_indices and remove separate nelt field.
7892 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7893 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7894 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7895 (aarch64_expand_vec_perm_const): Update accordingly.
7896 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7897 to vec_perm_indices.
7898 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7899 auto_vec_perm_indices and remove separate nelt field.
7900 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7901 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7902 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7904 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7905 to vec_perm_indices.
7906 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7907 sel to vec_perm_indices.
7908 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7909 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7910 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7912 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7915 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7918 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7919 on the FUNCTION_DECL function context if it has a DIE that is a
7922 2017-09-25 Richard Biener <rguenther@suse.de>
7924 PR tree-optimization/82285
7925 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7928 2017-09-25 Tom de Vries <tom@codesourcery.com>
7932 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7935 2017-09-25 Richard Biener <rguenther@suse.de>
7937 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7938 ISL errors other than isl_error_quota happen. Dump if the
7939 schedule is the same.
7940 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7941 errors instead of aborting inside ISL.
7943 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7946 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7947 of libgcc_eh for m64.
7948 * config/i386/darwin64.h: Likewise.
7950 2017-09-25 Richard Biener <rguenther@suse.de>
7953 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7954 attribute for incomplete types nor twice for complete ones.
7956 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7959 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7960 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7963 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7966 * config/i386/i386.opt (mprefer-avx256): Use
7967 ix86_target_flags variable.
7968 * config/i386/i386.c (ix86_target_string): Move
7969 -mprefer-avx256 to flag2_opts.
7971 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7974 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7975 and x != -1 | y != -1 into (x & y) != -1.
7977 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7979 * config.gcc: Add new case statement to set
7980 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7981 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7982 s390x-*-linux* case statements. Added aarch64 to the list of
7983 supported architectures.
7985 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7987 PR tree-optimization/82289
7988 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7989 STMT_VINFO_RELEVANT_P.
7991 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7992 Alan Hayward <alan.hayward@arm.com>
7993 David Sherwood <david.sherwood@arm.com>
7995 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7996 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7997 (extract_range_from_binary_expr_1): Don't call
7998 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8000 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8001 Alan Hayward <alan.hayward@arm.com>
8002 David Sherwood <david.sherwood@arm.com>
8004 * target.def (preferred_vector_alignment): New hook.
8005 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8007 * doc/tm.texi: Regenerate.
8008 * targhooks.h (default_preferred_vector_alignment): Declare.
8009 * targhooks.c (default_preferred_vector_alignment): New function.
8010 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8012 (DR_TARGET_ALIGNMENT): New macro.
8013 (aligned_access_p): Update commentary.
8014 (vect_known_alignment_in_bytes): New function.
8015 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8017 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8018 Calculate the misalignment based on the target alignment rather than
8020 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8021 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8022 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8023 the target alignment, rather than masking the element misalignment
8024 with the number of elements in a vector. Also use the target
8025 alignment when calculating the maximum number of peels.
8026 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8027 instead of TYPE_ALIGN_UNIT.
8028 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8029 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8030 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8031 (vect_create_data_ref_ptr): Likewise.
8032 (vect_setup_realignment): Realign by ANDing with
8033 -DR_TARGET_MISALIGNMENT.
8034 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8035 the number of peels based on DR_TARGET_ALIGNMENT.
8036 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8037 with the guaranteed alignment boundary when deciding whether
8039 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8040 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8041 (ensure_base_align): Remove stmt_info parameter. Get the
8042 target base alignment from DR_TARGET_ALIGNMENT.
8043 (vectorizable_store): Update call accordingly. Interpret
8044 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8046 (vectorizable_load): Likewise.
8048 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8049 Alan Hayward <alan.hayward@arm.com>
8050 David Sherwood <david.sherwood@arm.com>
8052 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8053 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8054 (vect_enhance_data_refs_alignment): Likewise.
8056 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8058 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8059 error. Only quit immediately if parsing is complete.
8060 (BEGIN): Initialize fatal_err and parse_done.
8061 (begin fpu, end fpu): Check number of arguments.
8062 (begin arch, end arch): Likewise.
8063 (begin cpu, end cpu): Likewise.
8064 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8065 (optalias): Likewise.
8067 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8069 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8070 * config/arm/arm-isa.h: Delete. Move definitions to ...
8071 * arm-cpus.in: ... here. Use new feature and fgroup values.
8072 * config/arm/arm.c (arm_option_override): Use lower case for feature
8074 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8075 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8076 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8078 (print_isa_bits_for): New function.
8079 (gen_isa): New function.
8080 (gen_comm_data): Use print_isa_bits_for.
8081 (define feature): New keyword.
8082 (define fgroup): New keyword.
8083 * config/arm/t-arm (TM_H): Remove.
8084 (GTM_H): Add arm-isa.h.
8085 (arm-isa.h): Add rule to generate file.
8086 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8087 case for feature bit names.
8089 2017-09-22 Richard Biener <rguenther@suse.de>
8091 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8093 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8094 print of no dependency loops ...
8095 * graphite.c (graphite_transform_loops): ... here.
8096 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8098 (same_close_phi_node, remove_duplicate_close_phi,
8099 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8100 (canonicalize_loop_closed_ssa): ... here and simplify.
8101 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8102 (optimize_isl): Use dump_printf_loc to tell when we stopped
8103 optimizing because of an ISL timeout.
8105 2017-09-22 Richard Biener <rguenther@suse.de>
8107 PR tree-optimization/82291
8108 * tree-if-conv.c (predicate_mem_writes): Make sure to
8109 remove writes in blocks predicated with false.
8111 2017-09-22 Richard Biener <rguenther@suse.de>
8113 * sese.c: Include cfganal.h.
8114 (if_region_set_false_region): Remove.
8115 (create_if_region_on_edge): Likewise.
8116 (move_sese_in_condition): Re-implement without destroying
8119 2017-09-22 Richard Biener <rguenther@suse.de>
8121 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8122 Verify both BBs contain loop PHI nodes before dispatching to
8124 (graphite_regenerate_ast_isl): Do not recompute dominators,
8125 do not verify three times. Restructure for clarity.
8126 * graphite-scop-detection.c (same_close_phi_node,
8127 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8128 defined_in_loop_p, canonicalize_loop_closed_ssa,
8129 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8130 checking and SSA rewrite, move to ...
8131 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8132 (graphite_initialize): Do not pass in ctx, do not reset the
8133 SCEV cache, compute only dominators.
8134 (graphite_transform_loops): Allocate ISL ctx after
8135 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8136 Maintain post-dominators only around build_scops.
8137 * sese.c (if_region_set_false_region): Make static. Free
8138 and recompute dominators.
8139 (move_sese_in_condition): Assert we don't get called with
8140 post-dominators computed.
8141 * sese.h (if_region_set_false_region): Remove.
8143 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8145 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8146 mode attribute for TARGET_AVX512VL.
8148 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8150 * config/i386/i386.opt (mprefer-avx256): New option.
8151 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8153 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8154 for TARGET_PREFER_AVX256.
8155 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8157 2017-09-21 Jeff Law <law@redhat.com>
8159 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8160 Fix dump output if the only stack space is for pushed registers.
8162 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8164 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8167 2017-09-21 Martin Sebor <msebor@redhat.com>
8170 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8171 code (in C++) or code that triggers warnings.
8173 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8175 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8177 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8179 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8180 * sched-rgn.c: Ditto.
8181 * sel-sched-ir.c: Ditto.
8183 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8185 * toplev.h (set_random_seed): Adjust return type.
8186 * toplev.c (init_local_tick): Move eager initialization of random_seed
8187 to get_random_seed. Adjust comment.
8188 (init_random_seed): Inline to get_random_seed, delete.
8189 (get_random_seed): Initialize random_seed lazily.
8190 (set_random_seed): Do not return previous value.
8191 (print_switch_value): Do not call get_random_seed.
8193 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8195 * cgraph.c (delete_function_version): New, broken out from...
8196 (cgraph_node::delete_function_version): ...here. Rename to
8197 cgraph_node::delete_function_version_by_decl. Update all uses.
8198 (cgraph_node::remove): Call delete_function_version.
8200 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8203 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8204 VAR_DECLs to which addressable non-volatile parameters are mapped
8205 and for id->retvar after the return value assignment. Clear
8206 id->retval and id->retbnd after inlining.
8208 2017-09-21 Richard Biener <rguenther@suse.de>
8210 PR tree-optimization/82276
8211 PR tree-optimization/82244
8212 * tree-vrp.c (build_assert_expr_for): Set
8213 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8215 (remove_range_assertions): Revert earlier change.
8217 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8220 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8222 2017-09-21 Richard Biener <rguenther@suse.de>
8224 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8225 Restore valid IL after code generation errors.
8226 * graphite.c (graphite_transform_loops): Diagnose code
8227 generation issues as MSG_MISSED_OPTIMIZATION and continue
8228 with processing SCOPs.
8230 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8231 Alan Hayward <alan.hayward@arm.com>
8232 David Sherwood <david.sherwood@arm.com>
8234 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8235 rather than choosing between plus_constant and gen_rtx_<CODE>.
8236 * expr.c (emit_push_insn): Likewise.
8237 (expand_expr_real_2): Likewise.
8239 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8240 Alan Hayward <alan.hayward@arm.com>
8241 David Sherwood <david.sherwood@arm.com>
8243 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8245 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8246 Alan Hayward <alan.hayward@arm.com>
8247 David Sherwood <david.sherwood@arm.com>
8249 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8250 calling tree_to_uhwi.
8252 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8253 Alan Hayward <alan.hayward@arm.com>
8254 David Sherwood <david.sherwood@arm.com>
8256 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8257 INTEGER_CST rather than a negative test for ADDR_EXPR.
8259 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8260 Alan Hayward <alan.hayward@arm.com>
8261 David Sherwood <david.sherwood@arm.com>
8263 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8264 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8266 2017-09-21 Richard Biener <rguenther@suse.de>
8268 PR tree-optimization/71351
8269 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8270 graphite_create_new_loop_guard): Remove, fold remaining parts
8272 (translate_isl_ast_node_for): ... here and simplify.
8274 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8277 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8278 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8279 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8280 alternative always use QI mode, for -Os imov (=R,R) alternative
8281 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8284 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8285 Jeff Law <law@redhat.com>
8287 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8288 (allocate_stack_space): New function, partially extracted from
8290 (s390_emit_prologue): Track offset to most recent stack probe.
8291 Code to allocate space moved into allocate_stack_space.
8292 Dump actions when no stack is allocated.
8293 (s390_prologue_plus_offset): New function.
8294 (s390_emit_stack_probe): Likewise.
8296 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8298 * common.opt (Wa, Wl, Wp, g, gz=): Add
8300 (gno-column-info): Remove.
8301 (gcolumn-info): Drop RejectNegative.
8303 (gno-record-gcc-switches): Remove.
8304 (grecord-gcc-switches): Drop RejectNegative.
8305 (gno-split-dwarf): Remove.
8306 (gsplit-dwarf): Drop RejectNegative.
8307 (gno-strict-dwarf): Remove.
8308 (gstrict-dwarf): Drop RejectNegative.
8309 * config/darwin.opt (gfull, gused): Add RejectNegative.
8310 * dwarf2out.c (gen_producer_string): Drop
8311 gno-record-gcc-switches handler.
8312 * optc-gen.awk: Add g to prefixes with negative forms.
8313 * opts-common.c (remapping_prefix_p): New.
8314 (find_opt): Check it.
8315 (generate_canonical_option): Test g prefix.
8316 (option_map): Add -gno- mapping.
8317 (add_misspelling_candidates): Check remapping_prefix_p.
8319 2017-09-20 Jeff Law <law@redhat.com>
8321 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8322 thinko in stack clash protection support.
8324 * explow.c (compute_stack_clash_protection_loop_data): Use
8325 CONST_INT_P instead of explicit test. Verify object is a
8326 CONST_INT_P before looking at INTVAL.
8327 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8328 instead of explicit test.
8330 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8333 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8334 address instead of to r1 and r11.
8336 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8338 * config.gcc: Support "knm".
8339 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8340 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8342 * config/i386/i386.c (m_KNM): Define.
8343 (processor_target_table): Add "knm".
8345 (ix86_option_override_internal): Add "knm".
8346 (ix86_issue_rate): Add PROCESSOR_KNM.
8347 (ix86_adjust_cost): Ditto.
8348 (ia32_multipass_dfa_lookahead): Ditto.
8349 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8350 (fold_builtin_cpu): Add M_INTEL_KNM.
8351 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8352 (processor_type): Add PROCESSOR_KNM.
8353 * config/i386/x86-tune.def: Add m_KNM.
8354 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8356 2017-09-20 Richard Biener <rguenther@suse.de>
8358 PR tree-optimization/80213
8359 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8360 are allowed in empty BBs as well.
8361 (canonicalize_loop_closed_ssa): Also look for other complex
8363 (scop_detection::get_sese): Include the loop-closed PHI block
8365 (scop_detection::merge_sese): Remove code adding extra blocks.
8366 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8367 (build_scops): Assert the final returned scop is invalid.
8369 2017-09-20 Richard Biener <rguenther@suse.de>
8371 PR tree-optimization/82264
8372 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8374 (vn_phi_lookup): Likewise.
8375 (vn_phi_insert): Likewise.
8377 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8379 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8380 that fits into uhwi or shwi, add DW_AT_const_value regardless
8381 of early_dwarf without going through RTL, using add_AT_unsigned
8384 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8385 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8386 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8387 (DEBUG_MACRO_SECTION): Likewise.
8388 (DEBUG_LTO_MACRO_SECTION): Likewise.
8389 (DEBUG_STR_DWO_SECTION): Likewise.
8390 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8391 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8392 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8393 (DEBUG_LTO_LINE_STR_SECTION): Define.
8394 (init_sections_and_labels): Initialize debug_line_str_section
8395 variable. Initialize debug_loc_section for -gdwarf-5 to
8396 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8398 2017-09-20 Richard Biener <rguenther@suse.de>
8400 * graphite-sese-to-poly.c (extract_affine): Properly handle
8401 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8403 2017-09-20 Richard Biener <rguenther@suse.de>
8405 PR tree-optimization/81373
8406 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8407 Force SESE live-out defs to be handled even if they are
8410 2017-09-19 Jeff Law <law@redhat.com>
8412 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8413 nothing for stack adjustments with REG_STACK_CHECK.
8414 * sched-deps.c (parse_add_or_inc): Reject insns with
8415 REG_STACK_CHECK from dependency breaking.
8416 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8417 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8418 * reg-notes.def (STACK_CHECK): New note.
8420 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8421 (ix86_expand_prologue): Dump stack clash info as needed.
8422 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8424 * function.c (dump_stack_clash_frame_info): New function.
8425 * function.h (dump_stack_clash_frame_info): Prototype.
8426 (enum stack_clash_probes): New enum.
8428 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8429 flag_stack_clash_protection.
8430 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8431 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8432 (arm_frame_pointer_required): Likewise.
8433 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8434 (ia64_expand_prologue): Likewise.
8435 * config/mips/mips.c (mips_expand_prologue): Likewise.
8436 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8437 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8438 (sparc_flat_expand_prologue): Likewise.
8439 * config/spu/spu.c (spu_expand_prologue): Likewise.
8441 * explow.c: Include "params.h".
8442 (anti_adjust_stack_and_probe_stack_clash): New function.
8443 (get_stack_check_protect): Likewise.
8444 (compute_stack_clash_protection_loop_data): Likewise.
8445 (emit_stack_clash_protection_loop_start): Likewise.
8446 (emit_stack_clash_protection_loop_end): Likewise.
8447 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8448 Use anti_adjust_stack_and_probe_stack_clash.
8449 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8450 (emit_stack_clash_protection_loop_start): Likewise.
8451 (emit_stack_clash_protection_loop_end): Likewise.
8452 * rtl.h (get_stack_check_protect): Prototype.
8453 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8454 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8455 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8457 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8459 * doc/tm.texi: Rebuilt.
8460 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8461 get_stack_check_protect.
8462 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8463 * config/arm/arm.c (arm_expand_prologue): Likewise.
8464 (arm_frame_pointer_required): Likewise.
8465 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8466 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8467 * config/mips/mips.c (mips_expand_prologue): Likewise.
8468 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8469 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8470 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8471 (sparc_flat_expand_prologue): Likewise.
8473 * common.opt (-fstack-clash-protection): New option.
8474 * flag-types.h (enum stack_check_type): Note difference between
8475 -fstack-check= and -fstack-clash-protection.
8476 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8477 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8478 * toplev.c (process_options): Issue warnings/errors for cases
8479 not handled with -fstack-clash-protection.
8480 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8481 (-fstack-check): Note additional problem with -fstack-check=generic.
8482 Note that -fstack-check is primarily for Ada and refer users
8483 to -fstack-clash-protection for stack-clash-protection.
8484 Document new params for stack clash protection.
8486 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8488 * config/i386/i386.c (ix86_split_long_move): Do not handle
8489 address used for LEA in a special way.
8491 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8493 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8495 2017-09-19 Martin Sebor <msebor@redhat.com>
8498 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8499 of incompatible types.
8501 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8504 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8505 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8506 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8508 2017-09-19 Richard Biener <rguenther@suse.de>
8510 PR tree-optimization/82244
8511 * tree-vrp.c (remove_range_assertions): Do not propagate
8512 a constant to abnormals but replace the assert with a copy.
8514 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8516 PR rtl-optimization/57878
8517 PR rtl-optimization/68988
8518 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8519 avoidance test involving non_reload_pseudos. Move frequency test
8520 below the general fragmentation avoidance test.
8522 2017-09-19 Richard Biener <rguenther@suse.de>
8524 PR tree-optimization/69728
8525 * graphite-sese-to-poly.c (schedule_error): New global.
8526 (add_loop_schedule): Handle empty domain by failing the
8528 (build_original_schedule): Handle schedule_error.
8530 2017-09-19 Richard Biener <rguenther@suse.de>
8532 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8533 Do not iterate to sibling loops but only to siblings of inner
8536 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8539 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8540 that the registers are different.
8542 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8544 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8545 to processor_model and "amdfam17h" to arch_names_table.
8546 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8548 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8551 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8553 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8554 Alan Hayward <alan.hayward@arm.com>
8555 David Sherwood <david.sherwood@arm.com>
8557 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8559 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8561 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8562 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8563 vect_schedule_slp_instance.
8564 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8565 Update call to vect_slp_analyze_node_operations. Simplify return
8567 (vect_slp_analyze_bb_1): Update call accordingly.
8568 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8569 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8570 (vect_schedule_slp): Update call accordingly.
8572 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8573 Alan Hayward <alan.hayward@arm.com>
8574 David Sherwood <david.sherwood@arm.com>
8576 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8577 with types that aren't in fact scalar.
8579 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8581 * tree-vect-slp.c (vect_record_max_nunits): New function,
8583 (vect_build_slp_tree_1): ...here.
8584 (vect_build_slp_tree_2): Call it for phis too.
8586 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8588 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8589 to vect_get_vec_def_for_operand when getting the mask operand.
8591 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8592 Alan Hayward <alan.hayward@arm.com>
8593 David Sherwood <david.sherwood@arm.com>
8595 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8598 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8599 Alan Hayward <alan.hayward@arm.com>
8600 David Sherwood <david.sherwood@arm.com>
8602 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8603 calculation for vector booleans.
8605 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8606 Alan Hayward <alan.hayward@arm.com>
8607 David Sherwood <david.sherwood@arm.com>
8609 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8611 (vect_transform_stmt): ...here.
8612 (vect_analyze_stmt): Use it instead of calling
8613 vectorizable_live_operation directly.
8615 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8617 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8618 non-SIMT targets in acc vector loops.
8620 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8622 * configure.ac: Add arc and check if assembler supports gdwarf2.
8623 * configure: Regenerate.
8625 2017-09-18 Richard Biener <rguenther@suse.de>
8627 PR tree-optimization/82220
8628 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8629 epilogue niters from the min_profitable_iters compute.
8631 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8634 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8635 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8636 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8638 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8641 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8642 switching to a new text section.
8644 2017-09-18 Richard Biener <rguenther@suse.de>
8646 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8648 (build_alias_set): Reject aliases with no access function.
8650 2017-09-18 Richard Biener <rguenther@suse.de>
8652 PR tree-optimization/79622
8653 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8655 (build_cross_bb_scalars_use): Likewise.
8657 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8659 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8661 2017-09-18 Alan Modra <amodra@gmail.com>
8664 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8665 stack_pointer_rtx for count 0. Update comments. Break up
8666 large rtl expression.
8668 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8670 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8671 Increase to 20 bytes.
8672 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8673 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8674 or avx version of the stub.
8676 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8679 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8680 compute the minimum stack alignment. Also update preferred stack
8681 boundary for leaf functions.
8683 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8685 PR tree-optimization/82228
8686 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8689 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8691 * common/config/nds32/nds32-common.c
8692 (nds32_option_optimization_table): Refine formatting.
8693 (nds32_option_optimization_table): Use -fsched-pressure and
8694 -fomit-frame-pointer for specific optimization level.
8696 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8698 * config/nds32/nds32.c: Refine formatting and comments.
8699 * config/nds32/nds32.h: Likewise.
8700 * config/nds32/nds32.md: Likewise.
8701 * config/nds32/nds32-cost.c: Likewise.
8702 * config/nds32/nds32-isr.c: Likewise.
8703 * config/nds32/nds32-md-auxiliary.c: Likewise.
8704 * config/nds32/nds32-multiple.md: Likewise.
8705 * config/nds32/nds32-predicates.c: Likewise.
8707 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8708 Jakub Jelinek <jakub@redhat.com>
8710 Add support for -std=c++2a.
8711 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8713 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8715 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8718 * doc/extend.texi (Common Function Attributes): Add
8719 references to ARM, AArch64, and S/390 specific attributes.
8720 (Function Specific Option Pragmas): Add AArch64 and S/390
8721 to list of back ends that support the target pragma.
8723 2017-09-15 Nathan Sidwell <nathan@acm.org>
8725 * doc/standards.texi: Fix C++17 description. Update URLs for
8728 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8730 * common.opt (Wcast-align=strict): New warning option.
8731 * doc/invoke.texi: Document -Wcast-align=strict.
8733 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8735 * cgraph.h (cgraph_thunk_info): Add comments.
8736 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8737 assert for VIRTUAL_* arguments stricter.
8739 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8741 PR tree-optimization/71026
8742 * match.pd: Move RDIV patterns from fold-const.c
8743 * fold-const.c (distribute_real_division): Removed.
8744 (fold_binary_loc): Remove calls to distribute_real_divison.
8746 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8748 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8749 c++1z and gnu++1z as deprecated. Change other references to
8750 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8751 Change -Wc++1z-compat to -Wc++17-compat.
8752 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8753 * dwarf2out.c (highest_c_language): Handle C++17.
8754 (gen_compile_unit_die): Likewise.
8756 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8758 PR rtl-optimization/82192
8759 * combine.c (make_extraction): Don't look through non-paradoxical
8760 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8763 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8764 Alan Hayward <alan.hayward@arm.com>
8765 David Sherwood <david.sherwood@arm.com>
8767 * target.def (function_arg_offset): New hook.
8768 * targhooks.h (default_function_arg_offset): Declare.
8769 * targhooks.c (default_function_arg_offset): New function.
8770 * function.c (locate_and_pad_parm): Use
8771 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8772 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8773 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8774 * doc/tm.texi: Regenerate.
8775 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8776 * config/spu/spu.c (spu_function_arg_offset): New function.
8777 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8778 * system.h (FUNCTION_ARG_OFFSET): Poison.
8780 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8781 Alan Hayard <alan.hayward@arm.com>
8782 David Sherwood <david.sherwood@arm.com>
8784 * target.def (truly_noop_truncation): New hook.
8785 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8786 than TRULY_NOOP_TRUNCATION.
8787 * hooks.h (hook_bool_uint_uint_true): Declare.
8788 * hooks.c (hook_bool_uint_uint_true): New function.
8789 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8790 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8791 * doc/tm.texi: Regenerate.
8792 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8793 rather than TRULY_NOOP_TRUNCATION in comments.
8794 (simplify_comparison): Likewise.
8795 (record_truncated_value): Likewise.
8796 * expmed.c (extract_bit_field_1): Likewise.
8797 (extract_split_bit_field): Likewise.
8798 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8799 instead of TRULY_NOOP_TRUNCATION.
8800 * function.c (assign_parm_setup_block): Likewise.
8801 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8802 * rtlhooks.c: Include target.h.
8803 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8804 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8805 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8806 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8807 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8808 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8809 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8810 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8811 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8812 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8813 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8814 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8815 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8816 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8817 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8818 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8819 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8820 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8821 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8822 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8823 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8824 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8825 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8826 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8827 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8828 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8829 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8830 rather than TRULY_NOOP_TRUNCATION in comments.
8831 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8832 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8833 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8834 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8835 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8836 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8837 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8838 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8839 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8840 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8841 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8842 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8843 rather than TRULY_NOOP_TRUNCATION in comments.
8844 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8845 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8846 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8847 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8848 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8849 TRULY_NOOP_TRUNCATION condition.
8850 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8851 (TRULY_NOOP_TRUNCATION): Delete.
8852 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8853 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8854 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8855 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8856 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8857 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8858 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8859 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8860 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8861 rather than TRULY_NOOP_TRUNCATION in comments.
8862 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8863 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8864 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8865 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8866 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8867 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8869 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8872 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8873 (*cmp_ior): Likewise.
8874 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8875 (*ior_scc_scc_cmp): Likewise.
8876 (*and_scc_scc): Likewise.
8877 (*and_scc_scc_cmp): Likewise.
8879 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8880 Alan Hayard <alan.hayward@arm.com>
8881 David Sherwood <david.sherwood@arm.com>
8883 * target.def (can_change_mode_class): New hook.
8884 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8885 (hard_regno_nregs): Likewise.
8886 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8887 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8888 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8889 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8890 (LOAD_EXTEND_OP): Update accordingly.
8891 * doc/tm.texi: Regenerate.
8892 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8893 CANNOT_CHANGE_MODE_CLASS.
8894 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8895 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8896 * combine.c (simplify_set): Update accordingly.
8897 * emit-rtl.c (validate_subreg): Likewise.
8898 * recog.c (general_operand): Likewise.
8899 * regcprop.c (mode_change_ok): Likewise.
8900 * reload1.c (choose_reload_regs): Likewise.
8901 (inherit_piecemeal_p): Likewise.
8902 * rtlanal.c (simplify_subreg_regno): Likewise.
8903 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8904 instead of CANNOT_CHANGE_MODE_CLASS.
8905 (reload_cse_simplify_operands): Likewise.
8906 * reload.c (push_reload): Use targetm.can_change_mode_class
8907 instead of CANNOT_CHANGE_MODE_CLASS.
8908 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8909 REG_CANNOT_CHANGE_MODE_P.
8910 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8911 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8912 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8913 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8914 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8915 (arm_can_change_mode_class): New function.
8916 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8917 than CANNOT_CHANGE_MODE_CLASS in comments.
8918 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8919 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8920 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8921 (ix86_can_change_mode_class): ...this new function, inverting the
8922 sense of the return value.
8923 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8924 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8925 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8926 (ia64_can_change_mode_class): New function.
8927 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8928 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8929 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8930 (m32c_can_change_mode_class): ...this new function, inverting the
8931 sense of the return value.
8932 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8933 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8934 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8935 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8936 (mips_can_change_mode_class): ...this new function, inverting the
8937 sense of the return value.
8938 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8939 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8940 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8941 (msp430_can_change_mode_class): New function.
8942 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8943 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8944 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8945 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8946 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8947 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8948 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8949 (pa_cannot_change_mode_class): Replace with...
8950 (pa_can_change_mode_class): ...this new function, inverting the
8951 sense of the return value.
8952 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8953 than CANNOT_CHANGE_MODE_CLASS in comments.
8954 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8955 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8956 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8957 (pdp11_cannot_change_mode_class): Replace with...
8958 (pdp11_can_change_mode_class): ...this new function, inverting the
8959 sense of the return value.
8960 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8961 * config/powerpcspe/powerpcspe-protos.h
8962 (rs6000_cannot_change_mode_class_ptr): Delete.
8963 * config/powerpcspe/powerpcspe.c
8964 (rs6000_cannot_change_mode_class_ptr): Delete.
8965 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8966 (rs6000_option_override_internal): Assign to
8967 targetm.can_change_mode_class instead of
8968 rs6000_cannot_change_mode_class_ptr.
8969 (rs6000_cannot_change_mode_class): Replace with...
8970 (rs6000_can_change_mode_class): ...this new function, inverting the
8971 sense of the return value.
8972 (rs6000_debug_cannot_change_mode_class): Replace with...
8973 (rs6000_debug_can_change_mode_class): ...this new function.
8974 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8975 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8976 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8977 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8978 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8980 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8981 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8982 (rs6000_option_override_internal): Assign to
8983 targetm.can_change_mode_class instead of
8984 rs6000_cannot_change_mode_class_ptr.
8985 (rs6000_cannot_change_mode_class): Replace with...
8986 (rs6000_can_change_mode_class): ...this new function, inverting the
8987 sense of the return value.
8988 (rs6000_debug_cannot_change_mode_class): Replace with...
8989 (rs6000_debug_can_change_mode_class): ...this new function.
8990 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8991 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8992 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8993 (s390_can_change_mode_class): ...this new function, inverting the
8994 sense of the return value.
8995 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8996 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8997 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8998 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8999 (sh_cannot_change_mode_class): Replace with...
9000 (sh_can_change_mode_class): ...this new function, inverting the
9001 sense of the return value.
9002 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9003 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9004 (sparc_can_change_mode_class): New function.
9005 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9006 * config/spu/spu.c (spu_can_change_mode_class): New function.
9007 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9008 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9009 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9010 (visium_can_change_mode_class): New function.
9011 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9013 2017-09-15 Richard Biener <rguenther@suse.de>
9015 PR tree-optimization/82217
9016 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9017 but not undefined case.
9019 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9022 * postreload.c (reload_cse_simplify_operands): Skip
9023 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9025 2017-09-15 Richard Biener <rguenther@suse.de>
9027 PR tree-optimization/68823
9028 * graphite-scop-detection.c (build_alias_set): If we have a
9029 possible dependence check whether we can handle them by just
9030 looking at the DRs DR_ACCESS_FNs.
9031 (build_scops): If build_alias_set fails, fail the SCOP.
9033 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9035 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9036 to support float128 built-in functions that require the ISA 3.0
9038 (BU_FLOAT128_3_HW): Likewise.
9039 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9041 (FMAF128): Likewise.
9043 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9044 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9045 floating point instructions.
9046 (rs6000_invalid_builtin): Likewise.
9047 (rs6000_builtin_mask_names): Likewise.
9048 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9049 (RS6000_BTM_FLOAT128_HW): Likewise.
9050 (RS6000_BTM_COMMON): Likewise.
9051 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9053 * doc/extend.texi (RS/6000 built-in functions): Document the
9054 IEEE 128-bit floating point square root and fused multiply-add
9057 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9059 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9060 reg (r2) isn't in the set of registers defined in the prologue.
9062 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9063 Alan Hayward <alan.hayward@arm.com>
9064 David Sherwood <david.sherwood@arm.com>
9066 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9067 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9068 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9069 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9070 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9072 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9073 max_vectorization_factor.
9074 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9076 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9077 Alan Hayward <alan.hayward@arm.com>
9078 David Sherwood <david.sherwood@arm.com>
9080 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9081 (vect_worthwhile_without_simd_p): Declare.
9082 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9083 (vectorizable_reduction): Use it.
9084 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9085 (vectorizable_operation): Likewise.
9087 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9088 Alan Hayward <alan.hayward@arm.com>
9089 David Sherwood <david.sherwood@arm.com>
9091 * tree-vectorizer.h (vect_get_num_copies): New function.
9092 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9093 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9094 (vectorizable_induction): Likewise.
9095 (vectorizable_live_operation): Likewise.
9096 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9097 (vectorizable_bswap): Likewise.
9098 (vectorizable_call): Likewise.
9099 (vectorizable_conversion): Likewise.
9100 (vectorizable_assignment): Likewise.
9101 (vectorizable_shift): Likewise.
9102 (vectorizable_operation): Likewise.
9103 (vectorizable_store): Likewise.
9104 (vectorizable_load): Likewise.
9105 (vectorizable_condition): Likewise.
9106 (vectorizable_comparison): Likewise.
9107 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9109 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9110 Alan Hayward <alan.hayward@arm.com>
9111 David Sherwood <david.sherwood@arm.com>
9113 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9114 of vect_init_vector.
9116 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9117 Alan Hayward <alan.hayward@arm.com>
9118 David Sherwood <david.sherwood@arm.com>
9120 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9121 an inline wrapper that provides a location.
9122 (gimple_build_vector): Likewise.
9123 * gimple-fold.c (gimple_build_vector_from_val): New function.
9124 (gimple_build_vector): Likewise.
9125 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9126 functions to build the initial value. Always return a gimple value.
9127 (get_initial_defs_for_reduction): Likewise. Only compute
9129 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9130 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9131 (vectorizable_induction): Use gimple_build_vector rather than
9134 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9135 Alan Hayward <alan.hayward@arm.com>
9136 David Sherwood <david.sherwood@arm.com>
9138 * target.h (vec_perm_indices): New typedef.
9139 (auto_vec_perm_indices): Likewise.
9140 * optabs-query.h: Include target.h
9141 (can_vec_perm_p): Take a vec_perm_indices *.
9142 * optabs-query.c (can_vec_perm_p): Likewise.
9143 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9144 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9145 * tree-vect-generic.c (lower_vec_perm): Likewise.
9146 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9147 (vect_grouped_load_supported): Likewise.
9148 (vect_shift_permute_load_chain): Likewise.
9149 (vect_permute_store_chain): Use auto_vec_perm_indices.
9150 (vect_permute_load_chain): Likewise.
9151 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9152 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9153 Update uses of can_vec_perm_p.
9154 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9155 mode with a number of elements. Take a vec_perm_indices *.
9156 (vect_create_epilog_for_reduction): Update accordingly.
9157 Use auto_vec_perm_indices.
9158 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9159 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9160 (vect_transform_slp_perm_load): Likewise.
9161 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9162 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9163 (vect_gen_perm_mask_checked): Likewise.
9164 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9165 (vect_gen_perm_mask_checked): Likewise.
9166 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9167 (vectorizable_store): Likewise.
9168 (vectorizable_load): Likewise.
9169 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9170 (vectorizable_bswap): Likewise.
9172 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9173 Alan Hayward <alan.hayward@arm.com>
9174 David Sherwood <david.sherwood@arm.com>
9176 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9177 * tree.c (build_vector): Likewise.
9178 (build_vector_from_ctor): Update accordingly.
9179 (build_vector_from_val): Likewise.
9180 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9181 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9182 * tree-vect-generic.c (add_rshift): Likewise.
9183 (expand_vector_divmod): Likewise.
9184 (optimize_vector_constructor): Likewise.
9185 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9186 (vect_transform_slp_perm_load): Likewise.
9187 (vect_schedule_slp_instance): Likewise.
9188 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9189 (vectorizable_call): Likewise.
9190 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9191 * expmed.c (make_tree): Likewise.
9192 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9193 a vector passed to build_vector.
9194 (fold_convert_const): Likewise.
9195 (exact_inverse): Likewise.
9196 (fold_ternary_loc): Likewise.
9197 (fold_relational_const): Likewise.
9198 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9199 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9200 (const_unop): Likewise. Store the reduction accumulator in a
9201 variable rather than an array.
9202 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9203 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9204 the new vector, rather than constructing it after the input arrays.
9205 (native_interpret_vector): Use auto_vec<tree> when building
9206 a vector passed to build_vector. Add elements in order.
9207 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9208 auto_vec<tree> when building a vector passed to build_vector.
9209 (vect_create_epilog_for_reduction): Likewise.
9210 (vectorizable_induction): Likewise.
9211 (get_initial_def_for_reduction): Likewise. Fix indentation of
9213 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9215 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9216 passed to build_vector.
9218 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9219 Alan Hayward <alan.hayward@arm.com>
9220 David Sherwood <david.sherwood@arm.com>
9222 * tree-core.h (tree_base::u): Add an "nelts" field.
9223 (tree_vector): Use VECTOR_CST_NELTS as the length.
9224 * tree.c (tree_size): Likewise.
9225 (make_vector): Initialize VECTOR_CST_NELTS.
9226 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9227 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9228 TYPE_VECTOR_SUBPARTS.
9229 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9230 as the number of units, setting it from VECTOR_CST_NELTS.
9231 (const_vector_from_tree): Likewise.
9232 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9233 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9234 (fold_negate_expr_1): Likewise.
9235 (fold_convert_const): Likewise.
9236 (const_binop): Likewise. Differentiate the number of output and
9238 (const_unop): Likewise.
9239 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9240 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9241 in cases that did the opposite.
9243 2017-09-14 Richard Biener <rguenther@suse.de>
9245 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9248 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9250 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9252 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9255 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9256 if and where to split a bb, except for splitting before debug insn
9257 sequences followed by non-label real insn. Delete debug insns
9258 in between basic blocks.
9260 * combine.c (make_compound_operation_int): Formatting fixes.
9262 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9263 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9264 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9265 * config/sol2.h (LINK_EH_SPEC): Likewise.
9266 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9267 * config/s390/linux.h (LINK_SPEC): Likewise.
9268 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9269 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9270 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9271 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9272 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9273 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9274 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9275 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9277 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9279 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9281 (ENDFILE_LINUX_SPEC): Likewise.
9282 (LINK_EH_SPEC): Likewise.
9283 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9284 (LINK_OS_LINUX_SPEC32): Likewise.
9285 (LINK_OS_LINUX_SPEC64): Likewise.
9286 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9287 (LINK_OS_LINUX_SPEC): Likewise.
9289 2017-09-13 Martin Liska <mliska@suse.cz>
9292 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9293 CASE_HIGH is NULL_TREE.
9295 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9296 Alan Hayward <alan.hayward@arm.com>
9297 David Sherwood <david.sherwood@arm.com>
9299 * target.def (secondary_memory_needed): New hook.
9300 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9301 instead of SECONDARY_MEMORY_NEEDED.
9302 (secondary_memory_needed_mode): Likewise.
9303 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9304 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9305 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9306 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9307 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9308 * doc/tm.texi: Regenerate.
9309 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9310 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9311 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9312 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9313 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9314 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9315 mode argument first and change the reg_class arguments to reg_class_t.
9316 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9317 Make static. Update the call to inline_secondary_memory_needed.
9318 (ix86_register_move_cost): Update the call to
9319 inline_secondary_memory_needed.
9320 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9321 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9323 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9324 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9326 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9327 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9328 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9329 and match hook interface. Add comment from mips.h.
9330 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9331 * config/mmix/mmix.md (truncdfsf2): Refer to
9332 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9334 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9335 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9336 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9337 (pa_secondary_memory_needed): New function.
9338 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9339 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9340 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9341 (pdp11_secondary_memory_needed): Make static and match hook interface.
9342 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9343 * config/powerpcspe/powerpcspe-protos.h
9344 (rs6000_secondary_memory_needed_ptr): Delete.
9345 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9347 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9348 (rs6000_option_override_internal): Assign to
9349 targetm.secondary_memory_needed rather than
9350 rs6000_secondary_memory_needed_ptr.
9351 (rs6000_secondary_memory_needed): Match hook interface.
9352 (rs6000_debug_secondary_memory_needed): Likewise.
9353 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9354 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9355 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9356 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9357 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9358 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9360 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9361 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9362 (rs6000_option_override_internal): Assign to
9363 targetm.secondary_memory_needed rather than
9364 rs6000_secondary_memory_needed_ptr.
9365 (rs6000_secondary_memory_needed): Match hook interface.
9366 (rs6000_debug_secondary_memory_needed): Likewise.
9367 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9368 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9369 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9370 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9371 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9372 (sparc_secondary_memory_needed): New function.
9373 * lra-constraints.c (check_and_process_move): Refer to
9374 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9376 (curr_insn_transform): Likewise.
9377 (process_alt_operands): Use targetm.secondary_memory_needed
9378 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9379 (check_secondary_memory_needed_p): Likewise.
9380 (choose_split_class): Likewise.
9381 * reload.c: Unconditionally include code that was previously
9382 conditional on SECONDARY_MEMORY_NEEDED.
9383 (push_secondary_reload): Use targetm.secondary_memory_needed
9384 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9385 (push_reload): Likewise.
9386 * reload1.c: Unconditionally include code that was previously
9387 conditional on SECONDARY_MEMORY_NEEDED.
9388 (choose_reload_regs): Use targetm.secondary_memory_needed
9389 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9390 (gen_reload): Likewise.
9391 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9393 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9394 Alan Hayward <alan.hayward@arm.com>
9395 David Sherwood <david.sherwood@arm.com>
9397 * target.def (secondary_memory_needed_mode): New hook:
9398 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9399 * targhooks.h (default_secondary_memory_needed_mode): New function.
9400 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9401 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9402 * doc/tm.texi: Regenerate.
9403 * lra-constraints.c (check_and_process_move): Use
9404 targetm.secondary_memory_needed_mode instead of
9405 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9406 (curr_insn_transform): Likewise.
9407 * reload.c (get_secondary_mem): Likewise.
9408 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9409 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9411 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9412 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9413 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9414 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9415 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9417 * config/powerpcspe/powerpcspe-protos.h
9418 (rs6000_secondary_memory_needed_mode): Delete.
9419 * config/powerpcspe/powerpcspe.c
9420 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9421 (rs6000_secondary_memory_needed_mode): Make static.
9422 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9423 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9425 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9427 (rs6000_secondary_memory_needed_mode): Make static.
9428 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9429 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9430 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9431 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9432 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9434 (sparc_secondary_memory_needed_mode): New function.
9435 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9437 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9439 * config/aarch64/constraints.md (Umq): New constraint.
9440 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9442 (mov<mode>): Update condition.
9444 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9446 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9447 when bitposition is the same.
9449 2017-09-13 Richard Biener <rguenther@suse.de>
9451 * dwarf2out.c (output_die_symbol): Remove.
9452 (output_die): Do not output a DIEs symbol.
9454 2017-09-13 Richard Biener <rguenther@suse.de>
9457 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9458 default-def to avoid breaking iterator update with the weird
9459 interaction with cgraph_update_edges_for_call_stmt_node.
9461 2017-09-13 Richard Biener <rguenther@suse.de>
9463 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9464 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9465 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9466 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9468 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9470 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9471 Disable pc relative literal load irrespective of
9472 TARGET_FIX_ERR_A53_84341 for default.
9474 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9476 * config/sparc/sparc.c (output_return): Output the source location of
9477 the insn in the delay slot, if any.
9478 (output_sibcall): Likewise.
9480 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9483 * common.opt (-static-pie): New alias.
9484 (shared): Negate static-pie.
9485 (-no-pie): Update help text.
9487 (static-pie): New option.
9488 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9489 -static-pie support.
9490 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9491 (LINK_EH_SPEC): Likewise.
9492 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9493 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9494 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9495 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9496 (init_gcc_specs): Likewise.
9497 (init_spec): Likewise.
9498 (display_help): Update help message for -pie.
9499 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9502 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9504 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9505 (movdi_aarch64): Likewise.
9506 (movti_aarch64): Likewise.
9508 2017-09-12 Simon Wright <simon@pushface.org>
9511 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9512 calculation of the minor version, always output as 0.
9514 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9517 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9518 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9519 on it early, rather than manual conversion late. For
9520 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9521 instead of performing manual conversion.
9523 2017-09-12 Carl Love <cel@us.ibm.com>
9525 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9526 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9528 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9529 VMULOSW): Add definitions.
9530 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9531 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9532 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9533 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9534 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9536 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9538 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9540 (movti_aarch64): Likewise.
9541 (movdf_aarch64): Likewise.
9542 (movtf_aarch64): Likewise.
9543 (load_pairdi): Likewise.
9544 (store_pairdi): Likewise.
9545 (load_pairdf): Likewise.
9546 (store_pairdf): Likewise.
9547 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9548 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9549 (ldr_got_small_<mode>): Likewise.
9550 (ldr_got_small_28k_<mode>): Likewise.
9551 (ldr_got_tiny): Likewise.
9552 * config/aarch64/iterators.md (ldst_sz): New.
9553 (ldpstp_sz): Likewise.
9554 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9556 (thunderx_load): Split load_8 to load_16.
9557 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9559 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9560 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9561 (xgene1_store_pair): Split store_8 to store_16.
9562 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9563 (falkor_st_0_st_sd): Split store_8 to store_16.
9565 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9567 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9568 and store1/2/3/4 to store_4/8/12/16.
9569 * config/aarch64/aarch64.md: Update for rename.
9570 * config/arm/arm.md: Likewise.: Likewise.
9571 * config/arm/arm.c: Likewise.
9572 * config/arm/thumb1.md: Likewise.
9573 * config/arm/thumb2.md: Likewise.
9574 * config/arm/vfp.md: Likewise.
9575 * config/arm/arm-generic.md: Likewise.
9576 * config/arm/arm1020e.md: Likewise.
9577 * config/arm/arm1026ejs.md: Likewise.
9578 * config/arm/arm1136jfs.md: Likewise.
9579 * config/arm/arm926ejs.md: Likewise.
9580 * config/arm/cortex-a15.md: Likewise.
9581 * config/arm/cortex-a17.md: Likewise.
9582 * config/arm/cortex-a5.md: Likewise.
9583 * config/arm/cortex-a53.md: Likewise.
9584 * config/arm/cortex-a57.md: Likewise.
9585 * config/arm/cortex-a7.md: Likewise.
9586 * config/arm/cortex-a8.md: Likewise.
9587 * config/arm/cortex-a9.md: Likewise.
9588 * config/arm/cortex-m4.md: Likewise.
9589 * config/arm/cortex-m7.md: Likewise.
9590 * config/arm/cortex-r4.md: Likewise.
9591 * config/arm/exynos-m1.md: Likewise.
9592 * config/arm/fa526.md: Likewise.
9593 * config/arm/fa606te.md: Likewise.
9594 * config/arm/fa626te.md: Likewise.
9595 * config/arm/fa726te.md: Likewise.
9596 * config/arm/fmp626.md: Likewise.
9597 * config/arm/iwmmxt.md: Likewise.
9598 * config/arm/ldmstm.md: Likewise.
9599 * config/arm/marvell-pj4.md: Likewise.
9600 * config/arm/xgene1.md: Likewise.
9601 * config/aarch64/thunderx.md: Likewise.
9602 * config/aarch64/thunderx2t99.md: Likewise.
9603 * config/aarch64/falkor.md: Likewise.
9605 2017-09-12 Martin Liska <mliska@suse.cz>
9607 * attribs.c (private_lookup_attribute): New function.
9608 * attribs.h (private_lookup_attribute): Declared here.
9609 (lookup_attribute): Called from this place.
9611 2017-09-12 Richard Biener <rguenther@suse.de>
9613 PR tree-optimization/82157
9614 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9615 stmts with side-effects.
9617 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9618 Alan Hayward <alan.hayward@arm.com>
9619 David Sherwood <david.sherwood@arm.com>
9621 * target.def (hard_regno_nregs): New hook.
9622 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9623 * targhooks.h (default_hard_regno_nregs): Declare.
9624 * targhooks.c (default_hard_regno_nregs): New function.
9625 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9626 (TARGET_HARD_REGNO_NREGS): ...this hook.
9627 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9628 (CLASS_MAX_NREGS): Likewise.
9629 * doc/tm.texi: Regenerate.
9630 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9631 instead of HARD_REGNO_NREGS.
9632 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9633 HARD_REGNO_NREGS in the comment.
9634 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9635 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9636 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9637 Return an unsigned int.
9638 (TARGET_HARD_REGNO_NREGS): Redefine.
9639 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9640 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9641 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9642 (arc_hard_regno_nregs): New function.
9643 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9644 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9645 (arm_hard_regno_nregs): New function.
9646 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9647 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9648 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9649 (TARGET_HARD_REGNO_NREGS): Redefine.
9650 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9651 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9652 (HARD_REGNO_NREGS): Delete.
9653 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9654 (cr16_hard_regno_nregs): New function.
9655 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9656 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9657 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9658 (cris_hard_regno_nregs): New function.
9659 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9660 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9661 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9662 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9663 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9664 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9665 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9666 (frv_hard_regno_nregs): Make static. Take and return an
9668 (frv_class_max_nregs): Remove outdated copy of documentation.
9669 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9670 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9671 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9672 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9673 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9674 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9675 (TARGET_HARD_REGNO_NREGS): Redefine.
9676 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9677 (CLASS_MAX_NREGS): Update comment.
9678 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9679 (ia64_hard_regno_nregs): New function.
9680 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9681 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9682 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9683 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9684 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9686 (m32c_hard_regno_nregs): Likewise. Make static.
9687 (TARGET_HARD_REGNO_NREGS): Redefine.
9688 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9689 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9690 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9691 (m68k_hard_regno_nregs): New function.
9692 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9693 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9694 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9695 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9696 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9697 Take and return an unsigned int.
9698 (TARGET_HARD_REGNO_NREGS): Redefine.
9699 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9700 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9701 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9702 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9703 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9704 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9705 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9706 (msp430_hard_regno_nregs): Make static. Take and return an
9708 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9709 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9710 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9711 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9712 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9713 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9714 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9715 (TARGET_HARD_REGNO_NREGS): Redefine.
9716 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9717 (PA_HARD_REGNO_NREGS): ...this.
9718 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9719 (PA_HARD_REGNO_NREGS): ...this.
9720 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9721 (pa_hard_regno_nregs): New function.
9722 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9723 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9724 (pdp11_hard_regno_nregs): New function.
9725 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9726 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9727 (rs6000_hard_regno_nregs_hook): New function.
9728 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9729 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9730 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9731 Take and return an unsigned int. Move earlier in file.
9732 (TARGET_HARD_REGNO_NREGS): Redefine.
9733 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9734 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9735 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9736 (rl78_hard_regno_nregs): Make static. Take and return an
9738 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9739 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9740 (rs6000_hard_regno_nregs_hook): New function.
9741 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9742 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9743 (TARGET_HARD_REGNO_NREGS): Redefine.
9744 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9745 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9746 instead of HARD_REGNO_NREGS.
9747 (s390_hard_regno_nregs): New function.
9748 (s390_hard_regno_mode_ok): Add comment from s390.h.
9749 (TARGET_HARD_REGNO_NREGS): Redefine.
9750 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9751 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9752 (sh_hard_regno_nregs): New function.
9753 (sh_pass_in_reg_p): Use it.
9754 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9755 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9756 (sparc_hard_regno_nregs): New function.
9757 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9758 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9759 (spu_function_arg_advance): Use it, supplying a valid register number.
9760 (TARGET_HARD_REGNO_NREGS): Redefine.
9761 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9762 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9763 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9764 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9765 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9766 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9767 (CLASS_MAX_NREGS): Remove copy of old documentation.
9768 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9769 (visium_hard_regno_nregs): New function.
9770 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9771 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9772 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9773 xtensa_hard_regno_nregs): New function.
9774 * system.h (HARD_REGNO_NREGS): Poison.
9776 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9778 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9779 hard_regno_nregs instead of HARD_REGNO_NREGS.
9780 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9781 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9782 (c6x_expand_epilogue): Likewise.
9783 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9784 (frv_read_iacc_argument): Likewise.
9785 * config/sh/sh.c: Include regs.h.
9786 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9787 (regs_used): Likewise.
9788 (output_stack_adjust): Likewise.
9789 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9790 * expmed.c: Include regs.h.
9791 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9792 * ree.c: Include regs.h.
9793 (combine_reaching_defs): Use hard_regno_nregs instead of
9795 (add_removable_extension): Likewise.
9797 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9799 * regs.h (hard_regno_nregs): Turn into a function.
9800 (end_hard_regno): Update accordingly.
9801 * caller-save.c (setup_save_areas): Likewise.
9802 (save_call_clobbered_regs): Likewise.
9803 (replace_reg_with_saved_mem): Likewise.
9804 (insert_restore): Likewise.
9805 (insert_save): Likewise.
9806 * combine.c (can_change_dest_mode): Likewise.
9807 (move_deaths): Likewise.
9808 (distribute_notes): Likewise.
9809 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9810 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9811 (rs6000_split_multireg_move): Likewise.
9812 (rs6000_register_move_cost): Likewise.
9813 (rs6000_memory_move_cost): Likewise.
9814 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9815 (rs6000_split_multireg_move): Likewise.
9816 (rs6000_register_move_cost): Likewise.
9817 (rs6000_memory_move_cost): Likewise.
9818 * cselib.c (cselib_reset_table): Likewise.
9819 (cselib_lookup_1): Likewise.
9820 * emit-rtl.c (set_mode_and_regno): Likewise.
9821 * function.c (aggregate_value_p): Likewise.
9822 * ira-color.c (setup_profitable_hard_regs): Likewise.
9823 (check_hard_reg_p): Likewise.
9824 (calculate_saved_nregs): Likewise.
9825 (assign_hard_reg): Likewise.
9826 (improve_allocation): Likewise.
9827 (calculate_spill_cost): Likewise.
9828 * ira-emit.c (modify_move_list): Likewise.
9829 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9830 (ira_hard_reg_in_set_p): Likewise.
9831 * ira.c (setup_reg_mode_hard_regset): Likewise.
9832 (clarify_prohibited_class_mode_regs): Likewise.
9833 (check_allocation): Likewise.
9834 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9835 (lra_setup_reg_renumber): Likewise.
9836 (setup_try_hard_regno_pseudos): Likewise.
9837 (spill_for): Likewise.
9838 (assign_hard_regno): Likewise.
9839 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9840 * lra-constraints.c (in_class_p): Likewise.
9841 (lra_constraint_offset): Likewise.
9842 (simplify_operand_subreg): Likewise.
9843 (lra_constraints): Likewise.
9844 (split_reg): Likewise.
9845 (split_if_necessary): Likewise.
9846 (invariant_p): Likewise.
9847 (inherit_in_ebb): Likewise.
9848 * lra-lives.c (process_bb_lives): Likewise.
9849 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9850 (get_hard_regs): Likewise.
9851 (do_remat): Likewise.
9852 * lra-spills.c (assign_spill_hard_regs): Likewise.
9853 * mode-switching.c (create_pre_exit): Likewise.
9854 * postreload.c (reload_combine_recognize_pattern): Likewise.
9855 * recog.c (peep2_find_free_register): Likewise.
9856 * regcprop.c (kill_value_regno): Likewise.
9857 (set_value_regno): Likewise.
9858 (copy_value): Likewise.
9859 (maybe_mode_change): Likewise.
9860 (find_oldest_value_reg): Likewise.
9861 (copyprop_hardreg_forward_1): Likewise.
9862 * regrename.c (check_new_reg_p): Likewise.
9863 (regrename_do_replace): Likewise.
9864 * reload.c (push_reload): Likewise.
9865 (combine_reloads): Likewise.
9866 (find_dummy_reload): Likewise.
9867 (operands_match_p): Likewise.
9868 (find_reloads): Likewise.
9869 (find_equiv_reg): Likewise.
9870 (reload_adjust_reg_for_mode): Likewise.
9871 * reload1.c (count_pseudo): Likewise.
9872 (count_spilled_pseudo): Likewise.
9873 (find_reg): Likewise.
9874 (clear_reload_reg_in_use): Likewise.
9875 (free_for_value_p): Likewise.
9876 (allocate_reload_reg): Likewise.
9877 (choose_reload_regs): Likewise.
9878 (reload_adjust_reg_for_temp): Likewise.
9879 (emit_reload_insns): Likewise.
9880 (delete_output_reload): Likewise.
9881 * rtlanal.c (subreg_get_info): Likewise.
9882 * sched-deps.c (sched_analyze_reg): Likewise.
9883 * sel-sched.c (init_regs_for_mode): Likewise.
9884 (mark_unavailable_hard_regs): Likewise.
9885 (choose_best_reg_1): Likewise.
9886 (verify_target_availability): Likewise.
9887 * valtrack.c (dead_debug_insert_temp): Likewise.
9888 * var-tracking.c (track_loc_p): Likewise.
9889 (emit_note_insn_var_location): Likewise.
9890 * varasm.c (make_decl_rtl): Likewise.
9891 * reginfo.c (choose_hard_reg_mode): Likewise.
9892 (init_reg_modes_target): Refer directly to
9893 this_target_regs->x_hard_regno_nregs.
9895 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9897 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9898 instead of hard_regno_nregs.
9900 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9902 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9903 end_hard_regno instead of hard_regno_nregs.
9904 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9905 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9906 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9907 * ira-color.c (improve_allocation): Likewise.
9908 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9909 * lra-lives.c (mark_regno_live): Likewise.
9910 (mark_regno_dead): Likewise.
9911 * lra-remat.c (operand_to_remat): Likewise.
9912 * lra.c (collect_non_operand_hard_regs): Likewise.
9913 * postreload.c (reload_combine_note_store): Likewise.
9914 (move2add_valid_value_p): Likewise.
9915 * reload.c (regno_clobbered_p): Likewise.
9917 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9919 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9921 * config/v850/v850.c (v850_reorg): Likewise.
9922 * reload.c (refers_to_regno_for_reload_p): Likewise.
9923 (find_equiv_reg): Likewise.
9924 * reload1.c (reload_reg_reaches_end_p): Likewise.
9926 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9928 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9930 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9931 * config/arm/arm.c (output_move_neon): Likewise.
9932 (arm_attr_length_move_neon): Likewise.
9933 (neon_split_vcombine): Likewise.
9934 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9935 (c6x_mark_reg_written): Likewise.
9936 (c6x_dwarf_register_span): Likewise.
9937 * config/i386/i386.c (ix86_save_reg): Likewise.
9938 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9939 (rws_access_reg): Likewise.
9940 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9941 * mode-switching.c (create_pre_exit): Likewise.
9942 * ree.c (combine_reaching_defs): Likewise.
9943 (add_removable_extension): Likewise.
9944 * regcprop.c (find_oldest_value_reg): Likewise.
9945 (copyprop_hardreg_forward_1): Likewise.
9946 * reload.c (reload_inner_reg_of_subreg): Likewise.
9947 (push_reload): Likewise.
9948 (combine_reloads): Likewise.
9949 (find_dummy_reload): Likewise.
9950 (reload_adjust_reg_for_mode): Likewise.
9951 * reload1.c (find_reload_regs): Likewise.
9952 (forget_old_reloads_1): Likewise.
9953 (reload_reg_free_for_value_p): Likewise.
9954 (reload_adjust_reg_for_temp): Likewise.
9955 (emit_reload_insns): Likewise.
9956 (delete_output_reload): Likewise.
9957 * sel-sched.c (choose_best_reg_1): Likewise.
9958 (choose_best_pseudo_reg): Likewise.
9960 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9961 Alan Hayward <alan.hayward@arm.com>
9962 David Sherwood <david.sherwood@arm.com>
9964 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9965 * target.def (slow_unaligned_access): New hook.
9966 * targhooks.h (default_slow_unaligned_access): Declare.
9967 * targhooks.c (default_slow_unaligned_access): New function.
9968 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9969 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9970 * doc/tm.texi: Regenerate.
9971 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9972 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9973 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9975 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9976 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9978 (rs6000_slow_unaligned_access): New function.
9979 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9980 (expand_block_compare): Likewise.
9981 (expand_strn_compare): Likewise.
9982 (rs6000_rtx_costs): Likewise.
9983 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9984 (riscv_slow_unaligned_access): Likewise.
9985 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9986 (riscv_slow_unaligned_access_p): ...this and make static.
9987 (riscv_option_override): Update accordingly.
9988 (riscv_slow_unaligned_access): New function.
9989 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9990 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9991 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9992 (rs6000_slow_unaligned_access): New function.
9993 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9994 (rs6000_rtx_costs): Likewise.
9995 * config/rs6000/rs6000-string.c (expand_block_compare)
9996 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9997 of SLOW_UNALIGNED_ACCESS.
9998 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9999 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10000 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10001 of SLOW_UNALIGNED_ACCESS.
10002 * expmed.c (simple_mem_bitfield_p): Likewise.
10003 * expr.c (alignment_for_piecewise_move): Likewise.
10004 (emit_group_load_1): Likewise.
10005 (emit_group_store): Likewise.
10006 (copy_blkmode_from_reg): Likewise.
10007 (emit_push_insn): Likewise.
10008 (expand_assignment): Likewise.
10009 (store_field): Likewise.
10010 (expand_expr_real_1): Likewise.
10011 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10012 * lra-constraints.c (simplify_operand_subreg): Likewise.
10013 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10014 * gimple-ssa-store-merging.c: Likewise in block comment at start
10016 * tree-ssa-strlen.c: Include target.h.
10017 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10018 of SLOW_UNALIGNED_ACCESS.
10019 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10021 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10023 PR rtl-optimization/82185
10024 * expmed.c (emit_store_flag_int): Only test tem if it has been
10027 2017-09-12 Richard Biener <rguenther@suse.de>
10029 PR middle-end/82149
10030 * match.pd ((FTYPE) N CMP CST): Fix typo.
10032 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10034 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10036 (mips_near_type_p): Add 'short_call' attribute as a synonym
10038 * doc/extend.texi (short_call): Document new function attribute.
10040 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10043 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10044 assertion check that in the condition.
10045 (get_atomic_generic_size): Likewise. Before testing if parameter
10046 has pointer type, if it has array type, call for C++
10047 default_conversion to perform array-to-pointer conversion.
10049 2017-09-12 Richard Biener <rguenther@suse.de>
10051 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10052 for operations we cannot scalarize.
10054 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10056 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10057 vectors heap vectors. Clean up comments.
10058 Make visited_bbs a reference.
10059 (profitable_jump_thread_path): Make GC
10060 vectors heap vectors. Clean up comments.
10062 (convert_and_register_jump_thread_path): Make GC vectors heap
10064 (check_subpath_and_update_thread_path): Same. Clean up comments.
10065 Make visited_bbs a reference.
10066 (handle_phi): Abstract common code to to
10067 register_jump_thread_path_if_profitable.
10068 Rename VAR_BB to DEF_BB.
10070 Make GC vectors heap vectors.
10071 Make visited_bbs a reference.
10072 (handle_assignment): Same.
10073 (register_jump_thread_path_if_profitable): New.
10074 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10076 Make GC vectors heap vectors. Clean up comments.
10077 Make visited_bbs a reference.
10078 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10079 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10082 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10085 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10086 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10088 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10090 Revert r251800 and r251799.
10092 2017-09-11 Martin Jambor <mjambor@suse.cz>
10095 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10096 arguments in advance.
10097 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10098 use it to find predecessor edges.
10099 (naive_outof_ssa): Collect vector of predecessors.
10101 2017-09-08 Jason Merrill <jason@redhat.com>
10103 PR c++/70029 - ICE with ref-qualifier and -flto
10104 * langhooks.h (struct lang_hooks_for_types): Add
10105 copy_lang_qualifiers.
10106 * attribs.c (build_type_attribute_qual_variant): Use it.
10107 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10109 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10110 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10112 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10115 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10116 (*mulsi3_sp64): New instruction.
10117 (mulsi3): New expander.
10119 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10121 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10123 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10125 * sancov.c: Include memmodel.h.
10127 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10130 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10133 2017-09-07 Carl Love <cel@us.ibm.com>
10135 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10136 the sldi instruction.
10138 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10140 * sancov.c: Include tm_p.h.
10142 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10145 * output.h (switch_to_other_text_partition): New declaration.
10146 * varasm.c (switch_to_other_text_partition): New function.
10147 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10148 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10149 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10150 to the other text partition before emitting LCL label and switch back
10151 after emitting the word after it.
10153 2017-09-07 Richard Biener <rguenther@suse.de>
10155 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10156 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10157 critical edges here, after loop init.
10158 (pass_data_pre): Remove PROP_no_crit_edges flags.
10159 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10160 for valueization of call args to avoid leaking VN_TOP.
10161 (visit_use): Assert we do not visit default defs.
10162 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10163 Use error_mark_node to more easily detect leaking VN_TOP.
10164 All default-defs are varying, not VN_TOP. Mark them visited.
10165 (run_scc_vn): Make code match comment.
10167 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10169 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10170 OPTION_MASK_FLOAT128_KEYWORD.
10171 (POWERPC_MASKS): Likewise.
10172 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10173 support for the -mfloat128-type option, and make -mfloat128
10174 default on PowerPC Linux systems. Define or undefine
10175 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10176 Define __float128 to be __ieee128 if IEEE 128-bit support is
10177 enabled, or undefine it.
10178 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10179 Delete defining __FLOAT128_TYPE__.
10180 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10181 -mfloat128-type option and make -mfloat128 default on PowerPC
10183 (TARGET_FLOAT128_TYPE): Likewise.
10184 (-mfloat128-type): Likewise.
10185 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10186 Delete the -mfloat128-type option and make -mfloat128 default on
10187 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10188 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10189 128-bit floating point is enabled. Change tests from using
10190 -mfloat128-type to -mfloat128.
10191 (rs6000_mangle_type): Use the correct mangling for the __float128
10192 type even if normal long double is restricted to 64-bits.
10193 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10195 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10196 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10197 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10199 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10200 documentation for -mfloat128.
10202 2017-09-06 Olivier Hainque <hainque@adacore.com>
10204 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10206 2017-09-06 Wish Wu <wishwu007@gmail.com>
10207 Jakub Jelinek <jakub@redhat.com>
10209 * asan.c (initialize_sanitizer_builtins): Add
10210 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10211 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10212 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10213 BT_FN_VOID_UINT64_PTR variables.
10214 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10215 (BT_FN_VOID_UINT16_UINT16): Likewise.
10216 (BT_FN_VOID_UINT32_UINT32): Likewise.
10217 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10218 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10219 (BT_FN_VOID_UINT64_PTR): Likewise.
10220 * common.opt (flag_sanitize_coverage): New variable.
10221 (fsanitize-coverage=trace-pc): Remove.
10222 (fsanitize-coverage=): Add.
10223 * flag-types.h (enum sanitize_coverage_code): New enum.
10224 * fold-const.c (fold_range_test): Disable non-short-circuit
10225 optimization if flag_sanitize_coverage.
10226 (fold_truth_andor): Likewise.
10227 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10228 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10229 (coverage_sanitizer_opts): New array.
10230 (get_closest_sanitizer_option): Add OPTS argument, handle also
10231 OPT_fsanitize_coverage_.
10232 (parse_sanitizer_options): Adjusted to also handle
10233 OPT_fsanitize_coverage_.
10234 (common_handle_option): Add OPT_fsanitize_coverage_.
10235 * sancov.c (instrument_comparison, instrument_switch): New function.
10236 (sancov_pass): Add trace-cmp support.
10237 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10238 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10239 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10240 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10241 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10242 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10243 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10244 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10245 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10246 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10248 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10250 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10251 error. Only quit immediately if parsing is complete.
10252 (BEGIN): Initialize fatal_err and parse_done.
10253 (begin fpu, end fpu): Check number of arguments.
10254 (begin arch, end arch): Likewise.
10255 (begin cpu, end cpu): Likewise.
10256 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10257 (optalias): Likewise.
10259 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10261 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10262 * config/arm/arm-isa.h: Delete. Move definitions to ...
10263 * arm-cpus.in: ... here. Use new feature and fgroup values.
10264 * config/arm/arm.c (arm_option_override): Use lower case for feature
10266 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10267 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10268 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10270 (print_isa_bits_for): New function.
10271 (gen_isa): New function.
10272 (gen_comm_data): Use print_isa_bits_for.
10273 (define feature): New keyword.
10274 (define fgroup): New keyword.
10275 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10276 (arm-isa.h): Add rule to generate file.
10277 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10278 case for feature bit names.
10280 2017-09-06 Richard Biener <rguenther@suse.de>
10282 * tree-ssa-pre.c (NECESSARY): Remove.
10283 (create_expression_by_pieces): Do not touch pass-local flags.
10284 (insert_into_preds_of_block): Likewise.
10285 (do_pre_regular_insertion): Likewise.
10286 (eliminate_insert): Likewise.
10287 (eliminate_dom_walker::before_dom_children): Likewise.
10288 (fini_eliminate): Do not look at inserted_exprs.
10289 (mark_operand_necessary): Remove.
10290 (remove_dead_inserted_code): Replace with simple work-list
10291 algorithm based on inserted_exprs and SSA uses.
10292 (pass_pre::execute): Re-order fini_eliminate and
10293 remove_dead_inserted_code.
10295 2017-09-06 Olivier Hainque <hainque@adacore.com>
10297 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10298 for VxWorks 7. Adjust surrounding comments.
10300 2017-09-06 Richard Biener <rguenther@suse.de>
10302 * gimple-ssa-strength-reduction.c
10303 (find_candidates_dom_walker::before_dom_children): Also allow
10306 2017-09-06 Richard Biener <rguenther@suse.de>
10308 PR tree-optimization/82108
10309 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10310 for gap in the non-permutation SLP case.
10312 2017-09-06 Martin Jambor <mjambor@suse.cz>
10314 PR tree-optimization/82078
10315 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10316 add_access_to_work_queue...
10317 (build_accesses_from_assign): ...here.
10318 (propagate_all_subaccesses): Make sure racc is the group
10319 representative, if there is one.
10321 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10323 PR middle-end/82095
10324 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10327 2017-09-06 Richard Biener <rguenther@suse.de>
10329 * gimple-ssa-strength-reduction.c
10330 (find_candidates_dom_walker::before_doom_children): Use a
10331 type and not a mode check.
10333 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10336 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10338 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10340 2017-09-05 Jeff Law <law@redhat.com>
10342 PR tree-optimization/64910
10343 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10344 cases where we have 3 or more operands.
10346 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10348 PR middle-end/81768
10349 * omp-low.c (lower_omp_for): Recompute tree invariant if
10350 gimple_omp_for_initial/final is ADDR_EXPR.
10352 PR middle-end/81768
10353 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10354 into gimple val before gimplification fo the COND_EXPR.
10356 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10358 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10359 REGION_COPY argument.
10360 (thread_through_all_blocks): Remove unused argument to
10361 duplicate_thread_path.
10363 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10364 Alan Hayward <alan.hayward@arm.com>
10365 David Sherwood <david.sherwood@arm.com>
10367 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10368 Take a scalar_mode rather than a machine_mode.
10369 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10370 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10371 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10372 (aarch64_gen_adjusted_ldpstp): Likewise.
10373 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10375 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10376 Alan Hayward <alan.hayward@arm.com>
10377 David Sherwood <david.sherwood@arm.com>
10379 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10380 Take a scalar_int_mode instead of a machine_mode.
10381 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10382 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10383 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10384 (aarch64_simd_attr_length_rglist): Delete.
10385 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10386 a scalar_int_mode instead of a machine_mode.
10387 (aarch64_add_offset): Likewise.
10388 (aarch64_internal_mov_immediate): Likewise
10389 (aarch64_add_constant_internal): Likewise.
10390 (aarch64_add_constant): Likewise.
10391 (aarch64_movw_imm): Likewise.
10392 (aarch64_rtx_arith_op_extract_p): Likewise.
10393 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10394 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10395 Remove assert that the mode isn't a vector.
10396 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10397 (aarch64_expand_mov_immediate): Update calls after above changes.
10398 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10399 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10400 (aarch64_move_imm): Likewise.
10401 (aarch64_can_const_movi_rtx_p): Likewise.
10402 (aarch64_strip_extend): Likewise.
10403 (aarch64_extr_rtx_p): Likewise.
10404 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10405 a CONST_INT when the mode parameter is VOIDmode.
10406 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10408 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10410 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10411 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10412 (bitwise_type_for_mode): Update accordingly.
10414 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10416 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10417 * stor-layout.c (mode_for_size_tree): Likewise.
10418 (mode_for_array): Update accordingly.
10419 (layout_decl): Likewise.
10420 (compute_record_mode): Likewise. Only set the mode once.
10422 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10424 * target.def (get_mask_mode): Change return type to opt_mode.
10426 * doc/tm.texi: Regenerate.
10427 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10428 * targhooks.c (default_get_mask_mode): Likewise.
10429 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10430 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10431 targetm.get_mask_mode.
10432 * tree.c (build_truth_vector_type): Likewise.
10434 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10436 * machmode.h (mode_for_vector): Return an opt_mode.
10437 * stor-layout.c (mode_for_vector): Likewise.
10438 (mode_for_int_vector): Update accordingly.
10439 (layout_type): Likewise.
10440 * config/i386/i386.c (emit_memmov): Likewise.
10441 (ix86_expand_set_or_movmem): Likewise.
10442 (ix86_expand_vector_init): Likewise.
10443 (ix86_get_mask_mode): Likewise.
10444 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10446 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10447 * expmed.c (extract_bit_field_1): Likewise.
10448 * expr.c (expand_expr_real_2): Likewise.
10449 * optabs-query.c (can_vec_perm_p): Likewise.
10450 (can_vec_mask_load_store_p): Likewise.
10451 * optabs.c (expand_vec_perm): Likewise.
10452 * targhooks.c (default_get_mask_mode): Likewise.
10453 * tree-vect-stmts.c (vectorizable_store): Likewise.
10454 (vectorizable_load): Likewise.
10455 (get_vectype_for_scalar_type_and_size): Likewise.
10457 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10459 * machmode.h (mode_for_int_vector): New function.
10460 * stor-layout.c (mode_for_int_vector): Likewise.
10461 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10462 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10463 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10464 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10465 (s390_expand_vcond): Likewise.
10467 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10469 * machmode.h (opt_machine_mode): New type.
10470 (opt_mode<T>): Allow construction from anything that can be
10472 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10473 (mode_for_size): Return an opt_machine_mode.
10474 * stor-layout.c (mode_for_size): Likewise.
10475 (mode_for_size_tree): Update call accordingly.
10476 (bitwise_mode_for_mode): Likewise.
10477 (make_fract_type): Likewise.
10478 (make_accum_type): Likewise.
10479 * caller-save.c (replace_reg_with_saved_mem): Update call
10481 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10482 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10483 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10484 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10485 * expmed.c (extract_bit_field_1): Likewise.
10486 * reload.c (get_secondary_mem): Likewise.
10487 * varasm.c (assemble_integer): Likewise.
10488 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10491 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10493 * machmode.h (decimal_float_mode_for_size): New function.
10494 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10495 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10496 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10497 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10498 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10499 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10501 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10503 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10504 (get_builtin_sync_mode): Likewise.
10505 (expand_ifn_atomic_compare_exchange): Likewise.
10506 (expand_builtin_atomic_clear): Likewise.
10507 (expand_builtin_atomic_test_and_set): Likewise.
10508 (fold_builtin_atomic_always_lock_free): Likewise.
10509 * calls.c (compute_argument_addresses): Likewise.
10510 (emit_library_call_value_1): Likewise.
10511 (store_one_arg): Likewise.
10512 * combine.c (combine_instructions): Likewise.
10513 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10514 * config/arm/arm.c (arm_function_value): Likewise.
10515 (aapcs_allocate_return_reg): Likewise.
10516 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10517 * config/i386/i386.c (construct_container): Likewise.
10518 (ix86_gimplify_va_arg): Likewise.
10519 (ix86_expand_sse_cmp): Likewise.
10520 (emit_memmov): Likewise.
10521 (emit_memset): Likewise.
10522 (expand_small_movmem_or_setmem): Likewise.
10523 (ix86_expand_pextr): Likewise.
10524 (ix86_expand_pinsr): Likewise.
10525 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10526 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10528 * config/mips/mips.c (mips_function_value_1) Likewise.
10529 (mips_block_move_straight): Likewise.
10530 (mips_expand_ins_as_unaligned_store): Likewise.
10531 * config/powerpcspe/powerpcspe.c
10532 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10533 (rs6000_darwin64_record_arg_flush): Likewise.
10534 * config/rs6000/rs6000.c
10535 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10536 (rs6000_darwin64_record_arg_flush): Likewise.
10537 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10538 (sparc_function_value_1): Likewise.
10539 * config/spu/spu.c (adjust_operand): Likewise.
10540 (spu_emit_branch_or_set): Likewise.
10541 (arith_immediate_p): Likewise.
10542 * emit-rtl.c (gen_lowpart_common): Likewise.
10543 * expr.c (expand_expr_real_1): Likewise.
10544 * function.c (assign_parm_setup_block): Likewise.
10545 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10546 * reload1.c (alter_reg): Likewise.
10547 * stor-layout.c (mode_for_vector): Likewise.
10548 (layout_type): Likewise.
10550 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10552 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10553 (spu_convert_move): Likewise.
10554 * lower-subreg.c (resolve_simple_move): Likewise.
10556 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10559 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10560 define_insn to a define_expand.
10561 (altivec_vsum2sws_direct): New define_insn.
10562 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10564 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10566 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10569 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10574 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10575 realign stack if stack alignment needed is less than incoming
10578 2017-09-05 Marek Polacek <polacek@redhat.com>
10581 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10584 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10586 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10588 2017-09-05 Richard Biener <rguenther@suse.de>
10590 PR tree-optimization/82084
10591 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10593 2017-09-05 Richard Biener <rguenther@suse.de>
10595 PR tree-optimization/82102
10596 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10598 2017-09-05 Martin Liska <mliska@suse.cz>
10600 PR tree-optimization/82032
10601 * tree-cfg.c (generate_range_test): New function.
10602 * tree-cfg.h (generate_range_test): Declared here.
10603 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10604 (cleanup_control_expr_graph): Use it.
10605 * tree-switch-conversion.c (try_switch_expansion): Remove
10607 (emit_case_nodes): Use generate_range_test.
10609 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10612 * config/i386/i386.md (*<btsc><mode>_mask): Add
10613 TARGET_USE_BT to insn constraint.
10614 (*btr<mode>_mask): Ditto.
10616 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10618 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10619 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10621 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10624 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10625 TARGET_NEON and TARGET_IWMMXT.
10626 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10627 TARGET_NEON and TARGET_IWMMXT.
10628 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10630 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10632 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10633 (ix86_rewrite_tls_address): Ditto.
10634 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10635 (ix86_rewrite_tls_address_1): Ditto.
10636 (ix86_rewrite_tls_address): Ditto.
10637 * config/i386/predicates.md (tls_address_pattern): New predicate.
10638 * config/i386/i386.md (TLS address splitter): New splitter.
10640 2017-09-04 Richard Biener <rguenther@suse.de>
10642 PR tree-optimization/82084
10643 * fold-const.h (can_native_encode_string_p): Declare.
10644 * fold-const.c (can_native_encode_string_p): Factor out from ...
10645 (native_encode_string): ... here.
10646 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10647 vectorizing stores from constants we later cannot handle.
10649 2017-09-04 Marek Polacek <polacek@redhat.com>
10652 * doc/invoke.texi: Update -Wtautological-compare documentation.
10654 2017-09-04 Jeff Law <law@redhat.com>
10656 PR tree-optimization/64910
10657 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10658 swap the first and last operand if the last is a constant.
10660 2017-09-04 Marek Polacek <polacek@redhat.com>
10663 * convert.c (do_narrow): When sanitizing signed integer overflows,
10664 bail out for signed types.
10665 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10667 2017-09-04 Richard Biener <rguenther@suse.de>
10669 PR tree-optimization/82060
10670 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10671 Move devirtualization after stmt folding and before EH/AB/noreturn
10672 cleanup to get the stmt refs canonicalized. Use a bool instead
10673 of gimple_modified_p since that doesn't work for NOPs. Schedule
10674 NOPs generated by folding for removal.
10676 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10677 Alan Hayward <alan.hayward@arm.com>
10678 David Sherwood <david.sherwood@arm.com>
10680 * coretypes.h (pad_direction): New enum.
10681 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10682 (FUNCTION_ARG_PADDING): Likewise.
10683 * target.def (function_arg_padding): New hook.
10684 * targhooks.h (default_function_arg_padding): Declare.
10685 * targhooks.c (default_function_arg_padding): New function.
10686 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10687 (TARGET_FUNCTION_ARG_PADDING): ...this.
10688 * doc/tm.texi: Regenerate.
10689 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10690 instead of direction.
10691 (compute_argument_addresses): Likewise.
10692 (load_register_parameters): Likewise.
10693 (emit_library_call_value_1): Likewise.
10694 (store_one_arg): Use targetm.calls.function_arg_padding instead
10695 of FUNCTION_ARG_PADDING.
10696 (must_pass_in_stack_var_size_or_pad): Likewise.
10697 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10698 (emit_group_store): Likewise.
10699 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10700 instead of FUNCTION_ARG_PADDING.
10701 (emit_push_insn): Likewise, and propagate enum change throughout
10703 * function.h (direction): Delete.
10704 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10706 * function.c (assign_parm_find_stack_rtl): Likewise.
10707 (assign_parm_setup_block_p): Likewise.
10708 (assign_parm_setup_block): Likewise.
10709 (gimplify_parameters): Likewise.
10710 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10711 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10713 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10714 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10715 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10716 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10717 (aarch64_function_arg_padding): ...this new function.
10718 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10719 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10720 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10721 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10722 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10723 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10724 (arm_pad_arg_upward): Replace with...
10725 (arm_function_arg_padding): ...this new function.
10726 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10728 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10729 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10730 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10731 (ia64_hpux_function_arg_padding): Replace with...
10732 (ia64_function_arg_padding): ...this new function. Use pad_direction
10733 instead of direction. Check for TARGET_HPUX.
10734 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10735 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10736 (iq2000_function_arg_padding): New function.
10737 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10738 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10739 (mips_function_arg_padding): ...this new function.
10740 (mips_pad_reg_upward): Update accordingly.
10741 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10742 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10743 targetm.calls.function_arg_padding.
10744 (FUNCTION_ARG_PADDING): Delete.
10745 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10746 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10747 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10748 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10749 (nios2_block_reg_padding): Return pad_direction instead of direction.
10750 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10751 instead of direction.
10752 (nios2_function_arg_padding): Likewise. Make static.
10753 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10754 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10755 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10756 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10757 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10758 (pa_function_arg_padding): Make static. Return pad_direction instead
10760 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10761 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10762 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10763 instead of direction. Use targetm.calls.function_arg_padding.
10764 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10765 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10766 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10767 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10768 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10770 (function_arg_padding): Rename to...
10771 (rs6000_function_arg_padding): ...this. Make static. Return
10772 pad_direction instead of direction.
10773 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10774 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10775 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10776 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10777 instead of direction. Use targetm.calls.function_arg_padding.
10778 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10779 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10780 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10781 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10782 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10783 (function_arg_padding): Rename to...
10784 (rs6000_function_arg_padding): ...this. Make static. Return
10785 pad_direction instead of direction.
10786 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10787 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10788 * config/s390/s390.c (s390_function_arg_padding): New function.
10789 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10790 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10791 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10792 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10793 (function_arg_padding): Rename to...
10794 (sparc_function_arg_padding): ...this. Make static. Return
10795 pad_direction instead of direction.
10796 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10797 * config/spu/spu.c (spu_function_arg_padding): New function.
10798 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10799 * system.h (FUNCTION_ARG_PADDING): Poison.
10801 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10802 Alan Hayward <alan.hayward@arm.com>
10803 David Sherwood <david.sherwood@arm.com>
10805 * target.def (modes_tieable_p): New hook.
10806 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10807 (TARGET_MODES_TIEABLE_P): ...this.
10808 * doc/tm.texi.in: Regenerate.
10809 * hooks.h (hook_bool_mode_mode_true): Declare.
10810 * hooks.c (hook_bool_mode_mode_true): New function.
10811 * combine.c (subst): Use targetm.modes_tieable_p instead of
10813 * dse.c (find_shift_sequence): Likewise.
10814 * expmed.c (extract_low_bits): Likewise.
10815 * lower-subreg.c: Include target.h.
10816 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10818 * rtlanal.c (rtx_cost): Likewise.
10819 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10820 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10821 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10822 (TARGET_MODES_TIEABLE_P): Redefine.
10823 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10824 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10825 (TARGET_MODES_TIEABLE_P): Redefine.
10826 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10827 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10828 (arc_modes_tieable_p): New function.
10829 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10830 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10831 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10832 (arm_modes_tieable_p): Make static.
10833 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10834 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10835 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10836 (TARGET_MODES_TIEABLE_P): Redefine.
10837 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10838 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10839 (TARGET_MODES_TIEABLE_P): Redefine.
10840 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10841 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10842 (cr16_modes_tieable_p): New function.
10843 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10844 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10845 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10846 (TRULY_NOOP_TRUNCATION): Update comment.
10847 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10848 (TRULY_NOOP_TRUNCATION): Update comment.
10849 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10850 (frv_modes_tieable_p): New function.
10851 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10852 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10853 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10854 (TARGET_MODES_TIEABLE_P): Redefine.
10855 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10856 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10857 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10858 (TARGET_MODES_TIEABLE_P): Redefine.
10859 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10860 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10861 (ia64_modes_tieable_p): New function.
10862 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10863 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10864 (iq2000_modes_tieable_p): New function.
10865 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10866 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10867 (lm32_modes_tieable_p): New function.
10868 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10869 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10870 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10871 (TARGET_MODES_TIEABLE_P): Redefine.
10872 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10873 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10874 (m32r_modes_tieable_p): New function.
10875 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10876 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10877 (m68k_modes_tieable_p): New function.
10878 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10879 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10880 (mcore_modes_tieable_p): New function.
10881 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10882 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10884 (TARGET_MODES_TIEABLE_P): Redefine.
10885 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10886 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10887 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10888 (TARGET_MODES_TIEABLE_P): Redefine.
10889 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10890 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10891 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10892 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10893 (mn10300_modes_tieable_p): ...this and make static.
10894 (TARGET_MODES_TIEABLE_P): Redefine.
10895 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10896 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10897 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10898 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10899 (msp430_modes_tieable_p): Make static.
10900 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10901 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10902 (TARGET_MODES_TIEABLE_P): Redefine.
10903 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10904 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10905 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10906 (TARGET_MODES_TIEABLE_P): Redefine.
10907 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10908 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10909 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10910 (TARGET_MODES_TIEABLE_P): Redefine.
10911 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10912 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10913 (pdp11_modes_tieable_p): New function.
10914 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10915 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10916 (rs6000_modes_tieable_p): New function.
10917 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10918 * config/powerpcspe/powerpcspe.md: Update comment.
10919 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10920 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10921 (TARGET_MODES_TIEABLE_P): Redefine.
10922 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10923 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10924 (rl78_modes_tieable_p): New function.
10925 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10926 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10927 (rs6000_modes_tieable_p): New function.
10928 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10929 * config/rs6000/rs6000.md: Update comment.
10930 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10931 * config/rx/rx.c (rx_modes_tieable_p): New function.
10932 (TARGET_MODES_TIEABLE_P): Redefine.
10933 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10934 * config/s390/s390.c (s390_modes_tieable_p): New function.
10935 (TARGET_MODES_TIEABLE_P): Redefine.
10936 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10937 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10938 (sh_modes_tieable_p): New function.
10939 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10940 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10941 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10942 (sparc_modes_tieable_p): Make static.
10943 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10944 * config/spu/spu.c (spu_modes_tieable_p): New function.
10945 (TARGET_MODES_TIEABLE_P): Redefine.
10946 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10947 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10948 (TARGET_MODES_TIEABLE_P): Redefine.
10949 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10950 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10951 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10952 * config/v850/v850.c (v850_modes_tieable_p): New function.
10953 (TARGET_MODES_TIEABLE_P): Redefine.
10954 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10955 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10956 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10957 (visium_modes_tieable_p): New function.
10958 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10959 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10960 (xtensa_modes_tieable_p): New function.
10961 * system.h (MODES_TIEABLE_P): Poison.
10963 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10964 Alan Hayward <alan.hayward@arm.com>
10965 David Sherwood <david.sherwood@arm.com>
10967 * target.def (hard_regno_mode_ok): New hook.
10968 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10969 (TARGET_HARD_REGNO_MODE_OK): ...this.
10970 * doc/tm.texi.in: Regenerate.
10971 * hooks.h (hook_bool_uint_mode_true): Declare.
10972 * hooks.c (hook_bool_uint_mode_true): New function.
10973 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10974 HARD_REGNO_MODE_OK.
10975 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10976 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10977 instead of HARD_REGNO_MODE_OK.
10978 * caller-save.c: Include target.h.
10979 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10980 HARD_REGNO_MODE_OK.
10981 * combine.c (can_combine_p): Likewise.
10982 (combinable_i3pat): Likewise.
10983 (can_change_dest_mode): Likewise.
10984 * expr.c (init_expr_target): Likewise.
10985 (convert_move): Likewise.
10986 (convert_modes): Likewise.
10987 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10988 (setup_prohibited_mode_move_regs): Likewise.
10989 * ira.h (target_ira): Likewise.
10990 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10991 * lra-constraints.c (process_alt_operands): Likewise.
10992 (split_reg): Likewise.
10993 * recog.c (peep2_find_free_register): Likewise.
10994 * ree.c (combine_reaching_defs): Likewise.
10995 * regcprop.c (maybe_mode_change): Likewise.
10996 * reginfo.c (init_reg_sets_1): Likewise.
10997 (choose_hard_reg_mode): Likewise.
10998 (simplifiable_subregs): Likewise.
10999 * regrename.c (check_new_reg_p): Likewise.
11000 * reload.c (find_valid_class): Likewise.
11001 (find_valid_class_1): Likewise.
11002 (reload_inner_reg_of_subreg): Likewise.
11003 (push_reload): Likewise.
11004 (combine_reloads): Likewise.
11005 (find_dummy_reload): Likewise.
11006 (find_reloads): Likewise.
11007 * reload1.c (find_reg): Likewise.
11008 (set_reload_reg): Likewise.
11009 (allocate_reload_reg): Likewise.
11010 (choose_reload_regs): Likewise.
11011 (reload_adjust_reg_for_temp): Likewise.
11012 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11013 (simplify_subreg_regno): Likewise.
11014 * sel-sched.c (init_regs_for_mode): Likewise.
11015 * varasm.c (make_decl_rtl): Likewise.
11016 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11017 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11018 HARD_REGNO_MODE_OK.
11019 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11020 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11021 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11022 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11023 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11025 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11026 (arc_mode_class): Delete.
11027 (HARD_REGNO_MODE_OK): Delete.
11028 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11029 (arc_hard_regno_mode_ok): Rename old array to...
11030 (arc_hard_regno_mode_ok_modes): ...this.
11031 (arc_conditional_register_usage): Update accordingly.
11032 (arc_mode_class): Make static.
11033 (arc_hard_regno_mode_ok): New function.
11034 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11035 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11036 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11037 (arm_hard_regno_mode_ok): Make static.
11038 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11039 HARD_REGNO_MODE_OK.
11040 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11041 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11042 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11044 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11045 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11046 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11047 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11048 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11049 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11050 * config/bfin/predicates.md (valid_reg_operand): Use
11051 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11052 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11053 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11054 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11055 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11056 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11057 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11058 (cr16_hard_regno_mode_ok): Make static and return a bool.
11059 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11060 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11061 (cris_hard_regno_mode_ok): New function.
11062 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11063 (epiphany_mode_class): Delete.
11064 (HARD_REGNO_MODE_OK): Delete.
11065 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11066 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11067 (hard_regno_mode_ok): Rename to...
11068 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11069 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11070 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11071 HARD_REGNO_MODE_OK.
11072 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11073 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11074 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11075 (frv_hard_regno_mode_ok): Make static and return a bool.
11076 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11077 HARD_REGNO_MODE_OK.
11078 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11079 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11080 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11081 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11083 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11084 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11085 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11086 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11088 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11089 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11090 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11091 (ia64_hard_regno_mode_ok): New function.
11092 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11093 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11094 (iq2000_hard_regno_mode_ok): New function.
11095 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11096 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11097 (lm32_hard_regno_mode_ok): New function.
11098 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11099 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11100 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11101 instead of HARD_REGNO_MODE_OK.
11102 (m32c_hard_regno_ok): Rename to...
11103 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11104 (m32c_cannot_change_mode_class): Update accordingly.
11105 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11106 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11107 (m32r_mode_class): Delete.
11108 (HARD_REGNO_MODE_OK): Delete.
11109 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11110 (m32r_hard_regno_mode_ok): Rename to...
11111 (m32r_hard_regno_modes): ...this.
11112 (m32r_mode_class): Make static.
11113 (m32r_hard_regno_mode_ok): New function.
11114 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11115 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11116 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11117 (m68k_hard_regno_mode_ok): Make static.
11118 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11119 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11120 (mcore_hard_regno_mode_ok): New function.
11121 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11122 (HARD_REGNO_MODE_OK): Delete.
11123 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11125 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11126 (microblaze_hard_regno_mode_ok): New function.
11127 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11128 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11129 (mips_hard_regno_mode_ok): Delete.
11130 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11131 (mips_hard_regno_mode_ok_p): ...this and make static.
11132 (mips_hard_regno_mode_ok_p): Rename to...
11133 (mips_hard_regno_mode_ok_uncached): ...this.
11134 (mips_hard_regno_mode_ok): New function.
11135 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11136 of HARD_REGNO_MODE_OK.
11137 (mips_option_override): Update after above name changes.
11138 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11139 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11140 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11141 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11142 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11143 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11144 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11145 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11146 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11147 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11148 (msp430_hard_regno_mode_ok): Make static and return a bool.
11149 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11150 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11151 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11154 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11155 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11156 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11157 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11158 (PA_HARD_REGNO_MODE_OK): ...this
11159 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11160 (PA_HARD_REGNO_MODE_OK): ...this.
11161 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11162 (pa_hard_regno_mode_ok): New function.
11163 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11164 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11165 (pdp11_hard_regno_mode_ok): New function.
11166 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11167 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11169 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11171 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11172 (rs6000_hard_regno_mode_ok): Rename to...
11173 (rs6000_hard_regno_mode_ok_uncached): ...this.
11174 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11175 (rs6000_hard_regno_mode_ok): New function.
11176 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11177 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11178 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11179 (riscv_hard_regno_mode_ok): ...this and make static.
11180 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11181 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11182 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11183 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11184 (rl78_hard_regno_mode_ok): Make static and return bool.
11185 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11186 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11188 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11190 (rs6000_hard_regno_mode_ok): Rename to...
11191 (rs6000_hard_regno_mode_ok_uncached): ...this.
11192 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11193 (rs6000_hard_regno_mode_ok): New function.
11194 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11195 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11196 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11197 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11198 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11199 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11200 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11201 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11202 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11203 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11204 (sh_hard_regno_mode_ok): Make static.
11205 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11206 instead of HARD_REGNO_MODE_OK.
11207 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11208 (sparc_mode_class): Delete.
11209 (HARD_REGNO_MODE_OK): Delete.
11210 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11211 (hard_regno_mode_classes): Make static.
11212 (sparc_mode_class): Likewise.
11213 (sparc_hard_regno_mode_ok): New function.
11214 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11215 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11216 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11218 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11219 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11220 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11221 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11222 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11223 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11224 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11225 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11226 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11227 (visium_hard_regno_mode_ok): New function.
11228 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11229 instead of HARD_REGNO_MODE_OK.
11230 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11231 (HARD_REGNO_MODE_OK): Delete.
11232 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11233 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11234 (xtensa_option_override): Update accordingly.
11235 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11236 (xtensa_hard_regno_mode_ok): New function.
11237 * system.h (HARD_REGNO_MODE_OK): Poison.
11239 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11240 Alan Hayward <alan.hayward@arm.com>
11241 David Sherwood <david.sherwood@arm.com>
11243 * target.def (hard_regno_call_part_clobbered): New hook.
11244 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11245 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11246 * doc/tm.texi: Regenerate.
11247 * hooks.h (hook_bool_uint_mode_false): Declare.
11248 * hooks.c (hook_bool_uint_mode_false): New function.
11249 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11250 * cselib.c (cselib_process_insn): Use
11251 targetm.hard_regno_call_part_clobbered instead of
11252 HARD_REGNO_CALL_PART_CLOBBERED.
11253 * ira-conflicts.c (ira_build_conflicts): Likewise.
11254 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11255 * lra-constraints.c (need_for_call_save_p): Likewise.
11256 * lra-lives.c: Include target.h.
11257 (check_pseudos_live_through_calls): Use
11258 targetm.hard_regno_call_part_clobbered instead of
11259 HARD_REGNO_CALL_PART_CLOBBERED.
11260 * regcprop.c: Include target.h.
11261 (copyprop_hardreg_forward_1): Use
11262 targetm.hard_regno_call_part_clobbered instead of
11263 HARD_REGNO_CALL_PART_CLOBBERED.
11264 * reginfo.c (choose_hard_reg_mode): Likewise.
11265 * regrename.c (check_new_reg_p): Likewise.
11266 * reload.c (find_equiv_reg): Likewise.
11267 * reload1.c (emit_reload_insns): Likewise.
11268 * sched-deps.c (deps_analyze_insn): Likewise.
11269 * sel-sched.c (init_regs_for_mode): Likewise.
11270 (mark_unavailable_hard_regs): Likewise.
11271 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11272 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11273 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11275 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11276 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11277 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11279 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11281 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11282 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11283 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11285 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11286 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11287 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11289 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11290 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11292 * config/powerpcspe/powerpcspe.c
11293 (rs6000_hard_regno_call_part_clobbered): New function.
11294 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11295 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11296 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11298 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11299 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11300 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11302 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11303 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11304 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11306 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11307 Alan Hayward <alan.hayward@arm.com>
11308 David Sherwood <david.sherwood@arm.com>
11310 * rtl.h (subreg_memory_offset): Declare.
11311 * emit-rtl.c (subreg_memory_offset): New function.
11312 * expmed.c (store_bit_field_1): Use it.
11313 * expr.c (undefined_operand_subword_p): Likewise.
11314 * simplify-rtx.c (simplify_subreg): Likewise.
11316 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11318 PR rtl-optimization/57448
11321 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11322 using atomic_load pattern.
11323 (expand_atomic_store): Likewise.
11325 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11328 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11329 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11330 handling. Use replace_call_with_value with NULL instead of
11331 gsi_replace, unlink_stmt_vdef and release_defs.
11333 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11336 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11338 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11341 * rtl.h (emit_library_call_value_1): Declare.
11342 (emit_library_call): Replace declaration with a series of overloads.
11343 Remove the parameter count argument.
11344 (emit_library_call_value): Likewise.
11345 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11346 with an "rtx_mode_t *".
11347 (emit_library_call_value): Delete.
11348 (emit_library_call): Likewise.
11349 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11350 (asan_emit_allocas_unpoison): Likewise.
11351 * builtins.c (expand_builtin_powi): Likewise.
11352 (expand_asan_emit_allocas_unpoison): Likewise.
11353 * cfgexpand.c (expand_main_function): Likewise.
11354 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11355 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11356 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11357 * config/arm/arm.c (arm_trampoline_init): Likewise.
11358 (arm_call_tls_get_addr): Likewise.
11359 (arm_expand_divmod_libfunc): Likewise.
11360 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11361 (smulsi3_highpart): Likewise.
11362 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11363 (c6x_expand_compare): Likewise.
11364 (c6x_expand_movmem): Likewise.
11365 * config/frv/frv.c (frv_trampoline_init): Likewise.
11366 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11367 (ix86_expand_divmod_libfunc): Likewise.
11368 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11369 (ia64_expand_compare): Likewise.
11370 (ia64_profile_hook): Likewise.
11371 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11372 (nonlocal_goto): Likewise.
11373 (restore_stack_nonlocal): Likewise.
11374 * config/m32r/m32r.c (block_move_call): Likewise.
11375 (m32r_trampoline_init): Likewise.
11376 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11377 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11378 (m68k_call_m68k_read_tp): Likewise.
11379 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11380 (microblaze_expand_divide): Likewise.
11381 * config/mips/mips.h (mips_args): Likewise.
11382 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11383 (MIPS_ICACHE_SYNC): Likewise.
11384 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11385 (nios2_trampoline_init): Likewise.
11386 * config/pa/pa.c (hppa_tls_call): Likewise.
11387 (pa_trampoline_init): Likewise.
11388 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11389 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11390 (expand_strn_compare): Likewise.
11391 (rs6000_generate_compare): Likewise.
11392 (rs6000_expand_float128_convert): Likewise.
11393 (output_profile_hook): Likewise.
11394 (rs6000_trampoline_init): Likewise.
11395 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11396 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11397 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11398 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11399 (rs6000_generate_compare): Likewise.
11400 (rs6000_expand_float128_convert): Likewise.
11401 (output_profile_hook): Likewise.
11402 (rs6000_trampoline_init): Likewise.
11403 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11404 * config/sh/sh.c (sh_trampoline_init): Likewise.
11405 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11406 (sparc_emit_float_lib_cmp): Likewise.
11407 (sparc32_initialize_trampoline): Likewise.
11408 (sparc64_initialize_trampoline): Likewise.
11409 (sparc_profile_hook): Likewise.
11410 * config/spu/spu.c (ea_load_store): Likewise.
11411 * config/spu/spu.md (floatunssidf2): Likewise.
11412 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11413 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11414 * config/visium/visium.c (expand_block_move_4): Likewise.
11415 (expand_block_move_2): Likewise.
11416 (expand_block_move_1): Likewise.
11417 (expand_block_set_4): Likewise.
11418 (expand_block_set_2): Likewise.
11419 (expand_block_set_1): Likewise.
11420 (visium_trampoline_init): Likewise.
11421 (visium_profile_hook): Likewise.
11422 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11423 (xtensa_setup_frame_addresses): Likewise.
11424 (xtensa_trampoline_init): Likewise.
11425 * except.c (sjlj_emit_function_enter): Likewise.
11426 (sjlj_emit_function_exit): Likewise.
11427 * explow.c (allocate_dynamic_stack_space): Likewise.
11428 (probe_stack_range): Likewise.
11429 * expr.c (convert_mode_scalar): Likewise.
11430 * optabs.c (expand_binop): Likewise.
11431 (expand_twoval_binop_libfunc): Likewise.
11432 (expand_unop): Likewise.
11433 (prepare_cmp_insn): Likewise.
11434 (prepare_float_lib_cmp): Likewise.
11435 (expand_float): Likewise.
11436 (expand_fix): Likewise.
11437 (expand_fixed_convert): Likewise.
11438 (maybe_emit_sync_lock_test_and_set): Likewise.
11439 (expand_atomic_compare_and_swap): Likewise.
11440 (expand_mem_thread_fence): Likewise.
11441 (expand_atomic_fetch_op): Likewise.
11443 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11445 * doc/generic.texi (OpenACC): Adjust URL.
11446 * doc/invoke.texi (C Dialect Options): Ditto.
11448 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11450 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11451 predicate for operand 1. Add (m,<S>) constraint.
11452 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11453 Prevent memory operand 1 with register operand 2.
11455 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11457 PR rtl-optimization/82024
11458 * combine.c (try_combine): If the combination result is a PARALLEL,
11459 and we only need to retain the SET in there that would be placed
11460 at I2, check that we can place that at I3 instead, before doing so.
11462 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11465 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11467 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11468 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11471 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11472 Jeff Law <law@redhat.com>
11474 * varasm.c (bss_initializer_p): Do not put constants into .bss
11475 (categorize_decl_for_section): Handle bss_initializer_p returning
11476 false when DECL_INITIAL is NULL.
11478 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11481 * config/s390/s390.c (s390_can_inline_p): New function.
11483 2017-09-01 Jeff Law <law@redhat.com>
11485 PR tree-optimization/82052
11486 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11487 Always initialize the returned slot after a hash table miss
11488 when INSERT is true.
11490 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11492 * config/s390/s390.md (mem_signal_fence): Remove.
11493 * doc/md.texi (mem_signal_fence): Remove.
11494 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11496 * target-insns.def (mem_signal_fence): Remove.
11498 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11501 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11504 * asan.c (create_odr_indicator): Strip name encoding from assembler
11505 name before appending it after __odr_asan_.
11507 2017-09-01 Martin Liska <mliska@suse.cz>
11509 PR tree-optimization/82059
11510 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11511 frequency only when an edge is redirected.
11513 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11515 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11516 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11517 (arc_conditional_register_usage): Remove ARC600 lp_count
11519 (arc_file_start): Emit Tag_ARC_CPU_variation.
11520 (arc_can_use_doloop_p): New conditions to use ZOLs.
11521 (hwloop_fail): New function.
11522 (hwloop_optimize): Likewise.
11523 (hwloop_pattern_reg): Likewise.
11524 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11525 (arc_reorg_loops): New function, calls reorg_loops.
11526 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11527 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11529 (arc_loop_hazard): Remove function, functionality moved into
11531 (arc_hazard): Remove arc_loop_hazard call.
11532 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11533 into hwloop_optimize.
11534 (arc_label_align): Remove ZOL handling.
11535 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11536 * config/arc/arc.md (doloop_begin): Remove pattern.
11537 (doloop_begin_i): Likewise.
11538 (doloop_end_i): Likewise.
11539 (doloop_fallback): Likewise.
11540 (doloop_fallback_m): Likewise.
11541 (doloop_end): Reimplement expand.
11542 (arc_lp): New pattern for LP instruction.
11543 (loop_end): New pattern.
11544 (loop_fail): Likewise.
11545 (decrement_and_branch_until_zero): Likewise.
11546 * config/arc/arc.opt (mlpc-width): New option.
11547 * doc/invoke.texi (mlpc-width): Document option.
11549 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11551 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11552 (arc_ccfsm_advance): Fix checking for delay slots.
11553 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11555 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11557 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11559 (movhi_insn): Update store instruction constraint which are saving
11560 6-bit short immediates.
11561 (movsi_insn): Consider also short scaled load operations.
11562 (zero_extendhisi2_i): Use Usd constraint instead of T.
11563 (extendhisi2_i): Add q constraint.
11564 (arc_clzsi2): Add type and length attributes.
11565 (arc_ctzsi2): Likewise.
11566 * config/arc/constraints.md (Usc): Update constraint, the
11567 assembler can parse two relocations for a single instruction.
11569 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11571 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11572 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11574 2017-08-31 Olivier Hainque <hainque@adacore.com>
11576 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11577 match as powerpc-wrs-vxworks*.
11579 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11581 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11582 register constraint for by-element operand.
11583 (aarch64_mls_elt_merge<mode>): Likewise.
11585 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11587 * config/arc/arc.c (arc_can_follow_jump): Check for short
11590 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11592 * config.gcc: Use g.opt for arc.
11593 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11594 functionality moved to ...
11595 (legitimate_scaled_address_p): New function, ...here.
11596 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11597 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11598 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11600 (arc_override_options): Handle G option.
11601 (arc_output_pic_addr_const): Correct function definition.
11602 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11603 (arc_decl_anon_ns_mem_p): Delete.
11604 (arc_in_small_data_p): Overhaul this function to take into
11605 consideration the value given via G option.
11606 (arc_rewrite_small_data_1): Renamed and corrected old
11607 arc_rewrite_small_data function.
11608 (arc_rewrite_small_data): New function.
11609 (small_data_pattern): Don't use pic_offset_table_rtx.
11610 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11611 * config/arc/simdext.md (movmisalignv2hi): Use
11612 prepare_move_operands function.
11614 (movmisalign*): Likewise.
11615 * doc/invoke.texi (ARC options): Document -G option.
11617 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11619 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11621 * config/arc/arc.c (arc_print_operand): Output scalled address for
11622 sdata whenever is possible.
11623 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11624 load/stores are available.
11625 (compact_sda_memory_operand): Check for the alignment required by
11626 code density instructions.
11627 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11629 * config/arc/constraints.md (Usd): Update constraint.
11630 (Us0): New constraint.
11631 (Usc): Update constraint.
11633 2017-08-31 Richard Biener <rguenther@suse.de>
11635 PR middle-end/82054
11636 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11637 function only once.
11639 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11641 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11642 Resize type_signature.
11644 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11645 Alan Hayward <alan.hayward@arm.com>
11646 David Sherwood <david.sherwood@arm.com>
11648 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11649 subregs whose inner modes can be stored in GPRs.
11650 (aarch64_classify_index): Likewise.
11652 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11653 Alan Hayward <alan.hayward@arm.com>
11654 David Sherwood <david.sherwood@arm.com>
11656 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11657 (V_INT_EQUIV): ...this.
11658 (v_cmp_result): Rename to...
11659 (v_int_equiv): ...this.
11660 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11661 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11662 (copysign<mode>3): Likewise.
11663 (aarch64_simd_bsl<mode>_internal): Likewise.
11664 (aarch64_simd_bsl<mode>): Likewise.
11665 (vec_cmp<mode><mode>): Likewise.
11666 (vcond<mode><mode>): Likewise.
11667 (vcond<v_cmp_mixed><mode>): Likewise.
11668 (vcondu<mode><v_cmp_mixed>): Likewise.
11669 (aarch64_cm<optab><mode>): Likewise.
11670 (aarch64_cmtst<mode>): Likewise.
11671 (aarch64_fac<optab><mode>): Likewise.
11672 (vec_perm_const<mode>): Likewise.
11673 (vcond_mask_<mode><v_cmp_result>): Rename to...
11674 (vcond_mask_<mode><v_int_equiv>): ...this.
11675 (vec_cmp<mode><v_cmp_result>): Rename to...
11676 (vec_cmp<mode><v_int_equiv>): ...this.
11678 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11679 Alan Hayward <alan.hayward@arm.com>
11680 David Sherwood <david.sherwood@arm.com>
11682 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11684 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11685 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11686 (UNSPEC_LD4_DREG): New unspecs.
11687 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11688 (aarch64_ld2<mode>_dreg_be): Replace with...
11689 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11691 (aarch64_ld3<mode>_dreg_le)
11692 (aarch64_ld3<mode>_dreg_be): Replace with...
11693 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11695 (aarch64_ld4<mode>_dreg_le)
11696 (aarch64_ld4<mode>_dreg_be): Replace with...
11697 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11700 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11702 PR tree-optimization/81987
11703 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11704 insert an initializer in a location not dominated by the stride
11707 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11709 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11710 on the entire header of the finally block in the fallthru case.
11712 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11714 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11716 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11718 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11719 rs6000_emit_move_from_cr and call renamed function.
11720 (rs6000_emit_prologue): Call renamed functions.
11721 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11722 movesi_from_cr, remove volatile CRs.
11724 2017-08-30 Jon Beniston <jon@beniston.com>
11725 Richard Biener <rguenther@suse.de>
11727 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11728 of VECTOR_MODE_P check.
11729 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11730 element vector types.
11732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11734 * df.h (df_read_modify_subreg_p): Remove in favor of...
11735 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11736 const_rtx instead of an rtx.
11737 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11738 * df-problems.c (df_word_lr_mark_ref): Likewise.
11739 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11740 (mark_pseudo_reg_dead): Likewise.
11741 (mark_ref_dead): Likewise.
11742 * reginfo.c (init_subregs_of_mode): Likewise.
11743 * sched-deps.c (sched_analyze_1): Likewise.
11744 * df-scan.c (df_def_record_1): Likewise.
11745 (df_uses_record): Likewise.
11746 (df_read_modify_subreg_p): Remove in favor of...
11747 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11748 const_rtx instead of an rtx.
11750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11751 Alan Hayward <alan.hayward@arm.com>
11752 David Sherwood <david.sherwood@arm.com>
11754 * rtl.h (partial_subreg_p): New function.
11755 * caller-save.c (save_call_clobbered_regs): Use it.
11756 * calls.c (expand_call): Likewise.
11757 * combine.c (combinable_i3pat): Likewise.
11758 (simplify_set): Likewise.
11759 (make_extraction): Likewise.
11760 (make_compound_operation_int): Likewise.
11761 (gen_lowpart_or_truncate): Likewise.
11762 (force_to_mode): Likewise.
11763 (make_field_assignment): Likewise.
11764 (reg_truncated_to_mode): Likewise.
11765 (record_truncated_value): Likewise.
11766 (move_deaths): Likewise.
11767 * cse.c (record_jump_cond): Likewise.
11768 (cse_insn): Likewise.
11769 * cselib.c (cselib_lookup_1): Likewise.
11770 * expmed.c (extract_bit_field_using_extv): Likewise.
11771 * function.c (assign_parm_setup_reg): Likewise.
11772 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11773 * ira-build.c (create_insn_allocnos): Likewise.
11774 * lra-coalesce.c (merge_pseudos): Likewise.
11775 * lra-constraints.c (match_reload): Likewise.
11776 (simplify_operand_subreg): Likewise.
11777 (curr_insn_transform): Likewise.
11778 * lra-lives.c (process_bb_lives): Likewise.
11779 * lra.c (new_insn_reg): Likewise.
11780 (lra_substitute_pseudo): Likewise.
11781 * regcprop.c (mode_change_ok): Likewise.
11782 (maybe_mode_change): Likewise.
11783 (copyprop_hardreg_forward_1): Likewise.
11784 * reload.c (push_reload): Likewise.
11785 (find_reloads): Likewise.
11786 (find_reloads_subreg_address): Likewise.
11787 * reload1.c (alter_reg): Likewise.
11788 (eliminate_regs_1): Likewise.
11789 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11791 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11793 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11794 back to if statements, including unpack.
11796 2017-08-30 Martin Liska <mliska@suse.cz>
11798 PR inline-asm/82001
11799 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11801 (func_checker::compare_asm_inputs_outputs): ... this function.
11802 (func_checker::compare_gimple_asm): Use the function to compare
11803 also ASM constrains.
11804 * ipa-icf-gimple.h: Rename the function.
11806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11807 Alan Hayward <alan.hayward@arm.com>
11808 David Sherwood <david.sherwood@arm.com>
11810 * coretypes.h (complex_mode): New type.
11811 * gdbhooks.py (build_pretty_printer): Handle it.
11812 * machmode.h (complex_mode): New class.
11813 (complex_mode::includes_p): New function.
11814 (is_complex_int_mode): Likewise.
11815 (is_complex_float_mode): Likewise.
11816 * genmodes.c (get_mode_class): Handle complex mode classes.
11817 * function.c (expand_function_end): Use is_complex_int_mode.
11819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11820 Alan Hayward <alan.hayward@arm.com>
11821 David Sherwood <david.sherwood@arm.com>
11823 * coretypes.h (scalar_mode_pod): New typedef.
11824 * gdbhooks.py (build_pretty_printer): Handle it.
11825 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11826 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11827 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11828 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11829 as_a <scalar_mode>.
11831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11832 Alan Hayward <alan.hayward@arm.com>
11833 David Sherwood <david.sherwood@arm.com>
11835 * machmode.h (mode_for_vector): Take a scalar_mode instead
11837 * stor-layout.c (mode_for_vector): Likewise.
11838 * explow.c (promote_mode): Use as_a <scalar_mode>.
11839 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11842 Alan Hayward <alan.hayward@arm.com>
11843 David Sherwood <david.sherwood@arm.com>
11845 * target.def (preferred_simd_mode): Take a scalar_mode
11846 instead of a machine_mode.
11847 * targhooks.h (default_preferred_simd_mode): Likewise.
11848 * targhooks.c (default_preferred_simd_mode): Likewise.
11849 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11850 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11851 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11852 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11853 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11854 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11855 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11856 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11858 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11859 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11860 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11861 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11862 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11863 * doc/tm.texi: Regenerate.
11864 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11868 Alan Hayward <alan.hayward@arm.com>
11869 David Sherwood <david.sherwood@arm.com>
11871 * target.def (scalar_mode_supported_p): Take a scalar_mode
11872 instead of a machine_mode.
11873 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11874 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11875 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11876 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11877 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11878 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11879 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11880 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11881 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11882 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11883 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11884 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11885 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11886 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11888 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11889 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11890 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11891 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11892 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11894 * doc/tm.texi: Regenerate.
11896 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11897 Alan Hayward <alan.hayward@arm.com>
11898 David Sherwood <david.sherwood@arm.com>
11900 * coretypes.h (opt_scalar_mode): New typedef.
11901 * gdbhooks.py (build_pretty_printers): Handle it.
11902 * machmode.h (mode_iterator::get_2xwider): Add overload for
11904 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11906 * expr.c (convert_mode_scalar): Likewise.
11907 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11908 * optabs.c (expand_float): Likewise.
11909 (expand_fix): Likewise.
11910 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11913 Alan Hayward <alan.hayward@arm.com>
11914 David Sherwood <david.sherwood@arm.com>
11916 * optabs.c (expand_float): Explicitly check for scalars before
11917 using a branching expansion.
11918 (expand_fix): Likewise.
11920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11921 Alan Hayward <alan.hayward@arm.com>
11922 David Sherwood <david.sherwood@arm.com>
11924 * expr.c (convert_mode): Split scalar handling out into...
11925 (convert_mode_scalar): ...this new function. Treat the modes
11928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11929 Alan Hayward <alan.hayward@arm.com>
11930 David Sherwood <david.sherwood@arm.com>
11932 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11934 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11937 Alan Hayward <alan.hayward@arm.com>
11938 David Sherwood <david.sherwood@arm.com>
11940 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11941 rather than a machine_mode.
11942 (fixed_from_string): Likewise.
11943 (fixed_convert): Likewise.
11944 (fixed_convert_from_int): Likewise.
11945 (fixed_convert_from_real): Likewise.
11946 (real_convert_from_fixed): Likewise.
11947 * fixed-value.c (fixed_from_double_int): Likewise.
11948 (fixed_from_string): Likewise.
11949 (fixed_convert): Likewise.
11950 (fixed_convert_from_int): Likewise.
11951 (fixed_convert_from_real): Likewise.
11952 (real_convert_from_fixed): Likewise.
11953 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11956 Alan Hayward <alan.hayward@arm.com>
11957 David Sherwood <david.sherwood@arm.com>
11959 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11960 of separate mode class checks. Do not allow vector modes here.
11961 (immed_wide_int_const): Use as_a <scalar_mode>.
11962 * explow.c (trunc_int_for_mode): Likewise.
11963 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11964 (wi::shwi): Likewise.
11965 (wi::min_value): Likewise.
11966 (wi::max_value): Likewise.
11967 * dwarf2out.c (loc_descriptor): Likewise.
11968 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11969 for CONST_WIDE_INT.
11971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11972 Alan Hayward <alan.hayward@arm.com>
11973 David Sherwood <david.sherwood@arm.com>
11975 * tree.h (SCALAR_TYPE_MODE): New macro.
11976 * expr.c (expand_expr_addr_expr_1): Use it.
11977 (expand_expr_real_2): Likewise.
11978 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11979 (fold_convert_const_fixed_from_int): Likewise.
11980 (fold_convert_const_fixed_from_real): Likewise.
11981 (native_encode_fixed): Likewise
11982 (native_encode_complex): Likewise
11983 (native_encode_vector): Likewise.
11984 (native_interpret_fixed): Likewise.
11985 (native_interpret_real): Likewise.
11986 (native_interpret_complex): Likewise.
11987 (native_interpret_vector): Likewise.
11988 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11989 (simd_clone_adjust_argument_types): Likewise.
11990 (simd_clone_init_simd_arrays): Likewise.
11991 (simd_clone_adjust): Likewise.
11992 * stor-layout.c (layout_type): Likewise.
11993 * tree.c (build_minus_one_cst): Likewise.
11994 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11995 * tree-inline.c (estimate_move_cost): Likewise.
11996 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11997 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11998 (vectorizable_reduction): Likewise.
11999 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12000 (vect_recog_mixed_size_cond_pattern): Likewise.
12001 (check_bool_pattern): Likewise.
12002 (adjust_bool_pattern): Likewise.
12003 (search_type_for_mask_1): Likewise.
12004 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12005 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12006 (vectorizable_load): Likewise.
12007 (vectorizable_store): Likewise.
12008 * ubsan.c (ubsan_encode_value): Likewise.
12009 * varasm.c (output_constant): Likewise.
12011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12012 Alan Hayward <alan.hayward@arm.com>
12013 David Sherwood <david.sherwood@arm.com>
12015 * coretypes.h (scalar_mode): New class.
12016 * machmode.h (scalar_mode): Likewise.
12017 (scalar_mode::includes_p): New function.
12018 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12019 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12020 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12021 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12022 * expmed.c (store_bit_field_1): Likewise.
12023 (extract_bit_field_1): Likewise.
12024 * expr.c (write_complex_part): Likewise.
12025 (read_complex_part): Likewise.
12026 (emit_move_complex_push): Likewise.
12027 (expand_expr_real_2): Likewise.
12028 * function.c (assign_parm_setup_reg): Likewise.
12029 (assign_parms_unsplit_complex): Likewise.
12030 * optabs.c (expand_binop): Likewise.
12031 * rtlanal.c (subreg_get_info): Likewise.
12032 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12033 * varasm.c (output_constant_pool_2): Likewise.
12035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12036 Alan Hayward <alan.hayward@arm.com>
12037 David Sherwood <david.sherwood@arm.com>
12039 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12041 (expmed_mult_highpart_optab): Likewise.
12042 (expmed_mult_highpart): Likewise.
12044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12045 Alan Hayward <alan.hayward@arm.com>
12046 David Sherwood <david.sherwood@arm.com>
12048 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12049 instead of a machine_mode.
12050 (builtin_memset_read_str): Likewise.
12051 * builtins.c (c_readstr): Likewise.
12052 (builtin_memcpy_read_str): Likewise.
12053 (builtin_strncpy_read_str): Likewise.
12054 (builtin_memset_read_str): Likewise.
12055 (builtin_memset_gen_str): Likewise.
12056 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12057 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12058 instead of a machine_mode.
12059 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12061 (make_extraction): Likewise.
12062 (try_widen_shift_mode): Take and return scalar_int_modes instead
12064 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12065 a scalar_int_mode instead of a machine_mode.
12066 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12067 (avr_addr_space_pointer_mode): Likewise.
12068 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12069 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12070 (msp430_unwind_word_mode): Likewise.
12071 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12072 (spu_addr_space_pointer_mode): Likewise.
12073 (spu_addr_space_address_mode): Likewise.
12074 (spu_libgcc_cmp_return_mode): Likewise.
12075 (spu_libgcc_shift_count_mode): Likewise.
12076 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12077 (rl78_addr_space_pointer_mode): Likewise.
12078 (fl78_unwind_word_mode): Likewise.
12079 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12081 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12082 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12083 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12084 (mips_valid_pointer_mode): Likewise.
12085 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12086 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12087 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12089 (ft32_addr_space_address_mode): Likewise.
12090 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12091 scalar_int_mode instead of a machine_mode.
12092 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12094 (m32c_addr_space_address_mode): Likewise.
12095 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12096 (rs6000_eh_return_filter_mode): Likewise.
12097 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12098 (rs6000_eh_return_filter_mode): Likewise.
12099 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12100 (s390_libgcc_shift_count_mode): Likewise.
12101 (s390_unwind_word_mode): Likewise.
12102 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12104 * target.def (mode_rep_extended): Likewise.
12105 (valid_pointer_mode): Likewise.
12106 (addr_space.valid_pointer_mode): Likewise.
12107 (eh_return_filter_mode): Return a scalar_int_mode rather than
12109 (libgcc_cmp_return_mode): Likewise.
12110 (libgcc_shift_count_mode): Likewise.
12111 (unwind_word_mode): Likewise.
12112 (addr_space.pointer_mode): Likewise.
12113 (addr_space.address_mode): Likewise.
12114 * doc/tm.texi: Regenerate.
12115 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12117 (do_jump): Use scalar_int_mode for local variables.
12118 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12119 rather than a machine_mode.
12120 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12121 (scompare_loc_descriptor_wide): Likewise.
12122 (scompare_loc_descriptor_narrow): Likewise.
12123 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12125 * except.c (sjlj_emit_dispatch_table): Likewise.
12126 (expand_builtin_eh_copy_values): Likewise.
12127 * explow.c (convert_memory_address_addr_space_1): Likewise.
12128 Take a scalar_int_mode rather than a machine_mode.
12129 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12130 than a machine_mode.
12131 (memory_address_addr_space): Use scalar_int_mode for local variables.
12132 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12133 rather than a machine_mode.
12134 * expmed.c (mask_rtx): Likewise.
12135 (init_expmed_one_conv): Likewise.
12136 (expand_mult_highpart_adjust): Likewise.
12137 (extract_high_half): Likewise.
12138 (expmed_mult_highpart_optab): Likewise.
12139 (expmed_mult_highpart): Likewise.
12140 (expand_smod_pow2): Likewise.
12141 (expand_sdiv_pow2): Likewise.
12142 (emit_store_flag_int): Likewise.
12143 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12145 (extract_low_bits): Likewise.
12146 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12148 * expr.c (pieces_addr::adjust): Likewise.
12149 (can_store_by_pieces): Likewise.
12150 (store_by_pieces): Likewise.
12151 (clear_by_pieces_1): Likewise.
12152 (expand_expr_addr_expr_1): Likewise.
12153 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12154 (expand_expr_real_1): Likewise.
12155 (try_casesi): Likewise.
12156 * final.c (shorten_branches): Likewise.
12157 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12158 type of "mode" to machine_mode.
12159 * internal-fn.c (expand_arith_overflow_result_store): Take a
12160 scalar_int_mode rather than a machine_mode.
12161 (expand_mul_overflow): Use scalar_int_mode for local variables.
12162 * loop-doloop.c (doloop_modify): Likewise.
12163 (doloop_optimize): Likewise.
12164 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12165 than a machine_mode.
12166 (expand_doubleword_shift_condmove): Likewise.
12167 (expand_doubleword_shift): Likewise.
12168 (expand_doubleword_clz): Likewise.
12169 (expand_doubleword_popcount): Likewise.
12170 (expand_doubleword_parity): Likewise.
12171 (expand_absneg_bit): Use scalar_int_mode for local variables.
12172 (prepare_float_lib_cmp): Likewise.
12173 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12174 rather than a machine_mode.
12175 (convert_memory_address_addr_space): Likewise.
12176 (get_mode_bounds): Likewise.
12177 (get_address_mode): Return a scalar_int_mode rather than a
12179 * rtlanal.c (get_address_mode): Likewise.
12180 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12181 than a machine_mode.
12182 * targhooks.c (default_mode_rep_extended): Likewise.
12183 (default_valid_pointer_mode): Likewise.
12184 (default_addr_space_valid_pointer_mode): Likewise.
12185 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12186 than a machine_mode.
12187 (default_libgcc_cmp_return_mode): Likewise.
12188 (default_libgcc_shift_count_mode): Likewise.
12189 (default_unwind_word_mode): Likewise.
12190 (default_addr_space_pointer_mode): Likewise.
12191 (default_addr_space_address_mode): Likewise.
12192 * targhooks.h (default_eh_return_filter_mode): Likewise.
12193 (default_libgcc_cmp_return_mode): Likewise.
12194 (default_libgcc_shift_count_mode): Likewise.
12195 (default_unwind_word_mode): Likewise.
12196 (default_addr_space_pointer_mode): Likewise.
12197 (default_addr_space_address_mode): Likewise.
12198 (default_mode_rep_extended): Take a scalar_int_mode rather than
12200 (default_valid_pointer_mode): Likewise.
12201 (default_addr_space_valid_pointer_mode): Likewise.
12202 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12204 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12205 rather than a machine_mode.
12206 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12207 for local variables.
12208 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12209 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12210 than a machine_mode.
12212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12213 Alan Hayward <alan.hayward@arm.com>
12214 David Sherwood <david.sherwood@arm.com>
12216 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12217 the mode argument to scalar_int_mode.
12218 (do_jump_by_parts_zero_rtx): Likewise.
12219 (do_jump_by_parts_equality_rtx): Likewise.
12220 (do_jump_by_parts_greater): Take a mode argument.
12221 (do_jump_by_parts_equality): Likewise.
12222 (do_jump_1): Update calls accordingly.
12224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12225 Alan Hayward <alan.hayward@arm.com>
12226 David Sherwood <david.sherwood@arm.com>
12228 * is-a.h (safe_dyn_cast): New function.
12229 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12230 (jump_table_for_label): Likewise.
12231 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12232 instead of an rtx_insn *.
12233 (shorten_branches): Use dyn_cast instead of LABEL_P and
12234 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12235 rtx_jump_table_data::get_data_mode.
12236 (final_scan_insn): Likewise.
12238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12239 Alan Hayward <alan.hayward@arm.com>
12240 David Sherwood <david.sherwood@arm.com>
12242 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12243 trying to combine a full-register integer set with a subreg
12246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12247 Alan Hayward <alan.hayward@arm.com>
12248 David Sherwood <david.sherwood@arm.com>
12250 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12251 that is always either address_mode or pointer_mode.
12253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12254 Alan Hayward <alan.hayward@arm.com>
12255 David Sherwood <david.sherwood@arm.com>
12257 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12258 when the two are known to be equal.
12260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12261 Alan Hayward <alan.hayward@arm.com>
12262 David Sherwood <david.sherwood@arm.com>
12264 * simplify-rtx.c (simplify_const_unary_operation): Use
12265 is_a <scalar_int_mode> instead of checking for a nonzero
12266 precision. Forcibly convert op_mode to a scalar_int_mode
12267 in that case. More clearly differentiate the operand and
12268 result modes and use the former when deciding what the value
12269 of a count-bits operation should be. Use is_int_mode instead
12270 of checking for a MODE_INT. Remove redundant check for whether
12271 this mode has a zero precision.
12273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12274 Alan Hayward <alan.hayward@arm.com>
12275 David Sherwood <david.sherwood@arm.com>
12277 * optabs.c (widen_leading): Change the type of the mode argument
12278 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12279 (widen_bswap): Likewise.
12280 (expand_parity): Likewise.
12281 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12282 (expand_ffs): Likewise.
12283 (epand_unop): Check for scalar integer modes before calling the
12286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12287 Alan Hayward <alan.hayward@arm.com>
12288 David Sherwood <david.sherwood@arm.com>
12290 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12292 (expand_expr_real_1): Update call accordingly.
12294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12295 Alan Hayward <alan.hayward@arm.com>
12296 David Sherwood <david.sherwood@arm.com>
12298 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12299 value_mode arguments. Use scalar_int_mode internally.
12300 (store_bit_field_1): Rename the new integer mode from imode
12301 to op0_mode and use it instead of GET_MODE (op0). Update calls
12302 to store_split_bit_field, store_bit_field_using_insv and
12303 store_fixed_bit_field.
12304 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12305 Use scalar_int_mode internally. Use a bit count rather than a mode
12306 when calculating the largest bit size for get_best_mode.
12307 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12308 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12309 Remove assertion that OP0 has a scalar integer mode.
12310 (store_split_bit_field): Add op0_mode and value_mode arguments.
12311 Update calls to extract_fixed_bit_field.
12312 (extract_bit_field_using_extv): Add an op0_mode argument.
12313 Use scalar_int_mode internally.
12314 (extract_bit_field_1): Rename the new integer mode from imode to
12315 op0_mode and use it instead of GET_MODE (op0). Update calls to
12316 extract_split_bit_field, extract_bit_field_using_extv and
12317 extract_fixed_bit_field.
12318 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12319 to extract_split_bit_field and extract_fixed_bit_field_1.
12320 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12321 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12322 on the target mode.
12323 (extract_split_bit_field): Add an op0_mode argument. Update call
12324 to extract_fixed_bit_field.
12326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12327 Alan Hayward <alan.hayward@arm.com>
12328 David Sherwood <david.sherwood@arm.com>
12330 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12331 * explow.c (hard_function_value): Likewise.
12332 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12333 convert_to_mode call outside the loop.
12334 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12335 for the mode iterator. Require the mode specified by max_pieces
12337 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12339 (copy_blkmode_to_reg): Likewise.
12340 (set_storage_via_setmem): Likewise.
12341 * optabs.c (prepare_cmp_insn): Likewise.
12342 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12343 * stor-layout.c (finish_bitfield_representative): Likewise.
12345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12346 Alan Hayward <alan.hayward@arm.com>
12347 David Sherwood <david.sherwood@arm.com>
12349 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12350 * expr.c (convert_move): Use them.
12351 (convert_modes): Likewise.
12352 (store_expr_with_bounds): Likewise.
12354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12355 Alan Hayward <alan.hayward@arm.com>
12356 David Sherwood <david.sherwood@arm.com>
12358 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12359 parameter for the mode of "x". Remove the "known_x", "known_mode"
12360 and "known_ret" arguments. Change the type of the mode argument
12361 to scalar_int_mode.
12362 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12363 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12364 (reg_num_sign_bit_copies_for_combine): Likewise.
12365 * rtlanal.c (nonzero_bits1): Likewise.
12366 (num_sign_bit_copies1): Likewise.
12367 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12368 (reg_num_sign_bit_copies_general): Likewise.
12369 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12370 (reg_nonzero_bits_general): Likewise.
12372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12373 Alan Hayward <alan.hayward@arm.com>
12374 David Sherwood <david.sherwood@arm.com>
12376 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12377 than in subroutines. Return 1 for non-integer modes.
12378 (cached_num_sign_bit_copies): Change the type of the mode parameter
12379 to scalar_int_mode.
12380 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12381 classes. Handle CONST_INT_P first and then check whether X also
12382 has a scalar integer mode. Check the same thing for inner registers
12383 of a SUBREG and for values that are being extended or truncated.
12385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12386 Alan Hayward <alan.hayward@arm.com>
12387 David Sherwood <david.sherwood@arm.com>
12389 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12390 in subroutines. Return the mode mask for non-integer modes.
12391 (cached_nonzero_bits): Change the type of the mode parameter
12392 to scalar_int_mode.
12393 (nonzero_bits1): Likewise. Remove early exit for other mode
12394 classes. Handle CONST_INT_P first and then check whether X
12395 also has a scalar integer mode.
12397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12398 Alan Hayward <alan.hayward@arm.com>
12399 David Sherwood <david.sherwood@arm.com>
12401 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12402 Return a scalar_int_mode and assert that the size is greater than
12404 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12405 (op_by_pieces_d::op_by_pieces_d): Likewise.
12406 (op_by_pieces_d::run): Likewise.
12407 (can_store_by_pieces): Likewise.
12409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12410 Alan Hayward <alan.hayward@arm.com>
12411 David Sherwood <david.sherwood@arm.com>
12413 * combine.c (extract_left_shift): Add a mode argument and update
12415 (make_compound_operation_int): Change the type of the mode parameter
12416 to scalar_int_mode and update the call to extract_left_shift.
12418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12419 Alan Hayward <alan.hayward@arm.com>
12420 David Sherwood <david.sherwood@arm.com>
12422 * combine.c (simplify_and_const_int): Change the type of the mode
12423 parameter to scalar_int_mode.
12424 (simplify_and_const_int_1): Likewise. Update recursive call.
12426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12427 Alan Hayward <alan.hayward@arm.com>
12428 David Sherwood <david.sherwood@arm.com>
12430 * combine.c (simplify_compare_const): Check that the mode is a
12431 scalar_int_mode (rather than VOIDmode) before testing its
12433 (simplify_comparison): Move COMPARISON_P handling out of the
12434 loop and restrict the latter part of the loop to scalar_int_modes.
12435 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12436 and when considering SUBREG_REGs. Use is_int_mode instead of
12437 checking GET_MODE_CLASS against MODE_INT.
12439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12440 Alan Hayward <alan.hayward@arm.com>
12441 David Sherwood <david.sherwood@arm.com>
12443 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12444 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12445 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
12452 (force_to_mode): ...here. Keep xmode up-to-date and use it
12453 instead of GET_MODE (x).
12455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12456 Alan Hayward <alan.hayward@arm.com>
12457 David Sherwood <david.sherwood@arm.com>
12459 * optabs-query.h (extraction_insn::struct_mode): Change type to
12460 opt_scalar_int_mode and update comment.
12461 (extraction_insn::field_mode): Change type to scalar_int_mode.
12462 (extraction_insn::pos_mode): Likewise.
12463 * combine.c (make_extraction): Update accordingly.
12464 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12465 (get_optab_extraction_insn): Likewise.
12466 * recog.c (simplify_while_replacing): Likewise.
12467 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12468 parameter to opt_scalar_int_mode.
12470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12471 Alan Hayward <alan.hayward@arm.com>
12472 David Sherwood <david.sherwood@arm.com>
12474 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12475 to a scalar_int_mode instead of a machine_mode.
12476 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12477 (get_best_mode): Return a boolean and use a pointer argument to store
12478 the selected mode. Replace the limit mode parameter with a bit limit.
12479 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12480 for the values returned by bit_field_mode_iterator::next_mode.
12481 (store_bit_field): Update call to get_best_mode.
12482 (store_fixed_bit_field): Likewise.
12483 (extract_fixed_bit_field): Likewise.
12484 * expr.c (optimize_bitfield_assignment_op): Likewise.
12485 * fold-const.c (optimize_bit_field_compare): Likewise.
12486 (fold_truth_andor_1): Likewise.
12487 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12488 Update for new type of m_mode.
12489 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12496 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12497 (store_bit_field): Check is_a <scalar_int_mode> before calling
12498 strict_volatile_bitfield_p.
12499 (extract_bit_field): Likewise.
12501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12502 Alan Hayward <alan.hayward@arm.com>
12503 David Sherwood <david.sherwood@arm.com>
12505 * target.def (cstore_mode): Return a scalar_int_mode.
12506 * doc/tm.texi: Regenerate.
12507 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12508 * targhooks.h (default_cstore_mode): Likewise.
12509 * targhooks.c (default_cstore_mode): Likewise, using a forced
12511 * expmed.c (emit_cstore): Expect the target of the cstore to be
12514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12515 Alan Hayward <alan.hayward@arm.com>
12516 David Sherwood <david.sherwood@arm.com>
12518 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12520 (niter_desc): Likewise mode.
12521 (iv_analyze): Add a mode parameter.
12523 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12524 and change its type to scalar_int_mode.
12525 * loop-iv.c: Update commentary at head of file.
12526 (iv_constant): Pass the mode paraeter before the rtx it describes
12527 and change its type to scalar_int_mode. Remove VOIDmode handling.
12528 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12529 (iv_extend): Likewise.
12530 (shorten_into_mode): Likewise.
12531 (iv_add): Use scalar_int_mode.
12532 (iv_mult): Likewise.
12533 (iv_shift): Likewise.
12534 (canonicalize_iv_subregs): Likewise.
12535 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12536 it describes and change its mode to scalar_int_mode. Also change
12537 the type of the returned inner_mode to scalar_int_mode.
12538 (get_biv_step): Likewise, turning outer_mode from a pointer
12539 into a direct parameter. Update call to get_biv_step_1.
12540 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12541 iv_constant and get_biv_step.
12542 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12543 and change its type to scalar_int_mode. Don't initialise iv->mode
12544 to VOIDmode and remove later checks for its still being VOIDmode.
12545 Update calls to iv_analyze_op and iv_analyze_expr. Check
12546 is_a <scalar_int_mode> when changing the mode under consideration.
12547 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12548 Update call to iv_analyze_expr.
12549 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12550 inner register is not also a scalar_int_mode. Update call to
12552 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12553 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12554 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12555 separate mode class checks. Update calls to iv_analyze. Remove
12556 fix-up of VOIDmodes after iv_analyze_biv.
12557 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12558 don't have a scalar_int_mode. Update call to biv_p.
12560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12561 Alan Hayward <alan.hayward@arm.com>
12562 David Sherwood <david.sherwood@arm.com>
12564 * cfgexpand.c (convert_debug_memory_address): Use
12565 as_a <scalar_int_mode>.
12566 * combine.c (expand_compound_operation): Likewise.
12567 (make_extraction): Likewise.
12568 (change_zero_ext): Likewise.
12569 (simplify_comparison): Likewise.
12570 * cse.c (cse_insn): Likewise.
12571 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12572 (mem_loc_descriptor): Likewise.
12573 (loc_descriptor): Likewise.
12574 * expmed.c (init_expmed_one_mode): Likewise.
12575 (synth_mult): Likewise.
12576 (emit_store_flag_1): Likewise.
12577 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12578 of a comparison with size.
12579 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12580 (reduce_to_bit_field_precision): Likewise.
12581 * function.c (expand_function_end): Likewise.
12582 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12583 * loop-doloop.c (doloop_modify): Likewise.
12584 * optabs.c (expand_binop): Likewise.
12585 (expand_unop): Likewise.
12586 (expand_copysign_absneg): Likewise.
12587 (prepare_cmp_insn): Likewise.
12588 (maybe_legitimize_operand): Likewise.
12589 * recog.c (const_scalar_int_operand): Likewise.
12590 * rtlanal.c (get_address_mode): Likewise.
12591 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12592 (simplify_cond_clz_ctz): Likewise.
12593 * tree-nested.c (get_nl_goto_field): Likewise.
12594 * tree.c (build_vector_type_for_mode): Likewise.
12595 * var-tracking.c (use_narrower_mode): Likewise.
12597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12598 Alan Hayward <alan.hayward@arm.com>
12599 David Sherwood <david.sherwood@arm.com>
12601 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12602 * builtins.c (expand_builtin_signbit): Use it.
12603 * cfgexpand.c (expand_debug_expr): Likewise.
12604 * dojump.c (do_jump): Likewise.
12605 (do_compare_and_jump): Likewise.
12606 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12607 * expmed.c (make_tree): Likewise.
12608 * expr.c (expand_expr_real_2): Likewise.
12609 (expand_expr_real_1): Likewise.
12610 (try_casesi): Likewise.
12611 * fold-const-call.c (fold_const_call_ss): Likewise.
12612 * fold-const.c (unextend): Likewise.
12613 (extract_muldiv_1): Likewise.
12614 (fold_single_bit_test): Likewise.
12615 (native_encode_int): Likewise.
12616 (native_encode_string): Likewise.
12617 (native_interpret_int): Likewise.
12618 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12619 * internal-fn.c (expand_addsub_overflow): Likewise.
12620 (expand_neg_overflow): Likewise.
12621 (expand_mul_overflow): Likewise.
12622 (expand_arith_overflow): Likewise.
12623 * match.pd: Likewise.
12624 * stor-layout.c (layout_type): Likewise.
12625 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12626 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12627 * tree-ssanames.c (get_range_info): Likewise.
12628 * tree-switch-conversion.c (array_value_type) Likewise.
12629 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12630 (vect_recog_divmod_pattern): Likewise.
12631 (vect_recog_mixed_size_cond_pattern): Likewise.
12632 * tree-vrp.c (extract_range_basic): Likewise.
12633 (simplify_float_conversion_using_ranges): Likewise.
12634 * tree.c (int_fits_type_p): Likewise.
12635 * ubsan.c (instrument_bool_enum_load): Likewise.
12636 * varasm.c (mergeable_string_section): Likewise.
12637 (narrowing_initializer_constant_valid_p): Likewise.
12638 (output_constant): Likewise.
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 * machmode.h (NARROWEST_INT_MODE): New macro.
12645 * expr.c (alignment_for_piecewise_move): Use it instead of
12646 GET_CLASS_NARROWEST_MODE (MODE_INT).
12647 (push_block): Likewise.
12648 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12650 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12653 Alan Hayward <alan.hayward@arm.com>
12654 David Sherwood <david.sherwood@arm.com>
12656 * postreload.c (move2add_valid_value_p): Change the type of the
12657 mode parameter to scalar_int_mode.
12658 (move2add_use_add2_insn): Add a mode parameter and use it instead
12660 (move2add_use_add3_insn): Likewise.
12661 (reload_cse_move2add): Update accordingly.
12663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12664 Alan Hayward <alan.hayward@arm.com>
12665 David Sherwood <david.sherwood@arm.com>
12667 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12669 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12670 * optabs.c (expand_unop): Likewise.
12672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12673 Alan Hayward <alan.hayward@arm.com>
12674 David Sherwood <david.sherwood@arm.com>
12676 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12677 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12678 (popcount_loc_descriptor): Likewise.
12679 (bswap_loc_descriptor): Likewise.
12680 (rotate_loc_descriptor): Likewise.
12681 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12682 calling the functions above.
12684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12685 Alan Hayward <alan.hayward@arm.com>
12686 David Sherwood <david.sherwood@arm.com>
12688 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12690 (try_combine): Likewise.
12691 (simplify_if_then_else): Likewise.
12692 * cse.c (cse_insn): Likewise.
12693 * dwarf2out.c (mem_loc_descriptor): Likewise.
12694 * emit-rtl.c (gen_lowpart_common): Likewise.
12695 * simplify-rtx.c (simplify_truncation): Likewise.
12696 (simplify_binary_operation_1): Likewise.
12697 (simplify_const_relational_operation): Likewise.
12698 (simplify_ternary_operation): Likewise.
12699 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12702 Alan Hayward <alan.hayward@arm.com>
12703 David Sherwood <david.sherwood@arm.com>
12705 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12706 * reload.c (push_reload): Likewise.
12707 (find_reloads): Likewise.
12709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12710 Alan Hayward <alan.hayward@arm.com>
12711 David Sherwood <david.sherwood@arm.com>
12713 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12714 (make_compound_operation_int): Likewise.
12715 (change_zero_ext): Likewise.
12716 * expr.c (convert_move): Likewise.
12717 (convert_modes): Likewise.
12718 * fwprop.c (forward_propagate_subreg): Likewise.
12719 * loop-iv.c (get_biv_step_1): Likewise.
12720 * optabs.c (widen_operand): Likewise.
12721 * postreload.c (move2add_valid_value_p): Likewise.
12722 * recog.c (simplify_while_replacing): Likewise.
12723 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12724 (simplify_binary_operation_1): Likewise. Remove redundant
12725 mode equality check.
12727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12728 Alan Hayward <alan.hayward@arm.com>
12729 David Sherwood <david.sherwood@arm.com>
12731 * combine.c (combine_simplify_rtx): Add checks for
12732 is_a <scalar_int_mode>.
12733 (simplify_if_then_else): Likewise.
12734 (make_field_assignment): Likewise.
12735 (simplify_comparison): Likewise.
12736 * ifcvt.c (noce_try_bitop): Likewise.
12737 * loop-invariant.c (canonicalize_address_mult): Likewise.
12738 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12741 Alan Hayward <alan.hayward@arm.com>
12742 David Sherwood <david.sherwood@arm.com>
12744 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12745 is_a <scalar_int_mode> instead of != BLKmode.
12747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12748 Alan Hayward <alan.hayward@arm.com>
12749 David Sherwood <david.sherwood@arm.com>
12751 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12752 instead of != VOIDmode.
12753 * combine.c (if_then_else_cond): Likewise.
12754 (change_zero_ext): Likewise.
12755 * dwarf2out.c (mem_loc_descriptor): Likewise.
12756 (loc_descriptor): Likewise.
12757 * rtlanal.c (canonicalize_condition): Likewise.
12758 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12761 Alan Hayward <alan.hayward@arm.com>
12762 David Sherwood <david.sherwood@arm.com>
12764 * simplify-rtx.c (simplify_binary_operation_1): Use
12765 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12768 Alan Hayward <alan.hayward@arm.com>
12769 David Sherwood <david.sherwood@arm.com>
12771 * wide-int.h (int_traits<unsigned char>) New class.
12772 (int_traits<unsigned short>) Likewise.
12773 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12774 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12776 * combine.c (set_nonzero_bits_and_sign_copies): Use
12777 is_a <scalar_int_mode>.
12778 (find_split_point): Likewise.
12779 (combine_simplify_rtx): Likewise.
12780 (simplify_logical): Likewise.
12781 (expand_compound_operation): Likewise.
12782 (expand_field_assignment): Likewise.
12783 (make_compound_operation): Likewise.
12784 (extended_count): Likewise.
12785 (change_zero_ext): Likewise.
12786 (simplify_comparison): Likewise.
12787 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12788 (ucompare_loc_descriptor): Likewise.
12789 (minmax_loc_descriptor): Likewise.
12790 (mem_loc_descriptor): Likewise.
12791 (loc_descriptor): Likewise.
12792 * expmed.c (init_expmed_one_mode): Likewise.
12793 * lra-constraints.c (lra_constraint_offset): Likewise.
12794 * optabs.c (prepare_libcall_arg): Likewise.
12795 * postreload.c (move2add_note_store): Likewise.
12796 * reload.c (operands_match_p): Likewise.
12797 * rtl.h (load_extend_op): Likewise.
12798 * rtlhooks.c (gen_lowpart_general): Likewise.
12799 * simplify-rtx.c (simplify_truncation): Likewise.
12800 (simplify_unary_operation_1): Likewise.
12801 (simplify_binary_operation_1): Likewise.
12802 (simplify_const_binary_operation): Likewise.
12803 (simplify_const_relational_operation): Likewise.
12804 (simplify_subreg): Likewise.
12805 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12806 * var-tracking.c (adjust_mems): Likewise.
12807 (prepare_call_arguments): Likewise.
12809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12810 Alan Hayward <alan.hayward@arm.com>
12811 David Sherwood <david.sherwood@arm.com>
12813 * machmode.h (is_int_mode): New fuction.
12814 * combine.c (find_split_point): Use it.
12815 (combine_simplify_rtx): Likewise.
12816 (simplify_if_then_else): Likewise.
12817 (simplify_set): Likewise.
12818 (simplify_shift_const_1): Likewise.
12819 (simplify_comparison): Likewise.
12820 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12821 * cse.c (notreg_cost): Likewise.
12822 (cse_insn): Likewise.
12823 * cselib.c (cselib_lookup_1): Likewise.
12824 * dojump.c (do_jump_1): Likewise.
12825 (do_compare_rtx_and_jump): Likewise.
12826 * dse.c (get_call_args): Likewise.
12827 * dwarf2out.c (rtl_for_decl_init): Likewise.
12828 (native_encode_initializer): Likewise.
12829 * expmed.c (emit_store_flag_1): Likewise.
12830 (emit_store_flag): Likewise.
12831 * expr.c (convert_modes): Likewise.
12832 (store_field): Likewise.
12833 (expand_expr_real_1): Likewise.
12834 * fold-const.c (fold_read_from_constant_string): Likewise.
12835 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12836 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12837 * optabs.c (expand_binop): Likewise.
12838 (expand_unop): Likewise.
12839 (expand_abs_nojump): Likewise.
12840 (expand_one_cmpl_abs_nojump): Likewise.
12841 * simplify-rtx.c (mode_signbit_p): Likewise.
12842 (val_signbit_p): Likewise.
12843 (val_signbit_known_set_p): Likewise.
12844 (val_signbit_known_clear_p): Likewise.
12845 (simplify_relational_operation_1): Likewise.
12846 * tree.c (vector_type_mode): Likewise.
12848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12849 Alan Hayward <alan.hayward@arm.com>
12850 David Sherwood <david.sherwood@arm.com>
12852 * machmode.h (smallest_mode_for_size): Fix formatting.
12853 (smallest_int_mode_for_size): New function.
12854 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12855 instead of smallest_mode_for_size.
12856 * combine.c (make_extraction): Likewise.
12857 * config/arc/arc.c (arc_expand_movmem): Likewise.
12858 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12859 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12860 * config/s390/s390.c (s390_expand_insv): Likewise.
12861 * config/sparc/sparc.c (assign_int_registers): Likewise.
12862 * config/spu/spu.c (spu_function_value): Likewise.
12863 (spu_function_arg): Likewise.
12864 * coverage.c (get_gcov_type): Likewise.
12865 (get_gcov_unsigned_t): Likewise.
12866 * dse.c (find_shift_sequence): Likewise.
12867 * expmed.c (store_bit_field_1): Likewise.
12868 * expr.c (convert_move): Likewise.
12869 (store_field): Likewise.
12870 * internal-fn.c (expand_arith_overflow): Likewise.
12871 * optabs-query.c (get_best_extraction_insn): Likewise.
12872 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12873 * stor-layout.c (layout_type): Likewise.
12874 (initialize_sizetypes): Likewise.
12875 * targhooks.c (default_get_mask_mode): Likewise.
12876 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12879 Alan Hayward <alan.hayward@arm.com>
12880 David Sherwood <david.sherwood@arm.com>
12882 * machmode.h (opt_mode::else_blk): New function.
12883 (int_mode_for_mode): Declare.
12884 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12885 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12887 * cfgexpand.c (expand_debug_expr): Likewise.
12888 * combine.c (gen_lowpart_or_truncate): Likewise.
12889 (gen_lowpart_for_combine): Likewise.
12890 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12891 * config/avr/avr.c (avr_to_int_mode): Likewise.
12892 (avr_out_plus_1): Likewise.
12893 (avr_out_plus): Likewise.
12894 (avr_out_round): Likewise.
12895 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12896 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12897 (s390_expand_vcond): Likewise.
12898 * config/spu/spu.c (spu_split_immediate): Likewise.
12899 (spu_expand_mov): Likewise.
12900 * dse.c (get_stored_val): Likewise.
12901 * expmed.c (store_bit_field_1): Likewise.
12902 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12904 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12905 (extract_low_bits): Likewise.
12906 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12907 handling rather than repeating the check.
12908 (emit_group_store): Likewise.
12909 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12910 * optabs.c (expand_absneg_bit): Likewise.
12911 (expand_copysign_absneg): Likewise.
12912 (expand_copysign_bit): Likewise.
12913 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12914 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12915 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12916 * var-tracking.c (prepare_call_arguments): Likewise.
12917 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12918 int_mode_for_mode instead of mode_for_size.
12919 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12922 Alan Hayward <alan.hayward@arm.com>
12923 David Sherwood <david.sherwood@arm.com>
12925 * machmode.h (int_mode_for_size): New function.
12926 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12927 instead of mode_for_size.
12928 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12930 * combine.c (expand_field_assignment): Use int_mode_for_size
12931 instead of mode_for_size.
12932 (make_extraction): Likewise.
12933 (simplify_shift_const_1): Likewise.
12934 (simplify_comparison): Likewise.
12935 * dojump.c (do_jump): Likewise.
12936 * dwarf2out.c (mem_loc_descriptor): Likewise.
12937 * emit-rtl.c (init_derived_machine_modes): Likewise.
12938 * expmed.c (flip_storage_order): Likewise.
12939 (convert_extracted_bit_field): Likewise.
12940 * expr.c (copy_blkmode_from_reg): Likewise.
12941 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12942 * internal-fn.c (expand_mul_overflow): Likewise.
12943 * lower-subreg.c (simple_move): Likewise.
12944 * optabs-libfuncs.c (init_optabs): Likewise.
12945 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12946 * tree.c (vector_type_mode): Likewise.
12947 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12948 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12949 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12950 * tree-vect-stmts.c (vectorizable_load): Likewise.
12951 (vectorizable_store): Likewise.
12953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12954 Alan Hayward <alan.hayward@arm.com>
12955 David Sherwood <david.sherwood@arm.com>
12957 * coretypes.h (pod_mode): New type.
12958 (scalar_int_mode_pod): New typedef.
12959 * machmode.h (pod_mode): New class.
12960 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12961 * genmodes.c (emit_mode_int_n): Update accordingly.
12962 * lower-subreg.h (target_lower_subreg): Change type to
12963 scalar_int_mode_pod.
12964 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12965 scalar_int_mode_pod.
12967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12968 Alan Hayward <alan.hayward@arm.com>
12969 David Sherwood <david.sherwood@arm.com>
12971 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12972 machine_mode to scalar_int_mode.
12973 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12974 (rs6000_option_override_internal): Remove cast to int.
12975 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12976 machine_mode to scalar_int_mode.
12977 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12978 (rs6000_option_override_internal): Remove cast to int.
12979 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12980 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12982 * config/s390/s390.c (s390_expand_builtin): Likewise.
12983 * coretypes.h (scalar_int_mode): New type.
12984 (opt_scalar_int_mode): New typedef.
12985 * machmode.h (scalar_int_mode): New class.
12986 (scalar_int_mode::includes_p): New function.
12987 (byte_mode): Change type to scalar_int_mode.
12988 (word_mode): Likewise.
12989 (ptr_mode): Likewise.
12990 * emit-rtl.c (byte_mode): Likewise.
12991 (word_mode): Likewise.
12992 (ptr_mode): Likewise.
12993 (init_derived_machine_modes): Update accordingly.
12994 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12995 and MODE_PARTIAL_INT.
12996 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12997 opt_scalar_int_mode.
12999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13000 Alan Hayward <alan.hayward@arm.com>
13001 David Sherwood <david.sherwood@arm.com>
13003 * target.def (libgcc_floating_mode_supported_p): Take a
13005 * doc/tm.texi: Regenerate.
13006 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13008 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13009 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13013 Alan Hayward <alan.hayward@arm.com>
13014 David Sherwood <david.sherwood@arm.com>
13016 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13017 * doc/tm.texi: Regenerate.
13018 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13019 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13020 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13021 * targhooks.h (default_floatn_mode): Likewise.
13022 * targhooks.c (default_floatn_mode): Likewise.
13023 * tree.c (build_common_tree_nodes): Update accordingly.
13025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13026 Alan Hayward <alan.hayward@arm.com>
13027 David Sherwood <david.sherwood@arm.com>
13029 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13030 (mode_iterator::iterate_p): Likewise.
13031 (mode_iterator::get_wider): Likewise.
13032 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13035 Alan Hayward <alan.hayward@arm.com>
13036 David Sherwood <david.sherwood@arm.com>
13038 * coretypes.h (opt_scalar_float_mode): New typedef.
13039 * machmode.h (float_mode_for_size): New function.
13040 * emit-rtl.c (double_mode): Delete.
13041 (init_emit_once): Use float_mode_for_size.
13042 * stor-layout.c (layout_type): Likewise.
13043 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13046 Alan Hayward <alan.hayward@arm.com>
13047 David Sherwood <david.sherwood@arm.com>
13049 * output.h (assemble_real): Take a scalar_float_mode.
13050 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13051 * config/arm/arm.md (consttable_4): Likewise.
13052 (consttable_8): Likewise.
13053 (consttable_16): Likewise.
13054 * config/mips/mips.md (consttable_float): Likewise.
13055 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13056 * varasm.c (assemble_real): Take a scalar_float_mode.
13057 (output_constant_pool_2): Update accordingly.
13058 (output_constant): Likewise.
13060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13061 Alan Hayward <alan.hayward@arm.com>
13062 David Sherwood <david.sherwood@arm.com>
13064 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13065 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13066 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13067 (native_encode_real): Likewise.
13068 (native_interpret_real): Likewise.
13069 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13070 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13073 Alan Hayward <alan.hayward@arm.com>
13074 David Sherwood <david.sherwood@arm.com>
13076 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13077 <scalar_float_mode>. Simplify.
13078 (gen_extend_conv_libfunc): Likewise.
13080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13081 Alan Hayward <alan.hayward@arm.com>
13082 David Sherwood <david.sherwood@arm.com>
13084 * coretypes.h (scalar_float_mode): New type.
13085 * machmode.h (mode_traits::from_int): Use machine_mode if
13086 USE_ENUM_MODES is defined.
13087 (is_a): New function.
13089 (dyn_cast): Likewise.
13090 (scalar_float_mode): New class.
13091 (scalar_float_mode::includes_p): New function.
13092 (is_float_mode): Likewise.
13093 * gdbhooks.py (MachineModePrinter): New class.
13094 (build_pretty_printer): Use it for scalar_float_mode.
13095 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13096 (format_helper::format_helper): Turn into a template.
13097 * genmodes.c (get_mode_class): New function.
13098 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13099 or machine_mode if none.
13100 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13101 as_a <scalar_float_mode>.
13102 * dwarf2out.c (mem_loc_descriptor): Likewise.
13103 (insert_float): Likewise.
13104 (add_const_value_attribute): Likewise.
13105 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13106 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13107 (expand_unop): Update accordingly.
13108 (expand_abs_nojump): Likewise.
13109 (expand_copysign_absneg): Take a scalar_float_mode.
13110 (expand_copysign_bit): Likewise.
13111 (expand_copysign): Update accordingly.
13113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13114 Alan Hayward <alan.hayward@arm.com>
13115 David Sherwood <david.sherwood@arm.com>
13117 * coretypes.h (opt_mode): New class.
13118 * machmode.h (opt_mode): Likewise.
13119 (opt_mode::else_void): New function.
13120 (opt_mode::require): Likewise.
13121 (opt_mode::exists): Likewise.
13122 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13123 (GET_MODE_2XWIDER_MODE): Likewise.
13124 (mode_iterator::get_wider): Update accordingly.
13125 (mode_iterator::get_2xwider): Likewise.
13126 (mode_iterator::get_known_wider): Likewise, turning into a template.
13127 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13128 forcing a wider mode to exist.
13129 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13130 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13131 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13132 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13133 * lower-subreg.c (init_lower_subreg): Likewise.
13134 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13135 on the final iteration.
13136 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13137 a wider mode exists before asking for a move pattern.
13138 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13139 forcing a wider mode to exist.
13140 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13141 returning false if no such mode exists.
13142 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13143 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13144 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13145 Avoid checking for a MODE_INT if we already know the mode is not a
13147 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13148 forcing a wider mode to exist.
13149 (expmed_mult_highpart_optab): Likewise.
13150 (expmed_mult_highpart): Likewise.
13151 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13153 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13154 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13155 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13156 * internal-fn.c (expand_mul_overflow): Update use of
13157 GET_MODE_2XWIDER_MODE.
13158 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13159 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13160 GET_MODE_WIDER_MODE.
13161 (convert_plusminus_to_widen): Likewise.
13162 * tree-switch-conversion.c (array_value_type): Likewise.
13163 * var-tracking.c (emit_note_insn_var_location): Likewise.
13164 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13165 Return false inside rather than outside the loop if no wider mode
13167 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13168 and GET_MODE_2XWIDER_MODE
13169 (can_compare_p): Use else_void.
13170 * gdbhooks.py (OptMachineModePrinter): New class.
13171 (build_pretty_printer): Use it for opt_mode.
13173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13174 Alan Hayward <alan.hayward@arm.com>
13175 David Sherwood <david.sherwood@arm.com>
13177 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13178 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13181 Alan Hayward <alan.hayward@arm.com>
13182 David Sherwood <david.sherwood@arm.com>
13184 * machmode.h (mode_traits): New structure.
13185 (get_narrowest_mode): New function.
13186 (mode_iterator::start): Likewise.
13187 (mode_iterator::iterate_p): Likewise.
13188 (mode_iterator::get_wider): Likewise.
13189 (mode_iterator::get_known_wider): Likewise.
13190 (mode_iterator::get_2xwider): Likewise.
13191 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13192 (FOR_EACH_MODE): Likewise.
13193 (FOR_EACH_MODE_FROM): Likewise.
13194 (FOR_EACH_MODE_UNTIL): Likewise.
13195 (FOR_EACH_WIDER_MODE): Likewise.
13196 (FOR_EACH_2XWIDER_MODE): Likewise.
13197 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13198 * combine.c (simplify_comparison): Likewise
13199 * config/i386/i386.c (type_natural_mode): Likewise.
13200 * cse.c (cse_insn): Likewise.
13201 * dse.c (find_shift_sequence): Likewise.
13202 * emit-rtl.c (init_derived_machine_modes): Likewise.
13203 (init_emit_once): Likewise.
13204 * explow.c (hard_function_value): Likewise.
13205 * expmed.c (extract_fixed_bit_field_1): Likewise.
13206 (extract_bit_field_1): Likewise.
13207 (expand_divmod): Likewise.
13208 (emit_store_flag_1): Likewise.
13209 * expr.c (init_expr_target): Likewise.
13210 (convert_move): Likewise.
13211 (alignment_for_piecewise_move): Likewise.
13212 (widest_int_mode_for_size): Likewise.
13213 (emit_block_move_via_movmem): Likewise.
13214 (copy_blkmode_to_reg): Likewise.
13215 (set_storage_via_setmem): Likewise.
13216 (compress_float_constant): Likewise.
13217 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13218 * optabs-query.c (get_best_extraction_insn): Likewise.
13219 * optabs.c (expand_binop): Likewise.
13220 (expand_twoval_unop): Likewise.
13221 (expand_twoval_binop): Likewise.
13222 (widen_leading): Likewise.
13223 (widen_bswap): Likewise.
13224 (expand_parity): Likewise.
13225 (expand_unop): Likewise.
13226 (prepare_cmp_insn): Likewise.
13227 (prepare_float_lib_cmp): Likewise.
13228 (expand_float): Likewise.
13229 (expand_fix): Likewise.
13230 (expand_sfix_optab): Likewise.
13231 * postreload.c (move2add_use_add2_insn): Likewise.
13232 * reg-stack.c (reg_to_stack): Likewise.
13233 * reginfo.c (choose_hard_reg_mode): Likewise.
13234 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13235 * stor-layout.c (mode_for_size): Likewise.
13236 (smallest_mode_for_size): Likewise.
13237 (mode_for_vector): Likewise.
13238 (finish_bitfield_representative): Likewise.
13239 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13240 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13241 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13242 * var-tracking.c (prepare_call_arguments): Likewise.
13244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13245 Alan Hayward <alan.hayward@arm.com>
13246 David Sherwood <david.sherwood@arm.com>
13248 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13249 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13250 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13251 * machmode.h (mode_size): Move earlier in file.
13252 (mode_precision): Likewise.
13253 (mode_inner): Likewise.
13254 (mode_nunits): Likewise.
13255 (mode_unit_size): Likewise.
13256 (unit_unit_precision): Likewise.
13257 (mode_wider): Likewise.
13258 (mode_2xwider): Likewise.
13259 (machine_mode): New class.
13260 (mode_to_bytes): New function.
13261 (mode_to_bits): Likewise.
13262 (mode_to_precision): Likewise.
13263 (mode_to_inner): Likewise.
13264 (mode_to_unit_size): Likewise.
13265 (mode_to_unit_precision): Likewise.
13266 (mode_to_nunits): Likewise.
13267 (GET_MODE_SIZE): Use mode_to_bytes.
13268 (GET_MODE_BITSIZE): Use mode_to_bits.
13269 (GET_MODE_PRECISION): Use mode_to_precision.
13270 (GET_MODE_INNER): Use mode_to_inner.
13271 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13272 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13273 (GET_MODE_NUNITS): Use mode_to_nunits.
13274 * system.h (ALWAYS_INLINE): New macro.
13275 * config/powerpcspe/powerpcspe-c.c
13276 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13277 int for arg1_mode and arg2_mode.
13279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13280 Alan Hayward <alan.hayward@arm.com>
13281 David Sherwood <david.sherwood@arm.com>
13283 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13284 Prefix mode names with E_ in case statements.
13285 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13286 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13287 (aarch64_split_simd_move): Likewise.
13288 (aarch64_gen_storewb_pair): Likewise.
13289 (aarch64_gen_loadwb_pair): Likewise.
13290 (aarch64_gen_store_pair): Likewise.
13291 (aarch64_gen_load_pair): Likewise.
13292 (aarch64_get_condition_code_1): Likewise.
13293 (aarch64_constant_pool_reload_icode): Likewise.
13294 (get_rsqrte_type): Likewise.
13295 (get_rsqrts_type): Likewise.
13296 (get_recpe_type): Likewise.
13297 (get_recps_type): Likewise.
13298 (aarch64_gimplify_va_arg_expr): Likewise.
13299 (aarch64_simd_container_mode): Likewise.
13300 (aarch64_emit_load_exclusive): Likewise.
13301 (aarch64_emit_store_exclusive): Likewise.
13302 (aarch64_expand_compare_and_swap): Likewise.
13303 (aarch64_gen_atomic_cas): Likewise.
13304 (aarch64_emit_bic): Likewise.
13305 (aarch64_emit_atomic_swap): Likewise.
13306 (aarch64_emit_atomic_load_op): Likewise.
13307 (aarch64_evpc_trn): Likewise.
13308 (aarch64_evpc_uzp): Likewise.
13309 (aarch64_evpc_zip): Likewise.
13310 (aarch64_evpc_ext): Likewise.
13311 (aarch64_evpc_rev): Likewise.
13312 (aarch64_evpc_dup): Likewise.
13313 (aarch64_gen_ccmp_first): Likewise.
13314 (aarch64_gen_ccmp_next): Likewise.
13315 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13316 (alpha_emit_xfloating_libcall): Likewise.
13317 (emit_insxl): Likewise.
13318 (alpha_arg_type): Likewise.
13319 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13320 (arc_preferred_simd_mode): Likewise.
13321 (arc_secondary_reload): Likewise.
13322 (get_arc_condition_code): Likewise.
13323 (arc_print_operand): Likewise.
13324 (arc_legitimate_constant_p): Likewise.
13325 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13326 * config/arc/arc.md (casesi_load): Likewise.
13327 (casesi_compact_jump): Likewise.
13328 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13329 (cc_use_register): Likewise.
13330 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13331 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13332 (arm_init_iwmmxt_builtins): Likewise.
13333 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13334 (neon_expand_vector_init): Likewise.
13335 (arm_attr_length_move_neon): Likewise.
13336 (maybe_get_arm_condition_code): Likewise.
13337 (arm_emit_vector_const): Likewise.
13338 (arm_preferred_simd_mode): Likewise.
13339 (arm_output_iwmmxt_tinsr): Likewise.
13340 (thumb1_output_casesi): Likewise.
13341 (thumb2_output_casesi): Likewise.
13342 (arm_emit_load_exclusive): Likewise.
13343 (arm_emit_store_exclusive): Likewise.
13344 (arm_expand_compare_and_swap): Likewise.
13345 (arm_evpc_neon_vuzp): Likewise.
13346 (arm_evpc_neon_vzip): Likewise.
13347 (arm_evpc_neon_vrev): Likewise.
13348 (arm_evpc_neon_vtrn): Likewise.
13349 (arm_evpc_neon_vext): Likewise.
13350 (arm_validize_comparison): Likewise.
13351 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13352 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13353 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13354 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13355 (c6x_preferred_simd_mode): Likewise.
13356 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13357 (epiphany_rtx_costs): Likewise.
13358 * config/epiphany/predicates.md (proper_comparison_operator):
13360 * config/frv/frv.c (condexec_memory_operand): Likewise.
13361 (frv_emit_move): Likewise.
13362 (output_move_single): Likewise.
13363 (output_condmove_single): Likewise.
13364 (frv_hard_regno_mode_ok): Likewise.
13365 (frv_matching_accg_mode): Likewise.
13366 * config/h8300/h8300.c (split_adds_subs): Likewise.
13367 (h8300_rtx_costs): Likewise.
13368 (h8300_print_operand): Likewise.
13369 (compute_mov_length): Likewise.
13370 (output_logical_op): Likewise.
13371 (compute_logical_op_length): Likewise.
13372 (compute_logical_op_cc): Likewise.
13373 (h8300_shift_needs_scratch_p): Likewise.
13374 (output_a_shift): Likewise.
13375 (compute_a_shift_length): Likewise.
13376 (compute_a_shift_cc): Likewise.
13377 (expand_a_rotate): Likewise.
13378 (output_a_rotate): Likewise.
13379 * config/i386/i386.c (classify_argument): Likewise.
13380 (function_arg_advance_32): Likewise.
13381 (function_arg_32): Likewise.
13382 (function_arg_64): Likewise.
13383 (function_value_64): Likewise.
13384 (ix86_gimplify_va_arg): Likewise.
13385 (ix86_legitimate_constant_p): Likewise.
13386 (put_condition_code): Likewise.
13387 (split_double_mode): Likewise.
13388 (ix86_avx256_split_vector_move_misalign): Likewise.
13389 (ix86_expand_vector_logical_operator): Likewise.
13390 (ix86_split_idivmod): Likewise.
13391 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13392 (ix86_build_const_vector): Likewise.
13393 (ix86_build_signbit_mask): Likewise.
13394 (ix86_match_ccmode): Likewise.
13395 (ix86_cc_modes_compatible): Likewise.
13396 (ix86_expand_branch): Likewise.
13397 (ix86_expand_sse_cmp): Likewise.
13398 (ix86_expand_sse_movcc): Likewise.
13399 (ix86_expand_int_sse_cmp): Likewise.
13400 (ix86_expand_vec_perm_vpermi2): Likewise.
13401 (ix86_expand_vec_perm): Likewise.
13402 (ix86_expand_sse_unpack): Likewise.
13403 (ix86_expand_int_addcc): Likewise.
13404 (ix86_split_to_parts): Likewise.
13405 (ix86_vectorize_builtin_gather): Likewise.
13406 (ix86_vectorize_builtin_scatter): Likewise.
13407 (avx_vpermilp_parallel): Likewise.
13408 (inline_memory_move_cost): Likewise.
13409 (ix86_tieable_integer_mode_p): Likewise.
13410 (x86_maybe_negate_const_int): Likewise.
13411 (ix86_expand_vector_init_duplicate): Likewise.
13412 (ix86_expand_vector_init_one_nonzero): Likewise.
13413 (ix86_expand_vector_init_one_var): Likewise.
13414 (ix86_expand_vector_init_concat): Likewise.
13415 (ix86_expand_vector_init_interleave): Likewise.
13416 (ix86_expand_vector_init_general): Likewise.
13417 (ix86_expand_vector_set): Likewise.
13418 (ix86_expand_vector_extract): Likewise.
13419 (emit_reduc_half): Likewise.
13420 (ix86_emit_i387_round): Likewise.
13421 (ix86_mangle_type): Likewise.
13422 (ix86_expand_round_sse4): Likewise.
13423 (expand_vec_perm_blend): Likewise.
13424 (canonicalize_vector_int_perm): Likewise.
13425 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13426 (expand_vec_perm_1): Likewise.
13427 (expand_vec_perm_interleave3): Likewise.
13428 (expand_vec_perm_even_odd_pack): Likewise.
13429 (expand_vec_perm_even_odd_1): Likewise.
13430 (expand_vec_perm_broadcast_1): Likewise.
13431 (ix86_vectorize_vec_perm_const_ok): Likewise.
13432 (ix86_expand_vecop_qihi): Likewise.
13433 (ix86_expand_mul_widen_hilo): Likewise.
13434 (ix86_expand_sse2_abs): Likewise.
13435 (ix86_expand_pextr): Likewise.
13436 (ix86_expand_pinsr): Likewise.
13437 (ix86_preferred_simd_mode): Likewise.
13438 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13439 * config/i386/sse.md (*andnot<mode>3): Likewise.
13440 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13441 (*<code><mode>3): Likewise.
13442 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13443 (ia64_expand_atomic_op): Likewise.
13444 (ia64_arg_type): Likewise.
13445 (ia64_mode_to_int): Likewise.
13446 (ia64_scalar_mode_supported_p): Likewise.
13447 (ia64_vector_mode_supported_p): Likewise.
13448 (expand_vec_perm_broadcast): Likewise.
13449 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13450 (iq2000_function_arg_advance): Likewise.
13451 (iq2000_function_arg): Likewise.
13452 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13453 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13454 (m68k_libcall_value): Likewise.
13455 (m68k_function_value): Likewise.
13456 (sched_attr_op_type): Likewise.
13457 * config/mcore/mcore.c (mcore_output_move): Likewise.
13458 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13460 (microblaze_function_arg): Likewise.
13461 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13462 (mips_print_operand): Likewise.
13463 (mips_mode_ok_for_mov_fmt_p): Likewise.
13464 (mips_vector_mode_supported_p): Likewise.
13465 (mips_preferred_simd_mode): Likewise.
13466 (mips_expand_vpc_loongson_even_odd): Likewise.
13467 (mips_expand_vec_unpack): Likewise.
13468 (mips_expand_vi_broadcast): Likewise.
13469 (mips_expand_vector_init): Likewise.
13470 (mips_expand_vec_reduc): Likewise.
13471 (mips_expand_msa_cmp): Likewise.
13472 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13473 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13474 (cc_flags_for_mode): Likewise.
13475 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13476 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13477 (nds32_output_casesi_pc_relative): Likewise.
13478 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13479 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13480 (nvptx_gen_unpack): Likewise.
13481 (nvptx_gen_pack): Likewise.
13482 (nvptx_gen_shuffle): Likewise.
13483 (nvptx_gen_wcast): Likewise.
13484 (nvptx_preferred_simd_mode): Likewise.
13485 * config/pa/pa.c (pa_secondary_reload): Likewise.
13486 * config/pa/predicates.md (base14_operand): Likewise.
13487 * config/powerpcspe/powerpcspe-c.c
13488 (altivec_resolve_overloaded_builtin): Likewise.
13489 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13491 (rs6000_preferred_simd_mode): Likewise.
13492 (output_vec_const_move): Likewise.
13493 (rs6000_expand_vector_extract): Likewise.
13494 (rs6000_split_vec_extract_var): Likewise.
13495 (reg_offset_addressing_ok_p): Likewise.
13496 (rs6000_legitimate_offset_address_p): Likewise.
13497 (rs6000_legitimize_address): Likewise.
13498 (rs6000_emit_set_const): Likewise.
13499 (rs6000_const_vec): Likewise.
13500 (rs6000_emit_move): Likewise.
13501 (spe_build_register_parallel): Likewise.
13502 (rs6000_darwin64_record_arg_recurse): Likewise.
13503 (swap_selector_for_mode): Likewise.
13504 (spe_init_builtins): Likewise.
13505 (paired_init_builtins): Likewise.
13506 (altivec_init_builtins): Likewise.
13507 (do_load_for_compare): Likewise.
13508 (rs6000_generate_compare): Likewise.
13509 (rs6000_expand_float128_convert): Likewise.
13510 (emit_load_locked): Likewise.
13511 (emit_store_conditional): Likewise.
13512 (rs6000_output_function_epilogue): Likewise.
13513 (rs6000_handle_altivec_attribute): Likewise.
13514 (rs6000_function_value): Likewise.
13515 (emit_fusion_gpr_load): Likewise.
13516 (emit_fusion_p9_load): Likewise.
13517 (emit_fusion_p9_store): Likewise.
13518 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13519 (fusion_gpr_mem_load): Likewise.
13520 (fusion_addis_mem_combo_load): Likewise.
13521 (fusion_addis_mem_combo_store): Likewise.
13522 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13523 (fusion_gpr_mem_load): Likewise.
13524 (fusion_addis_mem_combo_load): Likewise.
13525 (fusion_addis_mem_combo_store): Likewise.
13526 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13528 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13529 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13530 (rs6000_preferred_simd_mode): Likewise.
13531 (output_vec_const_move): Likewise.
13532 (rs6000_expand_vector_extract): Likewise.
13533 (rs6000_split_vec_extract_var): Likewise.
13534 (reg_offset_addressing_ok_p): Likewise.
13535 (rs6000_legitimate_offset_address_p): Likewise.
13536 (rs6000_legitimize_address): Likewise.
13537 (rs6000_emit_set_const): Likewise.
13538 (rs6000_const_vec): Likewise.
13539 (rs6000_emit_move): Likewise.
13540 (rs6000_darwin64_record_arg_recurse): Likewise.
13541 (swap_selector_for_mode): Likewise.
13542 (paired_init_builtins): Likewise.
13543 (altivec_init_builtins): Likewise.
13544 (rs6000_expand_float128_convert): Likewise.
13545 (emit_load_locked): Likewise.
13546 (emit_store_conditional): Likewise.
13547 (rs6000_output_function_epilogue): Likewise.
13548 (rs6000_handle_altivec_attribute): Likewise.
13549 (rs6000_function_value): Likewise.
13550 (emit_fusion_gpr_load): Likewise.
13551 (emit_fusion_p9_load): Likewise.
13552 (emit_fusion_p9_store): Likewise.
13553 * config/rx/rx.c (rx_gen_move_template): Likewise.
13554 (flags_from_mode): Likewise.
13555 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13556 (s390_slb_comparison): Likewise.
13557 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13558 (s390_vector_mode_supported_p): Likewise.
13559 (s390_cc_modes_compatible): Likewise.
13560 (s390_match_ccmode_set): Likewise.
13561 (s390_canonicalize_comparison): Likewise.
13562 (s390_emit_compare_and_swap): Likewise.
13563 (s390_branch_condition_mask): Likewise.
13564 (s390_rtx_costs): Likewise.
13565 (s390_secondary_reload): Likewise.
13566 (__SECONDARY_RELOAD_CASE): Likewise.
13567 (s390_expand_cs): Likewise.
13568 (s390_preferred_simd_mode): Likewise.
13569 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13570 * config/sh/sh.c (sh_print_operand): Likewise.
13571 (dump_table): Likewise.
13572 (sh_secondary_reload): Likewise.
13573 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13574 * config/sh/sh.md (casesi_worker_1): Likewise.
13575 (casesi_worker_2): Likewise.
13576 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13577 (fcc_comparison_operator): Likewise.
13578 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13579 (emit_soft_tfmode_cvt): Likewise.
13580 (sparc_preferred_simd_mode): Likewise.
13581 (output_cbranch): Likewise.
13582 (sparc_print_operand): Likewise.
13583 (sparc_expand_vec_perm_bmask): Likewise.
13584 (vector_init_bshuffle): Likewise.
13585 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13586 (spu_vector_mode_supported_p): Likewise.
13587 (spu_expand_insv): Likewise.
13588 (spu_emit_branch_or_set): Likewise.
13589 (spu_handle_vector_attribute): Likewise.
13590 (spu_builtin_splats): Likewise.
13591 (spu_builtin_extract): Likewise.
13592 (spu_builtin_promote): Likewise.
13593 (spu_expand_sign_extend): Likewise.
13594 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13595 (tilegx_simd_int): Likewise.
13596 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13597 (tilepro_simd_int): Likewise.
13598 * config/v850/v850.c (const_double_split): Likewise.
13599 (v850_print_operand): Likewise.
13600 (ep_memory_offset): Likewise.
13601 * config/vax/vax.c (vax_rtx_costs): Likewise.
13602 (vax_output_int_move): Likewise.
13603 (vax_output_int_add): Likewise.
13604 (vax_output_int_subtract): Likewise.
13605 * config/visium/predicates.md (visium_branch_operator): Likewise.
13606 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13607 (visium_print_operand_address): Likewise.
13608 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13609 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13610 (xtensa_expand_conditional_branch): Likewise.
13611 (xtensa_copy_incoming_a7): Likewise.
13612 (xtensa_output_literal): Likewise.
13613 * dfp.c (decimal_real_maxval): Likewise.
13614 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13617 Alan Hayward <alan.hayward@arm.com>
13618 David Sherwood <david.sherwood@arm.com>
13620 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13621 (mode_nunits_inline): Likewise.
13622 (mode_inner_inline): Likewise.
13623 (mode_unit_size_inline): Likewise.
13624 (mode_unit_precision_inline): Likewise.
13625 (emit_insn_modes_h): Likewise. Also emit a #define of the
13627 (emit_mode_wider): Add an E_ prefix to mode names.
13628 (emit_mode_complex): Likewise.
13629 (emit_mode_inner): Likewise.
13630 (emit_mode_adjustments): Likewise.
13631 (emit_mode_int_n): Likewise.
13632 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13633 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13634 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13635 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13636 (CRC32_BUILTIN, ENTRY): Likewise.
13637 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13638 (aarch64_pop_regs): Likewise.
13639 (aarch64_process_components): Likewise.
13640 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13641 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13642 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13643 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13644 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13645 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13647 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13648 (pcmpistr): Likewise.
13649 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13650 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13651 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13653 * config/rl78/rl78.c (mduc_regs): Likewise.
13654 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13655 (htm_expand_builtin): Likewise.
13656 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13657 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13658 * config/xtensa/xtensa.c (print_operand): Likewise.
13659 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13660 (NUM_MODE_VECTOR_INT): Likewise.
13661 * genoutput.c (null_operand): Likewise.
13662 (output_operand_data): Likewise.
13663 * genrecog.c (print_parameter_value): Likewise.
13664 * lra.c (debug_operand_data): Likewise.
13666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13668 * dwarf2out.c (scompare_loc_descriptor_wide)
13669 (scompare_loc_descriptor_narrow): New functions, split out from...
13670 (scompare_loc_descriptor): ...here.
13671 * expmed.c (emit_store_flag_int): New function, split out from...
13672 (emit_store_flag): ...here.
13674 2017-08-30 Richard Biener <rguenther@suse.de>
13676 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13677 (dwarf2out_early_finish): Move setting of AT_pubnames from
13678 early debug output to early finish.
13680 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13682 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13683 and -mdata-region to the assembler.
13685 2017-08-30 Richard Biener <rguenther@suse.de>
13687 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13689 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13690 (dwarf2out_early_global_decl): Only generate a DIE for the
13691 abstract origin if it doesn't already exist or is a declaration DIE.
13692 (resolve_addr): Do not add the linkage name twice when
13693 generating a stub DIE for the DW_TAG_GNU_call_site target.
13695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13697 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13698 Use machine_mode rather than int for arg1_mode.
13700 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13703 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13704 that the second argument of the built-in functions to unpack
13705 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13706 switch statement instead a lot of if statements.
13707 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13708 Allow 64-bit values to be in Altivec registers as well as
13709 traditional floating point registers.
13710 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13712 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13714 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13715 MAX_REGS_PER_ADDRESS == 1.
13717 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13719 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13720 * config/i386/i386.c (ix86_option_override_internal): Simplify
13721 setting of opts->x_flag_entry.
13723 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13724 Jakub Jelinek <jakub@redhat.com>
13725 Richard Biener <rguenther@suse.de>
13727 PR tree-optimization/81503
13728 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13729 folded constant fits in the target type; reorder tests for clarity.
13731 2017-08-29 Martin Liska <mliska@suse.cz>
13733 * passes.def: Include pass_lower_switch.
13734 * stmt.c (dump_case_nodes): Remove and move to
13735 tree-switch-conversion.
13736 (case_values_threshold): Likewise.
13737 (expand_switch_as_decision_tree_p): Likewise.
13738 (emit_case_decision_tree): Likewise.
13739 (expand_case): Likewise.
13740 (balance_case_nodes): Likewise.
13741 (node_has_low_bound): Likewise.
13742 (node_has_high_bound): Likewise.
13743 (node_is_bounded): Likewise.
13744 (emit_case_nodes): Likewise.
13745 (struct simple_case_node): New struct.
13746 (add_case_node): Remove.
13747 (emit_case_dispatch_table): Use vector instead of case_list.
13748 (reset_out_edges_aux): Remove.
13749 (compute_cases_per_edge): Likewise.
13750 (expand_case): Build list of simple_case_node.
13751 (expand_sjlj_dispatch_table): Use it.
13752 * tree-switch-conversion.c (struct case_node): Moved from
13753 stmt.c and adjusted.
13754 (emit_case_nodes): Likewise.
13755 (node_has_low_bound): Likewise.
13756 (node_has_high_bound): Likewise.
13757 (node_is_bounded): Likewise.
13758 (case_values_threshold): Likewise.
13759 (reset_out_edges_aux): Likewise.
13760 (compute_cases_per_edge): Likewise.
13761 (add_case_node): Likewise.
13762 (dump_case_nodes): Likewise.
13763 (balance_case_nodes): Likewise.
13764 (expand_switch_as_decision_tree_p): Likewise.
13765 (emit_jump): Likewise.
13766 (emit_case_decision_tree): Likewise.
13767 (try_switch_expansion): Likewise.
13768 (do_jump_if_equal): Likewise.
13769 (emit_cmp_and_jump_insns): Likewise.
13770 (fix_phi_operands_for_edge): New function.
13771 (record_phi_operand_mapping): Likewise.
13772 (class pass_lower_switch): New pass.
13773 (pass_lower_switch::execute): New function.
13774 (make_pass_lower_switch): Likewise.
13775 (conditional_probability):
13776 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13777 * tree-pass.h: Add make_pass_lower_switch.
13779 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13782 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13785 2017-08-29 Richard Biener <rguenther@suse.de>
13787 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13788 we do not add a DW_AT_inline attribute twice.
13789 (gen_subprogram_die): Remove code setting DW_AT_inline on
13790 DECL_ABSTRACT_P nodes.
13792 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13794 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13795 calls to internal functions.
13796 (gimplify_modify_expr): Likewise.
13797 * tree-call-cdce.c (use_internal_fn): Likewise.
13798 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13799 (convert_to_divmod): Set the nothrow flag.
13800 * tree-if-conv.c (predicate_mem_writes): Likewise.
13801 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13802 (vectorizable_call): Likewise.
13803 (vectorizable_store): Likewise.
13804 (vectorizable_load): Likewise.
13805 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13806 (vect_recog_mask_conversion_pattern): Likewise.
13808 2017-08-29 Martin Liska <mliska@suse.cz>
13811 * gcc.c (driver_handle_option): Add new argument.
13812 * opts-common.c (handle_option): Pass
13813 target_option_override_hook.
13814 * opts-global.c (lang_handle_option): Add new option.
13815 (set_default_handlers): Add new argument.
13816 (decode_options): Likewise.
13817 * opts.c (target_handle_option): Likewise.
13818 (common_handle_option): Call target_option_override_hook.
13819 * opts.h (struct cl_option_handler_func): Add hook for
13820 target option override.
13821 (struct cl_option_handlers): Likewise.
13822 (set_default_handlers): Add new argument.
13823 (decode_options): Likewise.
13824 (common_handle_option): Likewise.
13825 (target_handle_option): Likewise.
13826 * toplev.c (toplev::main): Pass targetm.target_option.override
13829 2017-08-29 Richard Biener <rguenther@suse.de>
13830 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13832 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13833 life to the active subtree.
13835 2017-08-28 Jeff Law <law@redhat.com>
13837 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13838 derive_equivalences.
13839 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13840 Code moved into....
13841 (edge_info::derive_equivalences): New private member function
13843 * tree-ssa-dom.c (class edge_info): Changed from a struct
13844 to a class. Add ctor/dtor, methods and data members.
13845 (edge_info::edge_info): Renamed from allocate_edge_info.
13846 Initialize additional members.
13847 (edge_info::~edge_info): New.
13848 (free_dom_edge_info): Delete the edge info.
13849 (record_edge_info): Use new class & associated member functions.
13850 Tighten forms for testing for edge equivalences.
13851 (record_temporary_equivalences): Iterate over the simple
13852 equivalences rather than assuming there's only one per edge.
13853 (cprop_into_successor_phis): Iterate over the simple
13854 equivalences rather than assuming there's only one per edge.
13855 (optimize_stmt): Use operand_equal_p rather than pointer
13856 equality for mini-DSE code.
13858 2017-08-28 Nathan Sidwell <nathan@acm.org>
13860 * gcc.c (execute): Fold SIGPIPE handling into switch
13861 statement. Adjust internal error message.
13863 2017-08-28 Richard Biener <rguenther@suse.de>
13866 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13867 Do nothing for removed DIEs.
13869 2017-08-28 Richard Biener <rguenther@suse.de>
13871 PR tree-optimization/81977
13872 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13875 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13878 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13879 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13880 using targetm.gen_mem_thread_fence.
13882 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13885 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13886 predicate to register_operand. Reorder operands.
13887 (*btr<mode>): Ditto.
13888 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13889 (*btr<mode>_mask): Ditto.
13891 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13893 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13894 * config/rs6000/xmmintrin.h: New file.
13895 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13897 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13900 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13901 parameter and_insn and return it.
13902 (recombine_lvx_pattern): Insert a copy to ensure availability of
13903 the base register of the copied masking operation at the point of
13904 the instruction replacement.
13905 (recombine_stvx_pattern): Likewise.
13907 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13909 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13910 undocumented switches.
13911 (-mpower9-dform-vector): Likewise.
13912 (-mpower9-dform): Likewise.
13913 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13914 comments to delete references to -mpower9-dform* switches.
13915 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13916 Delete reference to -mpower9-dform* switches, test for
13917 -mpower9-vector instead.
13918 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13919 (OTHER_P9_VECTOR_MASKS): Likewise.
13920 (POWERPC_MASKS): Likewise.
13921 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13922 tests against -mpower9-dform* to -mpower9-vector. Delete code
13923 that checked for -mpower9-dform* consistancy with other options.
13924 Add test for -mpower9-misc to enable other power9 switches.
13925 (rs6000_init_hard_regno_mode_ok): Likewise.
13926 (rs6000_option_override_internal): Likewise.
13927 (rs6000_emit_prologue): Likewise.
13928 (rs6000_emit_epilogue): Likewise.
13929 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13930 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13931 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13933 (emit_fusion_p9_store): Likewise.
13934 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13935 resetting these macros if the assembler does not support ISA 3.0
13937 (TARGET_P9_DFORM_VECTOR): Likewise.
13938 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13939 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13941 2017-08-25 Alan Modra <amodra@gmail.com>
13943 PR rtl-optimization/81747
13944 * cse.c (cse_extended_basic_block): Don't attempt to record
13945 equivalences for degenerate conditional branches that branch
13946 to their fall-through.
13948 2017-08-24 Martin Sebor <msebor@redhat.com>
13950 PR middle-end/81908
13951 * gimple-fold.c (size_must_be_zero_p): New function.
13952 (gimple_fold_builtin_memory_op): Call it.
13954 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13956 * config/rs6000/mm_malloc.h: New file.
13958 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13960 PR tree-optimization/81913
13961 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13962 analysis when either IVs in condition can wrap.
13964 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13966 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13967 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13969 2017-08-24 Richard Biener <rguenther@suse.de>
13972 * targhooks.c (default_target_can_inline_p): Properly
13973 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13974 is present and always compare.
13975 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13976 infer -mfpmath=sse from TARGET_SSE_P.
13977 (ix86_can_inline_p): Properly use target_option_default_node when
13978 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13980 2017-08-24 Richard Biener <rguenther@suse.de>
13983 * dwarf2out.c (output_die): Handle flag_generate_offload like
13985 (output_comp_unit): Likewise.
13986 (gen_array_type_die): Likewise.
13987 (dwarf2out_early_finish): Likewise.
13988 (note_variable_value_in_expr): Likewise.
13989 (dwarf2out_finish): Likewise. Adjust assert.
13990 * cgraphunit.c (symbol_table::compile): Move setting of
13991 flag_generate_offload earlier ...
13992 (symbol_table::finalize_compilation_unit): ... here, before
13993 early debug finalization.
13995 2017-08-24 Richard Biener <rguenther@suse.de>
13997 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13998 and ipa-fnsummary.h.
13999 (ix86_can_inline_p): When ix86_fpmath flags do not match
14000 check whether the callee uses FP math at all.
14002 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14004 PR middle-end/81931
14005 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14006 instead of TYPE_PRECISION.
14008 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14009 Alan Hayward <alan.hayward@arm.com>
14010 David Sherwood <david.sherwood@arm.com>
14012 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14014 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14016 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14018 (ix86_frame::stack_realign_allocate): New field.
14019 (struct machine_frame_state): Modify comments.
14020 (machine_frame_state::sp_realigned_fp_end): New field.
14021 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14022 layout calculation.
14023 (sp_valid_at): Add assertion to assure no attempt to access invalid
14024 offset of a realigned stack.
14025 (fp_valid_at): Likewise.
14026 (choose_baseaddr): Modify comments.
14027 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14028 ix86_expand_prologue.
14029 (ix86_expand_prologue): Modify stack realignment and allocation.
14030 (ix86_expand_epilogue): Modify comments.
14031 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14032 avx2_runtime, avx512f, and avx512f_runtime.
14034 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14036 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14037 (mstackrealign): Do not init to -1.
14038 * config/i386/i386.c (ix86_option_override_internal):
14039 Check opts_set, not opts when setting default value of
14040 opts->x_ix86_force_align_arg_pointer.
14042 2017-08-23 Richard Biener <rguenther@suse.de>
14044 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14045 lang_hooks.decl_printable_name.
14046 * print-rtl-function.c (print_rtx_function): Likewise.
14047 * tree-pretty-print.c (dump_function_header): Likewise.
14049 2017-08-23 Richard Biener <rguenther@suse.de>
14052 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14053 -g0 at compile-time.
14055 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14057 PR middle-end/19706
14058 * doc/sourcebuild.texi (Other hardware attributes):
14061 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14063 PR middle-end/19706
14064 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14065 Fix single-use check.
14067 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14069 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14071 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14073 * doc/install.texi: Modify to add more details on running selected
14076 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14078 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14079 is combined with -mabi=ms.
14080 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14083 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14085 PR tree-optimization/81488
14086 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14087 and cached_basis fields.
14088 (MAX_SPREAD): New constant.
14089 (alloc_cand_and_find_basis): Initialize new fields.
14090 (clear_visited): New function.
14091 (create_phi_basis_1): Rename from create_phi_basis, set visited
14092 and cached_basis fields.
14093 (create_phi_basis): New wrapper function.
14094 (phi_add_costs_1): Rename from phi_add_costs, add spread
14095 parameter, set visited field, short-circuit when limits reached.
14096 (phi_add_costs): New wrapper function.
14097 (record_phi_increments_1): Rename from record_phi_increments, set
14099 (record_phi_increments): New wrapper function.
14100 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14101 (phi_incr_cost): New wrapper function.
14102 (all_phi_incrs_profitable_1): Rename from
14103 all_phi_incrs_profitable, set visited field.
14104 (all_phi_incrs_profitable): New wrapper function.
14106 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14107 Alan Hayward <alan.hayward@arm.com>
14108 David Sherwood <david.sherwood@arm.com>
14110 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14111 that takes the outer and inner modes.
14112 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14113 comparison as the canonical test for a paradoxical subreg.
14114 * combine.c (simplify_set): Use paradoxical_subreg_p.
14115 (make_extraction): Likewise.
14116 (force_to_mode): Likewise.
14117 (rtx_equal_for_field_assignment_p): Likewise.
14118 (gen_lowpart_for_combine): Likewise.
14119 (simplify_comparison): Likewise.
14120 * cse.c (equiv_constant): Likewise.
14121 * expmed.c (store_bit_field_1): Likewise.
14122 * final.c (alter_subreg): Likewise.
14123 * fwprop.c (propagate_rtx): Likewise.
14124 (forward_propagate_subreg): Likewise.
14125 * ira-conflicts.c (ira_build_conflicts): Likewise.
14126 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14127 * lra-constraints.c (curr_insn_transform): Likewise.
14128 (split_reg): Likewise.
14129 * lra-eliminations.c (move_plus_up): Likewise.
14130 (lra_eliminate_regs_1): Likewise.
14131 * recog.c (general_operand): Likewise.
14132 * ree.c (combine_reaching_defs): Likewise.
14133 * reload.c (push_reload): Likewise.
14134 (find_reloads): Likewise.
14135 * reload1.c (elimination_effects): Likewise.
14136 (compute_reload_subreg_offset): Likewise.
14137 (choose_reload_regs): Likewise.
14138 * rtlanal.c (subreg_lsb_1): Likewise.
14139 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14140 (simplify_subreg): Likewise.
14141 * var-tracking.c (track_loc_p): Likewise.
14142 * emit-rtl.c (byte_lowpart_offset): Likewise.
14143 (paradoxical_subreg_p): Delete out-of-line definition.
14145 2017-08-22 Jeff Law <law@redhat.com>
14147 PR tree-optimization/81741
14148 PR tree-optimization/71947
14149 * tree-ssa-dom.c: Include tree-inline.h.
14150 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14151 equivalences if one is more expensive to compute than the other.
14152 * tree-ssa-scopedtables.h (class const_or_copies): Make
14153 record_const_or_copy_raw method private.
14154 (class avail_exprs_stack): New method simplify_binary_operation.
14155 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14156 avail_exprs_stack::simplify_binary_operation as needed.
14157 (avail_exprs_stack::simplify_binary_operation): New function.
14159 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14161 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14162 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14163 (DOT_SYMBOLS): Likewise.
14164 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14165 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14166 (RS6000_ABI_NAME): Likewise.
14167 (TARGET_CMODEL): Likewise.
14168 (TOC_SECTION_ASM_OP): Likewise.
14169 (SET_CMODEL): New macro.
14170 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14172 2017-08-22 Richard Biener <rguenther@suse.de>
14174 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14175 to allow for free-lang-data replacements similar to verify_type_variant.
14177 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14178 Alan Hayward <alan.hayward@arm.com>
14179 David Sherwood <david.sherwood@arm.com>
14181 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14182 VOIDmode for the LABEL_REF.
14184 2017-08-22 Richard Biener <rguenther@suse.de>
14186 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14188 2017-08-22 Richard Biener <rguenther@suse.de>
14190 * common.opt (feliminate-dwarf2-dups): Ignore.
14191 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14192 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14193 same_die_p_wrap, compute_section_prefix,
14194 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14195 (comdat_symbol_id, comdat_symbol_number): Likewise.
14196 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14198 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14199 (output_die): Mark unreachable path unreachable.
14200 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14201 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14202 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14203 (dwarf2out_early_finish): Likewise.
14205 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14207 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14209 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14212 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14213 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14214 (avr_attribute_table) <io, io_low, address>: Initialize
14215 .decl_required with true.
14217 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14219 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14220 undocumented debugging options.
14221 (-mvsx-scalar-double): Likewise.
14222 (-mallow-df-permute): Likewise.
14223 (-mvectorize-builtins): Likewise.
14224 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14225 (rs6000_builtin_vectorized_function): Likewise.
14226 (rs6000_builtin_md_vectorized_function): Likewise.
14227 (rs6000_opt_vars): Likewise.
14229 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14232 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14233 (*btrq_imm): Rename from *btrq.
14234 (*btcq_imm): Rename from *btcq.
14235 (btsc): New code attribute.
14236 (*<btsc><mode>): New insn pattern.
14237 (*btr<mode>): Ditto.
14238 (*<btsc><mode>_mask): New insn_and_split pattern.
14239 (*btr<mode>_mask): Ditto.
14241 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14242 Alan Hayward <alan.hayward@arm.com>
14243 David Sherwood <david.sherwood@arm.com>
14245 * function.c (pad_below): Simplify padding calculation.
14247 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14248 Alan Hayward <alan.hayward@arm.com>
14249 David Sherwood <david.sherwood@arm.com>
14251 * target.def (function_prologue): Remove frame size argument.
14252 (function_epilogue): Likewise.
14253 * doc/tm.texi: Regenerate.
14254 * final.c (final_start_function): Update call to function_prologue.
14255 (final_end_function): Update call to function_epilogue.
14256 (default_function_pro_epilogue): Remove frame size argument.
14257 * output.h (default_function_pro_epilogue): Likewise.
14258 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14259 (arm_output_function_prologue): Likewise.
14260 * config/frv/frv.c (frv_function_prologue): Likewise.
14261 (frv_function_epilogue): Likewise.
14262 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14263 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14264 (ia64_output_function_epilogue): Likewise.
14265 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14266 (m32r_output_function_epilogue): Likewise.
14267 * config/microblaze/microblaze.c (microblaze_function_prologue)
14268 (microblaze_function_epilogue): Likewise.
14269 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14270 (mips_output_function_epilogue): Likewise.
14271 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14272 (mmix_target_asm_function_epilogue): Likewise.
14273 * config/msp430/msp430.c (msp430_start_function): Likewise.
14274 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14275 (nds32_asm_function_epilogue): Likewise.
14276 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14277 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14278 (pa_output_function_epilogue): Likewise.
14279 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14280 (rs6000_output_function_epilogue): Likewise.
14281 * config/rl78/rl78.c (rl78_start_function): Likewise.
14282 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14283 (rs6000_output_function_epilogue): Likewise.
14284 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14285 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14286 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14287 (sparc_asm_function_epilogue): Likewise.
14289 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14291 * tree.h (type_has_mode_precision_p): New function.
14292 * convert.c (convert_to_integer_1): Use it.
14293 * expr.c (expand_expr_real_2): Likewise.
14294 (expand_expr_real_1): Likewise.
14295 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14296 * match.pd: Likewise.
14297 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14298 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14299 * tree-tailcall.c (process_assignment): Likewise.
14300 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14301 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14302 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14303 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14304 (vectorizable_assignment): Likewise.
14305 (vectorizable_shift): Likewise.
14306 (vectorizable_operation): Likewise.
14307 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14309 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14311 * match.pd: Add pow (C, x) simplification.
14313 2017-08-21 Richard Biener <rguenther@suse.de>
14315 PR tree-optimization/81900
14316 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14317 for blocks with abnormal predecessors.
14318 (compute_antic): Do not set visited flag prematurely.
14320 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14323 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14325 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14327 * stor-layout.h (vector_type_mode): Move to...
14328 * tree.h (vector_type_mode): ...here.
14329 * stor-layout.c (vector_type_mode): Move to...
14330 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14332 2017-08-21 Richard Biener <rguenther@suse.de>
14334 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14335 register_external_die hooks.
14336 (debug_false_tree_charstarstar_uhwistar): Declare.
14337 (debug_nothing_tree_charstar_uhwi): Likewise.
14338 * debug.c (do_nothing_debug_hooks): Adjust.
14339 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14340 (debug_nothing_tree_charstar_uhwi): Likewise.
14341 * dbxout.c (dbx_debug_hooks): Adjust.
14342 (xcoff_debug_hooks): Likewise.
14343 * sdbout.c (sdb_debug_hooks): Likewise.
14344 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14345 * dwarf2out.c (macinfo_label_base): New global.
14346 (dwarf2out_register_external_die): New function for the
14347 register_external_die hook.
14348 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14349 (dwarf2_debug_hooks): Use them.
14350 (dwarf2_lineno_debug_hooks): Adjust.
14351 (struct die_struct): Add with_offset flag.
14352 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14353 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14354 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14355 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14356 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14357 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14358 defining section names for the early LTO debug variants.
14359 (reset_indirect_string): New helper.
14360 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14361 (print_dw_val): Add support for offsetted symbol references.
14362 (get_ultimate_context): Split out from is_cxx.
14363 (is_cxx): Use get_ultimate_context.
14364 (is_fortran): Add decl overload.
14365 (compute_comp_unit_symbol): Split out worker from
14366 compute_section_prefix.
14367 (compute_section_prefix): Call compute_comp_unit_symbol and
14368 set comdat_type_p here.
14369 (output_die): Skip DIE symbol output for the LTO added one.
14370 Handle DIE symbol references with offset.
14371 (output_comp_unit): Guard section name mangling properly.
14372 For LTO debug sections emit a symbol at the section beginning
14373 which we use to refer to its DIEs.
14374 (add_abstract_origin_attribute): For DIEs registered via
14375 dwarf2out_register_external_die directly refer to the early
14376 DIE rather than indirectly through the shadow one we created.
14377 Remove obsolete call to dwarf2out_abstract_function for
14378 non-function/block origins.
14379 (gen_array_type_die): When generating early LTO debug do
14380 not emit DW_AT_string_length.
14381 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14382 late when in LTO. As suggested place a gcc_unreachable for
14383 the DECL_ABSTRACT_P case.
14384 (gen_subprogram_die): Avoid another specification DIE
14385 for early built declarations/definitions for the late LTO case.
14386 (gen_variable_die): Add type references for late duplicated VLA dies
14388 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14389 we have the abstract instance already.
14390 (process_scope_var): Adjust decl DIE contexts in LTO which
14391 first puts them in limbo.
14392 (gen_decl_die): Do not generate type DIEs late apart from
14393 types for VLAs or for decls we do not yet have a DIE. Do not
14394 call dwarf2out_abstract_function late.
14395 (dwarf2out_early_global_decl): Make sure to create DIEs
14396 for abstract instances of a decl first.
14397 (dwarf2out_late_global_decl): Adjust comment.
14398 (output_macinfo_op): With multiple macro sections use
14399 macinfo_label_base to distinguish labels.
14400 (output_macinfo): Likewise. Update macinfo_label_base.
14401 Pass in the line info label.
14402 (note_variable_value_in_expr): When generating LTO resolve
14403 all variable values here by generating DIEs as needed.
14404 (init_sections_and_labels): Add early LTO debug flag parameter
14405 and generate different sections and names if set. Add generation
14406 counter for the labels so we can have multiple of them.
14407 (reset_dies): Helper to allow DIEs to be output multiple times.
14408 (dwarf2out_finish): When outputting DIEs to the fat part of an
14409 LTO object first reset DIEs.
14410 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14411 (modified_type_die): Check for decl_ultimate_origin being self
14413 (gen_type_die_with_usage): Likewise.
14414 (gen_typedef_die): Allow decl_ultimate_origin being self.
14415 (set_decl_abstract_flags): Remove.
14416 (set_block_abstract_flags): Likewise.
14417 (dwarf2out_abstract_function): Treat the early generated DIEs
14418 as the abstract copy and only add DW_AT_inline and
14419 DW_AT_artificial here and call set_decl_origin_self.
14420 If the DIE has an abstract origin don't do anything.
14421 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14422 if we have none yet (Go fails to build one, PR78628).
14423 (variably_modified_type_p): Prevent endless recursion for Ada
14424 cyclic pointer types.
14425 * lto-streamer-in.c: Include debug.h.
14426 (dref_queue): New global.
14427 (lto_read_tree_1): Stream in DIE references.
14428 (lto_input_tree): Register DIE references.
14429 (input_function): Stream DECL_DEBUG_ARGS.
14430 * lto-streamer-out.c: Include debug.h.
14431 (lto_write_tree_1): Output DIE references.
14432 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14433 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14434 (output_function): Stream DECL_DEBUG_ARGS.
14435 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14436 Stream DECL_ABSTRACT_ORIGIN.
14437 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14438 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14439 DECL_CONTEXT for file-scope decls.
14440 * lto-streamer.h (struct dref_entry): Declare.
14441 (dref_queue): Likewise.
14442 * cfgexpand.c (pass_expand::execute): Do not call the
14443 outlining_inline_function hook here.
14444 * lto-wrapper.c (debug_obj): New global.
14445 (tool_cleanup): Unlink it if required.
14446 (debug_objcopy): New function.
14447 (run_gcc): Handle early debug sections in the IL files by
14448 extracting them to separate files, partially linkin them and
14449 feeding the result back as result to the linker.
14450 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14451 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14452 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14453 sections into a separate segment.
14454 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14456 (darwin_asm_dwarf_section): Likewise.
14457 (darwin_asm_output_dwarf_offset): Likewise.
14458 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
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 * read-md.h (md_reader::record_potential_iterator_use): Replace
14465 pointer argument with an rtx and an index.
14466 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14467 (apply_mode_iterator): Likewise.
14468 (apply_code_iterator): Likewise.
14469 (apply_int_iterator): Likewise.
14470 (apply_subst_iterator): Likewise.
14471 (record_iterator_use): Likewise.
14472 (record_attribute_use): Likewise.
14473 (md_reader::record_potential_iterator_use): Likewise. Update calls
14474 to record_iterator_use and apply_iterator.
14475 (iterator_use): Replace ptr with x and index.
14476 (attribute_use): Likewise.
14477 (apply_attribute_uses): Update calls to apply_iterator.
14478 (apply_iterators): Likewise. Update initialization of iterator_use.
14479 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14480 and record_potential_iterator_use.
14481 (rtx_reader::read_rtx_operand): Likewise.
14483 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14484 Alan Hayward <alan.hayward@arm.com>
14485 David Sherwood <david.sherwood@arm.com>
14487 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14490 2017-08-21 Richard Biener <rguenther@suse.de>
14492 PR middle-end/81884
14493 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14494 at struct end conservatively when comparing common bases.
14496 2017-08-21 Richard Biener <rguenther@suse.de>
14498 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14499 (mem_ref_in_stmt): Remove.
14500 (determine_max_movement): Use ref index to get at the reference.
14501 (invariantness_dom_walker::before_dom_children): Deal with
14502 lim data already initialized.
14503 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14505 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14507 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14508 (TARGET_ISA_ROUND): Ditto.
14509 (TARGET_ROUND): Ditto.
14510 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14511 * config/i386/i386.md: Ditto.
14512 * config/i386/sse.md: Ditto.
14513 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14514 with OPTION_MASK_ISA_SSE4_1.
14516 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14519 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14520 __builtin_ia32_lzcnt_u16.
14522 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14525 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14526 (rs6000_set_current_function): Rewrite function to use it.
14528 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14531 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14532 and TYPE_WARN_IF_NOT_ALIGN.
14533 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14534 (handle_warn_if_not_align): New.
14535 (place_union_field): Call handle_warn_if_not_align.
14536 (place_field): Call handle_warn_if_not_align.
14537 Copy TYPE_WARN_IF_NOT_ALIGN.
14538 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14539 (layout_type): Likewise.
14540 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14542 (tree_decl_common): Add warn_if_not_align.
14543 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14544 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14545 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14546 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14547 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14548 * doc/extend.texi: Document warn_if_not_aligned attribute.
14549 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14551 2017-08-17 Martin Liska <mliska@suse.cz>
14554 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14555 (get_data_dependence): Use it as pointer type.
14556 (distribute_loop): Likewise.
14558 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14560 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14561 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14562 (p8_vmrgow_<mode>_direct): New define_insn.
14563 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14564 handle endianness for vmrgew and vmrgow permute patterns.
14566 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14568 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14569 * config/rs6000/rs6000-cpus.def: Remove comment.
14570 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14571 (POWERPC_MASKS): Likewise.
14572 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14573 use of TARGET_VSX_TIMODE.
14574 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14575 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14576 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14577 (rs6000_option_override_internal): Remove dead code.
14578 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14579 (rs6000_legitimize_reload_address): Likewise.
14580 (rs6000_legitimate_address_p): Likewise.
14581 (rs6000_opt_masks): Delete "vsx-timode".
14582 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14583 from function comment.
14584 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14585 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14586 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14588 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14589 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14590 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14591 (VSX_TI): Likewise.
14593 (define_peephole2): Likewise.
14595 2017-08-17 Martin Sebor <msebor@redhat.com>
14598 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14599 past the end of an array.
14600 (test_pp_format): Add test cases.
14602 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14604 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14605 missing ECF_NOTHROW flags.
14607 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14610 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14611 operands residing in integer registers.
14612 (*vsx_le_perm_load_<mode>): Likewise.
14613 (*vsx_le_perm_store_<mode>): Likewise.
14614 (define_peephole2): Add peepholes to optimize the above.
14616 2017-08-17 Marek Polacek <polacek@redhat.com>
14618 PR middle-end/81814
14619 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14620 to mimic what shorten_compare did. Change the return type to bool.
14621 (fold_cond_expr_with_comparison): Update call to
14622 operand_equal_for_comparison_p.
14623 (fold_ternary_loc): Likewise.
14625 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14627 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14629 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14631 2017-08-17 Richard Biener <rguenther@suse.de>
14633 * tree-ssa-structalias.c (solve_graph): When propagating
14634 to successors update the graphs succ edges and avoid duplicate work.
14636 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14639 * config/i386/i386.c (ix86_option_override_internal): Save target
14640 specific options after ix86_stack_protector_guard_reg was changed.
14642 2017-08-17 Richard Biener <rguenther@suse.de>
14644 PR tree-optimization/81827
14645 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14646 (new_var_info): Initialize it conservatively.
14647 (get_call_vi): Mark register vars.
14648 (new_scalar_tmp_constraint_exp): Likewise.
14649 (handle_rhs_call): Likewise.
14650 (handle_const_call): Likewise.
14651 (create_function_info_for): Likewise.
14652 (solve_constraints): Sort varinfos to separate register from
14653 non-register vars to pack points-to solution bitmaps during
14656 2017-08-17 Marek Polacek <polacek@redhat.com>
14658 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14660 2017-08-17 Richard Biener <rguenther@suse.de>
14662 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14663 to true when overflow is undefined and we saturated the result.
14665 2017-08-17 Alan Modra <amodra@gmail.com>
14668 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14669 Don't use store multiple if only one reg needs saving.
14670 (interesting_frame_related_regno): New function.
14671 (rs6000_frame_related): Don't emit frame info for regs that
14673 (rs6000_emit_epilogue): Likewise.
14675 2017-08-16 Nathan Sidwell <nathan@acm.org>
14677 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14678 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14679 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14680 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14681 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14682 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14683 (verify_type): Adjust for TYPE_BINFO move.
14684 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14685 process TYPE_BINFO directly.
14686 (hash_tree): Likewise.
14687 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14689 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14692 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14694 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14696 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14699 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14700 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14701 Add ix86_binary_operator_ok to insn constraint.
14702 (*iordi_1_bts): Ditto.
14703 (*xordi_1_btc): Ditto.
14704 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14705 Update corresponding peephole2 pattern.
14709 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14711 PR tree-optimization/81832
14712 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14713 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14715 2017-08-16 Marek Polacek <polacek@redhat.com>
14718 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14719 regarding size_zero_node.
14721 2017-08-16 Martin Liska <mliska@suse.cz>
14724 * config.gcc: Respect previously set extra_objs in case
14727 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14729 PR tree-optimization/81835
14730 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14731 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14732 not depend on the phi.
14734 2017-08-16 Alan Modra <amodra@gmail.com>
14736 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14739 2017-08-16 Alan Modra <amodra@gmail.com>
14741 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14742 (save_reg_p): ..into this. Update all callers.
14743 (first_reg_to_save): Simplify.
14745 2017-08-16 Alan Modra <amodra@gmail.com>
14747 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14750 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14754 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14755 constant count if that count is less than 32.
14757 2017-08-15 Nathan Sidwell <nathan@acm.org>
14759 * gcc.c (execute): Emit friendlier message if inferior is killed
14760 by an external cause.
14762 2017-08-15 Richard Biener <rguenther@suse.de>
14764 PR tree-optimization/81790
14765 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14766 CONSTRUCTORs from simplifying and VN.
14768 2017-08-14 Martin Sebor <msebor@redhat.com>
14770 * builtin-attrs.def: Add comments.
14772 2017-08-14 Martin Sebor <msebor@redhat.com>
14775 * doc/extend.texi (attribute nonstring): Document new attribute.
14777 2017-08-14 Martin Sebor <msebor@redhat.com>
14780 * tree-diagnostic.c (default_tree_printer): Handle %G.
14781 * gimple-pretty-print.h (percent_G_format): Declare new function.
14782 * gimple-pretty-print.c (percent_G_format): Define.
14783 * tree-pretty-print.c (percent_K_format): Add argument.
14785 2017-08-14 Martin Sebor <msebor@redhat.com>
14787 PR translation/79998
14788 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14789 Remove a stray space.
14791 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14794 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14795 (*iordi_1_bts): Ditto.
14796 (*xordi_1_btc): Ditto.
14798 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14801 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14802 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14804 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14805 quoted strings, and make more translator-friendly.
14806 (darwin_rs6000_override_options): Likewise.
14807 (rs6000_option_override_internal): Likewise.
14808 (rs6000_return_in_memory): Fix overlong line.
14809 (init_cmulative_args): Use quoted strings, and make more
14810 translator-friendly.
14811 (rs6000_pass_by_reference): Fix overlong line.
14812 (def_builtin): Use quoted strings.
14813 (altivec_expand_predicate_builtin): Use quoted strings, and make
14814 more translator-friendly.
14815 (htm_expand_builtin): Use quoted strings.
14816 (cpu_expand_builtin): Use quoted strings, and make more
14817 translator-friendly.
14818 (altivec_expand_builtin): Likewise.
14819 (paired_expand_predicate_builtin): Likewise.
14820 (rs6000_invalid_builtin): Likewise.
14821 (builtin_function_type): Use quoted strings.
14822 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14823 more translator-friendly.
14824 (rs6000_trampoline_init): Likewise.
14825 (rs6000_handle_altivec_attribute): Likewise.
14826 (rs6000_inner_target_options): Use quoted strings.
14827 (rs6000_disable_incompatible_switches): Likewise.
14828 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14829 strings, and make more translator-friendly.
14830 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14832 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14834 PR tree-optimization/81799
14835 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14836 cond_expr to simple gimple operand.
14838 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14840 PR middle-end/46932
14841 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14843 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14847 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14848 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14849 TARGET_GASISR_PROLOGUES.
14850 * config/avr/avr.c (avr_option_override): Same.
14851 (avr_pass_pre_proep::execute): Same.
14853 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14856 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14857 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14859 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14861 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14862 define_insn_and_split. Split to a memory load from 0 in
14863 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14864 using PTR mode iterator.
14865 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14866 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14867 (*add_tp_<mode>): Redefine as define_insn_and_split.
14868 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14869 address space. Merge with *add_tp_x32 using PTR mode iterator.
14870 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14871 Split to an add with a memory load from 0 in
14872 DEFAULT_TLS_SEG_REG address space.
14874 2017-08-12 Andrew Pinski <apinski@cavium.com>
14876 * config/aarch64/aarch64-option-extensions.def (rdma):
14877 Fix feature string to what Linux prints out in /proc/cpuinfo.
14879 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14882 * dwarf2out.c (modified_type_die): For C typedef types that have
14883 an ultimate origin, process the ultimate origin instead of the
14885 (gen_typedef_die): Assert that input DECLs have no ultimate
14887 (gen_type_die_with_usage): For typedef variants that have an
14888 ultimate origin, just call gen_decl_die on the original DECL.
14889 (process_scope_var): Avoid creating DIEs for local typedefs and
14890 concrete static variables.
14892 2017-08-12 Alan Modra <amodra@gmail.com>
14896 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14897 match gnu-user.h startfile.
14898 (ENDFILE_LINUX_SPEC): Similarly.
14900 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14903 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14905 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14907 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14908 * config/aarch64/aarch64.md (mov<mode>): Change.
14909 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14910 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14911 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14913 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14915 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14916 for storage order barriers.
14918 2017-08-11 Martin Liska <mliska@suse.cz>
14921 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14922 variables of void type.
14924 2017-08-11 Martin Liska <mliska@suse.cz>
14926 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14927 TARGET_SUPPORTS_ALIASES.
14928 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14929 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14930 (optimize_weakref): Likewise.
14931 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14932 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14933 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14935 2017-08-11 Martin Liska <mliska@suse.cz>
14938 * config/i386/i386.c (make_resolver_func): Do complete
14939 refactoring of the function.
14941 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14944 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14945 * config/i386/i386.c (ix86_stack_protect_guard): Use
14946 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14947 * doc/invoke.texi (x86 Options): Document
14948 -mstack-protector-guard-symbol= option.
14950 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14952 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14953 * config/i386/i386.c (ix86_split_stack_guard): New function.
14954 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14955 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14956 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14957 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14958 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14959 (split_stack_space_check): Call ix86_split_stack_guard.
14961 2017-08-10 Martin Sebor <msebor@redhat.com>
14963 * print-tree.c (print_node): Print location using the established
14965 Replace spaces with colons.
14966 (debug_raw, debug): Ditto.
14968 2017-08-10 Martin Sebor <msebor@redhat.com>
14971 * pretty-print.c (pp_format): Correct the handling of %s precision.
14973 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14976 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14978 (ix86_finalize_stack_frame_flags): This. Also clear
14979 frame_pointer_needed if -fno-omit-frame-pointer is used without
14981 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14982 with ix86_finalize_stack_frame_flags.
14983 (ix86_expand_epilogue): Likewise.
14984 (ix86_expand_split_stack_prologue): Likewise.
14985 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14987 2017-08-10 Martin Liska <mliska@suse.cz>
14990 * c-attribs.c (handle_target_attribute):
14991 Report warning for an empty string argument of target attribute.
14993 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14996 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14998 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14999 or DECL_NONLOCAL labels.
15000 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15001 or DECL_NONLOCAL labels here.
15003 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15005 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15006 to indicate when early gimple folding has been disabled.
15007 (rs6000_gimple_fold_builtin): Add debug content.
15008 (rs6000_invalid_builtin): Fix whitespace.
15009 (rs6000_expand_builtin): Fix whitespace.
15010 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15012 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15015 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15016 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15018 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15020 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15021 (qdf24xx): Likewise.
15022 * config/aarch64/aarch64.md: Include falkor.md.
15023 * config/aarch64/falkor.md: New.
15025 2017-08-09 Marek Polacek <polacek@redhat.com>
15028 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15029 * diagnostic.c (emit_diagnostic): Add a comment.
15030 (emit_diagnostic_valist): New function.
15032 2017-08-09 Marek Polacek <polacek@redhat.com>
15035 * input.c (make_location): New overload.
15036 * input.h (make_location): Declare.
15038 2017-08-08 Alan Modra <amodra@gmail.com>
15039 H.J. Lu <hongjiu.lu@intel.com>
15042 * gcc.c (NO_PIE_SPEC): Delete.
15043 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15045 (LINK_PIE_SPEC): ..to here.
15046 (LINK_COMMAND_SPEC): Support -no-pie.
15047 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15048 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15050 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15051 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15052 (ENDFILE_CRTEND_SPEC): Similarly.
15054 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15057 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15058 (mstack-protector-guard-offset=): Ditto.
15059 * config/i386/i386.c (ix86_option_override): Handle
15060 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15062 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15063 ix86_stack_protect_guard_offset variables.
15064 (TARGET_STACK_PROTECT_GUARD): Always define.
15065 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15066 and -mstack-protector-guard-offset= options.
15068 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15070 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15071 boundary case for the last candidate.
15073 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15075 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15076 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15078 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15080 PR middle-end/19706
15081 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15082 * config/aarch64/aarch64-builtins.c
15083 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15084 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15085 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15087 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15088 Andrew Pinski <pinskia@gmail.com>
15090 PR middle-end/19706
15091 * internal-fn.def (XORSIGN): New.
15092 * optabs.def (xorsign_optab): New.
15093 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15094 (convert_expand_mult_copysign): New.
15095 (pass_optimize_widening_mul::execute): Call
15096 convert_expand_mult_copysign.
15098 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15100 PR tree-optimization/81354
15101 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15102 Insert on edges rather than explicitly creating landing pads.
15103 (analyze_candidates_and_replace): Commit edge inserts.
15105 2017-08-08 Richard Biener <rguenther@suse.de>
15107 PR middle-end/81719
15108 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15109 (expand_simple_operations): Also look through ADDR_EXPRs with
15110 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15112 2017-08-08 Richard Biener <rguenther@suse.de>
15114 PR tree-optimization/81723
15115 * tree-vect-slp.c (struct bst_traits): New hash traits.
15116 (bst_fail): New global.
15117 (vect_build_slp_tree_2): New worker, split out from ...
15118 (vect_build_slp_tree): ... this now wrapping it with using
15119 bst_fail set to cache SLP tree build fails. Properly handle
15121 (vect_analyze_slp_instance): Allocate and free bst_fail.
15123 2017-08-08 Martin Liska <mliska@suse.cz>
15126 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15127 LABEL_DECLs that can be from a different function.
15129 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15131 PR tree-optimization/81744
15132 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15133 loop's number of iterations.
15135 2017-08-08 Martin Liska <mliska@suse.cz>
15137 * asan.c: Include header files.
15138 * attribs.c (build_decl_attribute_variant): New function moved
15140 (build_type_attribute_qual_variant): Likewise.
15141 (cmp_attrib_identifiers): Likewise.
15142 (simple_cst_list_equal): Likewise.
15143 (omp_declare_simd_clauses_equal): Likewise.
15144 (attribute_value_equal): Likewise.
15145 (comp_type_attributes): Likewise.
15146 (build_type_attribute_variant): Likewise.
15147 (lookup_ident_attribute): Likewise.
15148 (remove_attribute): Likewise.
15149 (merge_attributes): Likewise.
15150 (merge_type_attributes): Likewise.
15151 (merge_decl_attributes): Likewise.
15152 (merge_dllimport_decl_attributes): Likewise.
15153 (handle_dll_attribute): Likewise.
15154 (attribute_list_equal): Likewise.
15155 (attribute_list_contained): Likewise.
15156 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15157 (lookup_attribute_by_prefix): Likewise.
15158 * bb-reorder.c: Include header files.
15159 * builtins.c: Likewise.
15160 * calls.c: Likewise.
15161 * cfgexpand.c: Likewise.
15162 * cgraph.c: Likewise.
15163 * cgraphunit.c: Likewise.
15164 * convert.c: Likewise.
15165 * dwarf2out.c: Likewise.
15166 * final.c: Likewise.
15167 * fold-const.c: Likewise.
15168 * function.c: Likewise.
15169 * gimple-expr.c: Likewise.
15170 * gimple-fold.c: Likewise.
15171 * gimple-pretty-print.c: Likewise.
15172 * gimple.c: Likewise.
15173 * gimplify.c: Likewise.
15174 * hsa-common.c: Likewise.
15175 * hsa-gen.c: Likewise.
15176 * internal-fn.c: Likewise.
15177 * ipa-chkp.c: Likewise.
15178 * ipa-cp.c: Likewise.
15179 * ipa-devirt.c: Likewise.
15180 * ipa-fnsummary.c: Likewise.
15181 * ipa-inline.c: Likewise.
15182 * ipa-visibility.c: Likewise.
15184 * lto-cgraph.c: Likewise.
15185 * omp-expand.c: Likewise.
15186 * omp-general.c: Likewise.
15187 * omp-low.c: Likewise.
15188 * omp-offload.c: Likewise.
15189 * omp-simd-clone.c: Likewise.
15190 * opts-global.c: Likewise.
15191 * passes.c: Likewise.
15192 * predict.c: Likewise.
15193 * sancov.c: Likewise.
15194 * sanopt.c: Likewise.
15195 * symtab.c: Likewise.
15196 * toplev.c: Likewise.
15197 * trans-mem.c: Likewise.
15198 * tree-chkp.c: Likewise.
15199 * tree-eh.c: Likewise.
15200 * tree-into-ssa.c: Likewise.
15201 * tree-object-size.c: Likewise.
15202 * tree-parloops.c: Likewise.
15203 * tree-profile.c: Likewise.
15204 * tree-ssa-ccp.c: Likewise.
15205 * tree-ssa-live.c: Likewise.
15206 * tree-ssa-loop.c: Likewise.
15207 * tree-ssa-sccvn.c: Likewise.
15208 * tree-ssa-structalias.c: Likewise.
15209 * tree-ssa.c: Likewise.
15210 * tree-streamer-in.c: Likewise.
15211 * tree-vectorizer.c: Likewise.
15212 * tree-vrp.c: Likewise.
15213 * tsan.c: Likewise.
15214 * ubsan.c: Likewise.
15215 * varasm.c: Likewise.
15216 * varpool.c: Likewise.
15217 * tree.c: Remove functions moved to attribs.[ch].
15218 * tree.h: Likewise.
15219 * config/aarch64/aarch64.c: Add attrs.h header file.
15220 * config/alpha/alpha.c: Likewise.
15221 * config/arc/arc.c: Likewise.
15222 * config/arm/arm.c: Likewise.
15223 * config/avr/avr.c: Likewise.
15224 * config/bfin/bfin.c: Likewise.
15225 * config/c6x/c6x.c: Likewise.
15226 * config/cr16/cr16.c: Likewise.
15227 * config/cris/cris.c: Likewise.
15228 * config/darwin.c: Likewise.
15229 * config/epiphany/epiphany.c: Likewise.
15230 * config/fr30/fr30.c: Likewise.
15231 * config/frv/frv.c: Likewise.
15232 * config/ft32/ft32.c: Likewise.
15233 * config/h8300/h8300.c: Likewise.
15234 * config/i386/winnt.c: Likewise.
15235 * config/ia64/ia64.c: Likewise.
15236 * config/iq2000/iq2000.c: Likewise.
15237 * config/lm32/lm32.c: Likewise.
15238 * config/m32c/m32c.c: Likewise.
15239 * config/m32r/m32r.c: Likewise.
15240 * config/m68k/m68k.c: Likewise.
15241 * config/mcore/mcore.c: Likewise.
15242 * config/microblaze/microblaze.c: Likewise.
15243 * config/mips/mips.c: Likewise.
15244 * config/mmix/mmix.c: Likewise.
15245 * config/mn10300/mn10300.c: Likewise.
15246 * config/moxie/moxie.c: Likewise.
15247 * config/msp430/msp430.c: Likewise.
15248 * config/nds32/nds32-isr.c: Likewise.
15249 * config/nds32/nds32.c: Likewise.
15250 * config/nios2/nios2.c: Likewise.
15251 * config/nvptx/nvptx.c: Likewise.
15252 * config/pa/pa.c: Likewise.
15253 * config/pdp11/pdp11.c: Likewise.
15254 * config/powerpcspe/powerpcspe.c: Likewise.
15255 * config/riscv/riscv.c: Likewise.
15256 * config/rl78/rl78.c: Likewise.
15257 * config/rx/rx.c: Likewise.
15258 * config/s390/s390.c: Likewise.
15259 * config/sh/sh.c: Likewise.
15260 * config/sol2.c: Likewise.
15261 * config/sparc/sparc.c: Likewise.
15262 * config/spu/spu.c: Likewise.
15263 * config/stormy16/stormy16.c: Likewise.
15264 * config/tilegx/tilegx.c: Likewise.
15265 * config/tilepro/tilepro.c: Likewise.
15266 * config/v850/v850.c: Likewise.
15267 * config/vax/vax.c: Likewise.
15268 * config/visium/visium.c: Likewise.
15269 * config/xtensa/xtensa.c: Likewise.
15271 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15274 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15275 constraints since the -mupper-regs-* switches have been
15277 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15278 into a vector from a double word element that was extracted from
15279 another vector, and eliminate extra XXPERMDI instructions.
15280 (vsx_concat_<mode>_2): Likewise.
15281 (vsx_concat_<mode>_3): Likewise.
15282 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15283 concat to allow optimizing inserts from previous extracts.
15285 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15287 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15288 memory reference to a SSP offset in TLS address space.
15289 (ix86_print_operand) <case '@'>: Remove.
15290 (ix86_print_operand_punct_valid_p): Remove '@' code.
15291 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15292 UNSPEC_SP_TLS_TEST.
15293 (stack_tls_protect_set_<mode>): Remove.
15294 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15295 (stack_tls_protect_test_<mode>): Remove.
15296 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15298 2017-08-07 Olivier Hainque <hainque@adacore.com>
15301 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15303 2017-08-07 Douglas Rupp <rupp@adacore.com>
15305 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15306 variable was referenced as multidir in command.
15308 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15311 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15314 2017-08-07 Martin Liska <mliska@suse.cz>
15316 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15317 * config/rl78/rl78.c: Add include of attribs.h.
15318 * config/sh/sh.c: Likewise.
15319 * config/v850/v850.c: Likewise.
15321 2017-08-07 Tom de Vries <tom@codesourcery.com>
15323 PR middle-end/78266
15324 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15326 2017-08-07 Martin Liska <mliska@suse.cz>
15328 * config/mips/mips.c: Include attribs.h.
15330 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15333 * doc/invoke.texi: Document change in behvaior for -Ofast for
15336 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15338 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15340 (aarch64_pop_regs): Likewise.
15341 (aarch64_gen_load_pair): Likewise.
15342 (aarch64_save_callee_saves): Likewise.
15343 (aarch64_restore_callee_saves): Likewise.
15345 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15347 * config/i386/i386.c: Revert the last change.
15349 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15352 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15354 (ix86_finalize_stack_frame_flags): This. Also clear
15355 frame_pointer_needed if -fno-omit-frame-pointer is used without
15357 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15358 with ix86_finalize_stack_frame_flags.
15359 (ix86_expand_epilogue): Likewise.
15360 (ix86_expand_split_stack_prologue): Likewise.
15362 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15365 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15366 to P_AES for Westmere.
15368 2017-08-07 Jonathan Yong <10walls@gmail.com>
15370 * config/i386/mingw.opt (fset-stack-executable): Removed.
15371 * config/i386/cygming.opt (fset-stack-executable): Moved
15373 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15375 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15377 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15379 2017-08-07 Marek Polacek <polacek@redhat.com>
15381 PR middle-end/81737
15382 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15384 2017-08-07 Martin Liska <mliska@suse.cz>
15386 * attribs.h (canonicalize_attr_name): New function.
15387 (cmp_attribs): Move from c-format.c and adjusted.
15388 (is_attribute_p): Moved from tree.h.
15389 * tree-inline.c: Add new includes.
15390 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15391 (private_is_attribute_p): Remove.
15392 (private_lookup_attribute): Likewise.
15393 (private_lookup_attribute_by_prefix): Simplify.
15394 (remove_attribute): Use is_attribute_p.
15395 * tree.h: Remove removed declarations.
15397 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15399 PR middle-end/81698
15400 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15401 instead of computing it in the function. Formatting fix.
15402 (expand_case): Don't rely on default_edge being the first edge,
15403 clear it if removing it, pass default_edge to
15404 emit_case_dispatch_table.
15405 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15408 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15410 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15411 insn in the function, emit NOP after the insn.
15413 2017-08-06 Tom de Vries <tom@codesourcery.com>
15415 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15418 2017-08-06 Tom de Vries <tom@codesourcery.com>
15420 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15423 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15425 PR tree-optimization/57371
15426 * match.pd: New pattern.
15428 2017-08-04 Marek Polacek <polacek@redhat.com>
15430 PR middle-end/81695
15431 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15432 perform the computation in offset_int.
15434 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15436 PR tree-optimization/81136
15437 * tree-vectorizer.h: Include tree-hash-traits.h.
15438 (vec_base_alignments): New typedef.
15439 (vec_info): Add a base_alignments field.
15440 (vect_record_base_alignments): Declare.
15441 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15443 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15444 (create_data_ref): Add an is_conditional_in_stmt argument.
15445 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15446 the is_conditional_in_stmt field.
15447 (data_ref_loc): Add an is_conditional_in_stmt field.
15448 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15449 (find_data_references_in_stmt): Update call to create_data_ref.
15450 (graphite_find_data_references_in_stmt): Likewise.
15451 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15452 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15453 (vect_record_base_alignment): New function.
15454 (vect_record_base_alignments): Likewise.
15455 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15456 for nested statements even if we fail to compute a misalignment.
15457 Use pooled base alignments for unconditional references.
15458 (vect_find_same_alignment_drs): Compare base addresses instead
15460 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15461 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15463 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15465 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15466 Add an explicit name for the enum. Use auto_vec for slp_instances
15467 and grouped_stores.
15468 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15470 (_bb_vec_info): Add a constructor and destructor.
15471 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15472 (destroy_loop_vec_info): Delete.
15473 (vect_destroy_datarefs): Likewise.
15474 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15475 (vec_info::vec_info): New function.
15476 (vec_info::~vec_info): Likewise.
15477 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15478 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15479 destroy_loop_vec_info.
15480 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15481 (_loop_vec_info::_loop_vec_info): ...this.
15482 (destroy_loop_vec_info): Replace with...
15483 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15484 the stmt_vec_infos. Leave handling of vec_info information to its
15485 destructor. Remove explicit vector releases.
15486 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15487 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15488 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15489 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15490 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15491 (destroy_bb_vec_info): Replace with...
15492 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15493 information to its destructor.
15494 (vect_slp_analyze_bb_1): Use new and delete instead of
15495 new_bb_vec_info and destroy_bb_vec_info.
15496 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15499 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15501 * tree-data-ref.h (subscript): Add access_fn field.
15502 (data_dependence_relation): Add could_be_independent_p.
15503 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15504 (same_access_functions): Move to tree-data-ref.c.
15505 * tree-data-ref.c (ref_contains_union_access_p): New function.
15506 (access_fn_component_p): Likewise.
15507 (access_fn_components_comparable_p): Likewise.
15508 (dr_analyze_indices): Add a reference to access_fn_component_p.
15509 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15511 (constant_access_functions): Likewise.
15512 (add_other_self_distances): Likewise.
15513 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15514 (initialize_data_dependence_relation): Use XCNEW and remove
15515 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15516 of access functions that have the same type. Allow the
15517 subsequence to end with different bases in some circumstances.
15518 Record the chosen access functions in SUB_ACCESS_FN.
15519 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15520 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15521 (subscript_dependence_tester_1): Likewise dra and drb.
15522 (build_classic_dist_vector): Update calls accordingly.
15523 (subscript_dependence_tester): Likewise.
15524 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15525 DDR_COULD_BE_INDEPENDENT_P.
15526 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15527 comp_alias_ddrs instead of may_alias_ddrs.
15528 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15530 (vect_analyze_data_ref_dependence): Use it if
15531 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15532 distance vectors if that fails.
15533 (dependence_distance_ge_vf): New function.
15534 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15535 LOOP_VINFO_MAY_ALIAS_DDRS.
15537 2017-08-04 Richard Biener <rguenther@suse.de>
15539 PR middle-end/81705
15540 * fold-const.c (fold_binary_loc): Properly restrict
15541 minus_var0 && minus_var1 case when associating undefined overflow
15544 2017-08-04 Tom de Vries <tom@codesourcery.com>
15546 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15548 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15550 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15551 Don't start diagnostic messages with a capital letter.
15552 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15554 (rs6000_invalid_builtin): Likewise.
15555 (rs6000_trampoline_init): Likewise.
15557 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15560 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15561 after setting changeable df flags.
15563 2017-08-03 Richard Biener <rguenther@suse.de>
15565 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15566 up if the use is in USE - X.
15568 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15570 * toplev.c (dumpfile.h): New include.
15571 (internal_error_reentered): New static function. Use it...
15572 (internal_error_function): ...here to handle reentered internal_error.
15574 2017-08-03 Richard Biener <rguenther@suse.de>
15576 PR middle-end/81148
15577 * fold-const.c (split_tree): Add minus_var and minus_con
15578 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15579 here but always use minus_*.
15580 (associate_trees): Assert we never associate with MINUS_EXPR
15581 and NULL first operand. Do not recurse for PLUS_EXPR operands
15582 when associating as MINUS_EXPR either.
15583 (fold_binary_loc): Track minus_var and minus_con.
15585 2017-08-03 Tom de Vries <tom@codesourcery.com>
15588 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15589 ACCEL_COMPILER, apply finish_options on
15590 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15592 2017-08-03 Tom de Vries <tom@codesourcery.com>
15595 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15596 function_entry_patch_area_size > 0.
15598 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15601 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15602 instead of 10??LU, perform unit multiplication in wide_int,
15603 don't change alloc_object_size_limit if the limit is larger
15606 PR tree-optimization/81655
15607 PR tree-optimization/81588
15608 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15609 the case when ranges[i].low and high are 1 for unsigned type with
15612 PR middle-end/81052
15613 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15614 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15616 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15618 * tree-vrp.h: Add include guard.
15620 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15623 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15624 (ud2): New insn pattern.
15625 * config/i386/i386.c (ix86_expand_epilogue):
15626 For naked functions, generate ud2 instead of trap insn.
15628 2017-08-02 Marek Polacek <polacek@redhat.com>
15631 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15633 2017-08-02 Tom de Vries <tom@codesourcery.com>
15634 Cesar Philippidis <cesar@codesourcery.com>
15636 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15637 Add missing edge probabilities.
15639 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15641 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15642 Correct endianness.
15644 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15646 PR middle-end/79499
15647 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15648 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15649 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15650 prologue_seq sequences - if any.
15652 2017-08-02 Richard Biener <rguenther@suse.de>
15654 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15655 via vectors if supported, integer extracts via punning if supported
15656 or otherwise vector extracts.
15658 2017-08-02 Richard Biener <rguenther@suse.de>
15660 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15662 (bitmap_insert_into_set): ... this.
15664 2017-08-02 Richard Biener <rguenther@suse.de>
15666 PR tree-optimization/81633
15668 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15670 PR tree-optimization/71752
15671 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15673 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15675 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15676 (machine_function::call_ms2sysv_pad_out): Remove field.
15677 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15678 (ix86_compute_frame_layout): Likewise.
15680 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15683 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15684 attribute with interrupt attribute.
15686 2017-08-01 Andrew Pinski <apinski@cavium.com>
15688 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15689 BIT_INSERT_EXPR's operand 1
15690 to see if the types precision matches.
15692 2017-08-01 Martin Liska <mliska@suse.cz>
15694 PR middle-end/70140
15695 * builtins.c (expand_builtin_memcpy_args): Remove.
15696 (expand_builtin_memcpy): Call newly added function
15697 expand_builtin_memory_copy_args.
15698 (expand_builtin_memcpy_with_bounds): Likewise.
15699 (expand_builtin_mempcpy): Remove last argument.
15700 (expand_builtin_mempcpy_with_bounds): Likewise.
15701 (expand_builtin_memory_copy_args): New function created from
15702 expand_builtin_mempcpy_args with small modifications.
15703 (expand_builtin_mempcpy_args): Remove.
15704 (expand_builtin_stpcpy): Remove unused argument.
15705 (expand_builtin): Likewise.
15706 (expand_builtin_with_bounds): Likewise.
15708 2017-08-01 Martin Liska <mliska@suse.cz>
15711 Make mempcpy more optimal (PR middle-end/70140).
15713 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15716 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15717 __builtin_vec_cmpne verify both arguments are compatible vectors
15718 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15719 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15720 move computation of aligned to after checking the argument types.
15724 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15725 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15728 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15729 Doug Rupp <rupp@adacore.com>
15730 Olivier Hainque <hainque@adacore.com>
15732 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15733 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15735 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15736 for TARGET_OS_CPP_BUILTIN.
15737 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15738 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15740 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15741 passing required abi options to the assembler for EABI configurations.
15742 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15743 of .text.hot and .text.unlikely sections for kernel modules when
15744 using ARM style exceptions.
15745 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15746 options. Add EXTRA_CC1_SPEC.
15747 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15749 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15751 (ARM_TARGET2_DWARF_FORMAT): Define.
15752 * config/arm/t-vxworks: Adjust multilib control to removal of the
15753 Diab command line options.
15755 2017-08-01 Martin Liska <mliska@suse.cz>
15757 PR gcov-profile/81561
15758 * gcov.c (unblock): Make unblocking safe as we need to preserve
15759 index correspondence of blocks and block_lists.
15761 2017-08-01 Richard Biener <rguenther@suse.de>
15763 PR tree-optimization/81181
15764 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15765 (compute_antic): ... end of iteration here.
15767 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15769 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15770 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15771 (ftree-slp-vectorize): Likewise.
15772 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15773 can no longer be set independent of flag_tree_loop_vectorize.
15774 * omp-general.c (emp_max_vf): Likewise.
15775 * opts.c (enable_fdo_optimizations): Remove references to
15776 flag_tree_vectorize, these are now implicit.
15777 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15778 and leave it for the options machinery.
15780 2017-08-01 Martin Liska <mliska@suse.cz>
15782 PR middle-end/70140
15783 * builtins.c (expand_builtin_memcpy_args): Remove.
15784 (expand_builtin_memcpy): Call newly added function
15785 expand_builtin_memory_copy_args.
15786 (expand_builtin_memcpy_with_bounds): Likewise.
15787 (expand_builtin_mempcpy): Remove last argument.
15788 (expand_builtin_mempcpy_with_bounds): Likewise.
15789 (expand_builtin_memory_copy_args): New function created from
15790 expand_builtin_mempcpy_args with small modifications.
15791 (expand_builtin_mempcpy_args): Remove.
15792 (expand_builtin_stpcpy): Remove unused argument.
15793 (expand_builtin): Likewise.
15794 (expand_builtin_with_bounds): Likewise.
15796 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15799 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15800 print "ds:" only for immediates in generic address space.
15802 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15805 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15806 (ix86_function_ok_for_sibcall): Return false for naked functions.
15808 2017-08-01 Richard Biener <rguenther@suse.de>
15810 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15811 (compute_antic): Seed worklist with exit block predecessors.
15812 * cfganal.c (dfs_find_deadend): For a cycle return the source
15813 of the edge closing it.
15815 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15817 * config/aarch64/aarch64.c
15818 (aarch64_can_const_movi_rtx_p): Move 0 check.
15820 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15822 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15823 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15825 * match.pd: Ditto in address comparison pattern.
15827 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15829 PR tree-optimization/81627
15830 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15831 closed ssa form for store-store chain.
15833 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15835 PR tree-optimization/81620
15836 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15837 for store-store chain.
15839 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15841 PR tree-optimization/81588
15842 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15843 ranges[i].in_p, invert comparison code ccode. For >/>=,
15844 swap rhs1 and rhs2 and comparison code unconditionally,
15845 for </<= don't do that. Don't swap rhs1/rhs2 again if
15846 ranges[i].in_p, instead invert comparison code ccode if
15847 opcode or oe->rank is BIT_IOR_EXPR.
15850 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15851 a direct optab to conversion optab.
15852 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15853 with GET_MODE_INNER as last argument instead of optab_handler.
15854 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15855 vector extraction if possible and optab is available.
15856 * expr.c (store_constructor): Use convert_optab_handler instead
15857 of optab_handler. Use vector initialization from smaller
15858 vectors if possible and optab is available.
15859 * tree-vect-stmts.c (vectorizable_load): Likewise.
15860 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15861 now have two modes.
15862 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15863 of vec_init from half-sized vectors with the same element mode.
15864 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15865 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15866 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15867 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15868 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15869 after mode in gen_vec_extract* calls.
15870 (vec_extract<mode>): Renamed to ...
15871 (vec_extract<mode><ssescalarmodelower>): ... this.
15872 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15873 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15874 element mode after mode in gen_vec_init* calls.
15875 (VEC_INIT_HALF_MODE): New mode iterator.
15876 (vec_init<mode>): Renamed to ...
15877 (vec_init<mode><ssescalarmodelower>): ... this.
15878 (vec_init<mode><ssehalfvecmodelower>): New expander.
15879 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15880 (vec_extractv2sfsf): ... this.
15881 (vec_initv2sf): Renamed to ...
15882 (vec_initv2sfsf): ... this.
15883 (vec_extractv2si): Renamed to ...
15884 (vec_extractv2sisi): ... this.
15885 (vec_initv2si): Renamed to ...
15886 (vec_initv2sisi): ... this.
15887 (vec_extractv4hi): Renamed to ...
15888 (vec_extractv4hihi): ... this.
15889 (vec_initv4hi): Renamed to ...
15890 (vec_initv4hihi): ... this.
15891 (vec_extractv8qi): Renamed to ...
15892 (vec_extractv8qiqi): ... this.
15893 (vec_initv8qi): Renamed to ...
15894 (vec_initv8qiqi): ... this.
15895 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15896 (vec_init<mode>): Renamed to ...
15897 (vec_init<mode><VEC_base_l>): ... this.
15898 (vec_extract<mode>): Renamed to ...
15899 (vec_extract<mode><VEC_base_l>): ... this.
15900 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15901 (vec_initv2sfsf): ... this.
15902 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15903 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15904 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15905 element mode after mode in gen_vec_init* calls.
15906 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15907 (vec_init<mode><Vel>): ... this.
15908 (vec_extract<mode>): Renamed to ...
15909 (vec_extract<mode><Vel>): ... this.
15910 * config/aarch64/iterators.md (Vel): New mode attribute.
15911 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15912 Add element mode after mode in gen_vec_extract* calls.
15913 * config/s390/vector.md (non_vec_l): New mode attribute.
15914 (vec_extract<mode>): Renamed to ...
15915 (vec_extract<mode><non_vec_l>): ... this.
15916 (vec_init<mode>): Renamed to ...
15917 (vec_init<mode><non_vec_l>): ... this.
15918 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15919 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15921 * config/arm/iterators.md (V_elem_l): New mode attribute.
15922 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15923 (vec_extract<mode><V_elem_l>): ... this.
15924 (vec_extractv2di): Renamed to ...
15925 (vec_extractv2didi): ... this.
15926 (vec_init<mode>): Renamed to ...
15927 (vec_init<mode><V_elem_l>): ... this.
15928 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15929 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15930 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15931 Add element mode after gen_vec_extract* calls.
15932 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15933 (vec_init<mode><unitmode>): ... this.
15934 (vec_extract<mode>): Renamed to ...
15935 (vec_extract<mode><unitmode>): ... this.
15936 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15937 (vec_init<mode><unitmode>): ... this.
15938 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15939 (vec_initv2sfsf): ... this.
15940 (vec_extractv2sf): Renamed to ...
15941 (vec_extractv2sfsf): ... this.
15942 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15943 Add element mode after gen_vec_extract* calls.
15944 * config/mips/mips.md (unitmode): New mode iterator.
15945 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15946 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15947 * config/spu/spu.md (inner_l): New mode attribute.
15948 (vec_init<mode>): Renamed to ...
15949 (vec_init<mode><inner_l>): ... this.
15950 (vec_extract<mode>): Renamed to ...
15951 (vec_extract<mode><inner_l>): ... this.
15952 * config/sparc/sparc.md (veltmode): New mode iterator.
15953 (vec_init<VMALL:mode>): Renamed to ...
15954 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15955 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15956 (vec_initv2sisi): ... this.
15957 (vec_initv2sf): Renamed to ...
15958 (vec_initv2sfsf): ... this.
15959 (vec_extractv2sf): Renamed to ...
15960 (vec_extractv2sfsf): ... this.
15961 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15962 (vec_init<mode>): Renamed to ...
15963 (vec_init<mode><VEC_base_l>): ... this.
15964 (vec_extract<mode>): Renamed to ...
15965 (vec_extract<mode><VEC_base_l>): ... this.
15966 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15967 (vec_initv2sfsf): ... this.
15968 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15969 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15970 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15971 gen_vec_init* calls.
15973 2017-08-01 Richard Biener <rguenther@suse.de>
15975 PR tree-optimization/81297
15976 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15977 TREE_OVERFLOW from INTEGER_CSTs.
15979 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15981 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15983 2017-07-31 Carl Love <cel@us.ibm.com>
15985 * config/rs6000/rs6000-c: Add support for built-in functions
15986 vector signed char vec_xl_be (signed long long, signed char *);
15987 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15988 vector signed int vec_xl_be (signed long long, signed int *);
15989 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15990 vector signed long long vec_xl_be (signed long long, signed long long *);
15991 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15992 vector signed short vec_xl_be (signed long long, signed short *);
15993 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15994 vector double vec_xl_be (signed long long, double *);
15995 vector float vec_xl_be (signed long long, float *);
15996 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15997 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15998 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16000 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16001 (altivec_expand_builtin): Add switch statement to call
16002 altivec_expand_xl_be for each builtin.
16003 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16004 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16005 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16006 __builtin_vsx_le_be_v16qi.
16007 * doc/extend.texi: Update the built-in documentation file for the
16008 new built-in functions.
16010 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16013 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16015 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16017 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16019 * config.gcc: Add z14.
16020 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16021 CPU model numbers for z13s and z14.
16022 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16024 * config/s390/s390-opts.h (enum processor_type): Rename
16025 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16026 * config/s390/s390.c (processor_table): Add field for CPU name to
16027 be passed to Binutils.
16028 (s390_asm_output_machine_for_arch): Use the new field in
16029 processor_table for Binutils.
16030 (s390_expand_builtin): Replace arch12 with z14.
16031 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16032 (s390_get_sched_attrmask): Likewise.
16033 (s390_get_unit_mask): Likewise.
16034 * config/s390/s390.opt: Add z14 to processor_type enum.
16036 2017-07-31 Martin Jambor <mjambor@suse.cz>
16039 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16040 regardless of optimization level.
16042 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16043 Martin Liska <mliska@suse.cz>
16045 * predict.def: Remove old comment and adjust probability.
16046 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16047 PREDICT statements.
16049 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16052 * config/i386/i386.c (ix86_function_naked): New function.
16053 (ix86_can_use_return_insn_p): Return false for naked functions.
16054 (ix86_expand_prologue): Skip prologue for naked functions.
16055 (ix86_expand_epilogue): Skip epilogue for naked functions
16056 and emit trap instruction.
16057 (ix86_warn_func_return): New function.
16058 (ix86_attribute_table): Add "naked" attribute specification.
16059 (TARGET_WARN_FUNC_RETURN): Define.
16060 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16062 2017-07-31 Martin Liska <mliska@suse.cz>
16064 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16065 (dump_gimple_bb_header): Always dump BB info.
16066 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16068 2017-07-31 Martin Liska <mliska@suse.cz>
16071 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16072 also with current_function_decl non-null equality.
16074 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16077 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16078 change type to the element type, instead add eltype variable and
16079 use it where we are interested in the element type.
16081 PR tree-optimization/81603
16082 * ipa-polymorphic-call.c
16083 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16084 offset arithmetic in offset_int, bail out if the resulting bit offset
16085 doesn't fit into shwi.
16087 2017-07-31 Martin Liska <mliska@suse.cz>
16089 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16090 (gimplify_save_expr): Fix comment.
16092 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16095 * config/i386/i386.c (ix86_function_arg): Update arguments for
16097 (ix86_compute_frame_layout): Set the initial stack offset to
16098 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16099 INCOMING_FRAME_SP_OFFSET.
16100 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16101 stack before exception handler returns.
16102 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16103 the 'ERROR_CODE' for exception handler.
16105 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16107 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16108 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16109 (ASM_OUTPUT_REG_POP): Ditto.
16110 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16111 instead of asm_fprintf to output pure string.
16113 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16115 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16116 to imported_module_or_decl hook.
16117 (debug_nothing_tree_tree_tree_bool): Remove.
16118 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16119 * debug.c (do_nothing_debug_hooks): Use
16120 debug_nothing_tree_tree_tree_bool_bool instead of
16121 debug_nothing_tree_tree_tree_bool.
16122 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16123 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16124 * sdbout.c (sdb_debug_hooks): Likewise.
16125 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16126 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16128 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16129 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16130 attribute, don't add anything.
16132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16134 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16135 (fold_build2_stat_loc): Likewise.
16136 (fold_build3_stat_loc): Likewise.
16137 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16138 (fold_build1_loc): Remove macro.
16139 (fold_build2_loc): Likewise.
16140 (fold_build3_loc): Likewise.
16142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16144 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16145 (gimple_build_debug_bind_source_stat): Likewise.
16146 * gimple.h (gimple_build_debug_bind): Remove macro.
16147 (gimple_build_debug_bind_source): Likewise.
16149 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16151 * bitmap.c (bitmap_alloc): Adjust.
16152 (bitmap_gc_alloc): Likewise.
16153 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16155 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16157 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16158 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16159 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16160 (bitmap_gc_alloc_stat): Likewise.
16161 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16163 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16165 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16166 * rtl.h (shallow_copy_rtx): Remove macro.
16168 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16170 * emit-rtl.c (gen_raw_REG): Adjust.
16171 * gengenrtl.c (gendef): Likewise.
16172 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16173 * rtl.h (rtx_alloc): Remove macro.
16175 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16177 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16178 (build_tree_list_stat): Likewise.
16179 * tree.h (build_tree_list): Remove macro.
16180 (build_tree_list_vec): Likewise.
16182 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16184 * tree.c (make_vector_stat): Remove _stat from name.
16185 (build_vector_stat): Likewise.
16186 * tree.h (make_vector_stat): Remove macro.
16187 (build_vector_stat): Likewise.
16189 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16191 * tree.h (build_var_debug_value): Remove prototype.
16193 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16195 * tree.c (tree_cons_stat): Remove _stat from name.
16196 * tree.h (tree_cons): Remove macro.
16198 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16200 * tree.c (build_vl_exp_stat): Remove _stat from name.
16201 * tree.h (build_vl_exp): Remove macro.
16203 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16205 * tree.c (build_decl_stat): Remove _stat from name.
16206 * tree.h (build_decl): Remove macro.
16208 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16210 * gimple.c (gimple_build_with_ops_stat): Adjust.
16211 (gimple_alloc_stat): Remove _stat from name.
16212 * gimple.h (gimple_alloc): Remove macro.
16214 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16216 * tree.c (make_tree_vec_stat): Remove _stat from name.
16217 (grow_tree_vec_stat): Likewise.
16218 * tree.h (make_tree_vec_stat): Adjust prototype.
16219 (grow_tree_vec_stat): Likewise.
16220 (make_tree_vec): Remove macro.
16221 (grow_tree_vec): Likewise.
16223 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16225 * fold-const.c (fold_build1_stat_loc): Adjust.
16226 (fold_build2_stat_loc): Likewise.
16227 (fold_build3_stat_loc): Likewise.
16228 * tree.c (build0_stat): Remove _stat from name.
16229 (build1_stat): Likewise.
16230 (build2_stat): Likewise.
16231 (build3_stat): Likewise.
16232 (build4_stat): Likewise.
16233 (build5_stat): Likewise.
16234 * tree.h (build1_loc): Remove macro, and rename _stat function
16236 (build2_loc): Likewise.
16237 (build3_loc): Likewise.
16238 (build4_loc): Likewise.
16239 (build5_loc): Likewise.
16241 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16243 * tree.c (make_int_cst_stat): Remove _stat from name.
16244 * tree.h (make_int_cst_stat): Adjust prototype.
16245 (make_int_cst): Remove macro.
16247 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16249 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16250 * tree.h (make_tree_binfo_stat): Adjust prototype.
16251 (make_tree_binfo): Remove.
16253 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16255 * tree.c (copy_node_stat): Rename to copy_node.
16256 (build_distinct_type_copy): Adjust.
16257 * tree.h (copy_node_stat): Adjust prototype.
16258 (copy_node): Remove macro.
16260 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16262 * tree.c (make_node_stat): rename to make_node.
16263 (build_tree_list_stat): Adjust.
16264 (build0_stat): Likewise.
16265 (build2_stat): Likewise.
16266 (build3_stat): Likewise.
16267 (build4_stat): Likewise.
16268 (build5_stat): Likewise.
16269 (build_decl_stat): Likewise.
16270 * tree.h (make_node_stat): Adjust prototype.
16271 (make_node): remove macro.
16273 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16275 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16276 (PPC_FEATURE2_SCV): Likewise.
16277 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16279 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16281 * config/aarch64/aarch64.c
16282 (aarch64_internal_mov_immediate): Add new special pattern.
16283 * config/aarch64/aarch64.md (*movdi_aarch64):
16284 Add reg/32bit const mov case.
16286 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16287 Richard Sandiford <richard.sandiford@linaro.org>
16289 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16290 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16291 Add integer and movi cases.
16292 (movi-split-hf-df-sf split, fp16): New.
16293 (enabled): Added TARGET_FP_F16INST.
16294 * config/aarch64/iterators.md (GPF_HF): New.
16295 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16297 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16299 * config/aarch64/aarch64.c
16300 (aarch64_simd_container_mode): Add prototype.
16301 (aarch64_expand_mov_immediate): Add HI support.
16302 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16303 (aarch64_can_const_movi_rtx_p): New.
16304 (aarch64_preferred_reload_class):
16305 Remove restrictions of using FP registers for certain SIMD operations.
16306 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16307 (aarch64_valid_floating_const): Add integer move validation.
16308 (aarch64_simd_imm_scalar_p): Remove.
16309 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16310 (aarch64_legitimate_constant_p): Expand list of supported cases.
16311 * config/aarch64/aarch64-protos.h
16312 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16313 (aarch64_reinterpret_float_as_int): New.
16314 (aarch64_simd_imm_scalar_p): Remove.
16315 * config/aarch64/constraints.md (Uvi): New.
16316 (Dd): Split into Ds and new Dd.
16317 * config/aarch64/aarch64.md (*movsi_aarch64):
16319 (*movdi_aarch64): Add SIMD mov case.
16321 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16323 * tree-predcom.c: (struct chain): Handle store-store chain in which
16324 stores for elimination only store loop invariant values.
16325 (execute_pred_commoning_chain): Ditto.
16326 (prepare_initializers_chain_store_elim): Ditto.
16327 (prepare_finalizers): Ditto.
16328 (is_inv_store_elimination_chain): New function.
16329 (initialize_root_vars_store_elim_1): New function.
16331 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16333 * tree-predcom.c: Revise general description of the pass.
16334 (enum chain_type): New enum type for store elimination.
16335 (struct chain): New field supporting store elimination.
16336 (struct component): Ditto.
16337 (dump_chain): Dump store-stores chain.
16338 (release_chain): Release resources.
16339 (split_data_refs_to_components): Compute and create component
16340 contains only stores for elimination.
16341 (get_chain_last_ref_at): New function.
16342 (make_invariant_chain): Initialization.
16343 (make_rooted_chain): Specify chain type in parameter and record it.
16344 (add_looparound_copies): Skip for store-stores chain.
16345 (determine_roots_comp): Compute type of chain and pass it to
16347 (initialize_root_vars_store_elim_2): New function.
16348 (finalize_eliminated_stores): New function.
16349 (remove_stmt): Handle store for elimination.
16350 (execute_pred_commoning_chain): Execute predictive commoning on
16351 store-store chains.
16352 (determine_unroll_factor): Skip unroll for store-stores chain.
16353 (prepare_initializers_chain_store_elim): New function.
16354 (prepare_initializers_chain): Hanlde store-store chain.
16355 (prepare_finalizers_chain, prepare_finalizers): New function.
16356 (tree_predictive_commoning_loop): Return integer value indicating
16357 if loop is unrolled or lcssa form is corrupted.
16358 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16360 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16362 * tree-predcom.c (initialize_root): Delete.
16363 (execute_pred_commoning_chain): Initialize root vars and replace
16364 reference of non-combined chain directly, rather than call above
16367 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16369 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16370 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16372 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16374 * tree-predcom.c (struct chain): New field init_seq.
16375 (release_chain): Release init_seq.
16376 (prepare_initializers_chain): Record intialization stmts in above
16378 (insert_init_seqs): New function.
16379 (tree_predictive_commoning_loop): Call insert_init_seqs.
16381 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16383 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16385 2017-07-28 Richard Biener <rguenther@suse.de>
16387 * match.pd: Remove superfluous :c.
16388 * genmatch.c (simplify::id): Add member.
16389 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16391 (current_id): New global.
16392 (dt_node::parent): Move from ...
16393 (dt_operand::parent): ... here. Add for_id member.
16394 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16395 (decision_tree::find_node): Relax order requirement when
16396 merging DT_TRUE nodes to ones inbetween the current simplify
16397 and the one we try to merge with. Add diagnostic whenever
16398 we need to enforce pattern order by not merging.
16399 (decision_tree::insert): Set current_id.
16400 (decision_tree::print_node): Dump parent node and for_id.
16401 (parser::last_id): Add member.
16402 (parser::push_simplify): Assign unique id.
16403 (parser::parser): Initialize last_id.
16405 2017-07-28 Martin Liska <mliska@suse.cz>
16408 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16409 gimple_build_debug_bind.
16411 2017-07-28 Richard Biener <rguenther@suse.de>
16413 PR tree-optimization/81502
16414 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16416 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16418 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16419 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16420 for BIT_FIELD_REF args.
16421 * fold-const.c (make_bit_field_ref): Likewise.
16422 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16424 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16427 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16428 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16429 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16430 Or it into SANITIZER_UNDEFINED.
16431 * ubsan.c: Include gimple-fold.h and varasm.h.
16432 (ubsan_expand_ptr_ifn): New function.
16433 (instrument_pointer_overflow): New function.
16434 (maybe_instrument_pointer_overflow): New function.
16435 (instrument_object_size): Formatting fix.
16436 (pass_ubsan::execute): Call instrument_pointer_overflow
16437 and maybe_instrument_pointer_overflow.
16438 * internal-fn.c (expand_UBSAN_PTR): New function.
16439 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16440 * sanitizer.def (__ubsan_handle_pointer_overflow,
16441 __ubsan_handle_pointer_overflow_abort): New builtins.
16442 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16443 * internal-fn.def (UBSAN_PTR): New internal function.
16444 * opts.c (sanitizer_opts): Add pointer-overflow.
16445 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16446 * fold-const.c (build_range_check): Compute pointer range check in
16447 integral type if pointer arithmetics would be needed. Formatting
16450 2017-07-28 Martin Liska <mliska@suse.cz>
16453 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16454 parameters that are of a variable-length.
16456 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16458 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16460 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16461 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16462 (CRT_CALL_STATIC_FUNCTION): Likewise.
16463 (ASM_DEFAULT_SPEC): New define.
16464 (ASM_SPEC32): Likewise.
16465 (ASM_SPEC64): Likewise.
16466 (ASM_SPEC_COMMON): Likewise.
16467 (ASM_SPEC): Likewise.
16468 (INVALID_64BIT): Likewise.
16469 (LINK_OS_DEFAULT_SPEC): Likewise.
16470 (LINK_OS_SPEC32): Likewise.
16471 (LINK_OS_SPEC64): Likewise.
16472 (POWERPC_LINUX): Likewise.
16473 (PTRDIFF_TYPE): Likewise.
16474 (RESTORE_FP_PREFIX): Likewise.
16475 (RESTORE_FP_SUFFIX): Likewise.
16476 (SAVE_FP_PREFIX): Likewise.
16477 (SAVE_FP_SUFFIX): Likewise.
16478 (SIZE_TYPE): Likewise.
16479 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16480 (TARGET_64BIT): Likewise.
16481 (TARGET_64BIT): Likewise.
16482 (TARGET_AIX): Likewise.
16483 (WCHAR_TYPE_SIZE): Likewise.
16484 (WCHAR_TYPE): Undefine.
16485 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16486 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16487 (CPP_OS_RTEMS_SPEC): Delete.
16488 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16489 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16491 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16493 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16495 PR tree-optimization/81578
16496 * tree-parloops.c (build_new_reduction): Bail out if
16497 reduction_code isn't one of the standard OpenMP reductions.
16498 Move the details printing after that decision.
16500 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16502 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16503 related to reload_in_progress.
16504 (splat_input_operand): Likewise.
16505 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16507 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16509 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16510 (TARGET_INSTANTIATE_DECLS): Likewise.
16511 (legitimate_indexed_address_p): Delete reload_in_progress code.
16512 (rs6000_debug_legitimate_address_p): Likewise.
16513 (rs6000_eliminate_indexed_memrefs): Likewise.
16514 (rs6000_emit_le_vsx_store): Likewise.
16515 (rs6000_emit_move_si_sf_subreg): Likewise.
16516 (rs6000_emit_move): Likewise.
16517 (register_to_reg_type): Likewise.
16518 (rs6000_pre_atomic_barrier): Likewise.
16519 (rs6000_machopic_legitimize_pic_address): Likewise.
16520 (rs6000_allocate_stack_temp): Likewise.
16521 (rs6000_address_for_fpconvert): Likewise.
16522 (rs6000_address_for_altivec): Likewise.
16523 (rs6000_secondary_memory_needed_rtx): Delete function.
16524 (rs6000_check_sdmode): Likewise.
16525 (rs6000_alloc_sdmode_stack_slot): Likewise.
16526 (rs6000_instantiate_decls): Likewise.
16527 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16528 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16529 Delete reload_in_progress.
16530 (*vec_reload_and_plus_<mptrsize>): Likewise.
16531 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16532 (vsx_div_v2di): Likewise.
16533 (vsx_udiv_v2di): Likewise.
16535 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16537 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16538 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16539 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16540 (rs6000_debug_reg_global): Delete print of LRA status.
16541 (rs6000_option_override_internal): Delete dead LRA related code.
16542 (rs6000_lra_p): Delete function.
16543 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16545 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16547 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16548 * config/riscv/rtems.h: New file.
16550 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16551 Sudakshina Das <sudi.das@arm.com>
16553 * config/aarch64/aarch64.md
16554 (define_split for and<mode>3nr_compare): Move
16555 non aarch64_logical_operand to a register.
16556 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16557 register immediate operand to a register.
16558 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16560 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16562 PR middle-end/81564
16563 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16565 2017-07-27 Richard Biener <rguenther@suse.de>
16567 PR tree-optimization/81573
16568 PR tree-optimization/81494
16569 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16570 multi defuse cycle case.
16572 2017-07-27 Richard Biener <rguenther@suse.de>
16574 PR tree-optimization/81571
16575 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16578 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16580 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16581 earlier and only if MASK_FPU is set. Adjust formatting.
16583 2017-07-27 Martin Liska <mliska@suse.cz>
16585 * opt-functions.awk: Add validation of value of Init.
16586 * optc-gen.awk: Pass new argument.
16588 2017-07-27 Martin Liska <mliska@suse.cz>
16590 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16591 Fix wrong condition.
16593 2017-07-27 Martin Liska <mliska@suse.cz>
16595 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16596 BBs and edges seen by autoFDO.
16598 2017-07-27 Richard Biener <rguenther@suse.de>
16600 PR tree-optimization/81502
16601 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16602 with incompatible but same sized type.
16603 (execute_update_addresses_taken): Likewise.
16605 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16607 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16608 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16610 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16613 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16614 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16615 Change s_operand to memory_operand.
16617 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16619 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16620 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16621 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16622 Emit instructions rather than returning an expression. Handle TFmode
16623 and KFmode by casting to TImode.
16624 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16625 (rs6000_emit_le_vsx_store): Likewise.
16626 * config/rs6000/vsx.md (VSX_TI): New iterator.
16627 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16628 (*vsx_le_undo_permute_<mode>): Likewise.
16629 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16630 emit the split sequence.
16631 (*vsx_le_perm_store_<mode>): Likewise.
16633 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16635 PR tree-optimization/81555
16636 PR tree-optimization/81556
16637 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16638 if true, force CHANGED for the recursive invocation.
16639 (reassociate_bb): Remember original length of ops array, pass
16640 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16642 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16643 attributes for noipa attribute. For naked attribute use
16644 lookup_attribute first before lookup_attribute_spec.
16645 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16647 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16648 for functions with noipa attribute.
16649 (cgraph_externally_visible_p): Return true for functions with noipa
16651 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16652 for functions with noipa attribute.
16653 * doc/extend.texi: Document noipa function attribute.
16654 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16655 also for functions with noipa attribute.
16656 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16658 2017-07-26 Andrew Pinski <apinski@cavium.com>
16660 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16661 vec_unalign_load_cost and vec_unalign_store_cost.
16663 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16665 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16666 -mvsx-small-integer option.
16667 (ISA_3_0_MASKS_IEEE): Likewise.
16668 (OTHER_VSX_VECTOR_MASKS): Likewise.
16669 (POWERPC_MASKS): Likewise.
16670 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16671 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16672 code, only testing for DImode being allowed in non-VSX floating
16674 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16675 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16677 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16678 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16679 TARGET_P8_VECTOR test.
16680 (rs6000_secondary_reload_simple_move): Likewise.
16681 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16682 since TARGET_P9_VECTOR was already tested.
16683 (rs6000_opt_masks): Remove -mvsx-small-integer.
16684 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16685 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16687 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16688 test for TARGET_VEXTRACTUB was used, and that uses
16690 (p9 extract splitter): Likewise.
16691 (vsx_extract_<mode>_di_p9): Likewise.
16692 (vsx_extract_<mode>_store_p9): Likewise.
16693 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16694 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16695 the elimination of TARGET_VSX_SMALL_INTEGER.
16696 (vsx_extract_<mode>_p8): Likewise.
16697 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16698 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16699 (vsx_set_<mode>_p9): Likewise.
16700 (vsx_set_v4sf_p9): Likewise.
16701 (vsx_set_v4sf_p9_zero): Likewise.
16702 (vsx_insert_extract_v4sf_p9): Likewise.
16703 (vsx_insert_extract_v4sf_p9_2): Likewise.
16704 * config/rs6000/rs6000.md (sign extend splitter): Change
16705 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16706 (floatsi<mode>2_lfiwax_mem): Likewise.
16707 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16708 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16709 since a test for TARGET_P9_VECTOR was used.
16710 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16711 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16712 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16713 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16714 TARGET_P8_VECTOR test.
16715 (fix_trunc<mode>si2_stfiwx): Likewise.
16716 (fix_trunc<mode>si2_internal): Likewise.
16717 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16718 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16720 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16721 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16722 TARGET_P8_VECTOR test.
16723 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16724 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16725 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16727 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16728 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16729 since a test for TARGET_P9_VECTOR was used.
16730 (splitter for loading small constants): Likewise.
16732 2017-07-26 Andrew Pinski <apinski@cavium.com>
16734 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16737 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16740 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16741 (fp_valid_at): Likewise.
16743 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16745 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16746 (qdf24xx_addrcost_table): Likewise.
16747 (cortexa57_tunings): Update to use generic_branch_cost.
16748 (cortexa72_tunings): Likewise.
16749 (cortexa73_tunings): Likewise.
16750 (qdf24xx_tunings): Likewise.
16752 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16754 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16755 (thunderx2t99_branch_cost): Likewise.
16756 (cortexa35_tunings): Update to use generic_branch_cost.
16757 (cortexa53_tunings): Likewise.
16758 (cortexa57_tunings): Likewise.
16759 (cortexa72_tunings): Likewise.
16760 (cortexa73_tunings): Likewise.
16761 (thunderx2t99_tunings): Likewise.
16763 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16765 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16766 (sparc_option_override): Honour MASK_FSMULD.
16767 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16768 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16769 * config/sparc/sparc.opt (mfsmuld): New option.
16770 * doc/invoke.texi (mfsmuld): Document option.
16772 2017-07-26 Marek Polacek <polacek@redhat.com>
16774 PR middle-end/70992
16775 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16777 2017-07-26 Richard Biener <rguenther@suse.de>
16779 * gimple-match-head.c (do_valueize): Return OP if valueize
16781 (get_def): New helper to get at the def stmt of a SSA name
16782 if valueize allows.
16783 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16784 do_valueize to get at the def stmt.
16785 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16787 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16789 PR middle-end/46932
16790 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16792 2017-07-26 Martin Liska <mliska@suse.cz>
16795 * function.c (expand_function_start): Make expansion of
16796 nonlocal_goto_save_area after parm_birth_insn.
16798 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16800 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16801 from all CPU target flags enable members.
16803 2017-07-26 Richard Biener <rguenther@suse.de>
16805 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16806 (decision_tree::gen): Make 'type' const.
16807 (write_predicate): Likewise.
16809 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16811 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16812 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16813 (rs6000_option_override_internal): Likewise.
16814 (rs6000_expand_vector_set): Likewise.
16815 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16816 (TARGET_UPPER_REGS_SF): Likewise.
16817 (TARGET_UPPER_REGS_DI): Likewise.
16818 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16819 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16820 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16821 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16822 (Splitters for DI constants in Altivec registers): Likewise.
16823 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16824 (vsx_set_v4sf_p9): Likewise.
16825 (vsx_set_v4sf_p9_zero): Likewise.
16826 (vsx_insert_extract_v4sf_p9): Likewise.
16827 (vsx_insert_extract_v4sf_p9_2): Likewise.
16829 2017-07-25 Carl Love <cel@us.ibm.com>
16831 * doc/extend.texi: Update the built-in documentation file for the
16832 existing built-in functions
16833 vector signed char vec_cnttz (vector signed char);
16834 vector unsigned char vec_cnttz (vector unsigned char);
16835 vector signed short vec_cnttz (vector signed short);
16836 vector unsigned short vec_cnttz (vector unsigned short);
16837 vector signed int vec_cnttz (vector signed int);
16838 vector unsigned int vec_cnttz (vector unsigned int);
16839 vector signed long long vec_cnttz (vector signed long long);
16840 vector unsigned long long vec_cnttz (vector unsigned long long);
16842 2017-07-25 Andrew Pinski <apinski@cavium.com>
16844 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16845 accesses where the use is for the first operand of a BIT_INSERT.
16847 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16850 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16851 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16853 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16855 * config/i386/gstabs.h: Delete.
16856 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16858 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16860 * config/i386/i386.c (ix86_decompose_address): Do not check for
16861 register RTX when looking at index_reg or base_reg.
16862 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16864 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16866 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16867 to update EH info here.
16869 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16871 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16873 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16875 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16877 2017-07-25 Torsten Duwe <duwe@suse.de>
16879 * common.opt: Introduce -fpatchable-function-entry
16880 command line option, and its variables function_entry_patch_area_size
16881 and function_entry_patch_area_start.
16882 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16883 including a two-value parser.
16884 * target.def (print_patchable_function_entry): New target hook.
16885 * targhooks.h (default_print_patchable_function_entry): New function.
16886 * targhooks.c (default_print_patchable_function_entry): Likewise.
16887 * toplev.c (process_options): Switch off IPA-RA if
16888 patchable function entries are being generated.
16889 * varasm.c (assemble_start_function): Look at the
16890 patchable-function-entry command line switch and current
16891 function attributes and maybe generate NOP instructions by
16892 calling the print_patchable_function_entry hook.
16893 * doc/extend.texi: Document patchable_function_entry attribute.
16894 * doc/invoke.texi: Document -fpatchable_function_entry
16895 command line option.
16896 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16898 * doc/tm.texi: Re-generate.
16900 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16903 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16904 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16906 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16908 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16910 2017-07-25 Richard Biener <rguenther@suse.de>
16912 PR tree-optimization/81455
16913 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16914 not walk in cycles when looking for guards.
16916 2017-07-25 Richard Biener <rguenther@suse.de>
16918 PR tree-optimization/81529
16919 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16920 when optimizing backedge uses.
16922 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16924 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16926 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16927 to dl_section_ref. On AIX, append an expression to subtract
16928 the size of the section length to dl_section_ref.
16930 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16932 * configure.ac: If any of the config.* scripts fail, exit 1.
16933 * configure: Regenerate.
16935 2017-07-25 Richard Biener <rguenther@suse.de>
16937 PR middle-end/81546
16938 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16939 of immediate uses, be more verbose on errors.
16941 2017-07-25 Richard Biener <rguenther@suse.de>
16943 PR tree-optimization/81510
16944 * tree-vect-loop.c (vect_is_simple_reduction): When the
16945 reduction stmt is not inside the loop bail out.
16947 2017-07-25 Richard Biener <rguenther@suse.de>
16949 PR tree-optimization/81303
16950 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16951 profitability check against LOOP_VINFO_NITERSM1.
16953 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16955 * domwalk.c (cmp_bb_postorder): Simplify.
16956 (sort_bbs_postorder): New function. Use it...
16957 (dom_walker::walk): ...here to optimize common cases.
16959 2017-07-25 Martin Liska <mliska@suse.cz>
16962 * ipa-visibility.c (function_and_variable_visibility): Make the
16963 redirection just on target that supports aliasing.
16964 Fix GNU coding style.
16966 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16969 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16971 * config/arm/rtems.h: Likewise.
16972 * config/bfin/rtems.h: Likewise.
16973 * config/i386/rtemself.h: Likewise.
16974 * config/lm32/rtems.h: Likewise.
16975 * config/m32c/rtems.h: Likewise.
16976 * config/m68k/rtemself.h: Likewise.
16977 * config/microblaze/rtems.h: Likewise.
16978 * config/mips/rtems.h: Likewise.
16979 * config/moxie/rtems.h: Likewise.
16980 * config/nios2/rtems.h: Likewise.
16981 * config/powerpcspe/rtems.h: Likewise.
16982 * config/rs6000/rtems.h: Likewise.
16983 * config/rtems.h: Likewise.
16984 * config/sh/rtems.h: Likewise.
16985 * config/sh/rtemself.h: Likewise.
16986 * config/sparc/rtemself.h: Likewise.
16988 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16991 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16992 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16993 * tree-ssa-structalias.c (alias_get_name): Same.
16995 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16998 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16999 instructions if no du chain is found.
17001 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17003 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17005 2017-07-25 Richard Biener <rguenther@suse.de>
17007 PR middle-end/81505
17008 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17011 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17013 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17014 upper-regs options.
17015 (ISA_2_7_MASKS_SERVER): Likewise.
17016 (ISA_3_0_MASKS_IEEE): Likewise.
17017 (OTHER_P8_VECTOR_MASKS): Likewise.
17018 (OTHER_VSX_VECTOR_MASKS): Likewise.
17019 (POWERPC_MASKS): Likewise.
17020 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17021 duplicate list of options.
17022 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17023 explicit -mupper-regs options.
17024 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17025 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17026 alias for -mupper-regs-df.
17027 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17028 (rs6000_init_hard_regno_mode_ok): Likewise.
17029 (rs6000_option_override_internal): Likewise.
17030 (rs6000_opt_masks): Likewise.
17031 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17032 options in terms of whether -mvsx or -mpower8-vector was used.
17033 (TARGET_UPPER_REGS_DI): Likewise.
17034 (TARGET_UPPER_REGS_SF): Likewise.
17035 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17036 -mupper-regs-* options.
17038 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17040 * passes.c (emergency_dump_function): Print some empty lines and a
17041 header before the RTL dump.
17043 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17045 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17047 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17050 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17051 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17053 2017-07-24 Carl Love <cel@us.ibm.com>
17055 * config/rs6000/rs6000-c.c: Add support for built-in functions
17056 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17057 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17058 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17059 vec_extract_fp_from_shortl): Add defines for the two builtins.
17060 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17061 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17063 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17064 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17065 * doc/extend.texi: Update the built-in documentation file for the
17066 new built-in function.
17068 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17071 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17073 * doc/generic.texi: Likewise.
17074 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17075 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17077 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17079 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17080 (aarch64_mls_elt_merge<mode>): Likewise.
17082 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17084 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17085 having __cxa_atexit.
17087 2017-07-23 Michael Collison <michael.collison@arm.com>
17089 * config/arm/arm.c (arm_option_override): Deprecate
17090 use of -mstructure-size-boundary.
17091 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17092 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17094 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17097 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17098 Reduce cost estimate for direct moves.
17100 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17103 * config/i386/i386.c (ix86_option_override_internal): Disable
17104 BMI, BMI2 and TBM instructions for -m16.
17106 2017-07-21 Carl Love <cel@us.ibm.com>
17108 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17109 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17110 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17111 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17112 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17113 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17114 VMULOSW): New enum "unspec" values.
17115 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17116 altivec_vmulosw): New patterns.
17117 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17118 VMULOSW): Add definitions.
17120 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17122 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17123 (qdf24xx): Likewise.
17124 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17125 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17126 (AARCH64_FL_V8_1): Renumber.
17127 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17128 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17129 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17130 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17131 rdma to feature modifiers list.
17133 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17135 PR middle-end/56727
17136 * ipa-visibility (function_and_variable_visibility): Convert
17137 recursive PLT call to direct call if appropriate.
17139 2017-07-21 Andrew Pinski <apinski@cavium.com>
17141 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17142 operand 1 to see if the types precision matches.
17143 * fold-const.c (operand_equal_p): Likewise.
17145 2017-07-21 Richard Biener <rguenther@suse.de>
17147 PR tree-optimization/81303
17148 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17149 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17150 (vect_peeling_hash_get_lowest_cost): Adjust.
17151 (vect_enhance_data_refs_alignment): Likewise. Use
17152 vect_get_peeling_costs_all_drs to compute the penalty for no
17153 peeling to match up costs.
17155 2017-07-21 Richard Biener <rguenther@suse.de>
17157 PR tree-optimization/81500
17158 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17159 we didn't identify a reduction path.
17161 2017-07-21 Tom de Vries <tom@codesourcery.com>
17162 Cesar Philippidis <cesar@codesourcery.com>
17164 PR gcov-profile/81442
17165 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17168 2017-07-21 Tom de Vries <tom@codesourcery.com>
17171 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17173 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17174 nvptx_override_options_after_change.
17176 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17178 * dwarf2out.c (output_file_names): Avoid double testing for
17179 dwarf_version >= 5.
17181 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17183 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17185 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17187 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17189 (try_crossjump_to_edge): Do not punt on partitioned functions.
17191 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17193 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17194 Put all BBs reachable only via paths crossing cold region to cold
17196 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17198 2016-07-21 Richard Biener <rguenther@suse.de>
17200 PR tree-optimization/81303
17201 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17202 into account prologue and epilogue iterations when raising
17203 min_profitable_iters to sth at least covering one vector iteration.
17205 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17207 * config/arm/arm.c (arm_test_cpu_arch_dat):
17210 2017-07-20 Nathan Sidwell <nathan@acm.org>
17212 Remove TYPE_METHODS.
17213 * tree.h (TYPE_METHODS): Delete.
17214 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17215 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17216 (dbxout_type_methods): Scan TYPE_FIELDS.
17217 (dbxout_type): Don't check TYPE_METHODS here.
17218 * function.c (use_register_for_decl): Always ignore register for
17219 class types when not optimizing.
17220 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17221 * tree.c (free_lang_data_in_type): Stitch out member functions and
17222 templates from TYPE_FIELDS.
17223 (build_distinct_type_copy, verify_type_variant,
17224 verify_type): Member fns are on TYPE_FIELDS.
17225 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17226 * tree-pretty-print.c (dump_generic_node): Likewise.
17228 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17231 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17232 V2TImode and V4TImode.
17233 (ix86_expand_vector_extract): Likewise.
17234 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17235 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17236 (ssescalarmode): Handle V4TImode and V2TImode.
17237 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17238 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17239 (VEXTRACTI128_MODE): New mode iterator.
17240 (splitter for *vec_extractv?ti first element): New.
17241 (VEC_INIT_MODE): New mode iterator.
17242 (vec_init<mode>): Consolidate 3 expanders into one using
17243 VEC_INIT_MODE mode iterator.
17245 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17247 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17248 non_spilled_static_chain_regno_p.
17250 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17252 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17254 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17256 * bb-reorder.c (connect_traces): Allow copying of blocks within
17259 2017-07-20 Richard Biener <rguenther@suse.de>
17261 * gimple.h (gimple_phi_result): Add gphi * overload.
17262 (gimple_phi_result_ptr): Likewise.
17263 (gimple_phi_arg): Likewise. Adjust index assert to only
17264 allow actual argument accesses rather than all slots available
17266 (gimple_phi_arg_def): Add gphi * overload.
17267 * tree-phinodes.c (make_phi_node): Initialize only actual
17269 (resize_phi_node): Clear memory not covered by old node,
17270 do not initialize excess argument slots.
17271 (reserve_phi_args_for_new_edge): Initialize new argument slot
17274 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17276 PR tree-optimization/81388
17278 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17280 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17281 by removing computation of may_be_zero.
17283 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17284 Tom de Vries <tom@codesourcery.com>
17286 PR middle-end/81030
17287 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17288 when gimple level profile disagrees with what RTL expander did.
17290 2017-07-20 Richard Biener <rguenther@suse.de>
17292 PR tree-optimization/61171
17293 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17294 (vect_analyze_stmt): Add slp instance parameter.
17295 (vectorizable_reduction): Likewise.
17296 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17297 (vect_is_simple_reduction): Deal with chains not detected
17298 as SLP reduction chain, specifically not properly associated
17299 chains containing a mix of plus/minus.
17300 (get_reduction_op): Remove.
17301 (get_initial_defs_for_reduction): Simplify, pass in whether
17302 this is a reduction chain, pass in the SLP node for the PHIs.
17303 (vect_create_epilog_for_reduction): Get the SLP instance as
17305 (vectorizable_reduction): Get the SLP instance as arg.
17306 During analysis remember the SLP node with the PHIs in the
17307 instance. Simplify getting at the vectorized reduction PHIs.
17308 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17309 through SLP instance.
17310 (vect_slp_analyze_operations): Likewise.
17311 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17312 (vect_transform_stmt): Likewise.
17314 2017-07-20 Tom de Vries <tom@codesourcery.com>
17316 PR tree-optimization/81489
17317 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17318 read of phi arg location to before loop that modifies phi.
17320 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17322 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17325 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17327 PR middle-end/81331
17328 * except.c (execute): Fix ordering issue.
17330 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17332 PR rtl-optimization/81423
17333 * combine.c (make_compound_operation_int): Don't try to optimize
17334 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17336 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17338 PR rtl-optimization/81423
17339 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17340 with a constant that is -1 in the truncated to mode.
17342 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17344 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17345 (determine_unlikely_bbs): ... here.
17346 * predict.h (propagate_unlikely_bbs_forward): Declare.
17347 * cfgexpand.c (pass_expand::execute): Use it.
17348 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17350 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17351 propagate_unlikely_bbs_forward.
17353 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17355 PR middle-end/81331
17356 * except.c (maybe_add_nop_after_section_switch): New function.
17359 2017-07-19 Tom de Vries <tom@codesourcery.com>
17361 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17363 2017-07-19 Tom de Vries <tom@codesourcery.com>
17365 * gimple.h (gimple_phi_arg): Make assert more strict.
17367 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17369 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17370 * config/rs6000/mmintrin.h: New file.
17371 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17373 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17375 PR tree-optimization/81346
17376 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17378 2017-07-19 Tom de Vries <tom@codesourcery.com>
17380 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17382 2017-07-19 Tom de Vries <tom@codesourcery.com>
17384 * config/nvptx/nvptx-modes.def: Add V2DImode.
17385 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17386 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17387 (nvptx_output_mov_insn): Handle lack of mov.b128.
17388 (nvptx_print_operand): Handle 'H' and 'L' codes.
17389 (nvptx_vector_mode_supported): Allow V2DImode.
17390 (nvptx_preferred_simd_mode): New function.
17391 (nvptx_data_alignment): New function.
17392 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17393 nvptx_preferred_simd_mode.
17394 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17396 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17398 2017-07-19 Tom de Vries <tom@codesourcery.com>
17400 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17401 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17402 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17403 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17404 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17405 (mov<VECIM>_insn): New define_insn.
17406 (define_expand "mov<VECIM>): New define_expand.
17408 2017-07-19 Tom de Vries <tom@codesourcery.com>
17410 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17412 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17414 PR tree-optimization/81346
17415 * fold-const.h (fold_div_compare, range_check_type): Declare.
17416 * fold-const.c (range_check_type): New function.
17417 (build_range_check): Use range_check_type.
17418 (fold_div_compare): No longer static, rewritten into
17419 a match.pd helper function.
17420 (fold_comparison): Don't call fold_div_compare here.
17421 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17422 as helper function.
17424 2017-07-19 Nathan Sidwell <nathan@acm.org>
17426 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17427 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17428 * tree.c (find_decls_types_r, verify_type): Use
17429 TYPE_{MIN,MAX}_VALUE_RAW.
17430 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17431 (hash_tree): Likewise.
17432 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17434 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17437 2017-07-18 Tom de Vries <tom@codesourcery.com>
17439 PR middle-end/81464
17440 * omp-expand.c (expand_omp_for_static_chunk): Handle
17441 equal-argument loop exit phi.
17443 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17446 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17447 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17448 operand 2 predicate.
17449 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17450 operand 2 predicate.
17451 (ror,rol -> rorx splitters): Use const_int_operand as
17452 operand 2 predicate.
17454 2017-06-18 Richard Biener <rguenther@suse.de>
17456 PR tree-optimization/81410
17457 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17458 the gap in the ! slp_perm SLP case after each group.
17460 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17462 PR middle-end/81463
17463 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17466 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17468 PR middle-end/81462
17469 * predict.c (set_even_probabilities): Cleanup; do not affect
17470 probabilities that are already known.
17471 (combine_predictions_for_bb): Call even when count is set.
17473 2017-07-18 Nathan Sidwell <nathan@acm.org>
17475 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17478 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17481 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17482 optimization for loop niter analysis.
17484 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17487 * config/avr/avr.c (avr_optimize_casesi): Don't use
17488 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17490 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17492 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17493 body_cost_vec from _vect_peel_extended_info.
17494 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17495 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17498 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17500 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17502 2017-07-18 Richard Biener <rguenther@suse.de>
17504 PR tree-optimization/80620
17505 PR tree-optimization/81403
17506 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17507 info when re-using a VN table entry.
17509 2017-07-18 Richard Biener <rguenther@suse.de>
17511 PR tree-optimization/81418
17512 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17513 vectype_in. Verify that with lane-reducing reduction operations
17514 we have a single def-use cycle.
17516 2017-07-17 Carl Love <cel@us.ibm.com>
17518 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17520 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17521 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17522 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17523 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17524 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17525 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17526 VMULOSW): New enum "unspec" values.
17527 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17528 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17529 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17530 altivec_vmulosw): New patterns.
17531 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17532 VMULOSW): Add definitions.
17533 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17535 * config/alpha/alpha.c: Include predict.h.
17537 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17539 * tree-vrp.c (compare_assert_loc): Fix comparison function
17540 to return predictable results.
17542 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17544 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17546 (subdi3): Likewise.
17547 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17548 * doc/invoke.texi (mexpand-adddi): Update text.
17550 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17552 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17553 that also clobbers the CC register. The old expand code is moved
17555 (*arc_clzsi2): ... here.
17556 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17557 the CC register. The old expand code is moved to ...
17558 (arc_ctzsi2): ... here.
17560 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17562 * config/arc/arc.opt (mindexed-loads): Use initial value
17563 TARGET_INDEXED_LOADS_DEFAULT.
17564 (mauto-modify-reg): Use initial value
17565 TARGET_AUTO_MODIFY_REG_DEFAULT.
17566 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17567 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17568 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17569 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17571 2017-07-17 Martin Liska <mliska@suse.cz>
17574 * opts.c (finish_options): Do not allow -fgnu-tm
17575 w/ -fsanitize={kernel-,}address. Say sorry.
17577 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17580 * tree-loop-distribution.c (classify_partition): Only assert on
17581 numer of iterations.
17582 (merge_dep_scc_partitions): Delete prameter. Update function call.
17583 (distribute_loop): Remove code handling loop with unknown niters.
17584 (pass_loop_distribution::execute): Skip loop with unknown niters.
17586 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17589 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17590 function sort_partitions_by_post_order.
17592 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17594 PR tree-optimization/81374
17595 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17596 the max index of basic blocks, rather than number of basic blocks.
17598 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17600 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17602 (arc_legitimate_pic_operand_p): Likewise.
17603 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17605 (arc_needs_pcl_p): Likewise.
17606 (arc_legitimate_pc_offset_p): Likewise.
17607 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17608 function is also used in constrains.md.
17609 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17610 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17611 PLUS. Only return true/false in known cases, otherwise assert.
17612 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17613 is already called in arc_legitimate_constant_p.
17614 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17616 (LEGITIMATE_PIC_OPERAND_P): Use
17617 arc_raw_symbolic_reference_mentioned_p function.
17618 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17623 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17624 Andrew Burgess <andrew.burgess@embecosm.com>
17626 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17627 (arc_return_address_register): New function.
17628 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17629 (arc_handle_fndecl_attribute): Add naked attribute.
17630 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17631 (TARGET_WARN_FUNC_RETURN): Likewise.
17632 (arc_allocate_stack_slots_for_args): New function.
17633 (arc_warn_func_return): Likewise.
17634 (machine_function): Change type fn_type.
17635 (arc_compute_function_type): Consider new naked function type,
17636 change function return type.
17637 (arc_must_save_register): Adapt to handle new
17638 arc_compute_function_type's return type.
17639 (arc_expand_prologue): Likewise.
17640 (arc_expand_epilogue): Likewise.
17641 (arc_return_address_regs): Delete.
17642 (arc_return_address_register): New function.
17643 (arc_epilogue_uses): Use above function.
17644 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17645 (arc_function_type): Change encoding, add naked type.
17646 (ARC_INTERRUPT_P): Change to handle the new encoding.
17647 (ARC_FAST_INTERRUPT_P): Likewise.
17648 (ARC_NORMAL_P): Define.
17649 (ARC_NAKED_P): Likewise.
17650 (arc_compute_function_type): Delete prototype.
17651 * config/arc/arc.md (in_ret_delay_slot): Use
17652 arc_return_address_register function.
17653 (simple_return): Likewise.
17654 (p_return_i): Likewise.
17656 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17658 PR tree-optimization/81428
17659 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17660 can't be built for those types.
17662 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17664 Remove stuff dead since r239246.
17666 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17667 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17668 (avr_inform_devices): Remove dead stuff.
17670 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17672 * config/arm/arm_neon.h: Fix softp typo.
17674 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17676 PR tree-optimization/81365
17677 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17678 aggregate moves onto bb predecessor edges, make sure there are no
17679 loads that could alias the lhs in between the start of bb and the
17682 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17685 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17686 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17687 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17689 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17691 PR tree-optimization/81396
17692 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17693 (init_symbolic_number): Initialize it to 1.
17694 (perform_symbolic_merge): Add n_ops from both operands into the new
17696 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17697 without base_addr as useless if they need more than one operation.
17698 (bswap_replace): Handle !bswap case for NULL base_addr.
17700 2017-07-17 Tom de Vries <tom@codesourcery.com>
17703 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17706 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17708 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17709 conditional builtin define __FIX_LEON3FT_B2BST.
17711 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17713 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17714 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17717 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17719 PR rtl-optimization/81424
17720 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17721 to remove potential trapping from operands if -fnon-call-exceptions.
17723 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17725 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17726 profile_proability for scalling.
17727 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17729 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17731 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17733 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17735 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17736 fixpoint arithmetics.
17738 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17740 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17741 fixpoint arithmetics.
17743 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17745 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17746 fixpoint arithmetics.
17748 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17750 * profile-count.h (profile_probability::from_reg_br_prob_note,
17751 profile_probability::to_reg_br_prob_note): New functions.
17752 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17753 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17754 * predict.c (probability_reliable_p): Update.
17755 (edge_probability_reliable_p): Update.
17756 (br_prob_note_reliable_p): Update.
17757 (invert_br_probabilities): Update.
17758 (add_reg_br_prob_note): New function.
17759 (combine_predictions_for_insn): Update.
17760 * asan.c (asan_clear_shadow): Update.
17761 * cfgbuild.c (compute_outgoing_frequencies): Update.
17762 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17763 (update_br_prob_note): Update.
17764 (rtl_verify_edges): Update.
17765 (purge_dead_edges): Update.
17766 (fixup_reorder_chain): Update.
17767 * emit-rtl.c (try_split): Update.
17768 * ifcvt.c (cond_exec_process_insns): Update.
17769 (cond_exec_process_if_block): Update.
17770 (dead_or_predicable): Update.
17771 * internal-fn.c (expand_addsub_overflow): Update.
17772 (expand_neg_overflow): Update.
17773 (expand_mul_overflow): Update.
17774 * loop-doloop.c (doloop_modify): Update.
17775 * loop-unroll.c (compare_and_jump_seq): Update.
17776 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17777 * predict.h: Update.
17778 * reorg.c (mostly_true_jump): Update.
17780 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17781 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17782 * config/arc/arc.c: (emit_unlikely_jump): Update.
17783 * config/arm/arm.c: (emit_unlikely_jump): Update.
17784 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17785 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17786 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17787 (ix86_print_operand): Update.
17788 (ix86_split_fp_branch): Update.
17789 (predict_jump): Update.
17790 * config/ia64/ia64.c (ia64_print_operand): Update.
17791 * config/mmix/mmix.c (mmix_print_operand): Update.
17792 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17793 (rs6000_expand_split_stack_prologue): Update.
17794 * config/rs6000/rs6000.c: Update.
17795 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17796 (s390_expand_vec_movstr): Update.
17797 (s390_expand_cs_tdsi): Update.
17798 (s390_expand_split_stack_prologue): Update.
17799 * config/sh/sh.c (sh_print_operand): Update.
17800 (expand_cbranchsi4): Update.
17801 (expand_cbranchdi4): Update.
17802 * config/sparc/sparc.c (output_v9branch): Update.
17803 * config/spu/spu.c (get_branch_target): Update.
17804 (ea_load_store_inline): Update.
17805 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17806 * config/tilepro/tilepro.c: Update.
17808 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17810 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17811 (gimplify_save_expr): Likewise.
17813 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17815 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17817 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17819 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17821 * timevar.def (TV_IPA_FNSUMMARY): Define.
17823 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17825 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17826 to back store errata sensitive sequence from being generated.
17827 (sqrtdf2_fix): Likewise.
17829 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17831 * tree-ssa-threadupdate.c (compute_path_counts,
17832 update_joiner_offpath_counts): Use profile_probability.
17834 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17837 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17839 * config/arm/arm-c.c (arm_cpu_builtins): Define
17840 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17842 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17844 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17845 array entries to represent __ieee128 versions of the
17846 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17847 scalar_extract_sig, and scalar_insert_exp built-in functions.
17848 (altivec_resolve_overloaded_builtin): Add special case handling
17849 for the __builtin_scalar_insert_exp function, as represented by
17850 the P9V_BUILTIN_VEC_VSIEDP constant.
17851 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17852 exponent support for __ieee128 argument.
17853 (VSESQP): Add scalar extract signature support for __ieee128
17855 (VSTDCNQP): Add scalar test negative support for __ieee128
17857 (VSIEQP): Add scalar insert exponent support for __int128 argument
17858 with __ieee128 result.
17859 (VSIEQPF): Add scalar insert exponent support for __ieee128
17860 argument with __ieee128 result.
17861 (VSTDCQP): Add scalar test data class support for __ieee128
17863 (VSTDCNQP): Add overload support for scalar test negative with
17864 __ieee128 argument.
17865 (VSTDCQP): Add overload support for scalar test data class
17866 __ieee128 argument.
17867 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17868 UNSPEC_VSX_SXSIGDP.
17869 (UNSPEC_VSX_SIEXPQP): New constant.
17870 (xsxexpqp): New insn for VSX scalar extract exponent quad
17872 (xsxsigqp): New insn for VSX scalar extract significand quad
17874 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17875 precision with floating point argument.
17876 (xststdcqp): New expand for VSX scalar test data class quad
17878 (xststdcnegqp): New expand for VSX scalar test negative quad
17880 (xststdcqp): New insn to match expansions for VSX scalar test data
17881 class quad precision and VSX scalar test negative quad precision.
17882 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17883 special case operand checking to enforce that second operand of
17884 VSX scalar test data class with quad precision argument is a 7-bit
17886 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17887 prototypes and descriptions of __ieee128 versions of
17888 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17889 scalar_test_data_class, and scalar_test_neg built-in functions.
17891 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17893 PR tree-optimization/81162
17894 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17895 replace a negate with an add.
17897 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17899 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17901 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17903 * config/arm/arm-c.c (arm_cpu_builtins): Define
17904 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17906 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17908 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17909 (armv8-r): Set ARM Cortex-R52 as default CPU.
17910 * config/arm/arm-tables.opt: Regenerate.
17911 * config/arm/arm-tune.md: Regenerate.
17912 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17914 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17915 extension for -mcpu=cortex-r52.
17917 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17919 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17920 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17921 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17922 (fp-armv8): Define it as FP_ARMv8 only.
17923 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17924 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17926 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17927 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17928 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17929 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17930 than TARGET_FPU_ARMV8.
17931 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17932 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17933 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17935 * config/arm/neon.md (neon_vrint): Likewise.
17936 (neon_vcvt): Likewise.
17937 (neon_<fmaxmin_op><mode>): Likewise.
17938 (<fmaxmin><mode>3): Likewise.
17939 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17940 * config/arm/predicates.md (arm_cond_move_operator): Check against
17941 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17943 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17945 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17946 to top of function.
17948 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17950 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17951 loop in comment with memset.
17953 2017-07-14 Martin Liska <mliska@suse.cz>
17955 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17956 * dwarf2out.c (is_java): Remove the function.
17957 (output_pubname): Remove usage of the function.
17958 (lower_bound_default): Remove usage of DW_LANG_Java.
17959 (gen_compile_unit_die): Likewise.
17960 * gcc.c: Remove compiler defaults for .java and .zip files.
17961 * gimple-expr.c (remove_suffix): Change as there's no longer
17962 extension than 4-letter one.
17963 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17964 (gimplify_save_expr): Likewise.
17965 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17966 as it's possible even for other languages than Java.
17967 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17968 * lto-opts.c (lto_write_options): Remove reference to Java.
17969 * opts.c (strip_off_ending): Update file extension handling.
17970 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17971 * tree-eh.c (lower_resx): Likewise.
17972 * tree.c (free_lang_data_in_type): Remove dead code.
17973 (find_decls_types_r): Likewise.
17974 (build_common_builtin_nodes): Remove Java from a comment.
17975 (verify_type): Remove dead code.
17976 * varasm.c (assemble_external): Remove Java from a comment.
17978 2017-07-14 Martin Liska <mliska@suse.cz>
17980 * opts.c (finish_options): Add quotes.
17981 (common_handle_option): Likewise.
17983 2017-07-14 Martin Liska <mliska@suse.cz>
17985 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17986 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17987 Remove N_SO_PASCAL.
17988 * dwarf2out.c (lower_bound_default): Do not handle
17990 (gen_compile_unit_die): Likewise.
17991 * gcc.c: Remove default extension binding for GNU Pascal.
17992 * stmt.c: Remove Pascal language from a comment.
17993 * xcoffout.c: Likewise.
17995 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17998 * diagnostic-show-locus.c (fixit_cmp): New function.
17999 (layout::layout): Sort m_fixit_hints.
18000 (column_range::column_range): Assert that the values are valid.
18001 (struct char_span): New struct.
18002 (correction::overwrite): New method.
18003 (struct source_line): New struct.
18004 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18005 calls in terms of classes source_line and char_span, and
18006 correction::overwrite.
18007 (selftest::test_overlapped_fixit_printing_2): New function.
18008 (selftest::diagnostic_show_locus_c_tests): Call it.
18010 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18012 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18013 early if there is no lhs.
18015 2017-07-13 Martin Liska <mliska@suse.cz>
18017 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18018 (gen_reference_type_die): Likewise.
18019 * stor-layout.c: Remove Pascal-related comment.
18021 2017-07-13 Martin Liska <mliska@suse.cz>
18023 * opts.c (finish_options): Add quotes to error messages.
18024 (parse_sanitizer_options): Likewise.
18026 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18028 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18030 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18032 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18034 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18036 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18038 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18040 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18043 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18044 provides the hardware capability bits, define the macro
18045 __BUILTIN_CPU_SUPPORTS__.
18046 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18047 if GLIBC does not provide the hardware capability bits. Add a
18048 gcc_unreachable call if the built-in cpu function is neither
18049 __builtin_cpu_is nor __builtin_cpu_supports.
18050 (rs6000_get_function_versions_dispatcher): Change the warning
18051 that an old GLIBC is used which does not export the capability
18052 bits to be an error.
18053 * doc/extend.texi (target_clones attribute): Document the
18054 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18055 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18056 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18057 the macros defined by GCC if the newer GLIBC is available.
18059 2017-07-12 Jeff Law <law@redhat.com>
18061 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18062 remaining includes slightly.
18063 * config/riscv/riscv-builtins.c: Include profile-count.h.
18065 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18068 * config/avr/avr.c (avr_set_current_function): In diagnostic
18069 messages: Quote keywords and (parts of) identifiers.
18070 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18073 2017-07-12 Carl Love <cel@us.ibm.com>
18075 * config/rs6000/rs6000-c.c: Add support for built-in functions
18076 vector bool char vec_revb (vector bool char);
18077 vector bool short vec_revb (vector short char);
18078 vector bool int vec_revb (vector bool int);
18079 vector bool long long vec_revb (vector bool long long);
18080 * doc/extend.texi: Update the built-in documentation file for the
18081 new built-in functions.
18083 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18085 * config/s390/s390.md: Remove movcc splitter.
18087 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18089 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18090 load/store on condition.
18092 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18095 * config/avr/avr.c (avr_encode_section_info)
18096 [progmem && !TREE_READONLY]: Error if progmem object needs
18099 2017-07-11 Michael Collison <michael.collison@arm.com>
18101 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18104 2017-07-11 Carl Love <cel@us.ibm.com>
18106 * config/rs6000/rs6000-c.c: Add support for builtins
18107 vector unsigned int vec_parity_lsbb (vector signed int);
18108 vector unsigned int vec_parity_lsbb (vector unsigned int);
18109 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18110 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18111 vector unsigned long long vec_parity_lsbb (vector signed long long);
18112 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18113 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18114 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18115 * doc/extend.texi: Update the built-in documentation file for the
18116 new built-in functions.
18118 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18120 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18121 (layout::m_primary_loc): New field.
18122 (layout::layout): Initialize new field. Move location filtering
18123 logic from here to...
18124 (layout::maybe_add_location_range): ...this new method. Add
18125 support for filtering to just the lines already specified by other
18127 (layout::will_show_line_p): New method.
18128 (gcc_rich_location::add_location_if_nearby): New method.
18129 (selftest::test_add_location_if_nearby): New test function.
18130 (selftest::diagnostic_show_locus_c_tests): Call it.
18131 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18134 2017-07-11 Tom de Vries <tom@codesourcery.com>
18136 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18137 (bb_first_real_insn): New function.
18138 (nvptx_single): Add extra initialization of broadcasted condition
18141 2017-07-11 Nathan Sidwell <nathan@acm.org>
18143 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18145 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18147 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18148 Binutils doc as TEXI will mess them up.
18149 * doc/invoke.texi (AVR Options): Same here.
18151 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18153 * config/sparc/sparc.opt (mfix-ut700): New option.
18154 (mfix-gr712rc): Likewise.
18155 (sparc_fix_b2bst): New variable.
18156 * doc/invoke.texi (SPARC options): Document them.
18157 (ARM options): Fix warnings.
18158 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18159 instructions to prevent sequences that can trigger the store-store
18160 errata for certain LEON3FT processors.
18161 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18162 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18163 * config/sparc/sparc.md (fix_b2bst): New attribute.
18164 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18166 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18169 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18171 (*rsqrtsf2_sse): Ditto.
18173 (div<mode>3): Macroize insn from divdf3 and divsf3
18174 using MODEF mode iterator.
18176 2017-07-10 Martin Sebor <msebor@redhat.com>
18178 PR tree-optimization/80397
18179 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18180 instead of testing for equality to INTEGER_TYPE.
18182 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18184 * config.gcc: Remove uclibc from arc target spec.
18186 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18188 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18190 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18193 * lto-wrapper.c (remove_option): New function.
18194 (merge_and_complain): Merge PIC/PIE options more realistically.
18196 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18198 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18202 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18203 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18204 * config.in: Regenerate.
18205 * configure: Regenerate.
18206 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18207 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18208 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18209 (TARGET_GASISR_PROLOGUES): ...target mask.
18210 * common/config/avr/avr-common.c
18211 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18212 Set -mgas-isr-prologues.
18213 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18214 INSERT_PASS_BEFORE for it.
18215 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18216 * config/avr/avr.c (avr_option_override)
18217 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18218 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18219 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18220 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18221 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18223 (make_avr_pass_pre_proep): New function.
18224 (emit_push_sfr) <treg>: Add argument to function and use it
18225 instead of TMP_REG.
18226 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18227 and set machine->gasisr.yes.
18228 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18229 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18230 __gcc_isr.n_pushed to .L__stack_usage.
18231 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18232 (avr_asm_final_postscan_insn): ...this new static function.
18233 * config/avr/avr.h (machine_function)
18234 <is_no_gccisr, use_L__stack_usage>: New fields.
18235 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18236 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18237 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18238 (gasisr, *gasisr): New expander and insn.
18239 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18240 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18241 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18243 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18245 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18248 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18250 Move jump-tables out of .text again.
18253 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18254 (ASM_OUTPUT_ADDR_VEC): New function.
18255 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18256 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18257 INSN_ADDRESSes as asm comment.
18258 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18259 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18260 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18261 * config/avr/avr.md (*tablejump): Adjust comment.
18262 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18263 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18265 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18266 (avr_output_addr_vec): New proto.
18267 (avr_log_t) <insn_addresses>: New field.
18269 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18272 * config/i386/i386.c (ix86_function_arg_advance): Set
18273 outgoing_args_on_stack to true if there are outgoing arguments
18275 (ix86_function_arg): Likewise.
18276 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18277 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18278 * config/i386/i386.h (machine_function): Add
18279 outgoing_args_on_stack.
18281 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18283 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18284 supporting pthreds.
18285 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18287 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18289 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18290 (REAL_H): Remove $(MACHMODE_H).
18291 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18293 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18294 $(MACHMODE_H) and double-int.h.
18295 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18297 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18300 2017-07-07 Andrew Pinski <apinski@cavium.com>
18302 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18303 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18305 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18307 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18308 Add warning if GCC was not configured to link against a GLIBC that
18309 exports the hardware capability bits.
18310 (make_resolver_func): Make resolver function private and not a
18311 COMDAT function. Create the name with clone_function_name instead
18312 of make_unique_name.
18315 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18316 correct operand in doing the split.
18318 2017-07-07 Carl Love <cel@us.ibm.com>
18320 * config/rs6000/rs6000-c: Add support for built-in function
18321 vector unsigned short vec_pack_to_short_fp32 (vector float,
18323 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18324 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18325 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18326 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18327 (convert_4f32_8i16): Add define_expand.
18328 * doc/extend.texi: Update the built-in documentation file for the
18329 new built-in function.
18331 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18333 * config/sparc/m8.md: New file.
18334 * config/sparc/sparc.md: Include m8.md.
18336 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18338 * config/sparc/sparc.opt: New option -mvis4b.
18339 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18340 (sparc_option_override): Handle VIS4B.
18341 (enum sparc_builtins): Define
18342 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18343 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18344 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18345 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18346 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18347 (check_constant_argument): New function.
18348 (sparc_vis_init_builtins): Define builtins
18349 __builtin_vis_dictunpack{8,16,32},
18350 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18351 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18352 __builtin_vis_fpcmpde{8,16,32}shl and
18353 __builtin_vis_fpcmpur{8,16,32}shl.
18354 (sparc_expand_builtin): Check that the constant operands to
18355 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18356 constant and in range.
18357 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18359 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18360 (SPARC_IMM5_P): Likewise.
18361 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18362 (enabled): Handle vis4b.
18363 (UNSPEC_DICTUNPACK): New unspec.
18364 (UNSPEC_FPCMPSHL): Likewise.
18365 (UNSPEC_FPUCMPSHL): Likewise.
18366 (UNSPEC_FPCMPDESHL): Likewise.
18367 (UNSPEC_FPCMPURSHL): Likewise.
18368 (cpu_feature): New CPU feature `vis4b'.
18369 (dictunpack{8,16,32}): New insns.
18370 (FPCSMODE): New mode iterator.
18371 (fpcscond): New code iterator.
18372 (fpcsucond): Likewise.
18373 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18374 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18375 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18376 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18377 * config/sparc/constraints.md: Define constraints `q' for unsigned
18378 2-bit integer constants and `t' for unsigned 5-bit integer
18380 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18382 (imm5_operand_dictunpack16): Likewise.
18383 (imm5_operand_dictunpack32): Likewise.
18384 (imm2_operand): Likewise.
18385 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18386 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18387 ditunpack* and fpcmp*shl builtins.
18389 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18391 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18392 * config.in: Add HAVE_AS_SPARC6 define.
18393 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18395 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18397 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18398 (CPP_CPU_SPEC): Handle m8.
18399 (ASM_CPU_SPEC): Likewise.
18400 * config/sparc/sparc-opts.h (enum processor_type): Add
18402 * config/sparc/sparc.c (m8_costs): New struct.
18403 (sparc_option_override): Handle TARGET_CPU_m8.
18404 (sparc32_initialize_trampoline): Likewise.
18405 (sparc64_initialize_trampoline): Likewise.
18406 (sparc_issue_rate): Likewise.
18407 (sparc_register_move_cost): Likewise.
18408 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18409 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18410 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18411 (CPP_CPU_SPEC): Handle M8.
18412 (ASM_CPU_SPEC): Likewise.
18413 (AS_M8_FLAG): Define.
18414 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18415 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18416 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18418 * configure: Regenerate.
18419 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18422 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18424 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18426 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18427 ("*movdi_insn_sp32"): Do not set v3pipe.
18428 ("*movsi_insn"): Likewise.
18429 ("*movdi_insn_sp64"): Likewise.
18430 ("*movsf_insn"): Likewise.
18431 ("*movdf_insn_sp32"): Likewise.
18432 ("*movdf_insn_sp64"): Likewise.
18433 ("*zero_extendsidi2_insn_sp64"): Likewise.
18434 ("*sign_extendsidi2_insn"): Likewise.
18435 ("*mov<VM32:mode>_insn"): Likewise.
18436 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18437 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18438 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18439 ("<vlop:code><VL:mode>3"): Likewise.
18440 ("*not_<vlop:code><VL:mode>3"): Likewise.
18441 ("*nand<VL:mode>_vis"): Likewise.
18442 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18443 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18444 ("one_cmpl<VL:mode>2"): Likewise.
18445 ("faligndata<VM64:mode>_vis"): Likewise.
18446 ("alignaddrsi_vis"): Likewise.
18447 ("alignaddrdi_vis"): Likweise.
18448 ("alignaddrlsi_vis"): Likewise.
18449 ("alignaddrldi_vis"): Likewise.
18450 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18451 ("bmaskdi_vis"): Likewise.
18452 ("bmasksi_vis"): Likewise.
18453 ("bshuffle<VM64:mode>_vis"): Likewise.
18454 ("cmask8<P:mode>_vis"): Likewise.
18455 ("cmask16<P:mode>_vis"): Likewise.
18456 ("cmask32<P:mode>_vis"): Likewise.
18457 ("pdistn<P:mode>_vis"): Likewise.
18458 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18460 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18462 * config/sparc/sparc.md ("subtype"): New insn attribute.
18463 ("*wrgsr_sp64"): Set insn subtype.
18464 ("*rdgsr_sp64"): Likewise.
18465 ("alignaddrsi_vis"): Likewise.
18466 ("alignaddrdi_vis"): Likewise.
18467 ("alignaddrlsi_vis"): Likewise.
18468 ("alignaddrldi_vis"): Likewise.
18469 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18470 ("fexpand_vis"): Likewise.
18471 ("fpmerge_vis"): Likewise.
18472 ("faligndata<VM64:mode>_vis"): Likewise.
18473 ("bshuffle<VM64:mode>_vis"): Likewise.
18474 ("cmask8<P:mode>_vis"): Likewise.
18475 ("cmask16<P:mode>_vis"): Likewise.
18476 ("cmask32<P:mode>_vis"): Likewise.
18477 ("fchksm16_vis"): Likewise.
18478 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18479 ("fmean16_vis"): Likewise.
18480 ("fp<plusminus_insn>64_vis"): Likewise.
18481 ("<plusminus_insn>v8qi3"): Likewise.
18482 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18483 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18484 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18485 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18486 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18487 ("*movqi_insn"): Likewise.
18488 ("*movhi_insn"): Likewise.
18489 ("*movsi_insn"): Likewise.
18490 ("movsi_pic_gotdata_op"): Likewise.
18491 ("*movdi_insn_sp32"): Likewise.
18492 ("*movdi_insn_sp64"): Likewise.
18493 ("movdi_pic_gotdata_op"): Likewise.
18494 ("*movsf_insn"): Likewise.
18495 ("*movdf_insn_sp32"): Likewise.
18496 ("*movdf_insn_sp64"): Likewise.
18497 ("*zero_extendhisi2_insn"): Likewise.
18498 ("*zero_extendqihi2_insn"): Likewise.
18499 ("*zero_extendqisi2_insn"): Likewise.
18500 ("*zero_extendqidi2_insn"): Likewise.
18501 ("*zero_extendhidi2_insn"): Likewise.
18502 ("*zero_extendsidi2_insn_sp64"): Likewise.
18503 ("ldfsr"): Likewise.
18504 ("prefetch_64"): Likewise.
18505 ("prefetch_32"): Likewise.
18506 ("tie_ld32"): Likewise.
18507 ("tie_ld64"): Likewise.
18508 ("*tldo_ldub_sp32"): Likewise.
18509 ("*tldo_ldub1_sp32"): Likewise.
18510 ("*tldo_ldub2_sp32"): Likewise.
18511 ("*tldo_ldub_sp64"): Likewise.
18512 ("*tldo_ldub1_sp64"): Likewise.
18513 ("*tldo_ldub2_sp64"): Likewise.
18514 ("*tldo_ldub3_sp64"): Likewise.
18515 ("*tldo_lduh_sp32"): Likewise.
18516 ("*tldo_lduh1_sp32"): Likewise.
18517 ("*tldo_lduh_sp64"): Likewise.
18518 ("*tldo_lduh1_sp64"): Likewise.
18519 ("*tldo_lduh2_sp64"): Likewise.
18520 ("*tldo_lduw_sp32"): Likewise.
18521 ("*tldo_lduw_sp64"): Likewise.
18522 ("*tldo_lduw1_sp64"): Likewise.
18523 ("*tldo_ldx_sp64"): Likewise.
18524 ("*mov<VM32:mode>_insn"): Likewise.
18525 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18526 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18528 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18530 * config/sparc/sparc.md ("type"): New insn type viscmp.
18531 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18533 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18534 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18535 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18536 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18538 ("n7_vis_logical_11cycle"): Likewise.
18539 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18540 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18541 * config/sparc/niagara.md ("niag_vis"): Likewise.
18542 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18543 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18545 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18547 * config/sparc/sparc.md: New instruction type `bmask'.
18548 (bmaskdi_vis): Use the `bmask' type.
18549 (bmasksi_vis): Likewise.
18550 * config/sparc/ultra3.md (us3_array): Likewise.
18551 * config/sparc/niagara7.md (n7_array): Likewise.
18552 * config/sparc/niagara4.md (n4_array): Likewise.
18553 * config/sparc/niagara2.md (niag2_vis): Likewise.
18554 (niag3_vis): Likewise.
18555 * config/sparc/niagara.md (niag_vis): Likewise.
18557 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18559 * ipa-comdats.c: Remove optimize check from gate.
18560 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18561 for functions not optimized.
18562 (ipa_fn_summary_read): Skip optimize check.
18563 (ipa_fn_summary_write): Likewise.
18564 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18566 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18568 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18569 (check_callers): Give up on caller which is not optimized.
18570 (inline_small_functions): Likewise.
18571 (ipa_inline): Do not give up when not optimizing.
18572 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18573 away unoptimizes cdtors.
18574 (whole_program_function_and_variable_visibility): Do
18575 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18576 * ipa.c (process_references): Do not check optimize.
18577 (symbol_table::remove_unreachable_nodes): Update optimize check.
18578 (set_writeonly_bit): Update optimize check.
18579 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18580 (pass_ipa_single_use::gate): Remove.
18582 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18584 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18585 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18586 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18587 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18588 permute_load, permute_store, adjust_extract, adjust_splat,
18589 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18590 replace_swap_with_copy, dump_swap_insn_table,
18591 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18592 recombine_lvx_pattern, recombine_stvx_pattern,
18593 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18594 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18595 to file rs6000-p8swap.c.
18596 * config/rs6000/rs6000-p8swap.c: New file.
18597 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18598 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18599 and rs6000*-*-* targets.
18601 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18603 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18605 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18607 * lto-wrapper.c (merge_and_complain): Do not merge
18608 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18609 fsigned_zeros, ftrapping_math, fwrapv.
18610 (append_compiler_options): Do not track these options.
18611 (append_linker_options): Likewie
18613 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18615 * cgraphunit.c (cgraph_node::finalize_function): When
18616 !flag_toplevel_reorde set no_reorder flag.
18617 (varpool_node::finalize_decl): Likewise.
18618 (symbol_table::compile): Drop no toplevel reorder path.
18620 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18622 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18623 edges; zero probability is not better than uninitialized.
18625 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18627 * asan.h (asan_sanitize_allocas_p): Declare.
18628 * asan.c (asan_sanitize_allocas_p): New function.
18629 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18630 (handle_builtin_alloca): Likewise.
18631 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18632 if !asan_sanitize_allocas_p.
18633 * params.def (asan-instrument-allocas): Add new option.
18634 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18635 * opts.c (common_handle_option): Disable allocas sanitization for
18638 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18640 * asan.c: Include gimple-fold.h.
18641 (get_last_alloca_addr): New function.
18642 (handle_builtin_stackrestore): Likewise.
18643 (handle_builtin_alloca): Likewise.
18644 (asan_emit_allocas_unpoison): Likewise.
18645 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18646 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18647 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18648 (instrument_builtin_call): Pass gimple iterator to
18649 get_mem_refs_of_builtin_call.
18650 (last_alloca_addr): New global.
18651 * asan.h (asan_emit_allocas_unpoison): Declare.
18652 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18653 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18654 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18655 if function calls alloca.
18656 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18657 * gimple-fold.h (replace_call_with_value): Declare.
18658 * internal-fn.c: Include asan.h.
18659 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18660 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18662 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18664 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18665 (C_SELFTEST_FLAGS): New.
18666 (CPP_SELFTEST_FLAGS): New.
18667 (SELFTEST_DEPS): New, from deps of s-selftest.
18668 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18669 (CPP_SELFTEST_DEPS): New.
18670 (selftest): Add dependency on s-selftest-c++.
18671 (s-selftest): Rename to...
18672 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18673 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18674 than SELFTEST_FLAGS.
18675 (selftest-gdb): Rename to...
18676 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18678 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18679 (selftest-valgrind): Rename to...
18680 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18682 (selftest-valgrind): Reintroduce as an alias for
18683 selftest-c-valgrind.
18684 (s-selftest-c++): New.
18685 (selftest-c++-gdb): New.
18686 (selftest-c++-valgrind): New.
18688 2017-07-06 Olivier Hainque <hainque@adacore.com>
18690 * gcc.c (process_command): When deciding if undefined variables
18691 should be ignored when processing specs, accept "gcc -v" as well.
18693 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18695 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18696 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18698 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18700 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18701 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18702 * config/arm/arm-tables.opt: Regenerate.
18703 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18705 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18707 2017-07-06 Carl Love <cel@us.ibm.com>
18709 * ChangeLog: Clean up from mid air collision
18711 2017-07-06 Carl Love <cel@us.ibm.com>
18713 * config/rs6000/rs6000-c.c: Add support for built-in functions
18714 vector signed int vec_subc (vector signed int, vector signed int);
18715 vector signed __int128 vec_subc (vector signed __int128,
18716 vector signed __int128);
18717 vector unsigned __int128 vec_subc (vector unsigned __int128,
18718 vector unsigned __int128);
18719 vector signed int vec_sube (vector signed int, vector signed int,
18720 vector signed int);
18721 vector unsigned int vec_sube (vector unsigned int,
18722 vector unsigned int,
18723 vector unsigned int);
18724 vector signed __int128 vec_sube (vector signed __int128,
18725 vector signed __int128,
18726 vector signed__int128);
18727 vector unsigned __int128 vec_sube (vector unsigned __int128,
18728 vector unsigned __int128,
18729 vector unsigned __int128);
18730 vector signed int vec_subec (vector signed int, vector signed int,
18731 vector signed int);
18732 vector unsigned int vec_subec (vector unsigned int,
18733 vector unsigned int,
18734 vector unsigned int);
18735 vector signed __int128 vec_subec (vector signed __int128,
18736 vector signed __int128,
18737 vector signed__int128);
18738 vector unsigned __int128 vec_subec (vector unsigned __int128,
18739 vector unsigned __int128,
18740 vector unsigned __int128);
18741 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18742 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18743 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18744 BU_ALTIVEC_OVERLOAD_X definitions.
18745 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18746 * doc/extend.texi: Update the built-in documentation file for the new
18747 built-in functions.
18749 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18752 * diagnostic-show-locus.c (layout::layout): Use start and finish
18753 spelling location for the start and finish of each range.
18754 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18755 Add unused aspect param.
18756 * input.c (expand_location_1): Add "aspect" param, and use it
18757 to access the correct part of the location.
18758 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18760 (expand_location_to_spelling_point): Likewise.
18761 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18762 param, and pass it to expand_location_1.
18764 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18766 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18767 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18768 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18769 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18770 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18771 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18772 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18773 _mm_maskz_getmant_ss): New intrinsics.
18774 (__builtin_ia32_getexpss128_mask): Changed to ...
18775 __builtin_ia32_getexpss128_round ... this.
18776 (__builtin_ia32_getexpsd128_mask): Changed to ...
18777 __builtin_ia32_getexpsd128_round ... this.
18778 * config/i386/i386-builtin-types.def
18779 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18780 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18781 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18782 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18783 __builtin_ia32_getmantss_mask_round): New builtins.
18784 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18785 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18786 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18787 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18788 * config/i386/sse.md
18789 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18790 avx512f_sgetexp<mode><mask_scalar_name>
18791 <round_saeonly_scalar_name> ... this.
18792 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18793 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18794 vgetexp<ssescalarmodesuffix>
18795 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18796 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18797 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18798 avx512f_vgetmant<mode><mask_scalar_name>
18799 <round_saeonly_scalar_name> ... this.
18800 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18801 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18802 vgetmant<ssescalarmodesuffix>
18803 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18804 %0<mask_scalar_operand4>, %1, %2
18805 <round_saeonly_scalar_mask_op4>, %3} ... this.
18806 * config/i386/subst.md (mask_scalar_operand4,
18807 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18808 round_saeonly_scalar_nimm_predicate): New subst attributes.
18810 2017-07-06 Julia Koval <julia.koval@intel.com>
18812 * config/i386/i386.c (ix86_erase_embedded_rounding):
18813 Remove code for old rounding pattern.
18815 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18817 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18819 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18821 * doc/sourcebuild.texi (Test Directives, Variants of
18822 dg-require-support): Add documentation for dg-require-stack-check.
18824 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18826 * config/i386/subst.md (mask_scalar, round_scalar,
18827 round_saeonly_scalar): New meta-templates.
18828 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18829 round_scalar_mask_operand3, round_scalar_mask_op3,
18830 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18831 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18832 round_saeonly_scalar_constraint,
18833 round_saeonly_scalar_prefix): New subst attribute.
18834 * config/i386/sse.md
18835 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18836 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18837 <round_scalar_name> ... this.
18838 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18839 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18840 <round_scalar_name> ... this.
18841 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18842 <sse>_vm<code><mode>3<mask_scalar_name>
18843 <round_saeonly_scalar_name> ... this.
18844 (v<plusminus_mnemonic><ssescalarmodesuffix>
18845 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18846 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18847 v<plusminus_mnemonic><ssescalarmodesuffix>
18848 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18849 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18850 (v<multdiv_mnemonic><ssescalarmodesuffix>
18851 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18852 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18853 v<multdiv_mnemonic><ssescalarmodesuffix>
18854 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18855 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18856 (v<maxmin_float><ssescalarmodesuffix>
18857 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18858 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18859 v<maxmin_float><ssescalarmodesuffix>
18860 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18861 %0<mask_scalar_operand3>, %1, %<iptr>2
18862 <round_saeonly_scalar_mask_op3>} ... this.
18864 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18866 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18867 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18869 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18870 Alan Hayward <alan.hayward@arm.com>
18871 David Sherwood <david.sherwood@arm.com>
18873 * combine.c (simplify_if_then_else): Remove "enum" before
18875 * compare-elim.c (can_eliminate_compare): Likewise.
18876 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18878 (aarch64_lookup_simd_builtin_type): Likewise.
18879 (aarch64_simd_builtin_type): Likewise.
18880 (aarch64_init_simd_builtin_types): Likewise.
18881 (aarch64_simd_expand_args): Likewise.
18882 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18884 (aarch64_reverse_mask): Likewise.
18885 (aarch64_simd_emit_reg_reg_move): Likewise.
18886 (aarch64_gen_adjusted_ldpstp): Likewise.
18887 (aarch64_ccmp_mode_to_code): Likewise.
18888 (aarch64_operands_ok_for_ldpstp): Likewise.
18889 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18890 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18892 (aarch64_min_divisions_for_recip_mul): Likewise.
18893 (aarch64_reassociation_width): Likewise.
18894 (aarch64_get_condition_code_1): Likewise.
18895 (aarch64_simd_emit_reg_reg_move): Likewise.
18896 (aarch64_simd_attr_length_rglist): Likewise.
18897 (aarch64_reverse_mask): Likewise.
18898 (aarch64_operands_ok_for_ldpstp): Likewise.
18899 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18900 (aarch64_gen_adjusted_ldpstp): Likewise.
18901 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18903 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18904 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18905 (arm_lookup_simd_builtin_type): Likewise.
18906 (arm_simd_builtin_type): Likewise.
18907 (arm_init_simd_builtin_types): Likewise.
18908 (arm_expand_builtin_args): Likewise.
18909 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18910 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18911 (ft32_setup_incoming_varargs): Likewise.
18912 (ft32_function_arg): Likewise.
18913 (ft32_function_arg_advance): Likewise.
18914 (ft32_pass_by_reference): Likewise.
18915 (ft32_arg_partial_bytes): Likewise.
18916 (ft32_valid_pointer_mode): Likewise.
18917 (ft32_addr_space_pointer_mode): Likewise.
18918 (ft32_addr_space_legitimate_address_p): Likewise.
18919 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18921 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18922 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18923 (iamcu_alignment): Likewise.
18924 (canonicalize_vector_int_perm): Likewise.
18925 (ix86_noce_conversion_profitable_p): Likewise.
18926 (ix86_mpx_bound_mode): Likewise.
18927 (ix86_operands_ok_for_move_multiple): Likewise.
18928 * config/microblaze/microblaze-protos.h
18929 (microblaze_expand_conditional_branch_reg): Likewise.
18930 * config/microblaze/microblaze.c
18931 (microblaze_expand_conditional_branch_reg): Likewise.
18932 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18934 (rs6000_reassociation_width): Likewise.
18935 (rs6000_invalid_binary_op): Likewise.
18936 (fusion_p9_p): Likewise.
18937 (emit_fusion_p9_load): Likewise.
18938 (emit_fusion_p9_store): Likewise.
18939 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18941 (riscv_hard_regno_mode_ok_p): Likewise.
18942 (riscv_address_insns): Likewise.
18943 (riscv_split_symbol): Likewise.
18944 (riscv_legitimize_move): Likewise.
18945 (riscv_function_value): Likewise.
18946 (riscv_hard_regno_nregs): Likewise.
18947 (riscv_expand_builtin): Likewise.
18948 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18949 (riscv_build_integer): Likewise.
18950 (riscv_split_integer): Likewise.
18951 (riscv_legitimate_constant_p): Likewise.
18952 (riscv_cannot_force_const_mem): Likewise.
18953 (riscv_regno_mode_ok_for_base_p): Likewise.
18954 (riscv_valid_base_register_p): Likewise.
18955 (riscv_valid_offset_p): Likewise.
18956 (riscv_valid_lo_sum_p): Likewise.
18957 (riscv_classify_address): Likewise.
18958 (riscv_legitimate_address_p): Likewise.
18959 (riscv_address_insns): Likewise.
18960 (riscv_load_store_insns): Likewise.
18961 (riscv_force_binary): Likewise.
18962 (riscv_split_symbol): Likewise.
18963 (riscv_force_address): Likewise.
18964 (riscv_legitimize_address): Likewise.
18965 (riscv_move_integer): Likewise.
18966 (riscv_legitimize_const_move): Likewise.
18967 (riscv_legitimize_move): Likewise.
18968 (riscv_address_cost): Likewise.
18969 (riscv_subword): Likewise.
18970 (riscv_output_move): Likewise.
18971 (riscv_canonicalize_int_order_test): Likewise.
18972 (riscv_emit_int_order_test): Likewise.
18973 (riscv_function_arg_boundary): Likewise.
18974 (riscv_pass_mode_in_fpr_p): Likewise.
18975 (riscv_pass_fpr_single): Likewise.
18976 (riscv_pass_fpr_pair): Likewise.
18977 (riscv_get_arg_info): Likewise.
18978 (riscv_function_arg): Likewise.
18979 (riscv_function_arg_advance): Likewise.
18980 (riscv_arg_partial_bytes): Likewise.
18981 (riscv_function_value): Likewise.
18982 (riscv_pass_by_reference): Likewise.
18983 (riscv_setup_incoming_varargs): Likewise.
18984 (riscv_print_operand): Likewise.
18985 (riscv_elf_select_rtx_section): Likewise.
18986 (riscv_save_restore_reg): Likewise.
18987 (riscv_for_each_saved_reg): Likewise.
18988 (riscv_register_move_cost): Likewise.
18989 (riscv_hard_regno_mode_ok_p): Likewise.
18990 (riscv_hard_regno_nregs): Likewise.
18991 (riscv_class_max_nregs): Likewise.
18992 (riscv_memory_move_cost): Likewise.
18993 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18994 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18995 (rl78_addr_space_address_mode): Likewise.
18996 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18998 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18999 (rs6000_reassociation_width): Likewise.
19000 (rs6000_invalid_binary_op): Likewise.
19001 (fusion_p9_p): Likewise.
19002 (emit_fusion_p9_load): Likewise.
19003 (emit_fusion_p9_store): Likewise.
19004 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19005 (ok_for_simple_move_operands): Likewise.
19006 (ok_for_simple_move_strict_operands): Likewise.
19007 (ok_for_simple_arith_logic_operands): Likewise.
19008 (visium_legitimize_reload_address): Likewise.
19009 (visium_select_cc_mode): Likewise.
19010 (output_cbranch): Likewise.
19011 (visium_split_double_move): Likewise.
19012 (visium_expand_copysign): Likewise.
19013 (visium_expand_int_cstore): Likewise.
19014 (visium_expand_fp_cstore): Likewise.
19015 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19016 (visium_function_arg): Likewise.
19017 (visium_function_arg_advance): Likewise.
19018 (visium_libcall_value): Likewise.
19019 (visium_setup_incoming_varargs): Likewise.
19020 (visium_legitimate_constant_p): Likewise.
19021 (visium_legitimate_address_p): Likewise.
19022 (visium_legitimize_address): Likewise.
19023 (visium_secondary_reload): Likewise.
19024 (visium_register_move_cost): Likewise.
19025 (visium_memory_move_cost): Likewise.
19026 (prepare_move_operands): Likewise.
19027 (ok_for_simple_move_operands): Likewise.
19028 (ok_for_simple_move_strict_operands): Likewise.
19029 (ok_for_simple_arith_logic_operands): Likewise.
19030 (visium_function_value_1): Likewise.
19031 (rtx_ok_for_offset_p): Likewise.
19032 (visium_legitimize_reload_address): Likewise.
19033 (visium_split_double_move): Likewise.
19034 (visium_expand_copysign): Likewise.
19035 (visium_expand_int_cstore): Likewise.
19036 (visium_expand_fp_cstore): Likewise.
19037 (visium_split_cstore): Likewise.
19038 (visium_select_cc_mode): Likewise.
19039 (visium_split_cbranch): Likewise.
19040 (output_cbranch): Likewise.
19041 (visium_print_operand_address): Likewise.
19042 * expmed.c (flip_storage_order): Likewise.
19043 * expmed.h (emit_cstore): Likewise.
19044 (flip_storage_order): Likewise.
19045 * genrecog.c (validate_pattern): Likewise.
19046 * hsa-gen.c (gen_hsa_addr): Likewise.
19047 * internal-fn.c (expand_arith_overflow): Likewise.
19048 * ira-color.c (allocno_copy_cost_saving): Likewise.
19049 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19050 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19051 (process_invariant_for_inheritance): Likewise.
19052 * lra-eliminations.c (move_plus_up): Likewise.
19053 * omp-low.c (lower_oacc_reductions): Likewise.
19054 * simplify-rtx.c (simplify_subreg): Likewise.
19055 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19056 (TARGET_CHKP_BOUND_MODE): Likewise..
19057 * targhooks.c (default_chkp_bound_mode): Likewise.
19058 (default_setup_incoming_vararg_bounds): Likewise.
19059 * targhooks.h (default_chkp_bound_mode): Likewise.
19060 (default_setup_incoming_vararg_bounds): Likewise.
19061 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19062 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19063 (have_whole_vector_shift): Likewise.
19064 * tree-vect-stmts.c (vectorizable_load): Likewise.
19065 * doc/tm.texi: Regenerate.
19067 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19069 Graceful degrade if Binutils PR21472 is not available.
19072 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19073 .rodata in flash test fails.
19074 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19075 * confgure: Regenerate.
19076 * config.in: Regenerate.
19077 * config/avr/avr.c (avr_asm_named_section)
19078 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19079 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19080 (avr_asm_init_sections): Same.
19082 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19084 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19085 (fma<VH:mode>4_intrinsic): Likewise.
19086 (*fmsub<VCVTF:mode>4): Likewise.
19087 (*fmsub<VH:mode>4_intrinsic): Likewise.
19089 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19092 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19093 Don't depend on "optimize > 0".
19094 (out_movhi_r_mr, out_movqi_mr_r): Same.
19095 (out_movhi_mr_r, out_movqi_r_mr): Same.
19096 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19097 io_address_operand on "optimize > 0".
19099 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19101 * tree-loop-distribution.c: Add general explanantion on the pass.
19102 (generate_loops_for_partition): Mark distributed loop.
19103 (pg_add_dependence_edges): New parameter. Handle alias data
19104 dependence specially and record it in the parameter if asked.
19105 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19106 (init_partition_graph_vertices, add_partition_graph_edge): New.
19107 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19108 (free_partition_graph_vdata, build_partition_graph): New.
19109 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19110 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19111 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19112 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19113 (version_for_distribution_p, finalize_partitions): New.
19114 (distribute_loop): Handle alias data dependence specially. Factor
19115 out loop fusion code as functions and call these functions.
19117 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19119 * tree-loop-distribution.c (classify_partition): New parameter and
19120 better handle reduction statement.
19121 (rdg_build_partitions): Revise comment.
19122 (distribute_loop): Compute statements in all partitions and pass it
19123 to classify_partition.
19125 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19127 * tree-loop-distribution.c (enum partition_type): New.
19128 (struct partition): New field type.
19129 (partition_merge_into): Add parameter. Update partition type.
19130 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19131 (build_rdg_partition_for_vertex): Compute partition type.
19132 (rdg_build_partitions): Dump partition type.
19133 (distribute_loop): Update calls to partition_merge_into.
19135 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19137 * tree-loop-distribution.c (struct ddr_hasher): New.
19138 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19140 (classify_partition): Call get_data_dependence.
19141 (pg_add_dependence_edges): Ditto.
19142 (distribute_loop): Release data dependence hash table.
19144 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19146 * tree-loop-distribution.c (ref_base_address): Delete.
19147 (similar_memory_accesses): Rename ...
19148 (share_memory_accesses): ... to this. Check if partitions access
19149 the same memory reference.
19150 (distribute_loop): Call share_memory_accesses.
19152 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19154 * tree-loop-distribution.c (struct partition): New field recording
19155 its data reference.
19156 (partition_alloc, partition_free): Init and release data refs.
19157 (partition_merge_into): Merge data refs.
19158 (build_rdg_partition_for_vertex): Collect data refs for partition.
19159 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19161 (distribute_loop): Remve data refs from vertice data of partition
19164 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19166 * tree-loop-distribution.c (params.h): Include header file.
19167 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19168 (datarefs_vec): New global var.
19169 (create_rdg_vertices): Use datarefs_vec directly.
19170 (free_rdg): Don't free data references.
19171 (build_rdg): Update use. Don't free data references.
19172 (distribute_loop): Compute global variable for data references.
19173 Bail out if there are too many data references.
19175 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19177 * tree-loop-distribution.c (loop_nest): New global var.
19178 (build_rdg): Use loop directly, rather than loop nest.
19179 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19181 (distribute_loop): Compute global variable loop nest. Update use.
19183 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19185 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19186 (partition_merge_into): New parameter. Dump reason for fusion.
19187 (distribute_loop): Update use of partition_merge_into.
19189 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19191 * tree-loop-distribution.c (bb_top_order_index): New.
19192 (bb_top_order_index_size, bb_top_order_cmp): New.
19193 (stmts_from_loop): Use topological order.
19194 (pass_loop_distribution::execute): Compute and release topological
19195 order for basic blocks.
19197 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19199 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19202 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19204 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19205 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19206 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19207 * internal-fn.def (LOOP_DIST_ALIAS): New.
19208 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19209 (fold_loop_internal_call): ... this.
19210 (vect_loop_dist_alias_call): New function.
19211 (set_uid_loop_bbs): Call fold_loop_internal_call.
19212 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19215 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19218 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19219 Require dead FLAGS_REG at the beginning of a peephole.
19221 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19224 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19225 arguments in the call to __builtin_ia32_sbb_u32.
19226 (_subborrow_u64): Swap _X and _Y arguments in the call to
19227 __builtin_ia32_sbb_u64.
19229 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19232 * tree-vrp.c (compare_assert_loc): Turn into a function template
19233 with stable template parameter. Only test if a->e is NULL,
19234 !a->e == !b->e has been verified already. Use e == NULL or
19235 e != NULL instead of e or ! e tests. If stable is true, don't use
19236 iterative_hash_expr, on the other side allow a or b or both NULL
19237 and sort the NULLs last.
19238 (process_assert_insertions): Sort using compare_assert_loc<false>
19239 instead of compare_assert_loc, later sort using
19240 compare_assert_loc<true> before calling process_assert_insertions_for
19241 in a loop. Use break instead of continue once seen NULL pointer.
19243 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19245 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19246 Cortex-R7 and Cortex-R8 processors.
19248 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19250 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19251 uninitialized while src is not.
19253 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19255 * common/config/arm/arm-common.c: Adjust include path for
19257 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19258 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19259 (arm-cpu-data.h): Likewise.
19260 (arm-cpu-cdata.h): Likewise.
19261 * config/arm/arm-cpu.h: Delete.
19262 * config/arm/arm-cpu-cdata.h: Delete.
19263 * config/arm/arm-cpu-data.h: Delete.
19265 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19267 * config/arm/arm-cpus.in (cortex-a55): New.
19268 (cortex-a75): Likewise.
19269 (cortex-a75.cortex-a55): Likewise.
19270 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19272 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19273 * config/arm/arm-cpu-cdata.h: Regenerate.
19274 * config/arm/arm-cpu-data.h: Regenerate.
19275 * config/arm/arm-cpu.h: Regenerate.
19276 * config/arm/arm-tables.opt: Regenerate.
19277 * config/arm/arm-tune.md: Regenerate.
19279 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19281 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19283 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19285 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19288 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19290 PR tree-optimization/81292
19291 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19292 full_string_p, also call adjust_related_strinfos if the adjustment
19293 is simple, otherwise invalidate related strinfos.
19295 2017-07-04 Martin Liska <mliska@suse.cz>
19298 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19299 newly created variable as DECL_IGNORED_P.
19301 2017-07-04 Martin Liska <mliska@suse.cz>
19304 * ipa-inline.c (inline_small_functions):
19305 Use xstrdup_for_dump.
19307 2017-07-04 Tom de Vries <tom@codesourcery.com>
19309 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19311 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19314 * config/darwin.c (darwin_function_switched_text_sections):
19317 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19319 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19321 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19323 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19325 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19327 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19328 min_profitable_iters, and th as inclusive lower bounds.
19329 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19330 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19331 for min_profitable_iters and min_profitable_estimate.
19332 (vect_transform_loop): Treat th as an inclusive lower bound.
19333 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19335 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19338 * config/darwin.c (darwin_function_switched_text_sections):
19339 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19340 in two pieces, and suppress the use of buf.
19342 2017-07-03 Nathan Sidwell <nathan@acm.org>
19344 * hash-table.h (hash_table_mod1): Fix indentation.
19346 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19348 PR middle-end/81290
19349 * predict.c (force_edge_cold): Be more careful about propagation
19351 * profile-count.h (profile_probability::guessed,
19352 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19354 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19356 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19358 * doc/invoke.texi (rcpc architecture extension): Document it.
19360 2017-07-03 Richard Biener <rguenther@suse.de>
19362 PR tree-optimization/60510
19363 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19364 the scalar reduction PHI and use it.
19365 (vectorizable_reduction): Properly guard the single_defuse_cycle
19366 path for non-SLP reduction chains where we cannot use it.
19367 Rework reduc_def/index and vector type deduction. Rework
19368 vector operand gathering during reduction op code-gen.
19369 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19370 chains dissolve the chain and leave it to non-SLP reduction
19373 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19375 * tree-data-ref.h (dr_alignment): Declare.
19376 * tree-data-ref.c (dr_alignment): New function.
19377 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19378 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19380 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19382 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19384 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19385 and base_misalignment fields.
19386 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19387 * tree-data-ref.c: Include builtins.h.
19388 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19389 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19390 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19391 * tree-vect-data-refs.c: Include tree-cfg.h.
19392 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19393 fields instead of calculating an alignment here.
19394 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19395 innermost_loop_behavior fields.
19397 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19399 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19401 (DR_STEP_ALIGNMENT): New macro.
19402 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19403 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19404 (create_data_ref): Print it.
19405 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19406 to tell whether the step preserves vector (mis)alignment.
19407 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19408 Move the check for an integer step and generalise to all INTEGER_CST.
19409 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19410 Print the outer step alignment.
19412 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19414 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19415 with offset_alignment.
19416 (DR_ALIGNED_TO): Delete.
19417 (DR_OFFSET_ALIGNMENT): New macro.
19418 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19419 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19420 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19421 (create_data_ref): Likewise.
19422 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19423 (vect_analyze_data_refs): Likewise.
19424 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19425 creating dummy innermost behavior.
19427 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19429 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19430 with a "innermost_loop_behavior *" and refeence tree.
19431 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19432 (create_data_ref): Update call accordingly.
19433 * tree-predcom.c (find_looparound_phi): Likewise.
19435 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19437 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19438 fields with dr_wrt_vec_loop.
19439 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19440 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19441 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19442 (vect_dr_behavior): New function.
19443 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19444 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19445 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19446 track whether the step preserves the misalignment.
19447 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19448 Use vect_dr_behavior.
19449 (vect_setup_realignment): Update call accordingly.
19450 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19451 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19452 call to vect_create_addr_base_for_vector_ref.
19453 (vect_create_cond_for_align_checks): Likewise.
19454 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19455 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19456 (vect_recog_mask_conversion_pattern): Likewise.
19457 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19458 (new_stmt_vec_info): Remove redundant zeroing.
19460 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19462 * common/config/arm/arm-common.c (arm_be8_option): New function.
19463 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19464 (ISA_ARMv6): Add isa_bit_be8.
19465 * config/arm/arm.h (arm_be8_option): Add prototype.
19466 (BE8_SPEC_FUNCTION): New define.
19467 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19468 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19469 (mlittle-endian): Similarly.
19470 (mbe8, mbe32): New options.
19471 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19472 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19474 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19476 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19478 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19480 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19481 (cleanup_tree_cfg_bb): Use it.
19482 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19484 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19486 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19489 * loop-doloop.c (add_test): Update profile.
19491 2017-07-03 Martin Liska <mliska@suse.cz>
19494 * sanopt.c (rewrite_usage_of_param): New function.
19495 (sanitize_rewrite_addressable_params): Likewise.
19496 (pass_sanopt::execute): Call rewrite_usage_of_param.
19498 2017-07-03 Richard Biener <rguenther@suse.de>
19500 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19501 back to using VIEW_CONVERT_EXPR.
19503 2017-07-03 Martin Liska <mliska@suse.cz>
19506 * doc/extend.texi: Document when a resolver function is
19507 generated for target_clones.
19509 2017-07-03 Martin Liska <mliska@suse.cz>
19511 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19512 and shadow memory of auto variables which are subject of
19513 use-after-scope sanitization.
19514 (asan_expand_mark_ifn): Add do set only when is_poison.
19516 2016-07-03 Richard Biener <rguenther@suse.de>
19518 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19520 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19521 (vectorizable_reduction): Perform reduction PHI creation when
19522 visiting a reduction PHI and adjust and simplify code generation
19523 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19524 (vect_transform_loop): Visit reduction PHIs.
19525 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19526 defs into the SLP tree.
19527 (vect_build_slp_tree): Reduction defs terminate the recursion.
19528 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19530 (vect_get_vec_defs_for_stmt_copy): Export.
19531 (vect_get_vec_defs): Likewise.
19532 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19534 (vect_get_vec_defs_for_stmt_copy): Declare.
19535 (vect_get_vec_defs): Likewise.
19537 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19539 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19540 parameter with a "loop" parameter and use it instead of the
19541 loop containing DR_STMT. Don't check simple_iv when doing
19542 BB analysis. Describe the two analysis modes in the comment.
19544 2017-07-03 Tom de Vries <tom@codesourcery.com>
19546 PR tree-optimization/69468
19547 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19548 (find_same_succ_bb): Handle ignore_edge_flags.
19550 2017-07-03 Tom de Vries <tom@codesourcery.com>
19552 PR tree-optimization/81192
19553 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19555 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19557 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19559 2017-07-03 Tom de Vries <tom@codesourcery.com>
19561 PR tree-optimization/81192
19562 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19563 BB_SAME_SUCC (bb) == NULL.
19565 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19567 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19570 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19572 * dumpfile.c: Include profile-count.h
19573 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19575 (insert_cond_bb): Update profile.
19576 * tree-cfg.h (insert_cond_bb): Update prototype.
19577 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19578 * tree-dump.c: Do not include tree-cfg.
19580 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19582 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19584 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19586 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19589 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19591 * tree-complex.c (expand_complex_div_wide): update profile.
19593 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19594 Alan Hayward <alan.hayward@arm.com>
19595 David Sherwood <david.sherwood@arm.com>
19597 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19598 (CORETYPES_H): New define.
19599 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19600 (insn-modes-inline.h, s-modes-inline-h): New rules.
19601 (generated_files): Add insn-modes-inline.h.
19602 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19603 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19604 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19605 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19606 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19607 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19608 (build/gencodes.o, build/genconditions.o): Likewise.
19609 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19610 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19611 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19612 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19613 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19614 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19615 * coretypes.h: Include everything up to real.h for generators.
19616 Include insn-modes.h first. Include wide-int-print.h after
19617 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19618 * machmode.h: Don't include insn-modes.h here.
19619 * function-tests.c: Remove includes of signop.h, machmode.h,
19620 double-int.h and wide-int.h.
19622 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19624 * optc-save-gen.awk: Likewise.
19625 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19626 * godump.c: Remove include of wide-int-print.h.
19627 * pretty-print.h: Likewise.
19628 * wide-int-print.cc: Likewise.
19629 * wide-int.cc: Likewise.
19630 * hash-map-tests.c: Remove include of signop.h.
19631 * hash-set-tests.c: Likewise.
19632 * rtl-tests.c: Likewise.
19633 * mkconfig.sh: Remove include of machmode.h.
19634 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19636 (emit_insn_modes_inline_h): ...this new function. Emit the code
19637 into an insn-modes-inline.h header file, adding appropriate
19638 include guards and end comments.
19639 (emit_insn_modes_c_header): Remove include of machmode.h.
19640 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19642 (main): Handle -i flag and call emit_insn_modes_inline_h when
19645 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19647 * tree-ssa-strlen.c (strinfo): Rename the length field to
19648 nonzero_chars. Add a full_string_p field.
19649 (compare_nonzero_chars, zero_length_string_p): New functions.
19650 (get_addr_stridx): Add an offset_out parameter.
19651 Use compare_nonzero_chars.
19652 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19653 (new_strinfo): Update after above changes to strinfo.
19654 (set_endptr_and_length): Set full_string_p.
19655 (get_string_length): Update after above changes to strinfo.
19656 (unshare_strinfo): Update call to new_strinfo.
19657 (maybe_invalidate): Likewise.
19658 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19659 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19660 as a uhwi instead of an shwi. Update after above changes to
19661 strinfo and new_strinfo.
19662 (zero_length_string): Assert that chainsi contains full strings.
19663 Use zero_length_string_p. Update call to new_strinfo.
19664 (adjust_related_strinfos): Update after above changes to strinfo.
19665 Copy full_string_p from origsi.
19666 (adjust_last_stmt): Use zero_length_string_p.
19667 (handle_builtin_strlen): Update after above changes to strinfo and
19668 new_strinfo. Install the lhs as the string length if the previous
19669 entry didn't describe a full string.
19670 (handle_builtin_strchr): Update after above changes to strinfo
19672 (handle_builtin_strcpy): Likewise.
19673 (handle_builtin_strcat): Likewise.
19674 (handle_builtin_malloc): Likewise.
19675 (handle_pointer_plus): Likewise.
19676 (handle_builtin_memcpy): Likewise. Track nonzero characters
19677 that aren't necessarily followed by a nul terminator.
19678 (handle_char_store): Likewise.
19680 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19682 PR tree-optimization/80769
19683 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19684 for malloc and calloc. Document the new invariant that all related
19685 strinfos have delayed lengths or none do.
19686 (verify_related_strinfos): Move earlier in file.
19687 (set_endptr_and_length): New function, split out from...
19688 (get_string_length): ...here. Also set the lengths of related
19690 (zero_length_string): Assert that chainsi has known (rather than
19692 (adjust_related_strinfos): Likewise.
19694 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19696 PR tree-optimization/81136
19697 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19698 assert that two references with the same misalignment have the same
19699 compile-time misalignment if those compile-time misalignments
19702 2017-07-01 Andi Kleen <ak@linux.intel.com>
19704 * print-tree.c (print_node): Print all attributes.
19706 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19708 * cfg.c (scale_bbs_frequencies): New function.
19709 * cfg.h (scale_bbs_frequencies): Declare it.
19710 * cfgloopanal.c (single_likely_exit): Cleanup.
19711 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19713 (scale_loop_profile): Likewise.
19714 (loop_version): Likewise.
19715 (create_empty_loop_on_edge): Update.
19716 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19717 scale_loop_frequencies, scale_loop_profile, loopify,
19718 loop_version): Update prototypes.
19719 * modulo-sched.c (sms_schedule): Update.
19720 * predict.c (unlikely_executed_edge_p): Also check probability.
19721 (probably_never_executed_edge_p): Fix typo.
19722 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19723 * tree-parloops.c (gen_parallel_loop): Update.
19724 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19725 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19726 * tree-ssa-loop-split.c (split_loop): Update.
19727 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19728 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19729 (vect_loop_versioning): Update.
19730 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19732 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19734 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19736 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19738 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19739 to keep profile consistent.
19741 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19743 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19744 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19745 * profile-count.h (max_safe_multiplier): Make unsigned.
19746 (profile_count::guessed_zero): New.
19748 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19750 * bb-reorder.c (fix_up_crossing_landing_pad,
19751 fix_crossing_conditional_branches): Use make_single_succ_edge
19752 to keep profile consistent.
19754 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19756 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19759 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19762 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19763 the right scopes, make sure cond_jump isn't preserved between multiple
19764 iterations. Search for fallthru edge whenever there are 3+ edges and
19765 use find_fallthru_edge for it.
19767 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19769 Patch by Alexander Monakov <amonakov@ispras.ru>
19770 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19771 probabilities consistently.
19773 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19775 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19776 emit_cmp_and_jump_insns.
19778 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19781 * tree-inline.c (expand_call_inline): Combine profile statuses.
19783 2017-06-30 Andrew Pinski <apinski@cavium.com>
19785 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19786 fold_stmt returned true.
19788 2017-06-30 Nathan Sidwell <nathan@acm.org>
19790 * ggc.h (empty_string): Delete.
19791 * cfgexpand.c (expand_asm_stmt): Use plain "".
19792 * optabs.c (expand_asm_memory_barrier): Likewise.
19793 * stringpool.c (empty_string): Delete.
19794 (digit_vector, digit_string): Delete.
19795 (ggc_alloc_string): Use plain "", don't optimize single digit
19796 strings. Use ggc_alloc_atomic.
19798 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19800 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19801 comparison set and one other set, use the cost of the non-comparison
19804 2017-06-30 Nathan Sidwell <nathan@acm.org>
19806 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19809 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19811 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19812 loops. Remove now unneeded calls to gimple_switch_set_label() that
19813 just set removed labels to NULL_TREE.
19815 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19817 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19818 (set_range_info): ...here. Only call set_range_info_raw if domain
19820 (set_nonzero_bits): Call set_range_info_raw.
19821 * tree-ssanames.h (set_range_info_raw): New.
19823 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19826 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19827 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19828 of nonimmediate_operand and <store_mask_constraint> instead of m
19829 for the input operand. For V8FI iterator, always split if input
19830 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19831 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19832 <store_mask_predicate> instead of register_operand and
19833 <store_mask_constraint> instead of v for the input operand. Make
19834 sure both operands aren't MEMs for if not <mask_applied>.
19836 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19838 * lto-wrapper.c (copy_file) Close both file descriptors before
19841 2017-06-30 Martin Liska <mliska@suse.cz>
19844 * multiple_target.c (create_dispatcher_calls): Make ifunc
19845 also for function that don't have calls or are not referenced.
19847 2017-06-30 Richard Biener <rguenther@suse.de>
19849 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19850 analyze the first scalar stmt. Move vector type computation
19851 for the BB case here from ...
19852 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19853 live operation processing in the SLP case properly.
19855 2017-06-30 Richard Biener <rguenther@suse.de>
19857 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19859 2017-06-30 Martin Liska <mliska@suse.cz>
19862 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19863 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19865 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19867 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19870 2017-06-30 Richard Biener <rguenther@suse.de>
19872 PR tree-optimization/81249
19873 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19874 condition reduction result to original scalar type.
19876 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19878 * profile-count.h (enum profile_quality): Fix typos and whitespace
19881 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19883 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19884 type for branch probabilities.
19886 2017-06-29 Julian Brown <julian@codesourcery.com>
19887 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19889 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19890 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19891 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19892 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19894 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19896 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19897 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19898 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19899 CC usage from generic code to here.
19900 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19901 CC usage into the target macros.
19903 2017-06-29 Maya Rashish <coypu@sdf.org>
19905 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19908 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19910 * arm/arm-builtins.c: Include profile-count.h
19911 * except.c (sjlj_emit_function_enter): Use
19912 profile_probability::unlikely.
19914 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19916 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19917 and tocrel_offset be pointer args rather than implicitly using
19919 (legitimate_constant_pool_address_p, rs6000_emit_move,
19920 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19921 tocrel_offset and use in toc_relative_expr_p call.
19922 (print_operand, print_operand_address): Use static tocrel_base_oac
19923 and tocrel_offset_oac.
19924 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19927 2017-06-29 Maya Rashish <coypu@sdf.org>
19929 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19931 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19933 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19934 objects, take into account only the alignment of 'op0' and 'mode1' if
19937 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19939 * ccmp.c (ccmp_tree_comparison_p): New function.
19940 (ccmp_candidate_p): Update to use above function.
19941 (get_compare_parts): New function.
19942 (expand_ccmp_next): Update to use new functions.
19943 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19945 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19946 take mode as argument.
19947 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19948 * expr.c (expand_expr_real_1): Pass mode as argument.
19950 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19952 * combine.c (combine_instructions): Print insns to dump_file, together
19955 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19957 * asan.c (asan_emit_stack_protection): Update.
19958 (create_cond_insert_point): Update.
19959 * auto-profile.c (afdo_propagate_circuit): Update.
19960 * basic-block.h (struct edge_def): Turn probability to
19961 profile_probability.
19962 (EDGE_FREQUENCY): Update.
19963 * bb-reorder.c (find_traces_1_round): Update.
19964 (better_edge_p): Update.
19965 (sanitize_hot_paths): Update.
19966 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19967 (make_single_succ_edge): Update.
19968 (check_bb_profile): Update.
19969 (dump_edge_info): Update.
19970 (update_bb_profile_for_threading): Update.
19971 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19972 probabilitycount to 0.
19973 * cfgbuild.c (compute_outgoing_frequencies): Update.
19974 * cfgcleanup.c (try_forward_edges): Update.
19975 (outgoing_edges_match): Update.
19976 (try_crossjump_to_edge): Update.
19977 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19978 (expand_gimple_tailcall): Update.
19979 (construct_init_block): Use make_single_succ_edge.
19980 (construct_exit_block): Use make_single_succ_edge.
19981 * cfghooks.c (verify_flow_info): Update.
19982 (redirect_edge_succ_nodup): Update.
19983 (split_edge): Update.
19984 (account_profile_record): Update.
19985 * cfgloopanal.c (single_likely_exit): Update.
19986 * cfgloopmanip.c (scale_loop_profile): Update.
19987 (set_zero_probability): Remove.
19988 (duplicate_loop_to_header_edge): Update.
19989 * cfgloopmanip.h (loop_version): Update prototype.
19990 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19991 (force_nonfallthru_and_redirect): Update.
19992 (update_br_prob_note): Update.
19993 (rtl_verify_edges): Update.
19994 (purge_dead_edges): Update.
19995 (rtl_lv_add_condition_to_bb): Update.
19996 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19997 * cgraphunit.c (init_lowered_empty_function): Update.
19998 (cgraph_node::expand_thunk): Update.
19999 * cilk-common.c: Include profile-count.h
20000 * dojump.c (inv): Remove.
20001 (jumpifnot): Update.
20002 (jumpifnot_1): Update.
20003 (do_jump_1): Update.
20005 (do_jump_by_parts_greater_rtx): Update.
20006 (do_compare_rtx_and_jump): Update.
20007 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20008 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20009 * dwarf2cfi.c: Include profile-count.h
20010 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20011 (sjlj_emit_dispatch_table): Likewise.
20012 * explow.c: Include profile-count.h
20013 * expmed.c (emit_store_flag_force): Update.
20014 (do_cmp_and_jump): Update.
20015 * expr.c (compare_by_pieces_d::generate): Update.
20016 (compare_by_pieces_d::finish_mode): Update.
20017 (emit_block_move_via_loop): Update.
20018 (store_expr_with_bounds): Update.
20019 (store_constructor): Update.
20020 (expand_expr_real_2): Update.
20021 (expand_expr_real_1): Update.
20022 * expr.h (try_casesi, try_tablejump): Update prototypes.
20023 * gimple-pretty-print.c (dump_probability): Update.
20024 (dump_profile): New.
20025 (dump_gimple_label): Update.
20026 (dump_gimple_bb_header): Update.
20027 * graph.c (draw_cfg_node_succ_edges): Update.
20028 * hsa-gen.c (convert_switch_statements): Update.
20029 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20030 (find_if_case_1): Update.
20031 (find_if_case_2): Update.
20032 * internal-fn.c (expand_arith_overflow_result_store): Update.
20033 (expand_addsub_overflow): Update.
20034 (expand_neg_overflow): Update.
20035 (expand_mul_overflow): Update.
20036 (expand_vector_ubsan_overflow): Update.
20037 * ipa-cp.c (good_cloning_opportunity_p): Update.
20038 * ipa-split.c (split_function): Use make_single_succ_edge.
20039 * ipa-utils.c (ipa_merge_profiles): Update.
20040 * loop-doloop.c (add_test): Update.
20041 (doloop_modify): Update.
20042 * loop-unroll.c (compare_and_jump_seq): Update.
20043 (unroll_loop_runtime_iterations): Update.
20044 * lra-constraints.c (lra_inheritance): Update.
20045 * lto-streamer-in.c (input_cfg): Update.
20046 * lto-streamer-out.c (output_cfg): Update.
20047 * mcf.c (adjust_cfg_counts): Update.
20048 * modulo-sched.c (sms_schedule): Update.
20049 * omp-expand.c (expand_omp_for_init_counts): Update.
20050 (extract_omp_for_update_vars): Update.
20051 (expand_omp_ordered_sink): Update.
20052 (expand_omp_for_ordered_loops): Update.
20053 (expand_omp_for_generic): Update.
20054 (expand_omp_for_static_nochunk): Update.
20055 (expand_omp_for_static_chunk): Update.
20056 (expand_cilk_for): Update.
20057 (expand_omp_simd): Update.
20058 (expand_omp_taskloop_for_outer): Update.
20059 (expand_omp_taskloop_for_inner): Update.
20060 * omp-simd-clone.c (simd_clone_adjust): Update.
20061 * optabs.c (expand_doubleword_shift): Update.
20062 (expand_abs): Update.
20063 (emit_cmp_and_jump_insn_1): Update.
20064 (expand_compare_and_swap_loop): Update.
20065 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20066 * predict.c (predictable_edge_p): Update.
20067 (edge_probability_reliable_p): Update.
20068 (set_even_probabilities): Update.
20069 (combine_predictions_for_insn): Update.
20070 (combine_predictions_for_bb): Update.
20071 (propagate_freq): Update.
20072 (estimate_bb_frequencies): Update.
20073 (force_edge_cold): Update.
20074 * profile-count.c (profile_count::dump): Add missing space into dump.
20075 (profile_count::debug): Add newline.
20076 (profile_count::differs_from_p): Explicitly convert to unsigned.
20077 (profile_count::stream_in): Update.
20078 (profile_probability::dump): New member function.
20079 (profile_probability::debug): New member function.
20080 (profile_probability::differs_from_p): New member function.
20081 (profile_probability::differs_lot_from_p): New member function.
20082 (profile_probability::stream_in): New member function.
20083 (profile_probability::stream_out): New member function.
20084 * profile-count.h (profile_count_quality): Rename to ...
20085 (profile_quality): ... this one.
20086 (profile_probability): New.
20087 (profile_count): Update.
20088 * profile.c (compute_branch_probabilities): Update.
20089 * recog.c (peep2_attempt): Update.
20090 * sched-ebb.c (schedule_ebbs): Update.
20091 * sched-rgn.c (find_single_block_region): Update.
20092 (compute_dom_prob_ps): Update.
20093 (schedule_region): Update.
20094 * sel-sched-ir.c (compute_succs_info): Update.
20095 * stmt.c (struct case_node): Update.
20096 (do_jump_if_equal): Update.
20097 (get_outgoing_edge_probs): Update.
20098 (conditional_probability): Update.
20099 (emit_case_dispatch_table): Update.
20100 (expand_case): Update.
20101 (expand_sjlj_dispatch_table): Update.
20102 (emit_case_nodes): Update.
20103 * targhooks.c: Update.
20104 * tracer.c (better_p): Update.
20105 (find_best_successor): Update.
20106 * trans-mem.c (expand_transaction): Update.
20107 * tree-call-cdce.c: Update.
20108 * tree-cfg.c (gimple_split_edge): Upate.
20109 (move_sese_region_to_fn): Upate.
20110 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20111 * tree-eh.c (lower_resx): Upate.
20112 (cleanup_empty_eh_move_lp): Upate.
20113 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20114 * tree-inline.c (copy_edges_for_bb): Update.
20115 (copy_cfg_body): Update.
20116 * tree-parloops.c (gen_parallel_loop): Update.
20117 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20118 (gimple_gen_time_profiler): Update.
20119 * tree-ssa-dce.c (remove_dead_stmt): Update.
20120 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20121 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20122 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20123 (unloop_loops): Update.
20124 (try_peel_loop): Update.
20125 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20126 * tree-ssa-loop-split.c (connect_loops): Update.
20127 (split_loop): Update.
20128 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20129 (hoist_guard): Update.
20130 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20131 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20132 (value_replacement): Update.
20133 * tree-ssa-reassoc.c (branch_fixup): Update.
20134 * tree-ssa-tail-merge.c (replace_block_by): Update.
20135 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20136 (create_edge_and_update_destination_phis): Update.
20137 (compute_path_counts): Update.
20138 (recompute_probabilities): Update.
20139 (update_joiner_offpath_counts): Update.
20140 (freqs_to_counts_path): Update.
20141 (duplicate_thread_path): Update.
20142 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20143 (struct switch_conv_info): Update.
20144 (gen_inbound_check): Update.
20145 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20146 (vect_do_peeling): Update.
20147 (vect_loop_versioning): Update.
20148 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20149 (optimize_mask_stores): Update.
20150 * ubsan.c (ubsan_expand_null_ifn): Update.
20151 * value-prof.c (gimple_divmod_fixed_value): Update.
20152 (gimple_divmod_fixed_value_transform): Update.
20153 (gimple_mod_pow2): Update.
20154 (gimple_mod_pow2_value_transform): Update.
20155 (gimple_mod_subtract): Update.
20156 (gimple_mod_subtract_transform): Update.
20157 (gimple_ic): Update.
20158 (gimple_stringop_fixed_value): Update.
20159 (gimple_stringops_transform): Update.
20160 * value-prof.h: Update.
20162 2017-06-29 Carl Love <cel@us.ibm.com>
20164 * config/rs6000/rs6000-c.c: Add support for built-in functions
20165 vector signed int vec_signed (vector float);
20166 vector signed long long vec_signed (vector double);
20167 vector signed int vec_signed2 (vector double, vector double);
20168 vector signed int vec_signede (vector double);
20169 vector signed int vec_signedo (vector double);
20170 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20171 instruction generator.
20172 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20173 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20174 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20176 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20177 vunsignede_v2df): Add define_expands.
20178 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20179 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20180 VEC_UNSIGNEDO): Add definitions.
20181 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20182 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20183 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20184 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20185 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20186 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20187 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20188 vec_unsignede, vec_unsignedo): Add builtin defines.
20189 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20191 * doc/extend.texi: Update the built-in documentation file for the
20192 new built-in functions.
20194 2017-06-29 Richard Biener <rguenther@suse.de>
20196 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20197 reduction chains to LOOP_VINFO_REDUCTIONS.
20198 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20199 SLP reductions after processing reduction chains.
20201 2017-06-29 Nathan Sidwell <nathan@acm.org>
20203 * builtins.c (fold_builtin_FUNCTION): Use
20204 lang_hooks.decl_printable_name.
20206 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20208 PR middle-end/81194
20209 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20210 with only one label.
20211 * stmt.c (expand_case): Assert NCASES is greater than one.
20213 2017-06-29 Richard Biener <rguenther@suse.de>
20215 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20217 (group_case_labels): Likewise.
20218 (find_taken_edge): Push sanity checking on val to workers...
20219 (find_taken_edge_cond_expr): ... here
20220 (find_taken_edge_switch_expr): ... and here, handle cases
20221 with just a default label.
20222 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20223 (group_case_labels): Likewise.
20224 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20225 group_case_labels does anything cleanup the CFG again.
20227 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20229 PR tree-optimization/81196
20230 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20231 exit condition comparing two IVs.
20233 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20235 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20236 profile to the dummy entry at the end of the list of architectures.
20237 * config/arm/arm-cpu-cdata.h: Regenerated.
20239 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20240 Michael Collison <michael.collison@arm.com>
20243 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20245 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20246 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20247 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20248 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20249 of shift when the shift amount is masked with constant equal to
20250 the size of the mode.
20251 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20254 2017-06-29 Martin Liska <mliska@suse.cz>
20256 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20258 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20261 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20264 (edge_growth_cache_entry::edge_growth_cache_entry): New
20266 (reset_edge_growth_cache): Update to use constructor.
20268 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20270 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20271 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20272 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20274 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20276 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20277 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20279 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20281 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20282 (*-linux-uclibc*): Add t-uclibc tmake_file.
20283 * config/t-musl: New.
20284 * config/t-uclibc: New.
20286 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20288 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20290 (gen_comm_data): Emit architectural setting of arch_prof.
20291 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20293 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20294 (armv8-m.base, armv8-m.main): Likewise.
20295 * arm-protos.h (arm_build_target): Add profile field.
20296 (arch_option): Likewise.
20297 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20299 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20300 arm_active_target.profile.
20302 2017-06-28 Richard Biener <rguenther@suse.de>
20304 PR middle-end/81227
20305 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20306 TYPE_OVERFLOW_WRAPS.
20307 * match.pd (negate_expr_p): Likewise.
20308 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20309 fold_build2, not fold_binary.
20311 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20313 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20314 Convert memory address to Pmode.
20315 (aarch64_print_operand): Assert MEM operands are always Pmode.
20317 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20320 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20321 Remove redundant if.
20322 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20323 * config/arm/aarch-common-protos.h
20324 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20325 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20327 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20330 * multiple_target.c (create_dispatcher_calls): Set the default
20331 clone to be static, not public.
20333 2017-06-28 Richard Biener <rguenther@suse.de>
20335 * tree-vect-loop.c (vectorizable_reduction): Move special
20336 cond reduction IV var creation ...
20337 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20338 parameter. Use STMT_VINFO_VECTYPE.
20339 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20342 2017-06-28 Martin Liska <mliska@suse.cz>
20345 * ipa-visibility.c (non_local_p): Handle visibility.
20347 2017-06-28 Martin Liska <mliska@suse.cz>
20350 * common.opt: Add IntegerRange to various options.
20351 * opt-functions.awk (integer_range_info): New function.
20352 * optc-gen.awk: Add integer_range_info to cl_options struct.
20353 * opts-common.c (decode_cmdline_option): Handle
20354 CL_ERR_INT_RANGE_ARG.
20355 (cmdline_handle_error): Likewise.
20356 * opts.c (print_filtered_help): Show valid interval in
20357 when --help is provided.
20358 * opts.h (struct cl_option): Add range_min and range_max fields.
20359 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20361 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20363 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20364 (x * C EQ/NE y * C): New transformation.
20366 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20368 * genmultilib (combination_space): Accept '+' in option names.
20370 2017-06-28 Martin Liska <mliska@suse.cz>
20373 * asan.c (instrument_derefs): Bail out inner references
20374 that are hard register variables.
20376 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20379 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20380 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20382 2017-06-28 Richard Biener <rguenther@suse.de>
20384 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20385 (vect_get_slp_defs): Adjust.
20386 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20388 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20390 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20391 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20392 (vectorizable_reduction): Adjust.
20393 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20395 (vect_get_slp_defs): Likewise.
20396 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20397 (vectorizable_bswap): Adjust.
20398 (vectorizable_call): Likewise.
20399 (vectorizable_conversion): Likewise.
20400 (vectorizable_assignment): Likewise.
20401 (vectorizable_shift): Likewise.
20402 (vectorizable_operation): Likewise.
20403 (vectorizable_store): Likewise.
20404 (vectorizable_condition): Likewise.
20405 (vectorizable_comparison): Likewise.
20407 2017-06-28 Michael Collison <michael.collison@arm.com>
20410 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20412 (arm_gen_movmemqi): Removed unused variable 'i'.
20413 Convert 'for' loop into 'while' loop.
20414 (arm_expand_prologue): Remove last unnecessary set of insn.
20415 (thumb_pop): Remove unused variable 'pushed_words'.
20416 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20418 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20420 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20421 * config/s390/s390-protos.h: Add prototype of
20422 s390_rel_address_ok_p.
20423 * config/s390/s390.c (s390_got_symbol): New function.
20424 (s390_rel_address_ok_p): New function.
20425 (legitimize_pic_address): Use s390_rel_address_ok_p.
20426 (s390_load_got): Use s390_got_symbol.
20427 (s390_option_override): Issue error if
20428 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20429 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20431 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20433 2017-06-27 Andrew Pinski <apinski@cavium.com>
20435 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20436 (X * copysign (1.0, X)): New pattern.
20437 (X * copysign (1.0, -X)): New pattern.
20438 (copysign (-1.0, CST)): New pattern.
20440 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20442 * genmultilib (combination_space): Remove variable.
20443 Validate reuse rules against regular expression for any sequence
20444 of multilib options in any order.
20446 2017-06-27 Michael Collison <michael.collison@arm.com>
20448 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20449 call aarch64_split_simd_combine.
20450 * (aarch64_combine_internal<mode>): Delete pattern.
20451 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20452 Allow register and subreg operands.
20454 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20456 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20457 specific need, just fallback on defaults.
20458 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20460 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20461 Olivier Hainque <hainque@adacore.com>
20463 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20465 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20466 targets. Pick a default if no particular attempt applied.
20467 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20470 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20472 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20473 (x86_64-wrs-vxworks7): Likewise.
20475 2017-06-27 Marek Polacek <polacek@redhat.com>
20478 * ubsan.c (instrument_null): Check get_base_address's result for null.
20480 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20482 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20484 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20486 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20487 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20488 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20489 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20490 New function types.
20491 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20492 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20493 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20494 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20495 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20496 BUILT_IN_FEUPDATEENV): New builtins.
20497 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20498 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20499 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20500 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20502 (builtin_structptr_types): Adjust size.
20503 * tree.c (builtin_structptr_types): Add four entries.
20505 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20506 Olivier Hainque <hainque@adacore.com>
20508 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20509 (TLS_SYM): New local macro, forcing reference to __tls__ on
20510 link command lines for VxWorks 7 RTPs, triggering initialization
20512 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20513 OS features TLS support, true for RTPs on VxWorks 7.
20514 * config/vxworks.c (vxworks_override_options): Setup emutls
20517 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20519 * predict.c (test_prediction_value_range): Use -1U instead of -1
20520 to avoid narrowing conversion warning.
20521 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20522 to avoid narrowing conversion warning.
20523 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20525 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20527 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20529 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20530 64bit configurations.
20531 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20532 (SIZE_TYPE): Likewise.
20533 * config/vxworks.c (vxworks_emutls_var_fields): Use
20534 long_unsigned_type_node instead of unsigned_type_node as the offset
20535 field type, which is "pointer" mode in emutls.c.
20537 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20540 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20542 PR middle-end/81207
20543 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20544 gimple_vuse copying separately from gimple_vdef copying.
20546 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20548 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20551 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20552 Olivier Hainque <hainque@adacore.com>
20554 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20555 for all vxworks7 targets.
20556 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20557 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20558 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20559 variations for VX6/VX7 and 32/64bits later on in ...
20560 (VXWORKS_LIB_SPEC): Leverage new macros.
20561 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20562 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20564 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20566 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20567 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20569 2017-06-26 Carl Love <cel@us.ibm.com>
20571 * config/rs6000/rs6000-c.c: Add support for built-in functions
20572 vector bool char vec_reve (vector bool char);
20573 vector signed char vec_reve (vector signed char);
20574 vector unsigned char vec_reve (vector unsigned char);
20575 vector bool int vec_reve (vector bool int);
20576 vector signed int vec_reve (vector signed int);
20577 vector unsigned int vec_reve (vector unsigned int);
20578 vector bool long long vec_reve (vector bool long long);
20579 vector signed long long vec_reve (vector signed long long);
20580 vector unsigned long long vec_reve (vector unsigned long long);
20581 vector bool short vec_reve (vector bool short);
20582 vector signed short vec_reve (vector signed short);
20583 vector double vec_reve (vector double);
20584 vector float vec_reve (vector float);
20585 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20586 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20587 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20588 (altivec_vreve): New pattern.
20589 * config/rs6000/altivec.h (vec_reve): New define.
20590 * doc/extend.texi (vec_rev): Update the built-in documentation file
20591 for the new built-in functions.
20593 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20595 PR tree-optimization/71815
20596 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20598 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20600 (slsr_process_phi): Likewise.
20601 (replace_uncond_cands_and_profitable_phis): Don't replace a
20602 multiply candidate with a stride of 1 (copy or cast).
20603 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20605 (lowest_cost_path): Likewise.
20606 (total_savings): Likewise.
20608 2017-06-26 Richard Biener <rguenther@suse.de>
20611 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20612 Use def_builtin_pure for all gather builtins.
20614 2017-06-26 Richard Biener <rguenther@suse.de>
20616 PR tree-optimization/81203
20617 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20618 non-dominating BBs.
20620 2017-06-26 Marek Polacek <polacek@redhat.com>
20623 * doc/invoke.texi: Document -Wmultistatement-macros.
20625 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20627 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20628 arm_neon_ok_no_float_abi effective target.
20630 2017-06-26 Richard Biener <rguenther@suse.de>
20632 PR tree-optimization/80928
20633 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20634 (copy_bbs): Set BB_DUPLICATED flag early.
20635 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20637 (execute_on_shrinking_pred): Likewise.
20638 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20639 BB_DUPLICATED blocks.
20640 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20641 iterate over all PHIs considering removal of *gsi.
20643 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20645 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20648 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20650 * config/rs6000/rs6000-string.c: (expand_block_clear,
20651 do_load_for_compare, select_block_compare_mode,
20652 compute_current_alignment, expand_block_compare,
20653 expand_strncmp_align_check, expand_strn_compare,
20654 expand_block_move, rs6000_output_load_multiple)
20655 Move functions related to string/block move/compare
20656 to a separate file.
20657 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20658 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20659 for this function which is now used in two files.
20660 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20661 * config.gcc: Add rs6000-string.o to extra_objs for
20662 targets powerpc*-*-* and rs6000*-*-*.
20664 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20667 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20668 32-bit, since indexed is not valid for DImode.
20669 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20670 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20671 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20672 (define_peephole2 for Altivec d-form store): Likewise.
20675 * multiple_target.c (create_dispatcher_calls): Only create the
20676 dispatcher call if the function is the default clone of a
20677 versioned function.
20679 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20681 PR middle-end/80902
20682 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20683 a call, force the call to not be a tail call.
20685 2017-06-23 Jeff Law <law@redhat.com>
20687 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20690 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20692 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20693 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20694 handling for early expansion of vector shifts (sl,sr,sra,rl).
20695 (builtin_function_type): Add vector shift right instructions
20696 to the unsigned argument list.
20698 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20700 rtl-optimizatoin/79286
20701 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20702 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20703 trap. PIC register plus a const unspec without offset can never trap.
20705 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20707 * tree.h (builtin_structptr_type): New type.
20708 (builtin_structptr_types): Declare new array.
20709 * tree.c (builtin_structptr_types): New array.
20710 (free_lang_data, build_common_tree_nodes): Use it.
20712 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20715 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20718 2017-06-22 Matt Turner <mattst88@gmail.com>
20720 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20721 Lake models to skylake case. Assume skylake for unknown
20722 models with clflushopt.
20724 2017-06-22 Jeff Law <law@redhat.com>
20726 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20727 frame sizes that do not satisfy aarch64_uimm12_shift.
20729 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20731 * profile-count.h (apply_probability,
20732 apply_scale, probability_in): Fix checks for zero.
20734 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20736 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20737 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20739 2016-06-22 Richard Biener <rguenther@suse.de>
20741 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20742 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20743 REDUC_MAX_EXPR support.
20744 (vectorizable_reduction): Likewise.
20745 (vect_create_epilog_for_reduction): Likewise.
20747 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20749 * match.pd (A / (1 << B) -> A >> B): New.
20750 * generic-match-head.c: Include optabs-tree.h.
20751 * gimple-match-head.c: Likewise.
20752 * optabs-tree.h (target_supports_op_p): New.
20753 * optabs-tree.c (target_supports_op_p): New.
20755 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20757 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20758 $gcc_cv_ld --help output.
20759 (gcc_cv_ld_demangle): Likewise.
20760 (gcc_cv_ld_eh_frame_hdr): Likewise.
20761 (gcc_cv_ld_pie): Likewise.
20762 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20763 (gcc_cv_ld_buildid): Likewise.
20764 (gcc_cv_ld_sysroot): Likewise.
20765 (ld_bndplt_support): Likewise.
20766 (ld_pushpopstate_support): Likewise.
20767 * configure: Regenerate.
20768 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20770 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20773 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20774 operands indexes to avoid gap between operands and match_dups.
20776 2017-06-21 Andrew Pinski <apinski@cavium.com>
20778 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20779 Increment Arith_shift and Arith_shift_reg by 1.
20780 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20782 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20783 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20784 (aarch64_strip_extend): Add new argument and test for it.
20785 (aarch64_cheap_mult_shift_p): New function.
20786 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20787 add a cost if it is true.
20788 Update calls to aarch64_strip_extend.
20789 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20791 2017-06-21 Andrew Pinski <apinski@cavium.com>
20793 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20795 (thunderxt88): Likewise.
20796 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20797 (thunderx_prefetch_tune): New variable.
20798 (thunderx2t99_prefetch_tune): Update for the correct values.
20799 (thunderxt88_tunings): New variable.
20800 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20801 generic_prefetch_tune.
20802 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20804 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20806 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20807 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20808 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20809 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20810 (aarch64_atomic_cas<mode>, GPI): Likewise.
20812 2017-06-21 Martin Liska <mliska@suse.cz>
20814 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20815 statements on cold and hot labels.
20816 * predict.c (tree_estimate_probability_bb): Remove the
20817 prediction from this place.
20819 2017-06-21 Martin Liska <mliska@suse.cz>
20821 PR tree-optimization/79489
20822 * gimplify.c (maybe_add_early_return_predict_stmt): New
20824 (gimplify_return_expr): Call the function.
20825 * predict.c (tree_estimate_probability_bb): Remove handling
20827 * predict.def: Update comment about early return predictor.
20828 * gimple-predict.h (is_gimple_predict): New function.
20829 * predict.def: Change default value of early return to 66.
20830 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20832 * passes.def: Put pass_strip_predict_hints to the beginning of
20835 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20837 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20838 FUNCTION_DECL declarations.
20839 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20841 (dwaf2out_decl): Likewise.
20842 * godump.c (go_early_global_decl): Skip call to the real debug hook
20843 for FUNCTION_DECL declarations.
20844 * passes.c (rest_of_decl_compilation): Skip call to the
20845 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20846 -fdump-go-spec is passed.
20848 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20850 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20851 Reorder for compactness.
20852 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20853 (def_builtin_pure, def_builtin_pure2): New functions.
20854 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20856 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20858 * match.pd (nop_convert): New predicate.
20859 ((A +- CST1) +- CST2): Allow some NOP conversions.
20861 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20864 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20865 with ctors/dtors if GOVD_SHARED is set.
20867 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20869 * config/aarch64/aarch64.md (movti_aarch64):
20870 Emit mov rather than orr.
20871 (movtf_aarch64): Likewise.
20872 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20873 Emit mov rather than orr.
20875 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20877 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20878 Swap alternatives, make integer dup more expensive.
20880 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20882 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20883 Return true for non-tls symbols.
20885 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20887 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20888 (cortex-a75): Likewise.
20889 (cortex-a75.cortex-a55): Likewise.
20890 * config/aarch64/aarch64-tune.md: Regenerate.
20891 * doc/invoke.texi (-mtune): Document new values for -mtune.
20893 2017-06-21 Tom de Vries <tom@codesourcery.com>
20895 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20896 stack_size feature.
20897 (Effective-Target Keywords, Other attributes): Suggest using
20898 dg-add-options stack_size feature to get stack limit in stack_size
20899 effective target documentation.
20901 2017-06-21 Julian Brown <julian@codesourcery.com>
20902 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20904 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20905 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20906 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20908 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20909 attribute type list for neon_multiply.
20910 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20911 attribute type list for neon_multiply.
20912 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20913 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20914 attribute type list for neon_multiply.
20915 * config/arm/types.md (crypto_pmull): Add.
20916 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20917 attribute type list.
20919 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20921 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20924 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20926 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20927 to make sure not to dereference a NULL cost_classes_ptr pointer.
20929 2017-06-20 Carl Love <cel@us.ibm.com>
20931 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20932 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20933 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20934 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20935 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20936 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20937 VMULOSW): New enum "unspec" values.
20938 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20939 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20940 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20941 altivec_vmulosw): New patterns.
20942 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20943 VMULOSW): Add definitions.
20945 2017-06-20 Julia Koval <julia.koval@intel.com>
20947 * config/i386/i386.c: Fix rounding expand for new pattern.
20948 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20950 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20952 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20953 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20955 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20957 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20960 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20962 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20963 architecture, then by implementer ID.
20964 * config/aarch64/aarch64-tune.md: Regenerate.
20966 2017-06-20 Richard Biener <rguenther@suse.de>
20968 PR middle-end/81097
20969 * fold-const.c (split_tree): Fold to type before negating.
20971 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20973 * diagnostic-show-locus.c
20974 (selftest::test_fixit_deletion_affecting_newline): New function.
20975 (selftest::diagnostic_show_locus_c_tests): Call it.
20977 2017-06-20 Andreas Schwab <schwab@suse.de>
20980 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20983 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20985 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20986 __ARM_FEATURE_COPROC according to support.
20988 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20990 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20991 Rewritten to avoid overflow for > 32-bit pointers.
20994 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20995 by removing enum keyword.
20996 (ubsan_type_descriptor): Likewise. Formatting fix.
20999 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21000 splitter): Require TARGET_SSE2 in the condition.
21002 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21005 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21006 for doing vector set of SFmode on ISA 3.0.
21007 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21008 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21010 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21011 SFmode value into a V4SF variable that was extracted from another
21012 V4SF variable without converting the element to double precision
21013 and back to single precision vector format.
21014 (vsx_insert_extract_v4sf_p9_2): Likewise.
21016 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21018 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21019 in UWHI to avoid undefined overflow.
21022 * ubsan.h (enum ubsan_encode_value_phase): New.
21023 (ubsan_encode_value): Change second argument to
21024 enum ubsan_encode_value_phase with default value of
21025 UBSAN_ENCODE_VALUE_GENERIC.
21026 * ubsan.c (ubsan_encode_value): Change second argument to
21027 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21028 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21029 create_tmp_var_raw instead of create_tmp_var and use a
21031 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21032 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21033 ubsan_encode_value callers.
21036 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21037 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21038 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21040 2017-06-19 Richard Biener <rguenther@suse.de>
21042 PR middle-end/81118
21043 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21044 estimates if we changed anything.
21046 2017-06-19 Richard Biener <rguenther@suse.de>
21048 PR tree-optimization/80887
21049 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21050 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21051 simplified lookups, then reset mprts_hook.
21052 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21054 (try_to_simplify): Likewise.
21056 2017-06-19 Martin Liska <mliska@suse.cz>
21059 * gimplify.c (gimplify_switch_expr):
21060 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21062 2017-06-19 Martin Liska <mliska@suse.cz>
21064 * doc/install.texi: Document that PGO runs in 4 stages.
21066 2017-06-19 Martin Liska <mliska@suse.cz>
21069 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21070 to dispatcher function name.
21071 * multiple_target.c (replace_function_decl): New function.
21072 (create_dispatcher_calls): Redirect both edges and references.
21074 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21076 * profile-count.c (profile_count::dump): Dump quality.
21077 (profile_count::differs_from_p): Update for unsigned val.
21078 * profile-count.h (profile_count_quality): New enum.
21079 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21081 2017-06-19 Richard Biener <rguenther@suse.de>
21083 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21084 struct function as arg.
21085 (estimate_numbers_of_iterations): Export overload with loop arg.
21086 (free_numbers_of_iterations_estimates_loop): Use an overload of
21087 free_numbers_of_iterations_estimates instead.
21088 * tree-cfg.c (remove_bb): Adjust.
21089 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21090 * tree-parloops.c (gen_parallel_loop): Likewise.
21091 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21093 (tree_unroll_loops_completely): Likewise.
21094 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21095 Use an overload instead and export.
21096 (estimated_loop_iterations): Adjust.
21097 (max_loop_iterations): Likewise.
21098 (likely_max_loop_iterations): Likewise.
21099 (estimate_numbers_of_iterations): Take struct function as arg
21101 (loop_exits_before_overflow): Adjust.
21102 (free_numbers_of_iterations_estimates_loop): Use an overload.
21103 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21104 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21106 2017-06-19 Richard Biener <rguenther@suse.de>
21109 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21110 RANGE_EXPR conservatively.
21112 2017-06-16 Carl Love <cel@us.ibm.com>
21114 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21115 definitions for vec_float, vec_float2, vec_floato,
21116 vec_floate built-ins.
21117 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21118 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21120 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21121 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21122 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21123 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21124 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21125 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21126 vec_floato): Add builtin defines.
21127 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21128 Update the built-in documentation file for the new built-in
21131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21133 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21134 (mthumb): Mark as the negative of -marm.
21136 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21138 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21140 (ARM Options, -mtune): Document that this accepts the same
21141 extension options as -mcpu.
21142 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21146 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21147 permitted extensions.
21149 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21151 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21152 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21153 (armv8-m.main): Add option +nodsp.
21154 * config/arm/arm-cpu-cdata.h: Regenerated.
21156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21158 * config/arm/t-fuchsia: New file.
21159 * config.gcc (arm*-*-fuchsia*): Use it.
21161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21163 * config/arm/t-symbian: Rewrite for new option infrastructure.
21165 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21167 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21168 (MULTILIB_REQUIRED): Likewise.
21170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21172 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21173 (MULTILIB_RESUE): Likewise.
21174 (MULTILIB_MATCHES): Likewise.
21175 (MULTLIB_REQUIRED): Likewise.
21177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21179 * config/arm/t-rtems: Rewrite for new option framework.
21181 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21183 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21184 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21185 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21186 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21187 * config/arm/t-multilib: ... here.
21188 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21189 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21190 armv7-a and armv8*-a when A-profile libraries have not been built.
21191 * config/arm/t-rmprofile: Rewrite.
21193 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21195 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21196 with a backslash. Remove the backslash after substituting unescaped
21198 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21202 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21203 the driver to pass through the default setting of -mfloat-abi.
21204 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21206 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21207 (all_feat_combs): New rule.
21208 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21210 * config/arm/t-aprofile: Rewrite.
21212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21214 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21215 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21216 fpu is not specified by the user/command-line.
21217 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21218 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21219 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21220 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21221 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21222 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21224 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21226 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21227 * config/arm/t-arm-elf: Rewritten.
21229 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21231 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21232 have some floating-point instructions.
21233 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21234 (TARGET_MAYBE_HARD_FLOAT): New macro.
21235 * config/arm/arm-builtins.c (arm_init_builtins): Use
21236 TARGET_MAYBE_HARD_FLOAT.
21237 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21241 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21242 (configargs.h): Include it.
21243 (arm_print_hint_for_fpu_option): New function.
21244 (arm_parse_fpu_option): New function.
21245 (candidate_extension): New class.
21246 (arm_canon_for_multilib): New function.
21247 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21248 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21249 (ARCH_CANONICAL_SPECS): New macro.
21250 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21254 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21255 are set after handling multilib fragments. Set target_cpu_default2
21258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21260 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21262 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21263 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21265 (check_arch): Likewise.
21266 * config/arm/arm.c (arm_configure_build_target): Handle
21267 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21268 options in the default.
21270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21272 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21273 when an option is an alias of another.
21274 * config/arm/parsecpu.awk (optalias): New parser token.
21275 (gen_comm_data): Mark non-alias options as such. Emit entries
21276 for extension aliases.
21277 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21278 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21279 (armv6kz, armv6zk, armv6t2): Likewise.
21280 (armv7): Make vfpv3-d16 an alias.
21281 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21283 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21284 Sort in canonical order.
21285 (armv8-a): Sort in canonical order.
21286 (armv8.1-a, armv8.2-a): Likewise.
21287 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21289 (cortex-a9): Sort in canonical order.
21290 * config/arm/arm.c (selftests.h): Include it.
21291 (arm_test_cpu_arch_data): New function.
21292 (arm_run_self_tests): New function.
21293 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21294 (targetm): Move declaration to the end of the file.
21295 * arm-cpu-cdata.h: Regenerated.
21297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21299 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21300 call to target_mode_check describing the type of option passed.
21301 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21302 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21303 arm_parse_cpu_option_name to match parameters against list of
21305 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21306 arm_arch_core_flags data structure.
21307 * config/arm/arm-cpu_cdata.h: Regenerated.
21309 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21311 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21313 (arm_print_hint_for_cpu_option): Likewise.
21314 (arm_print_hint_for_arch_option): Likewise.
21315 (arm_parse_cpu_option_name): Likewise.
21316 (arm_parse_arch_option_name): Likewise.
21317 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21318 of entries in the all_fpus list.
21319 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21320 (arm_parse_cpu_option_name): Declare.
21321 (arm_parse_arch_option_name): Declare.
21322 (arm_parse_option_features): Declare.
21323 (arm_intialize_isa): Declare.
21324 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21326 (gen_comm_data): ... here. Make definitions non-static.
21327 * config/arm/arm-cpu-data.h: Regenerated.
21328 * config/arm/arm-cpu-cdata.h: Regenerated.
21330 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21332 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21333 (cpu_arch_extension): New structure.
21334 (cpu_arch_option, arch_option, cpu_option): New structures.
21335 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21336 architecture types.
21337 (gen_data): Generate new format data tables.
21338 * config/arm/arm.c (cpu_tune): New structure.
21339 (cpu_option, processors): Delete.
21340 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21341 (arm_print_hint_for_cpu_option): ... this and ...
21342 (arm_print_hint_for_arch_option): ... this.
21343 (arm_parse_arch_cpu_name): Delete. Replace with ...
21344 (arm_parse_cpu_option_name): ... this and ...
21345 (arm_parse_arch_option_name): ... this.
21346 (arm_unrecognized_feature): Change type of target parameter to
21348 (arm_parse_arch_cpu_features): Delete. Replace with ...
21349 (arm_parse_option_features): ... this.
21350 (arm_configure_build_target): Rework to use new configuration data
21352 (arm_print_tune_info): Rework for new configuration data tables.
21353 * config/arm/arm-cpu-data.h: Regenerated.
21354 * config/arm/arm-cpu.h: Regenerated.
21356 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21358 * Makefile.in (OBJS): Move sbitmap.o from here ...
21359 (OBJS-libcommon): ... to here.
21361 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21363 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21364 (ISA_ALL_CRYPTO): New macro.
21365 (ISA_ALL_SIMD): New macro
21366 (ISA_ALL_FP): New macro.
21367 * config/arm/arm.c (fpu_bitlist): Update initializer.
21368 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21370 (arm9e): Add fpu. Add option for nofp
21371 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21372 (arm926ej-s, arm1026ej-s): Likewise.
21373 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21374 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21375 neon-fp16, neon-vfpv4, nofp and nosimd.
21376 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21377 (cortex-a8): Add fpu. Add option for nofp.
21378 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21379 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21380 (cortex-r4f): Add fpu.
21381 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21382 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21384 (cortex-r8): Likewise.
21385 (cortex-m4): Add fpu. Add option for nofp.
21386 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21387 (cortex-a17.cortex-a7): Likewise.
21388 (cortex-a32): Add fpu. Add options for crypto and nofp.
21389 (cortex-a35, cortex-a53): Likewise.
21390 (cortex-a57): Add fpu. Add option for crypto.
21391 (cortex-a72, cortex-a73): Likewise.
21392 (exynos-m1): Likewise.
21393 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21394 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21395 (cortex-m33): Add fpu. Add option for nofp.
21396 * config/arm/arm-cpu-cdata.h: Regenerated
21397 * config/arm/arm-cpu-data.h: Regenerated.
21399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21401 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21402 (armv5te, armv5tej): Likewise.
21403 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21404 (armv7): Add options fp and vfpv3-d16.
21405 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21406 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21408 (armv7ve): Likewise.
21409 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21410 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21411 (armv8-a): Add nocrypto option.
21412 (armv8.1-a, armv8.2-a): Likewise.
21413 (armv8-m.main): add options fp, fp.dp and nofp.
21415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21417 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21419 (armv8-a+crc): Delete.
21420 (armv8.1-a): Add options simd, crypto and nofp.
21421 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21422 (armv8.2-a+fp16): Delete.
21423 (armv8-m.main): Add option dsp.
21424 (armv8-m.main+dsp): Delete.
21425 (cortex-a8): Add fpu. Add nofp option.
21426 (cortex-a9): Add fpu. Add nofp and nosimd options.
21427 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21428 link to main cpu and architecture data structures.
21429 (gen_comm_data): Only put isa attributes from the main architecture
21431 (option): New statement for architecture and CPU entries.
21432 * arm.c (struct cpu_option): New structure.
21433 (struct processors): Add entry for options.
21434 (arm_unrecognized_feature): New function.
21435 (arm_parse_arch_cpu_name): Ignore any characters after the first
21437 (arm_parse_arch_cpu_feature): New function.
21438 (arm_configure_build_target): Separate out any CPU and architecture
21439 features and parse separately. Don't error out if -mfpu=auto is
21440 used with only an architecture string.
21441 (arm_print_asm_arch_directives): New function.
21442 (arm_file_start): Call it.
21443 * config/arm/arm-cpu-cdata.h: Regenerated.
21444 * config/arm/arm-cpu-data.h: Likewise.
21445 * config/arm/arm-tables.opt: Likewise.
21447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21449 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21450 assembler when it is not -mfpu=auto.
21452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21454 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21455 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21456 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21457 (ASM_CPU_SPEC): Rewrite.
21458 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21459 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21460 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21461 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21462 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21463 copied string is NUL-terminated. Also strip any characters prefixed
21465 (arm_rewrite_selected_arch): New function.
21466 (arm_rewrite_march): New function.
21468 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21470 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21471 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21472 (march, mcpu, mtune): Convert to string-based options.
21473 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21474 (arm_parse_arch_cpu_name): New function.
21475 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21476 identify selected architecture or CPU.
21477 (arm_option_save): New function.
21478 (TARGET_OPTION_SAVE): Redefine.
21479 (arm_option_restore): Restore string options.
21480 (arm_option_print): Print string options.
21482 2017-06-16 Martin Sebor <msebor@redhat.com>
21484 PR tree-optimization/80933
21485 PR tree-optimization/80934
21486 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21487 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21488 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21489 (gimple_fold_builtin): Call them.
21491 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21493 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21494 as unlikely; update profile.
21496 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21498 * predict.c (force_edge_cold): Handle declaring edges impossible
21501 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21503 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21505 (try_unroll_loop_completely): Fix reporting.
21507 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21509 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21511 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21514 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21515 if given a non-constant argument for an intrinsic which requires a
21518 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21520 * profile.c (compare_freqs): New function.
21521 (branch_prob): Sort edge list.
21522 (find_spanning_tree): Assume that the list is priority sorted.
21524 2017-06-16 Richard Biener <rguenther@suse.de>
21526 PR tree-optimization/81090
21527 * passes.def (pass_record_bounds): Remove.
21528 * tree-pass.h (make_pass_record_bounds): Likewise.
21529 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21530 make_pass_record_bounds): Likewise.
21531 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21532 not free niter estimates at the beginning but at the end.
21533 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21535 2017-06-16 Richard Biener <rguenther@suse.de>
21537 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21538 initializer to workaround ICE in host GCC 4.8.
21540 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21542 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21544 (clone_inlined_nodes): Update.
21546 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21548 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21549 prefetch settings, and enable prefetching by default at -O3.
21551 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21553 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21554 Set flag_prefetch_loop_arrays according to tuning data.
21556 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21558 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21559 New tune structure.
21560 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21561 [Unrelated to main purpose of the patch] Place the pointer field last
21562 to enable type checking errors when tune structure are wrongly merged.
21563 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21564 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21565 (thunderx2t99_prefetch_tune): New tune constants.
21566 (tune_params *_tunings): Update all tunings (no functional change).
21567 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21568 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21569 from tunings structures.
21571 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21574 * ubsan.c (instrument_null): Add T argument, use it instead
21575 of computing it based on IS_LHS.
21576 (instrument_object_size): Likewise.
21577 (pass_ubsan::execute): Adjust instrument_null and
21578 instrument_object_size callers to pass gimple_get_lhs or
21579 gimple_assign_rhs1 result to it. Use instrument_null instead of
21580 calling get_base_address and instrument_mem_ref. Handle
21581 aggregate call arguments for object-size sanitization.
21583 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21585 PR tree-optimization/81089
21586 * tree-vrp.c (is_masked_range_test): Validate operands of
21589 2017-06-15 Martin Sebor <msebor@redhat.com>
21592 * dumpfile.c (dump_register): Avoid calling memset to initialize
21593 a class with a default ctor.
21594 * gcc.c (struct compiler): Remove const qualification.
21595 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21596 * hash-table.h: Ditto.
21597 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21599 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21600 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21602 * params.h (struct param_info): Make struct members non-const.
21603 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21604 with default initialization.
21605 * vec.h (vec_copy_construct, vec_default_construct): New helper
21607 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21608 with vec_copy_construct.
21609 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21610 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21611 * doc/invoke.texi (-Wclass-memaccess): Document.
21613 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21615 * emit-rtl.h (is_leaf): Update comment about local
21616 register allocator.
21618 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21621 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21622 for a variable to have a section before checking if the section has a
21624 Set section to.persistent if persistent attribute is set.
21625 Warn if .persistent attribute is used on an automatic variable.
21627 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21629 PR rtl-optimization/80474
21630 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21632 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21634 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21637 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21639 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21640 (casesi): Emit bounds checking as RTL.
21641 (casesi_internal_mips16_<mode>): Remove bounds checking.
21643 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21645 * config/xtensa/xtensa.c (xtensa_option_override): Append
21646 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21647 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21648 xtensa_doloop_hooks): Define unconditionally.
21649 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21651 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21652 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21653 for it in xtensa_option_override.
21654 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21655 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21657 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21659 * doc/cppopts.texi: Document '-' special value to -MF.
21661 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21663 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21664 (cortex_a53_fconst): Likewise.
21665 (cortex_a53_fpmul): Likewise.
21666 (cortex_a53_f_load_64): Likewise.
21667 (cortex_a53_f_load_many): Likewise.
21668 (cortex_a53_advsimd_alu): Likewise.
21669 (cortex_a53_advsimd_alu_q): Likewise.
21670 (cortex_a53_advsimd_mul): Likewise.
21671 (cortex_a53_advsimd_mul_q): Likewise.
21672 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21673 Add missing fmul, r2f_cvt and fconst cases.
21675 2017-06-14 Richard Biener <rguenther@suse.de>
21677 PR middle-end/81088
21678 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21680 (fold_binary_loc): When associating do not treat pre-existing
21681 TREE_OVERFLOW on literal constants as a reason to allow
21682 TREE_OVERFLOW on associated literal constants.
21684 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21686 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21687 (MASK_FEATURES): New macro.
21688 * config/sparc/sparc.c (sparc_option_override): Remove the special
21689 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21691 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21693 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21694 a division of 0 if non-call exceptions are enabled.
21696 2017-06-14 Andrew Pinski <apinski@cavium.com>
21697 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21700 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21701 Improve vector initialization code gen for only variable case.
21703 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21705 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21707 2017-06-14 Richard Biener <rguenther@suse.de>
21709 PR tree-optimization/81083
21710 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21713 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21715 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21716 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21717 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21718 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21719 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21720 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21722 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21724 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21725 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21727 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21729 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21731 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21733 * config/rs6000/t-rtems: Don't handle SPE.
21735 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21737 * config/rs6000/t-linux: Don't handle SPE.
21739 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21741 * config/rs6000/eabispe.h: Delete file.
21743 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21745 * config/rs6000/t-spe: Delete file.
21747 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21749 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21750 (rs6000_legitimate_offset_address_p): Return false for anything in
21751 V2SImode or V2SFmode.
21753 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21755 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21756 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21757 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21759 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21760 (rs6000_legitimate_offset_address_p): Ditto.
21761 (rs6000_emit_move): Ditto.
21762 (rs6000_init_builtins): Remove V4HI_type_node.
21764 2017-06-13 Martin Liska <mliska@suse.cz>
21767 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21768 (gate_asan): Likewise.
21769 * asan.h (asan_no_sanitize_address_p): Remove the function.
21770 (sanitize_flags_p): New function.
21771 * builtins.def: Fix coding style.
21772 * common.opt: Use renamed enum value.
21773 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21774 * doc/extend.texi: Document no_sanitize attribute.
21775 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21776 to SANITIZE_UNDEFINED_NONDEFAULT.
21777 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21778 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21779 Use sanitize_flags_p.
21780 * gimplify.c (gimplify_function_tree): Likewise.
21781 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21782 * opts.c (parse_no_sanitize_attribute): New function.
21783 (common_handle_option): Use renamed enum value.
21784 * opts.h (parse_no_sanitize_attribute): Declare.
21785 * tree.c (sanitize_flags_p): New function.
21786 * tree.h: Declared here.
21787 * tsan.c: Use sanitize_flags_p.
21788 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21789 (instrument_mem_ref): Likewise.
21790 (instrument_bool_enum_load): Likewise.
21791 (do_ubsan_in_current_function): Remove the function.
21792 (pass_ubsan::execute): Use sanitize_flags_p.
21793 * ubsan.h: Remove do_ubsan_in_current_function
21794 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21795 (dump_function_to_file): Use it here.
21797 2017-06-13 Martin Jambor <mjambor@suse.cz>
21799 PR tree-optimization/80803
21800 PR tree-optimization/81063
21801 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21802 (propagate_subaccesses_across_link): Enqueue subtree whenever
21803 necessary instead of relying on the caller.
21805 2017-06-13 Martin Jambor <mjambor@suse.cz>
21807 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21808 that have a first_link.
21809 (sort_and_splice_var_accesses): Do not check first_link before
21811 (subtree_mark_written_and_enqueue): Likewise.
21812 (propagate_all_subaccesses): Likewise and do not stop at first
21813 parent with a first_link.
21815 2017-06-13 Martin Jambor <mjambor@suse.cz>
21817 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21820 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21822 * match.pd: New pattern.
21824 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21826 * tree-vrp.c (is_masked_range_test): New function.
21827 (register_edge_assert_for): Determine ranges for
21830 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21832 PR tree-optimization/67328
21833 * fold-const.c (maskable_range_p): New function.
21834 (build_range_check): Generate bittests if possible.
21836 2017-06-13 Martin Liska <mliska@suse.cz>
21838 * gimple-pretty-print.c (dump_probability): Add new argument.
21839 (dump_edge_probability): Dump both probability and count.
21840 (dump_gimple_label): Likewise.
21841 (dump_gimple_bb_header): Likewise.
21843 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21846 * config/avr/avr-devices.c: Fix indentation.
21847 * config/avr/gen-avr-mmcu-specs.c: Dito.
21849 2017-06-13 Richard Biener <rguenther@suse.de>
21851 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21852 instead get vector type from stmt_info.
21853 (vectorizable_reduction): Adjust. Remove dead code.
21855 2017-06-13 Richard Biener <rguenther@suse.de>
21857 PR middle-end/81065
21858 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21859 case of C * (x * C2 + C3).
21860 (fold_addr_of_array_ref_difference): Properly fold index difference.
21862 2017-06-12 David S. Miller <davem@davemloft.net>
21865 * config/sparc/sparc.md (return expander): Emit frame blockage if
21866 function uses alloca.
21868 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21870 * combine.c (make_field_assignment): Check len rather than the mode
21871 precision when calling force_to_mode.
21873 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21875 Support multilibs and devices that see flash in RAM address range.
21878 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21879 (avr_mcu_t) <flash_pm_offset>: New field.
21880 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21881 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21882 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21883 (AVR_TINY_PM_OFFSET): Remove macro.
21884 * config/avr/avr.opt (-mshort-calls): New option.
21885 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21886 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21887 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21888 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21889 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21890 instead of avr_arch->have_jmp_call.
21891 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21892 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21893 avr_arch->flash_pm_offset to define.
21894 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21895 new field flash_pm_offset. Add entry for avrxmega3.
21896 (avr_texinfo): Add entry for avrxmega3.
21897 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21898 attiny212, attiny214,
21899 attiny412, attiny414, attiny416, attiny417,
21900 attiny814, attiny816, attiny817,
21901 attiny1614, attiny1616, attiny1617,
21902 attiny3214, attiny3216, attiny3217.
21903 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21904 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21905 (avr_print_operand_address) [AVR_TINY]: Same.
21906 (avr_asm_init_sections) <readonly_data_section>: Only patch
21907 callback if avr_arch->flash_pm_offset = 0.
21908 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21909 for rodata if avr_arch->flash_pm_offset != 0.
21910 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21911 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21912 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21913 (m_options): Append opt_rcall.
21914 (m_dirnames): Append dir_rcall.
21915 * config/avr/t-multilib: Regenerate.
21917 * configure.ac [target=avr]: Check whether avrxmega3 default
21918 linker description file works as needed.
21919 * configure: Regenerate.
21920 * doc/avr-mmcu.texi: Regenerate.
21921 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21922 <__AVR_ARCH__>: Document avrxmega3 and 103.
21923 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21924 <__AVR_SHORT_CALLS__>: Document it.
21925 <__AVR_PM_BASE_ADDRESS__>: Document it.
21926 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21927 (AVR Variable Attributes) <progmem>: Document this is
21928 not needed for avrxmega3.
21929 (AVR Named Address Spaces) <__flash>: Dito.
21931 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21933 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21935 2017-06-12 Doug Rupp <rupp@adacore.com>
21937 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21938 Append vxworks-stdint.h to the tm_file list.
21939 * config/vxworks-stdint.h: New file.
21941 2017-06-12 Martin Liska <mliska@suse.cz>
21943 PR tree-optimization/81041
21944 * tree-profile.c (gimple_gen_ic_func_profiler):
21945 Create an extra BB in profile-generate
21946 (gimple_gen_time_profiler): Likewise.
21948 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21950 PR tree-optimization/81003
21951 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21952 (update_range_test): Use it instead of force_gimple_operand_gsi.
21954 2017-06-12 Richard Biener <rguenther@suse.de>
21956 PR tree-optimization/81053
21957 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21958 with backedge value not defined in loop. Simplify def stmt
21961 2017-06-11 Tom de Vries <tom@codesourcery.com>
21964 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21966 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21967 nvptx_cannot_force_const_mem.
21969 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21971 * opts.c (finish_options): Move test for flag_split_stack after
21972 it has been initialized.
21974 2017-06-11 Jason Merrill <jason@redhat.com>
21976 * tree.h (id_equal): New.
21977 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21978 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21979 instead of strcmp of IDENTIFIER_POINTER.
21981 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21983 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21984 (mark_all_inlined_calls_cdtor): Fix formating.
21985 (inline_transform): Rescale profile before inlining.
21987 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21989 * cgraph.h (cgraph_edge::clone): Update prototype.
21990 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21991 (cgraph_node::create_clone): Update.
21992 (cgraph_node::create_version_clone): Update.
21993 * tree-inline.c (copy_bb): Update.
21994 (expand_call_inline): Update.
21996 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21998 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21999 factored out from ...
22000 (rs6000_emit_prologue): ... here.
22002 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22004 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22005 factored out from ...
22006 (rs6000_emit_prologue): ... here.
22008 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22010 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22012 (handle_missing_profiles): Fix computation of tp_first_run.
22013 (counts_to_freqs): Do not touch freqs when count is 0.
22015 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22017 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22020 2017-06-10 Tom de Vries <tom@codesourcery.com>
22022 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22023 attributes): Document signal effective target.
22025 2017-06-10 Tom de Vries <tom@codesourcery.com>
22027 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22028 Document effective target stack_size.
22030 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22032 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22033 to the edit_context if they can be auto-applied.
22035 2017-06-9 Ian Lance Taylor <iant@golang.org>
22037 * opts.c (finish_options): If -fsplit-stack, disable implicit
22038 -forder-blocks-and-partition.
22039 * doc/invoke.texi (Optimize Options): Document that when using
22040 -fsplit-stack -forder-blocks-and-partition is not implicitly
22043 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22045 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22046 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22047 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22048 * builtins.def (abort, trap, unreachable): Declare cold.
22049 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22050 * tree-core.h (ECF_COLD): New.
22051 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22052 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22054 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22056 * predict.c (unlikely_executed_stmt_p): Cleanup.
22058 2017-06-09 Richard Biener <rguenther@suse.de>
22060 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22061 model if the ref is always written to.
22063 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22065 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22067 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22069 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22072 2017-06-09 Tom de Vries <tom@codesourcery.com>
22075 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22076 "target cannot support label values" when encountering LABEL_REF.
22078 2017-06-09 Martin Liska <mliska@suse.cz>
22080 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22081 (gimple_gen_ic_func_profiler): Emit direct comparison
22082 of __gcov_indirect_call_callee with NULL.
22083 (gimple_gen_time_profiler): Change probability from
22084 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22086 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22088 * profile.c (edge_gcov_counts): Turn to pointer.
22089 (compute_branch_probabilities, compute_branch_probabilities): Update.
22090 (branch_prob): Do not clear edge_gcov_count.
22091 * profile.h (edge_gcov_counts): Turn to pointer.
22092 (edge_gcov_count): Update.
22094 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22096 * gimple.h (gimple_check_failed): Mark cold.
22098 2017-06-09 Richard Biener <rguenther@suse.de>
22100 PR tree-optimization/66623
22101 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22102 refactor check_reduction into two parts, properly computing
22103 whether we have to check reduction validity for outer loop
22106 2017-06-09 Richard Biener <rguenther@suse.de>
22108 PR tree-optimization/79483
22109 * graphite-scop-detection.c (order): New global.
22110 (get_order): Compute bb to order mapping that satisfies code
22111 generation constraints.
22112 (cmp_pbbs): New helper.
22113 (build_scops): Start domwalk at entry block, sort generated
22116 2017-06-09 Richard Biener <rguenther@suse.de>
22118 PR middle-end/81007
22119 * ipa-polymorphic-call.c
22120 (ipa_polymorphic_call_context::restrict_to_inner_class):
22121 Skip FIELD_DECLs with error_mark_node type.
22122 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22125 2017-06-09 Martin Liska <mliska@suse.cz>
22127 * predict.c (struct branch_predictor): New struct.
22128 (test_prediction_value_range): New test.
22129 (predict_c_tests): New function.
22130 * selftest-run-tests.c (selftest::run_tests): Run the function.
22131 * selftest.h: Declare new tests.
22133 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22136 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22137 gen_add3_insn did not fail.
22138 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22139 r0, construct that number in a temporary reg and add that reg to r0.
22140 If asked to put the result in r0 as well, fail.
22142 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22144 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22145 for early expansion of vec_eqv.
22147 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22149 PR middle-end/81005
22150 * ubsan.c (instrument_null): Avoid pointless code temporary.
22151 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22153 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22157 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22160 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22163 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22165 * predict.c (unlikely_executed_edge_p): Move ahead.
22166 (probably_never_executed_edge_p): Use it.
22168 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22170 PR middle-end/79988
22171 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22172 gimple_call_builtin_p call.
22174 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22176 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22177 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22178 rtl_check_failed_type2, rtl_check_failed_code1,
22179 rtl_check_failed_code2, rtl_check_failed_code_mode,
22180 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22181 rtvec_check_failed_bounds, rtl_check_failed_flag,
22182 _fatal_insn_not_found, _fatal_insn): Likewise.
22183 * tree.h (tree_contains_struct_check_failed,
22184 tree_check_failed, tree_not_check_failed,
22185 tree_class_check_failed, tree_range_check_failed,
22186 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22187 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22188 tree_operand_check_failed, omp_clause_check_failed,
22189 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22192 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22194 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22195 flag_branch_probabilities.
22196 * ipa-inline.c (edge_badness): Likewise.
22197 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22198 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22199 * predict.c (maybe_hot_frequency_p): Likewise.
22200 (probably_never_executed): Likewise.
22201 * sched-ebb.c (schedule_ebbs): Likewise.
22202 * sched-rgn.c (find_single_block_region): Likewise.
22203 * tracer.c (tail_duplicate): Likewise.
22205 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22207 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22208 longer requires x_flag_profile_use.
22210 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22212 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22213 instead of flag_reorder_blocks_and_partition.
22214 * dbxout.c (dbxout_function_end): Likewise.
22215 * dwarf2out.c (gen_subprogram_die): Likewise.
22216 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22217 * hw-doloop.c (reorg_loops): Likewise.
22218 * varasm.c (assemble_start_function,
22219 assemble_end_function): Likewise.
22220 (decide_function_section): Do not check for
22221 flag_reorder_blocks_and_partition.
22223 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22225 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22227 (chkp_get_hard_register_fake_addr_expr): Ditto.
22228 (chkp_build_addr_expr): Add check for hard reg case.
22229 (chkp_parse_array_and_component_ref): Ditto.
22230 (chkp_find_bounds_1): Ditto.
22231 (chkp_process_stmt): Don't generate bounds store for
22234 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22236 * predict.c (maybe_hot_bb_p): Do not check profile status.
22237 (maybe_hot_edge_p): Likewise.
22238 (probably_never_executed): Check for zero counts even if profile
22240 (unlikely_executed_edge_p): New function.
22241 (unlikely_executed_stmt_p): New function.
22242 (unlikely_executed_bb_p): New function.
22243 (set_even_probabilities): Use unlikely predicates.
22244 (combine_predictions_for_bb): Likewise.
22245 (predict_paths_for_bb): Likewise.
22246 (predict_paths_leading_to_edge): Likewise.
22247 (determine_unlikely_bbs): New function.
22248 (estimate_bb_frequencies): Use it.
22249 (compute_function_frequency): Use zero counts even if profile is
22251 * profile-count.h: Fix typo.
22253 2017-08-08 Julia Koval <julia.koval@intel.com>
22255 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22256 _mm512_mask_cvtsepi16_storeu_epi8,
22257 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22258 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22259 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22260 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22261 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22262 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22263 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22264 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22265 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22266 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22267 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22268 __builtin_ia32_pmovuswb256mem_mask,
22269 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22270 __builtin_ia32_pmovwb512mem_mask): New builtins.
22272 2017-08-08 Julia Koval <julia.koval@intel.com>
22274 PR target/73350,80862
22275 * config/i386/subst.md (round): Fix round pattern.
22276 * config/i386/i386.c (ix86_erase_embedded_rounding):
22277 Fix erasing rounding for the fixed pattern.
22279 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22281 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22283 2017-06-08 Martin Liska <mliska@suse.cz>
22285 PR gcov-profile/80911
22286 * gcov.c (block_info::block_info): New constructor.
22288 2017-06-07 Carl Love <cel@us.ibm.com>
22290 * config/rs6000/rs6000-c: The return type of the following
22291 built-in functions was implemented as int not long long. Fix sign
22292 of return value for the unsigned version of vec_mulo and vec_mule.
22293 vector unsigned long long vec_bperm (vector unsigned long long,
22294 vector unsigned char)
22295 vector signed long long vec_mule (vector signed int,
22297 vector unsigned long long vec_mule (vector unsigned int,
22298 vector unsigned int)
22299 vector signed long long vec_mulo (vector signed int,
22301 vector unsigned long long vec_mulo (vector unsigned int,
22302 vector unsigned int)
22303 * doc/extend.texi: Fix the documentation for the built-in
22306 2017-06-07 Carl Love <cel@us.ibm.com>
22309 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22312 2017-06-07 Carl Love <cel@us.ibm.com>
22314 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22315 support, Generate doublehv for signed int/float for BE case only.
22317 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22319 * doc/invoke.texi (mcx16): Rewrite.
22321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22323 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22324 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22325 *mov<mode>_softfloat, and an anonymous splitter): Use
22326 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22330 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22332 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22333 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22334 (rs6000_debug_reg_global): Adjust.
22335 (rs6000_init_hard_regno_mode_ok): Adjust.
22336 (rs6000_dbx_register_number): Adjust.
22337 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22338 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22339 Remove SPE_ACC and SPEFSCR.
22340 (REG_ALLOC_ORDER): Ditto.
22341 (FRAME_POINTER_REGNUM): Change to 111.
22342 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22343 (REG_CLASS_NAMES): Ditto.
22344 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22345 (REGISTER_NAMES): Ditto.
22346 (ADDITIONAL_REG_NAMES): Ditto.
22347 (rs6000_reg_names): Ditto.
22348 * config/rs6000/rs6000.md: Renumber some register number
22351 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22353 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22355 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22356 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22358 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22359 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22360 Delete the SPE high registers.
22361 (REG_ALLOC_ORDER): Ditto.
22362 (enum reg_class): Remove SPE_HIGH_REGS.
22363 (REG_CLASS_NAMES): Ditto.
22364 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22365 (REGISTER_NAMES): Ditto.
22366 (rs6000_reg_names): Ditto.
22367 * doc/tm.texi.in: Remove SPE as example.
22368 * doc/tm.texi: Regenerate.
22370 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22372 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22373 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22374 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22375 * config/rs6000/rs6000.md (type): Remove "brinc".
22377 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22379 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22380 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22381 * config/rs6000/linuxspe.h: Delete file.
22382 * config/rs6000/rs6000.md: Don't include spe.md.
22383 * config/rs6000/spe.h: Delete file.
22384 * config/rs6000/spe.md: Delete file.
22385 * config/rs6000/t-rs6000: Remove spe.md.
22387 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22389 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22390 (reg_or_none500mem_operand): Delete.
22391 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22392 instead of reg_or_none500mem_operand.
22394 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22396 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22397 handling of SPE flags.
22398 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22400 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22402 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22404 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22405 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22406 paired_divv2sf3): Similar.
22407 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22408 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22409 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22411 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22412 Rename the paired_* instruction patterns.
22413 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22415 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22416 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22417 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22418 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22419 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22420 PAIRED_VECTOR_MODE.
22421 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22422 (spe_func_has_64bit_regs_p): Delete.
22423 (spe_expand_predicate_builtin): Delete.
22424 (spe_expand_evsel_builtin): Delete.
22425 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22426 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22427 (invalid_e500_subreg): Delete.
22428 (rs6000_legitimize_address): Always force_reg op2 as well, for
22429 paired single memory accesses.
22430 (rs6000_member_type_forces_blk): Delete.
22431 (rs6000_spe_function_arg): Delete.
22432 (rs6000_expand_unop_builtin): Delete SPE handling.
22433 (rs6000_expand_binop_builtin): Ditto.
22434 (spe_expand_stv_builtin): Delete.
22435 (bdesc_2arg_spe): Delete.
22436 (spe_expand_builtin): Delete.
22437 (spe_expand_predicate_builtin): Delete.
22438 (spe_expand_evsel_builtin): Delete.
22439 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22440 (spe_init_builtins): Delete.
22441 (spe_func_has_64bit_regs_p): Delete.
22442 (savres_routine_name): Delete "info" parameter. Adjust callers.
22443 (rs6000_emit_stack_reset): Ditto.
22444 (rs6000_dwarf_register_span): Delete.
22445 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22446 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22447 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22449 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22451 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22452 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22453 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22454 mulv2sf3, divv2sf3): Delete expanders.
22456 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22458 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22460 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22462 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22463 * config/rs6000/rs6000.c: Ditto.
22465 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22467 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22468 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22469 comparison_operator.
22471 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22473 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22474 * config/rs6000/rs6000.opt: Ditto.
22475 * config/rs6000/t-rtems: Ditto.
22477 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22479 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22480 TARGET_E500_SINGLE by 0, simplify.
22481 * config/rs6000/rs6000.c: Ditto.
22482 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22483 (spe_build_register_parallel): Delete.
22484 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22485 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22486 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22487 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22488 (E500_CONVERT): Delete.
22489 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22491 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22493 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22494 * config/rs6000/dfp.md: Ditto.
22495 (negdd2, *negdd2_fpr): Merge.
22496 (absdd2, *absdd2_fpr): Merge.
22497 (negtd2, *negtd2_fpr): Merge.
22498 (abstd2, *abstd2_fpr): Merge.
22499 * config/rs6000/e500.h: Delete file.
22500 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22501 TARGET_FPRS by 1 and simplify.
22502 * config/rs6000/rs6000-c.c: Ditto.
22503 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22504 TARGET_DF_SPE by 0.
22505 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22507 * config/rs6000/rs6000.md: Ditto.
22508 (floatdidf2, *floatdidf2_fpr): Merge.
22509 (move_from_CR_gt_bit): Delete.
22510 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22511 (E500_CR_IOR_COMPARE): Delete.
22512 (All patterns that require !TARGET_FPRS): Delete.
22513 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22515 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22517 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22519 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22521 * graphds.c (add_edge): Intitialize edge's attached data.
22522 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22523 pointer parameter. Call pointed function on each edge during
22524 graph traversing. Skip traversing the edge when the function
22526 (graphds_dfs, graphds_scc): Ditto.
22527 (for_each_edge): New parameter. Pass the new parameter to callback
22529 * graphds.h (skip_edge_callback): New function pointer type.
22530 (graphds_dfs, graphds_scc): New function pointer parameter.
22531 (graphds_edge_callback, for_each_edge): New parameter.
22533 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22535 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22536 out code checking if runtime alias check is possible to below ...
22537 Call the new function.
22538 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22539 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22541 2017-06-07 Marek Polacek <polacek@redhat.com>
22544 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22545 TYPE_OVERFLOW_WRAPS check.
22547 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22549 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22550 if versioning is required.
22551 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22552 peeling with the check for versioning.
22554 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22556 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22557 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22558 Set true to new parameter if new ssa variable is defined.
22559 (vect_gen_vector_loop_niters): Refactor. Set range information
22560 for the new vector loop bound variable.
22561 (vect_do_peeling): Ditto.
22563 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22565 * tree-affine.c (ssa.h): Include header file.
22566 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22567 has wrapping overflow behavior.
22569 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22571 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22573 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22575 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22576 (tree_to_aff_combination): ... here.
22578 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22580 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22581 reg_pressure model function.
22582 (ivopts_global_cost_for_size): Delete.
22583 (determine_set_costs, iv_ca_recount_cost): Call new model function
22584 ivopts_estimate_reg_pressure.
22586 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22588 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22589 expensive than udiv. Remove floating point cases from mod.
22591 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22593 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22594 Increase idiv cost.
22596 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22598 * config/aarch64/aarch64.md
22599 (copysignsf3): Fix mask generation.
22601 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22603 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22605 (class dump_manager): Add register_dumps method.
22606 * dumpfile.c: Include langhooks.h.
22607 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22608 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22609 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22610 (dump_manager::dump_register): Start with 512 entries instead of 32.
22611 (dump_manager::register_dumps): New method.
22612 * toplev.c (general_init): Instead of invoking register_dumps
22613 langhook, invoke register_dumps method on the dump manager.
22614 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22617 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22619 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22620 Say that # requires an associated define_split to exist, and that
22621 the define_split must be suitable for use after register allocation.
22623 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22625 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22626 (compute_outgoing_frequencies): Also initialize zero counts.
22627 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22628 around loops; preserve more of profile when nothing changes.
22630 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22632 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22634 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22635 * config/arm/arm-cpu-cdata.h: Regenerate.
22636 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22637 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22638 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22639 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22640 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22642 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22644 * config/arm/t-rmprofile: Likewise.
22645 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22647 2017-06-06 David S. Miller <davem@davemloft.net>
22650 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22651 blockage if function uses alloca.
22653 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22655 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22656 New "uid" fields to hold pretty-print IDs of group and ref.
22657 Memory references are now identified as <group_id>:<ref_id>
22658 instead of using [random] addresses.
22659 (dump_mem_details): Simplify, no functional change.
22660 (dump_mem_ref): Simplify and make output more concise.
22661 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22662 (find_or_create_group): Initialize group uid.
22663 (record_ref): Initialize ref uid. Improve debug output.
22664 (prune_group_by_reuse, should_issue_prefetch_p,)
22665 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22666 (mark_nontemporal_store, determine_loop_nest_reuse):
22667 Improve debug output.
22669 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22671 * dbgcnt.def (prefetch): New debug counter.
22672 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22673 (schedule_prefetches): Stop issueing prefetches if debug counter
22676 2017-06-06 Tom de Vries <tom@codesourcery.com>
22678 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22679 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22681 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22683 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22684 Use aarch64_reg_or_zero predicate for operand 4.
22685 (aarch64_compare_and_swap<mode> define_insn_and_split):
22686 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22687 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22689 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22691 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22692 (arm_compute_save_core_reg_mask): This.
22693 (thumb1_compute_save_reg_mask): Rename into ...
22694 (thumb1_compute_save_core_reg_mask): This.
22695 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22696 (arm_compute_frame_layout): Likewise.
22698 2017-06-06 Richard Biener <rguenther@suse.de>
22700 PR tree-optimization/80974
22701 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22702 keep or clear leaders SSA info.
22704 2017-06-06 Tom de Vries <tom@codesourcery.com>
22706 * config/nvptx/nvptx.c (split_mode_p): New function.
22707 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22709 2017-06-06 Tom de Vries <tom@codesourcery.com>
22711 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22713 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22716 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22719 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22721 * shrink-wrap.c (handle_simple_exit): Update profile.
22722 (try_shrink_wrapping): Upate profile.
22724 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22726 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22727 (tree_guess_outgoing_edge_probabilities): New.
22728 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22729 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22731 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22733 * ipa-split.c (split_function): Initialize return bb profile.
22735 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22737 * profile.c (compute_branch_probabilities): Also initialize
22738 EXIT_BLOCK profile.
22740 2017-06-06 Richard Biener <rguenther@suse.de>
22742 PR tree-optimization/80928
22743 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22744 (vect_analyze_loop_operations): Properly guard analysis for
22746 (vect_transform_loop): Likewise.
22747 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22748 (vect_model_induction_cost): Do not cost for pure SLP.
22749 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22750 of induction in inner loop vectorization.
22751 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22752 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22753 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22755 (vect_analyze_slp_cost_1): Cost induction.
22756 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22757 (vect_get_slp_vect_defs): Likewise.
22758 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22759 (vect_transform_stmt): Handle SLP reductions.
22760 * tree-vectorizer.h (vectorizable_induction): Adjust.
22762 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22764 * config/rs6000/rs6000.c (make_resolver_func): Update
22765 init_lowered_empty_function call.
22767 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22769 * doc/invoke.texi: Document the -fprofile-abs-path option.
22770 * common.opt (fprofile-abs-path): New option.
22771 * gcov-io.h (gcov_write_filename): Declare.
22772 * gcov-io.c (gcov_write_filename): New function.
22773 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22774 * profile.c (output_location): Likewise.
22776 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22778 * shring-wrap.c: Revert accidental commit.
22780 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22782 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22784 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22786 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22788 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22789 profile in callgraph edge.
22790 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22791 (apply_scale): Likewise.
22792 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22793 Also scale profile when inlining function with zero profile.
22794 (initialize_cfun): Update exit block profile even when it is zero.
22795 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22796 when profile is read.
22798 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22800 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22801 (CLONE_*): New constants to define the processors we can generate
22802 code for with the target_clone attribute.
22803 (rs6000_clone_map): New array to identify which clone processors
22804 the current program is running on.
22805 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22806 target_clone attribute.
22807 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22808 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22809 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22810 (cpu_expand_builtin): Add support for target_clone attribute.
22811 (rs6000_valid_attribute_p): Allow "default" attribute.
22812 (get_decl_name): New debug function to simplify printing the
22813 current function name in debugging statements.
22814 (rs6000_clone_priority): New functions to support the target_clone
22815 attribute, and be able to generate code to switch between ISA 2.05
22816 through ISA 3.0 (power6 through power9).
22817 (rs6000_compare_version_priority): Likewise.
22818 (rs6000_get_function_versions_dispatcher): Likewise.
22819 (make_resolver_func): Likewise.
22820 (add_condition_to_bb): Likewise.
22821 (dispatch_function_versions): Likewise.
22822 (rs6000_generate_version_dispatcher_body): Likewise.
22823 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22824 (fusion_gpr_load_p): Fix a spacing issue.
22825 * doc/extend.texi (Common Function Attributes): Document that the
22826 PowerPC supports the target_clone attribute.
22828 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22830 * config/arm/arm.h: explain F symbol found in description of ARM
22831 register allocation in its legend.
22833 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22835 * config/mips/frame-header-opt.c: Include profile-count.h.
22836 * config/riscv/riscv.c: Include profile-count.h
22838 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22840 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22842 (sm_set_flag_if_changed): Add bbs field.
22843 (execute_sm_if_changed_flag_set): Pass BBS.
22844 (execute_sm): Update.
22846 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22848 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22851 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22853 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22854 (peephole2): New peephole2 to emit the above.
22855 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22857 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22859 * config/aarch64/aarch64.c (define_peephole2 above
22860 *sub_<shift>_<mode>): New peephole.
22862 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22864 * config/i386/i386.c (make_resolver_func): Update.
22865 * Makefile.in: Add profile-count.h and profile-count.o
22866 * auto-profile.c (afdo_indirect_call): Update to new API.
22867 (afdo_set_bb_count): Update.
22868 (afdo_propagate_edge): Update.
22869 (afdo_propagate_circuit): Update.
22870 (afdo_calculate_branch_prob): Update.
22871 (afdo_annotate_cfg): Update.
22872 * basic-block.h: Include profile-count.h
22873 (struct edge_def): Turn count to profile_count.
22874 (struct basic_block_def): Likewie.
22875 (REG_BR_PROB_BASE): Move to profile-count.h
22876 (RDIV): Move to profile-count.h
22877 * bb-reorder.c (max_entry_count): Turn to profile_count.
22878 (find_traces): Update.
22879 (rotate_loop):Update.
22880 (connect_traces):Update.
22881 (sanitize_hot_paths):Update.
22882 * bt-load.c (migrate_btr_defs): Update.
22883 * cfg.c (RDIV): Remove.
22884 (init_flow): Use alloc_block.
22885 (alloc_block): Uninitialize count.
22886 (unchecked_make_edge): Uninitialize count.
22887 (check_bb_profile): Update.
22888 (dump_edge_info): Update.
22889 (dump_bb_info): Update.
22890 (update_bb_profile_for_threading): Update.
22891 (scale_bbs_frequencies_int): Update.
22892 (scale_bbs_frequencies_gcov_type): Update.
22893 (scale_bbs_frequencies_profile_count): New.
22894 * cfg.h (update_bb_profile_for_threading): Update.
22895 (scale_bbs_frequencies_profile_count): Declare.
22896 * cfgbuild.c (compute_outgoing_frequencies): Update.
22897 (find_many_sub_basic_blocks): Update.
22898 * cfgcleanup.c (try_forward_edges): Update.
22899 (try_crossjump_to_edge): Update.
22900 * cfgexpand.c (expand_gimple_tailcall): Update.
22901 (construct_exit_block): Update.
22902 * cfghooks.c (verify_flow_info): Update.
22903 (dump_bb_for_graph): Update.
22904 (split_edge): Update.
22905 (make_forwarder_block): Update.
22906 (duplicate_block): Update.
22907 (account_profile_record): Update.
22908 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22909 (get_estimated_loop_iterations): Update.
22910 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22911 (single_likely_exit): Update.
22912 * cfgloopmanip.c (scale_loop_profile): Update.
22914 (set_zero_probability): Update.
22915 (lv_adjust_loop_entry_edge): Update.
22916 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22917 (purge_dead_edges): Update.
22918 (rtl_account_profile_record): Update.
22919 * cgraph.c (cgraph_node::create): Uninitialize count.
22920 (symbol_table::create_edge): Uninitialize count.
22921 (cgraph_update_edges_for_call_stmt_node): Update.
22922 (cgraph_edge::dump_edge_flags): Update.
22923 (cgraph_node::dump): Update.
22924 (cgraph_edge::maybe_hot_p): Update.
22925 * cgraph.h: Include profile-count.h
22926 (create_clone), create_edge, create_indirect_edge): Update.
22927 (cgraph_node): Turn count to profile_count.
22928 (cgraph_edge0: Likewise.
22929 (make_speculative, clone): Update.
22930 (create_edge): Update.
22931 (init_lowered_empty_function): Update.
22932 * cgraphclones.c (cgraph_edge::clone): Update.
22933 (duplicate_thunk_for_node): Update.
22934 (cgraph_node::create_clone): Update.
22935 * cgraphunit.c (cgraph_node::analyze): Update.
22936 (cgraph_node::expand_thunk): Update.
22937 * final.c (dump_basic_block_info): Update.
22938 * gimple-streamer-in.c (input_bb): Update.
22939 * gimple-streamer-out.c (output_bb): Update.
22940 * graphite.c (print_global_statistics): Update.
22941 (print_graphite_scop_statistics): Update.
22942 * hsa-brig.c: Include basic-block.h.
22943 * hsa-dump.c: Include basic-block.h.
22944 * hsa-gen.c (T sum_slice): Update.
22945 (convert_switch_statements):Update.
22946 * hsa-regalloc.c: Include basic-block.h.
22947 * ipa-chkp.c (chkp_produce_thunks): Update.
22948 * ipa-cp.c (struct caller_statistics): Update.
22949 (init_caller_stats): Update.
22950 (gather_caller_stats): Update.
22951 (ipcp_cloning_candidate_p): Update.
22952 (good_cloning_opportunity_p): Update.
22953 (get_info_about_necessary_edges): Update.
22954 (dump_profile_updates): Update.
22955 (update_profiling_info): Update.
22956 (update_specialized_profile): Update.
22957 (perhaps_add_new_callers): Update.
22958 (decide_about_value): Update.
22959 (ipa_cp_c_finalize): Update.
22960 * ipa-devirt.c (struct odr_type_warn_count): Update.
22961 (struct decl_warn_count): Update.
22962 (struct final_warning_record): Update.
22963 (possible_polymorphic_call_targets): Update.
22964 (ipa_devirt): Update.
22965 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22966 * ipa-icf.c (sem_function::merge): Update.
22967 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22968 * ipa-inline.c (compute_uninlined_call_time): Update.
22969 (compute_inlined_call_time): Update.
22970 (want_inline_small_function_p): Update.
22971 (want_inline_self_recursive_call_p): Update.
22972 (edge_badness): Update.
22973 (lookup_recursive_calls): Update.
22974 (recursive_inlining): Update.
22975 (inline_small_functions): Update.
22976 (dump_overall_stats): Update.
22977 (dump_inline_stats): Update.
22978 * ipa-profile.c (ipa_profile_generate_summary): Update.
22979 (ipa_propagate_frequency): Update.
22980 (ipa_profile): Update.
22981 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22982 * ipa-utils.c (ipa_merge_profiles): Update.
22983 * loop-doloop.c (doloop_modify): Update.
22984 * loop-unroll.c (report_unroll): Update.
22985 (unroll_loop_runtime_iterations): Update.
22986 * lto-cgraph.c (lto_output_edge): Update.
22987 (lto_output_node): Update.
22988 (input_node): Update.
22989 (input_edge): Update.
22990 (merge_profile_summaries): Update.
22991 * lto-streamer-in.c (input_cfg): Update.
22992 * lto-streamer-out.c (output_cfg): Update.
22993 * mcf.c (create_fixup_graph): Update.
22994 (adjust_cfg_counts): Update.
22995 (sum_edge_counts): Update.
22996 * modulo-sched.c (sms_schedule): Update.
22997 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22998 * predict.c (maybe_hot_count_p): Update.
22999 (probably_never_executed): Update.
23000 (dump_prediction): Update.
23001 (combine_predictions_for_bb): Update.
23002 (propagate_freq): Update.
23003 (handle_missing_profiles): Update.
23004 (counts_to_freqs): Update.
23005 (rebuild_frequencies): Update.
23006 (force_edge_cold): Update.
23007 * predict.h: Include profile-count.h
23008 (maybe_hot_count_p, counts_to_freqs): UPdate.
23009 * print-rtl-function.c: Do not include cfg.h
23010 * print-rtl.c: Include basic-block.h
23011 * profile-count.c: New file.
23012 * profile-count.h: New file.
23013 * profile.c (is_edge_inconsistent): Update.
23014 (correct_negative_edge_counts): Update.
23015 (is_inconsistent): Update.
23016 (set_bb_counts): Update.
23017 (read_profile_edge_counts): Update.
23018 (compute_frequency_overlap): Update.
23019 (compute_branch_probabilities): Update; Initialize and deinitialize
23021 (branch_prob): Update.
23022 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23023 (edge_gcov_count): New.
23024 (bb_gcov_count): New.
23025 * shrink-wrap.c (try_shrink_wrapping): Update.
23026 * tracer.c (better_p): Update.
23027 * trans-mem.c (expand_transaction): Update.
23028 (ipa_tm_insert_irr_call): Update.
23029 (ipa_tm_insert_gettmclone_call): Update.
23030 * tree-call-cdce.c: Update.
23031 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23032 (gimple_duplicate_sese_tail): Update.
23033 (gimple_account_profile_record): Update.
23034 (execute_fixup_cfg): Update.
23035 * tree-inline.c (copy_bb): Update.
23036 (copy_edges_for_bb): Update.
23037 (initialize_cfun): Update.
23038 (freqs_to_counts): Update.
23039 (copy_cfg_body): Update.
23040 (expand_call_inline): Update.
23041 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23042 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23043 (try_unroll_loop_completely): Update.
23044 (try_peel_loop): Update.
23045 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23046 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23047 * tree-ssa-loop-split.c (connect_loops): Update.
23048 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23049 * tree-ssa-reassoc.c (branch_fixup): Update.
23050 * tree-ssa-tail-merge.c (replace_block_by): Update.
23051 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23052 (compute_path_counts): Update.
23053 (update_profile): Update.
23054 (recompute_probabilities): Update.
23055 (update_joiner_offpath_counts): Update.
23056 (estimated_freqs_path): Update.
23057 (freqs_to_counts_path): Update.
23058 (clear_counts_path): Update.
23059 (ssa_fix_duplicate_block_edges): Update.
23060 (duplicate_thread_path): Update.
23061 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23062 (struct switch_conv_info): Update.
23063 * tree-tailcall.c (decrease_profile): Update.
23064 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23065 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23066 * value-prof.c (check_counter): Update.
23067 (gimple_divmod_fixed_value): Update.
23068 (gimple_mod_pow2): Update.
23069 (gimple_mod_subtract): Update.
23070 (gimple_ic_transform): Update.
23071 (gimple_stringop_fixed_value): Update.
23072 * value-prof.h (gimple_ic): Update.
23074 2017-06-02 Carl Love <cel@us.ibm.com>
23076 * config/rs6000/rs6000-c: Add support for built-in functions
23077 vector double vec_doublee (vector signed int);
23078 vector double vec_doublee (vector unsigned int);
23079 vector double vec_doublee (vector float);
23080 vector double vec_doubleh (vector signed int);
23081 vector double vec_doubleh (vector unsigned int);
23082 vector double vec_doubleh (vector float);
23083 vector double vec_doublel (vector signed int);
23084 vector double vec_doublel (vector unsigned int);
23085 vector double vec_doublel (vector float);
23086 vector double vec_doubleo (vector signed int);
23087 vector double vec_doubleo (vector unsigned int);
23088 vector double vec_doubleo (vector float);.
23089 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23090 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23092 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23093 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23094 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23096 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23097 vec_doublel, vec_doubleh.
23098 * doc/extend.texi: Update the built-in documentation file for the
23099 new built-in functions.
23101 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23104 * ipa-inline-analysis.c (free_growth_caches): Set
23105 edge_removal_hook_holder to NULL after removing it.
23107 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23109 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23110 comparision with zero.
23112 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23113 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23114 for early expansion of vec_min and vec_max builtins.
23115 (builtin_function_type): Add min/max unsigned variants to those
23116 identified as having unsigned arguments.
23118 2017-06-02 Olivier Hainque <hainque@adacore.com>
23120 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23122 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23124 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23125 Use VALL_F16 iterator rather than VALL.
23127 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23129 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23130 Emit CBNZ inside loop when doing a strong exchange and comparing
23131 against zero. Generate the CC flags after the loop.
23133 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23135 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23136 (dl_section_ref): New.
23137 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23138 On AIX, append an expression to subtract the size of the
23139 section length to dl_section_ref.
23141 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23143 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23144 for early expansion of vector absolute builtins.
23146 2017-06-02 Richard Biener <rguenther@suse.de>
23148 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23149 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23151 2017-06-02 Richard Biener <rguenther@suse.de>
23153 PR tree-optimization/80948
23154 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23155 stmt order as well.
23157 2017-06-02 Richard Biener <rguenther@suse.de>
23159 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23161 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23162 for inductions as relevant.
23164 2017-06-02 Richard Biener <rguenther@suse.de>
23166 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23167 (vectorizable_induction): ... this. Remove dead code.
23169 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23171 * builtins. (expand_builtin_alloca): Remove second parameter and
23172 infer its value from the first parameter instead.
23173 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23175 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23177 PR rtl-optimization/80903
23178 * loop-doloop.c (add_test): Unshare sequence.
23180 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23182 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23184 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23186 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23188 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23189 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23190 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23191 (xlogue_layout::get_stub_rtx): Make static.
23192 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23193 (xlogue_layout::compute_stub_managed_regs): Rename to...
23194 (xlogue_layout::count_stub_managed_regs): ...this.
23195 (xlogue_layout::is_stub_managed_reg): New function.
23196 (xlogue_layout::m_stub_names): Rename to...
23197 (xlogue_layout::s_stub_names): ...this, make static.
23198 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23199 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23200 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23201 xlogue_layout::s_stub_names): Instantiate statics.
23202 (stub_managed_regs): Remove.
23203 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23204 (disable_call_ms2sysv_xlogues): Rename to...
23205 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23206 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23208 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23209 change after reload_completed.
23210 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23212 (ix86_expand_prologue): Likewise.
23213 (ix86_expand_epilogue): Likewise.
23214 (ix86_expand_split_stack_prologue): Likewise.
23215 (ix86_compute_frame_layout): Remove frame parameter ...
23216 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23217 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23219 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23220 (ix86_frame): Move from here ...
23221 * config/i386/i386.h (ix86_frame): ... to here.
23222 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23223 complete ix86_frame data structure instead. Remove some_ld_name.
23225 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23227 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23228 symbols that hold a DECL_VALUE_EXPR.
23230 2017-06-01 Martin Jambor <mjambor@suse.cz>
23232 PR tree-optimization/80898
23233 * tree-sra.c (process_subtree_disqualification): Removed.
23234 (disqualify_candidate): Do not acll
23235 process_subtree_disqualification.
23236 (subtree_mark_written_and_enqueue): New function.
23237 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23238 RHS has been disqualified and re-queue LHS if necessary. Apart
23239 from that, ignore disqualified RHS.
23241 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23243 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23244 address fetch for z10 or later.
23246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23248 * config/arc/arc.md (tst_movb): Add guard when splitting.
23250 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23252 * config/arc/arc.c (arc_can_eliminate): Test against
23253 arc_frame_pointer_needed.
23255 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23257 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23258 to prevent store reordering.
23259 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23260 (type): Add block type.
23261 (stack_tie): Define special instruction to be used in
23264 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23266 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23267 constraint. It is not valid for the pattern.
23268 (noncommutative_binary_comparison): Likewise.
23270 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23272 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23275 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23277 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23278 be used by the reg-alloc.
23280 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23282 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23283 reg-alloc when having mul64 or mul32x16 instructions.
23284 (mulsidi3): Likewise.
23285 (umulsidi3): Likewise.
23286 (mulsi32x16): New pattern.
23287 (mulsi64): Likewise.
23288 (mulsidi64): Likewise.
23289 (umulsidi64): Likewise.
23290 (MUL32x16_REG): Define.
23291 (mul64_600): Use MUL32x16_REG.
23292 (mac64_600): Likewise.
23293 (umul64_600): Likewise.
23294 (umac64_600): Likewise.
23296 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23298 * config/arc/arc.md (mulsi3_700): Make it commutative.
23300 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23302 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23304 (*sign_extendsidi2_insn): Likewise for movstosw.
23306 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23308 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23309 the type of the input discriminant value. Convert the
23310 discriminant value of signedness vary.
23312 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23314 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23315 Add to -Wall section.
23317 2017-06-01 Richard Biener <rguenther@suse.de>
23319 PR middle-end/66313
23320 * fold-const.c (fold_plusminus_mult_expr): If the factored
23321 factor may be zero use a wrapping type for the inner operation.
23322 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23323 and handle moved defs.
23324 (process_assignment): Properly guard the unary op case. Return a
23325 tri-state indicating that moving the stmt before the call may allow
23326 to continue. Pass through to_move.
23327 (find_tail_calls): Handle moving unrelated defs before
23330 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23333 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23334 splitter result in the canonical way.
23336 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23338 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23339 also for 32bit target. Update insn attributes.
23340 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23342 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23344 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23345 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23346 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23347 (_mm_maskz_min_ss): New intrinsics.
23349 2017-05-31 Martin Liska <mliska@suse.cz>
23351 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23352 Change comment style to one we normally use.
23353 (vectorizable_reduction): Likewise.
23354 (vectorizable_induction): Likewise.
23355 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23356 (vectorizable_call): Likewise.
23357 (vectorizable_simd_clone_call): Likewise.
23358 (vectorizable_conversion): Likewise.
23359 (vectorizable_assignment): Likewise.
23360 (vectorizable_shift): Likewise.
23361 (vectorizable_operation): Likewise.
23362 (vectorizable_store): Likewise.
23363 (vectorizable_load): Likewise.
23364 * tree-vectorizer.h: Likewise.
23366 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23368 * passes.c (emergency_dump_function): New.
23369 * tree-pass.h (emergency_dump_function): Declare.
23370 * plugin.c (plugins_internal_error_function): Remove.
23371 * plugin.h (plugins_internal_error_function): Remove declaration.
23372 * toplev.c (internal_error_function): New static function. Use it...
23373 (general_init): ...here.
23375 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23377 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23378 (arc_rtx_costs): Add costs for new patterns.
23379 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23380 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23382 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23384 * tree-ssa-strlen.c (get_next_strinfo): New function.
23385 (get_stridx_plus_constant): Use it.
23386 (zero_length_string): Likewise.
23387 (adjust_related_strinfos): Likewise.
23388 (adjust_last_stmt): Likewise.
23390 2017-05-31 Richard Biener <rguenther@suse.de>
23393 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23394 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23396 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23398 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23399 loop_vinfo argument and use of dependence distance vectors.
23400 Check instead whether the two references differ only in their
23401 initial value and assume that they have the same alignment if the
23402 difference is a multiple of the vector alignment.
23403 (vect_analyze_data_refs_alignment): Update call accordingly.
23405 2017-05-31 Martin Liska <mliska@suse.cz>
23408 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23410 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23412 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23413 (create_intersect_range_checks): Move from ...
23414 * tree-data-ref.c (create_intersect_range_checks_index)
23415 (create_intersect_range_checks): ... to here.
23416 (create_runtime_alias_checks): New function factored from ...
23417 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23418 here. Call above function.
23419 * tree-data-ref.h (create_runtime_alias_checks): New function.
23421 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23423 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23424 segment length for dr_b and compute it in wide_int.
23426 2017-05-31 Richard Biener <rguenther@suse.de>
23428 PR tree-optimization/80906
23429 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23430 and pass through iv_map.
23431 (copy_bb_and_scalar_dependences): Adjust.
23432 (translate_pending_phi_nodes): Likewise.
23433 (copy_loop_close_phi_args): Handle code-generating IVs instead
23436 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23438 * diagnostic-color.c (color_dict): Add "type-diff".
23439 (parse_gcc_colors): Update comment.
23440 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23441 -fdiagnostics-show-template-tree and -fno-elide-type.
23442 (GCC_COLORS): Add type-diff to example.
23444 (-fdiagnostics-show-template-tree): New.
23445 (-fno-elide-type): New.
23446 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23447 the pp_format_decoder callback. Call any m_format_postprocessor's
23449 (pretty_printer::pretty_printer): Initialize
23450 m_format_postprocessor.
23451 (pretty_printer::~pretty_printer): Delete any
23452 m_format_postprocessor.
23453 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23454 (class format_postprocessor): New class.
23455 (struct pretty_printer::format_decoder): Document the new parameters.
23456 (struct pretty_printer::m_format_postprocessor): New field.
23457 * tree-diagnostic.c (default_tree_printer): Update for new
23458 bool and const char ** params.
23459 * tree-diagnostic.h (default_tree_printer): Likewise.
23461 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23463 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23464 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23465 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23466 rs6000_gen_cell_microcode code.
23467 (rs6000_final_prescan_insn): Delete.
23468 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23469 "warn-cell-microcode" entries.
23470 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23471 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23472 throughout. Change cc_reg_not_micro_cr0_operand to
23473 cc_reg_not_cr0_operand throughout.
23474 (*extendhi<mode>2_noload): Delete.
23475 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23476 (mwarn-cell-microcode): Delete.
23477 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23478 -mgen-cell-microcode and -mwarn-cell-microcode.
23480 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23483 * config/i386/constraints.md (Yd): New constraint.
23485 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23486 and (?Yd, r) alternatives. Update insn attributes.
23487 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23488 and (?*Yd, r) alternatives. Update insn attributes.
23489 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23491 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23493 * gimplify.c (gimplify_modify_expr): Don't create a
23494 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23497 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23499 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23501 2017-05-30 Richard Biener <rguenther@suse.de>
23503 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23504 and reduc_def fields.
23505 (STMT_VINFO_REDUC_TYPE): New define.
23506 (STMT_VINFO_REDUC_DEF): Likewise.
23507 (vect_force_simple_reduction): Adjust prototype.
23508 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23509 (vect_is_simple_reduction): Remove check_reduction argument.
23510 (vect_force_simple_reduction): Adjust and set
23511 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23512 (vectorizable_reduction): Do not re-do reduction analysis
23513 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23514 * tree-parloops.c (gather_scalar_reductions): Adjust.
23516 2017-05-30 Richard Biener <rguenther@suse.de>
23518 PR middle-end/80901
23519 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23522 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23524 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23525 Introduce unknown_misalignment parameter and remove vf.
23526 (vect_peeling_hash_get_lowest_cost):
23527 Pass unknown_misalignment parameter.
23528 (vect_enhance_data_refs_alignment):
23529 Fix unsupportable data ref treatment.
23531 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23533 * tree-vect-data-refs.c (vect_get_data_access_cost):
23534 Workaround for SLP handling.
23535 (vect_enhance_data_refs_alignment):
23536 Compute costs for doing no peeling at all, compare to the best
23537 peeling costs so far and avoid peeling if cheaper.
23539 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23541 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23542 Return peeling info and set costs to zero for unlimited cost
23544 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23545 with unknown misalignment. Compute and costs for unknown
23546 misalignment, compare them to the costs for known misalignment
23547 and choose the cheapest for peeling.
23549 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23551 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23552 (vect_get_peeling_costs_all_drs): Create function.
23553 (vect_peeling_hash_get_lowest_cost):
23554 Use vect_get_peeling_costs_all_drs.
23555 (vect_peeling_supportable): Create function.
23556 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23558 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23560 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23561 DR_HAS_NEGATIVE_STEP.
23562 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23563 (vect_enhance_data_refs_alignment): Use.
23564 (vect_duplicate_ssa_name_ptr_info): Use.
23565 * tree-vectorizer.h (dr_misalignment): Use.
23566 (known_alignment_for_access_p): Use.
23568 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23571 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23573 (has_section_name): New function.
23575 2017-05-30 Martin Liska <mliska@suse.cz>
23578 * auto-profile.c (get_function_decl_from_block): Fix
23581 2017-05-30 Richard Biener <rguenther@suse.de>
23583 PR middle-end/80876
23584 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23586 2017-05-30 Martin Liska <mliska@suse.cz>
23588 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23589 * dumpfile.h (struct dump_file_info): Remove ctors.
23591 2017-05-30 Martin Liska <mliska@suse.cz>
23593 * predict.def: Fix GNU coding style.
23595 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23597 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23598 Mark 'to' argument with ATTRIBUTE_UNUSED.
23600 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23602 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23603 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23604 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23607 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23609 * doc/install.texi (Options specification): Restore entry of
23610 --enable-sjlj-exceptions.
23612 2017-05-27 Michael Eager <eager@eagercon.com>
23615 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23617 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23619 * config/microblaze/microblaze.h
23620 (FIXED_REGISTERS): Update in macro.
23621 (CALL_USED_REGISTERS): Update in macro.
23623 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23625 * doc/install.texi: Add links to macOS binary distributions.
23627 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23631 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23633 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23635 2017-05-26 Martin Liska <mliska@suse.cz>
23637 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23639 2017-05-26 Martin Liska <mliska@suse.cz>
23641 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23642 always leading ';; '.
23643 (dump_bb_info): Likewise.
23644 (brief_dump_cfg): Likewise.
23645 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23646 * dumpfile.c: Remove usage of TDF_VERBOSE.
23647 * dumpfile.h (enum dump_kind): Likewise.
23648 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23649 * print-tree.c (debug_verbose): Remove.
23650 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23651 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23652 * tree-diagnostic.c (default_tree_printer): Replace
23653 TDF_DIAGNOSTIC with TDF_SLIM.
23655 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23657 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23658 in parameter loop, rather than loop_vinfo.
23659 (create_intersect_range_checks): Ditto.
23660 (vect_create_cond_for_alias_checks): Update call to above functions.
23662 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23664 PR tree-optimization/80815
23665 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23666 for merging runtime alias checks. Handle negative DR_STEPs.
23668 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23670 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23672 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23673 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23674 out code pruning runtime alias checks.
23675 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23676 factored out from above.
23677 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23679 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23681 (prune_runtime_alias_test_list): New decalaration.
23683 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23685 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23686 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23687 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23688 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23689 (operator==, comp_dr_with_seg_len_pair): Ditto.
23690 (vect_prune_runtime_alias_test_list): Ditto.
23692 2017-05-26 Martin Liska <mliska@suse.cz>
23695 * params.def: Bound partial-inlining-entry-probability param.
23697 2017-05-26 Marek Polacek <polacek@redhat.com>
23700 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23703 2017-05-26 Richard Biener <rguenther@suse.de>
23705 PR tree-optimization/80842
23706 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23709 2017-05-26 Richard Biener <rguenther@suse.de>
23711 PR tree-optimization/80844
23712 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23714 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23716 * doc/md.texi (Machine Constraints): Update x86 family
23717 machine constraints section to match 'config/i386/constraints.md'.
23719 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23721 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23723 2017-05-25 Nathan Sidwell <nathan@acm.org>
23725 * doc/invoke.texi (--enable-languages): Update documentation.
23727 2017-05-25 Martin Liska <mliska@suse.cz>
23729 * dumpfile.c: Add TDF_FOLDING.
23730 * dumpfile.h (enum dump_kind): Likewise.
23731 * genmatch.c (dt_simplify::gen_1): Use it.
23733 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23735 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23737 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23739 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23740 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23742 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23744 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23745 * match.pd (X == C): Rewrite it here.
23746 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23747 with_certain_nonzero_bits2): New predicates.
23748 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23750 2017-05-24 Nathan Sidwell <nathan@acm.org>
23752 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23755 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23758 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23760 * config/powerpcspe: New port. Files are copied from the rs6000
23761 port, with "rs6000" in filenames replaced by "powerpcspe".
23763 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23765 PR rtl-optimization/80754
23766 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23768 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23770 * config/sparc/sparc.md (length): Return the correct value for -mflat
23771 sibcalls to match output_sibcall.
23773 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23777 * config/rs6000/rs6000.c (struct machine_function): Add new field
23779 (rs6000_get_separate_components): Init that field, use it.
23780 (rs6000_components_for_bb): Use the field.
23782 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23784 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23786 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23788 PR middle-end/80823
23789 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23791 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23794 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23795 address against address_operand predicate.
23796 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23798 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23800 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23801 parameters passed indirectly.
23803 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23805 * config/i386/i386.md (*movdi_internal): Remove SSE4
23806 alternative 18 (?r, *v). Update insn attributes.
23807 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23808 Update insn attributes.
23809 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23810 Update insn attributes.
23811 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23812 alternative 1 (r, v). Remove isa attribute.
23813 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23814 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23815 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23817 2017-05-23 Tom de Vries <tom@codesourcery.com>
23819 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23822 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23824 * cgraphunit.c (symbol_table::process_new_functions): Update.
23825 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23826 (inline_generate_summary): Rename to ...
23827 (ipa_fn_summary_generate): ... this one.
23828 (inline_read_summary): Rename to ...
23829 (ipa_fn_summary_read): ... this one.
23830 (inline_write_summary): Rename to ...
23831 (ipa_fn_summary_write): ... this one.
23832 (inline_free_summary): Rename to ...
23833 (ipa_free_fn_summary): ... this one.
23834 (pass_data_local_fn_summary, pass_local_fn_summary,
23835 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23836 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23837 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23838 make_pass_ipa_fn_summary): New.
23839 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23840 inline_write_summary, inline_free_summary): Remove.
23841 (ipa_free_fn_summary) : New.
23842 * ipa-inline.c (ipa_inline): Update.
23843 (pass_ipa_inline): Do not generate summaries.
23844 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23846 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23847 and add pass_ipa_fn_summary.
23848 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23850 (make_pass_inline_parameters): Remove.
23852 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23854 * omp-low.c (struct omp_context): Remove "default_kind" member.
23857 * omp-offload.c (execute_oacc_device_lower): Remove the
23858 parallelism dimensions function attributes for unparallelized
23859 OpenACC kernels constructs.
23861 2017-05-23 Martin Liska <mliska@suse.cz>
23863 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23865 (cgraph_edge::make_speculative): Likewise.
23866 (cgraph_edge::resolve_speculation): Likewise.
23867 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23868 (cgraph_node::dump): Likewise.
23869 * cgraph.h: Likewise.
23870 * cgraphunit.c (analyze_functions): Likewise.
23871 (symbol_table::compile): Likewise.
23872 * ipa-cp.c (print_all_lattices): Likewise.
23873 (determine_versionability): Likewise.
23874 (initialize_node_lattices): Likewise.
23875 (ipcp_verify_propagated_values): Likewise.
23876 (estimate_local_effects): Likewise.
23877 (update_profiling_info): Likewise.
23878 (create_specialized_node): Likewise.
23879 (perhaps_add_new_callers): Likewise.
23880 (decide_about_value): Likewise.
23881 (decide_whether_version_node): Likewise.
23882 (identify_dead_nodes): Likewise.
23883 (ipcp_store_bits_results): Likewise.
23884 * ipa-devirt.c (dump_targets): Likewise.
23885 (ipa_devirt): Likewise.
23886 * ipa-icf.c (sem_item::dump): Likewise.
23887 (sem_function::equals): Likewise.
23888 (sem_variable::equals): Likewise.
23889 (sem_item_optimizer::read_section): Likewise.
23890 (sem_item_optimizer::execute): Likewise.
23891 (congruence_class::dump): Likewise.
23892 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23893 (dump_inline_summary): Likewise.
23894 (estimate_node_size_and_time): Likewise.
23895 (inline_analyze_function): Likewise.
23896 * ipa-inline-transform.c (inline_call): Likewise.
23897 * ipa-inline.c (report_inline_failed_reason): Likewise.
23898 (want_early_inline_function_p): Likewise.
23899 (edge_badness): Likewise.
23900 (update_edge_key): Likewise.
23901 (inline_small_functions): Likewise.
23902 * ipa-profile.c (ipa_profile): Likewise.
23903 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23904 (ipa_make_edge_direct_to_target): Likewise.
23905 (remove_described_reference): Likewise.
23906 (ipa_impossible_devirt_target): Likewise.
23907 (propagate_controlled_uses): Likewise.
23908 (ipa_print_node_params): Likewise.
23909 (ipcp_transform_function): Likewise.
23910 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23911 (propagate_pure_const): Likewise.
23912 * ipa-reference.c (generate_summary): Likewise.
23913 (read_write_all_from_decl): Likewise.
23914 (propagate): Likewise.
23915 (ipa_reference_read_optimization_summary): Likewise.
23916 * ipa-utils.c (ipa_merge_profiles): Likewise.
23917 * ipa.c (walk_polymorphic_call_targets): Likewise.
23918 (symbol_table::remove_unreachable_nodes): Likewise.
23919 (ipa_single_use): Likewise.
23920 * passes.c (execute_todo): Likewise.
23921 * predict.c (drop_profile): Likewise.
23922 * symtab.c (symtab_node::get_dump_name): New function.
23923 (symtab_node::dump_name): Likewise.
23924 (symtab_node::dump_asm_name): Likewise.
23925 (symtab_node::dump_references): Likewise.
23926 (symtab_node::dump_referring): Likewise.
23927 (symtab_node::dump_base): Likewise.
23928 (symtab_node::debug_symtab): Likewise.
23929 * tree-sra.c (convert_callers_for_node): Likewise.
23930 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23931 * value-prof.c (init_node_map): Likewise.
23933 2017-05-23 Martin Liska <mliska@suse.cz>
23935 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23936 and symtab_node::debug_symtab to symbol_table::debug.
23937 * cgraphunit.c (analyze_functions): Use the renamed function.
23938 (symbol_table::compile): Likewise.
23939 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23940 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23941 * passes.c (execute_todo): Likewise.
23942 * symtab.c (symbol_table::dump): New function.
23943 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23945 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23947 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23948 that nonconst implies exec.
23950 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23952 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23953 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23954 (inline_edge_summary_vec): Turn into ...
23955 (ipa_call_summaries): ... this one.
23956 (redirect_to_unreachable, edge_set_predicate,
23957 evaluate_properties_for_edge, inline_summary_alloc,
23958 reset_ipa_call_summary, reset_inline_summary,
23959 inline_summary_t::duplicate): Update.
23960 (inline_edge_duplication_hook): Turn to ...
23961 (ipa_call_summary_t::duplicate): ... this one.
23962 (inline_edge_removal_hook): Turn to ...
23963 (ipa_call_summary_t::remove): ... this one.
23964 (dump_inline_edge_summary): Turn to ...
23965 (dump_ipa_call_summary): ... this one.
23966 (estimate_function_body_sizes): Update.
23967 (inline_update_callee_summaries): Update.
23968 (remap_edge_change_prob): Update.
23969 (remap_edge_summaries): Update.
23970 (inline_merge_summary): Update.
23971 (do_estimate_edge_time): Update.
23972 (inline_generate_summary): Update.
23973 (inline_read_section): Update.
23974 (inline_read_summary): Update.
23975 (inline_free_summary): Update.
23976 * ipa-inline.c (can_inline_edge_p): Update.
23977 (compute_inlined_call_time): Update.
23978 (want_inline_small_function_p): Update.
23979 (edge_badness): Update.
23980 (early_inliner): Update.
23981 * ipa-inline.h (inline_edge_summary): Turn to ...
23982 (ipa_call_summary): ... this one.
23983 (ipa_call_summary_t): New class.
23984 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23985 (ipa_call_summaries): New.
23986 (inline_edge_summary): Remove.
23987 (estimate_edge_growth): Update.
23988 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23989 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23990 * ipa-split.c (execute_split_functions): Update.
23991 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23993 2017-05-23 Tom de Vries <tom@codesourcery.com>
23995 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23996 attributes): Document rdrand effective target.
23998 2017-05-23 Tom de Vries <tom@codesourcery.com>
24000 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24001 attributes): Sort alphabetically.
24003 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24005 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24007 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24010 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24011 V2DF/V2DI splat into two separate patterns, one that handles
24012 registers, and the other that only handles memory. Drop support
24013 for splatting from a GPR on ISA 2.07 and then splitting the
24014 splat into direct move and splat.
24015 (vsx_splat_<mode>_reg): Likewise.
24016 (vsx_splat_<mode>_mem): Likewise.
24018 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24020 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24022 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24024 PR middle-end/80809
24025 * omp-low.c (finish_taskreg_remap): New function.
24026 (finish_taskreg_scan): If unit size of ctx->record_type
24027 is non-constant, unshare the size expression and replace
24028 decls in it with possible outer var refs.
24030 PR middle-end/80809
24031 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24032 GOVD_SHARED rather than GOVD_PRIVATE with it.
24033 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24034 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24036 PR middle-end/80853
24037 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24038 as last argument to build_outer_var_ref for pointer bases of array
24039 section reductions.
24041 2017-05-19 Martin Sebor <msebor@redhat.com>
24043 * print-tree.c (print_node): Print DECL_READ_P flag.
24045 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24047 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24048 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24049 * cgraph.c: Likewise.
24050 * cgraphunit.c: Likewise.
24051 * gengtype.c: Likewise.
24052 * ipa-cp.c: Likewise.
24053 * ipa-devirt.c: Likewise.
24054 * ipa-icf.c: Likewise.
24055 * ipa-predicate.c: Likewise.
24056 * ipa-profile.c: Likewise.
24057 * ipa-prop.c: Likewise.
24058 * ipa-split.c: Likewise.
24060 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24061 edge_predicate_pool, dump_inline_hints,
24062 inline_summary::account_size_time, redirect_to_unreachable,
24063 edge_set_predicate, set_hint_predicate,
24064 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24065 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24066 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24067 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24068 ipa_call_summary_t::remove, initialize_growth_caches,
24069 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24070 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24071 mark_modified, unmodified_parm_1, unmodified_parm,
24072 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24073 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24074 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24075 will_be_nonconstant_predicate, record_modified_bb_info,
24076 get_minimal_bb, record_modified, param_change_prob,
24077 phi_result_unknown_predicate, predicate_for_phi_result,
24078 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24079 estimate_function_body_sizes, compute_inline_parameters,
24080 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24081 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24082 inline_update_callee_summaries, remap_edge_change_prob,
24083 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24084 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24085 inline_analyze_function, inline_summary_t::insert,
24086 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24087 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24088 inline_free_summary): Move to ipa-fnsummary.h
24089 (predicate_t): Remove.
24090 * ipa-fnsummary.c: New file.
24091 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24092 (enum inline_hints_vals, inline_hints, agg_position_info,
24093 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24094 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24095 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24096 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24097 inline_read_summary, inline_write_summary, inline_free_summary,
24098 inline_analyze_function, initialize_inline_failed,
24099 inline_merge_summary, inline_update_overall_summary,
24100 compute_inline_parameters): Move to ipa-fnsummary.h
24101 * ipa-fnsummary.h: New file.
24102 * ipa-inline-transform.h: Include ipa-inline.h.
24103 * ipa-inline.c: LIkewise.
24105 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24107 * ipa-inline.c (edge_badness): Use inlined_time instead of
24108 inline_summaries->get.
24110 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24112 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24114 2017-05-22 Nathan Sidwell <nathan@acm.org>
24116 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24117 (fdump-lang): Document 'raw' option.
24118 * dumpfile.h (TDI_tu): Delete.
24119 * dumpfile.c (dump_files): Remove translation-unit.
24120 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24122 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24124 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24125 command option from $(AWK) call.
24126 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24127 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24128 [FORMAT]: Remove handling of variable.
24129 * config/avr/t-multilib: Regenerate.
24131 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24133 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24135 (dump_inline_summary): Do not print self_time.
24136 (estimate_function_body_sizes): Do not set self_time.
24137 (compute_inline_parameters): Likewise.
24138 (inline_read_section, inline_write_summary): Do not stream self_time.
24139 * ipa-inline.h (inline_summary): Drop self_time.
24141 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24143 * ipa-inline-analysis.c (account_size_time): Rename to ...
24144 (inline_summary::account_size_time): ... this one.
24145 (reset_ipa_call_summary): Turn to ...
24146 (ipa_call_summary::reset): ... this one.
24147 (reset_inline_summary): Turn to ...
24148 (inline_summary::reset): ... this one.
24149 (inline_summary_t::remove): Update.
24150 (inline_summary_t::duplicate): Update.
24151 (ipa_call_summary_t::remove): Update.
24152 (dump_inline_summary): Update.
24153 (estimate_function_body_sizes): Update.
24154 (compute_inline_parameters): Update.
24155 (estimate_node_size_and_time): Update.
24156 (inline_merge_summary): Update.
24157 (inline_update_overall_summary): Update.
24158 (inline_read_section): Update.
24159 (inline_write_summary): Update.
24160 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24161 add account_size_time and reset member functions.
24162 (ipa_call_summary): Add reset function.
24163 * ipa-predicate.h (predicate::operator &): Constify.
24165 2017-05-22 Richard Biener <rguenther@suse.de>
24167 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24169 2017-05-19 Jason Merrill <jason@redhat.com>
24171 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24173 2017-05-19 Marek Polacek <polacek@redhat.com>
24176 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24177 TYPE_OVERFLOW_WRAPS checks.
24179 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24181 * tree-core.h (enum omp_clause_default_kind): Add
24182 "OMP_CLAUSE_DEFAULT_PRESENT".
24183 * tree-pretty-print.c (dump_omp_clause): Handle it.
24184 * gimplify.c (enum gimplify_omp_var_data): Add
24185 "GOVD_MAP_FORCE_PRESENT".
24186 (gimplify_adjust_omp_clauses_1): Map it to
24187 "GOMP_MAP_FORCE_PRESENT".
24188 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24190 * gimplify.c (oacc_default_clause): Clarify.
24192 2017-05-19 Nathan Sidwell <nathan@acm.org>
24194 LANG_HOOK_REGISTER_DUMPS
24195 * toplev.c (general_init): Call register dump lang hook.
24196 * doc/invoke.texi: Document -fdump-lang option family.
24197 * dumpfile.c (dump_files): Remove class dump here.
24198 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24199 * dumpfile.h (tree_dump_index): Remove TDI_class.
24200 * langhooks-def.h (lhd_register_dumps): Declare.
24201 (LANG_HOOKS_REGISTER_DUMPS): Define.
24202 (LANG_HOOKS_INITIALIZER): Add it.
24203 * langhooks.c (lhd_register_dumps): Define.
24204 * langhooks.h (struct lang_hooks): Add register_dumps.
24206 2017-05-19 Nathan Sidwell <nathan@acm.org>
24208 * context.h (context::set_passes): New.
24209 * context.c (context::context): Do not create pass manager.
24210 * toplev.c (general_init): Create pass manager here.
24212 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24214 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24215 use this splitter if two add or or instructions would also work for
24216 the constant we want to generate.
24218 2017-05-19 Richard Biener <rguenther@suse.de>
24221 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24222 predicate evaluation.
24224 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24226 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24228 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24231 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24233 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24234 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24235 (inline_edge_summary_vec): Turn into ...
24236 (ipa_call_summaries): ... this one.
24237 (redirect_to_unreachable, edge_set_predicate,
24238 evaluate_properties_for_edge, inline_summary_alloc,
24239 reset_ipa_call_summary, reset_inline_summary,
24240 inline_summary_t::duplicate): Update.
24241 (inline_edge_duplication_hook): Turn to ...
24242 (ipa_call_summary_t::duplicate): ... this one.
24243 (inline_edge_removal_hook): Turn to ...
24244 (ipa_call_summary_t::remove): ... this one.
24245 (dump_inline_edge_summary): Turn to ...
24246 (dump_ipa_call_summary): ... this one.
24247 (estimate_function_body_sizes): Update.
24248 (inline_update_callee_summaries): Update.
24249 (remap_edge_change_prob): Update.
24250 (remap_edge_summaries): Update.
24251 (inline_merge_summary): Update.
24252 (do_estimate_edge_time): Update.
24253 (inline_generate_summary): Update.
24254 (inline_read_section): Update.
24255 (inline_read_summary): Update.
24256 (inline_free_summary): Update.
24257 * ipa-inline.c (can_inline_edge_p): Update.
24258 (compute_inlined_call_time): Update.
24259 (want_inline_small_function_p): Update.
24260 (edge_badness): Update.
24261 (early_inliner): Update.
24262 * ipa-inline.h (inline_edge_summary): Turn to ...
24263 (ipa_call_summary): ... this one.
24264 (ipa_call_summary_t): New class.
24265 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24266 (ipa_call_summaries): New.
24267 (inline_edge_summary): Remove.
24268 (estimate_edge_growth): Update.
24269 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24270 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24271 * ipa-split.c (execute_split_functions): Update.
24272 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24274 2017-05-19 Richard Biener <rguenther@suse.de>
24276 PR middle-end/80764
24277 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24279 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24281 * config/rs6000/rs6000.c (struct machine_function): Add field
24282 fpr_is_wrapped_separately.
24283 (rs6000_get_separate_components): Use 64 components. Handle the
24284 new FPR components.
24285 (rs6000_components_for_bb): Handle the FPR components.
24286 (rs6000_emit_prologue_components): Handle the FPR components.
24287 (rs6000_emit_epilogue_components): Handle the FPR components.
24288 (rs6000_set_handled_components): Handle the FPR components.
24289 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24290 that are already separately shrink-wrapped.
24291 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24292 that are already separately shrink-wrapped.
24294 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24297 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24300 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24301 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24302 cases where the register allocator uses a move and an offsettable
24303 memory operation to/from a FPR register on ISA 2.06/2.07.
24304 (define_peephole2 for Altivec d-form store): Likewise.
24306 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24309 * config/i386/mmx.md (*mov<mode>_internal): Enable
24310 alternatives 11, 12, 13 and 14 also for 32bit targets.
24311 Remove alternatives 15, 16, 17 and 18.
24312 * config/i386/sse.md (vec_concatv2di): Change
24313 alternative (!x, *y) to (x, ?!*Yn).
24315 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24317 * dumpfile.h (enum dump_kind): Remove stray comma.
24319 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24321 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24322 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24323 predicate::num_conditions
24324 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24325 (CHANGED): turn into predicate::changed.
24326 (agg_position_info): Move to ipa-predicate.h
24327 (add_condition, predicate::add_clause, predicate::operator &=,
24328 predicate::or_with, predicate::evaluate, predicate::probability,
24329 dump_condition, dump_clause, predicate::dump,
24330 predicate::remap_after_duplication, predicate::remap_after_inlining,
24331 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24332 (evaluate_conditions_for_known_args): Update.
24333 (set_cond_stmt_execution_predicate): Update.
24334 * ipa-inline.h: Include ipa-predicate.h
24335 (condition, inline_param_summary, conditions, agg_position_info,
24336 predicate): Move to ipa-predicate.h
24337 * ipa-predicate.c: New file.
24338 * ipa-predicate.h: New file.
24340 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24342 * final.c (leaf_function_p): Check we are not in a sequence.
24344 2017-05-18 Martin Liska <mliska@suse.cz>
24346 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24347 * dumpfile.c (dump_register): Use new enum dump_kind.
24348 (get_dump_file_name): Likewise.
24349 (dump_enable_all): Likewise.
24350 (dump_switch_p_1): Likewise.
24351 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24352 * dumpfile.h (enum dump_kind): New enum type.
24353 (struct dump_file_info): Create constructor and
24354 format fields and comments.
24355 * passes.c (pass_manager::register_one_dump_file):
24357 * statistics.c (statistics_early_init): Likewise.
24358 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24359 TDF_TREE with TDF_SLIM.
24360 (gather_memory_references_ref): Likewise.
24362 2017-05-18 Martin Liska <mliska@suse.cz>
24364 * vec.h (struct vnull): Use it.
24366 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24368 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24369 (true_predicate, false_predicate, true_predicate_p,
24370 false_predicate_p): Remove.
24371 (single_cond_predicate, not_inlined_predicate): Turn to member function
24373 (add_condition): Update.
24374 (add_clause): Turn to...
24375 (predicate::add_clause): ... this one; update; allow passing NULL
24377 (and_predicates): Turn to ...
24378 (predicate::operator &=): ... this one.
24379 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24380 (or_predicates): Turn to ...
24381 (predicate::or_with): ... this one.
24382 (evaluate_predicate): Turn to ...
24383 (predicate::evaluate): ... this one.
24384 (predicate_probability): Turn to ...
24385 (predicate::probability): ... this one.
24386 (dump_condition): Update.
24387 (dump_predicate): Turn to ...
24388 (predicate::dump): ... this one.
24389 (account_size_time): Update.
24390 (edge_set_predicate): Update.
24391 (set_hint_predicate): UPdate.
24392 (evaluate_conditions_for_known_args): Update.
24393 (evaluate_properties_for_edge): Update.
24394 (remap_predicate_after_duplication): Turn to...
24395 (predicate::remap_after_duplication): ... this one.
24396 (remap_hint_predicate_after_duplication): Update.
24397 (inline_summary_t::duplicate): UPdate.
24398 (dump_inline_edge_summary): Update.
24399 (dump_inline_summary): Update.
24400 (set_cond_stmt_execution_predicate): Update.
24401 (set_switch_stmt_execution_predicate): Update.
24402 (compute_bb_predicates): Update.
24403 (will_be_nonconstant_expr_predicate): Update.
24404 (will_be_nonconstant_predicate): Update.
24405 (phi_result_unknown_predicate): Update.
24406 (predicate_for_phi_result): Update.
24407 (array_index_predicate): Update.
24408 (estimate_function_body_sizes): Update.
24409 (estimate_node_size_and_time): Update.
24410 (estimate_ipcp_clone_size_and_time): Update.
24411 (remap_predicate): Rename to ...
24412 (predicate::remap_after_inlining): ... this one.
24413 (remap_hint_predicate): Update.
24414 (inline_merge_summary): Update.
24415 (inline_update_overall_summary): Update.
24416 (estimate_size_after_inlining): Update.
24417 (read_predicate): Rename to ...
24418 (predicate::stream_in): ... this one.
24419 (read_inline_edge_summary): Update.
24420 (write_predicate): Rename to ...
24421 (predicate::stream_out): ... this one.
24422 (write_inline_edge_summary): Update.
24423 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24424 (clause_t): Turn to uint32_t
24425 (predicate): Turn to class; implement constructor and operators
24427 (size_time_entry): Update.
24428 (inline_summary): Update.
24429 (inline_edge_summary): Update.
24431 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24433 * fold-const.c (fold_binary_loc): Move transformation...
24434 * match.pd (C - X CMP X): ... here.
24436 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24438 * config/sparc/sparc.c (sparc_option_override): Set function
24439 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24440 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24442 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24444 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24446 2017-05-18 Marek Polacek <polacek@redhat.com>
24449 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24450 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24451 gimple_assign_load_p.
24453 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24455 PR middle-end/80692
24456 * real.c (do_compare): Give decimal_do_compare preference over
24457 comparing just the signs.
24459 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24461 * doc/md.texi (Canonicalization of Instructions): Describe the
24462 canonical form of instructions that inherently set a condition
24465 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24467 PR middle-end/80775
24468 * tree-cfg.c: Move deletion of unreachable case statements to after
24469 the merging of consecutive case labels.
24471 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24473 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24474 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24475 restoring of callee-saved registers.
24477 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24479 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24480 * config/visium/visium.c (single_set_and_flags): Likewise.
24481 * config/visium/visium.md (Substitutions): Likewise.
24483 2017-05-17 Martin Liska <mliska@suse.cz>
24485 * cfg.c: Introduce dump_flags_t type and
24486 use it instead of int type.
24488 * cfghooks.c: Likewise.
24489 * cfghooks.h (struct cfg_hooks): Likewise.
24490 * cfgrtl.c: Likewise.
24491 * cfgrtl.h: Likewise.
24492 * cgraph.c (cgraph_node::get_body): Likewise.
24493 * coretypes.h: Likewise.
24494 * domwalk.c: Likewise.
24495 * domwalk.h: Likewise.
24496 * dumpfile.c (struct dump_option_value_info): Likewise.
24497 (dump_enable_all): Likewise.
24498 (dump_switch_p_1): Likewise.
24499 (opt_info_switch_p): Likewise.
24500 * dumpfile.h (enum tree_dump_index): Likewise.
24501 (struct dump_file_info): Likewise.
24502 * genemit.c: Likewise.
24503 * generic-match-head.c: Likewise.
24504 * gengtype.c (open_base_files): Likewise.
24505 * gimple-pretty-print.c: Likewise.
24506 * gimple-pretty-print.h: Likewise.
24507 * graph.c (print_graph_cfg): Likewise.
24508 * graphite-scop-detection.c (dot_all_sese): Likewise.
24509 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24510 * loop-unroll.c (report_unroll): Likewise.
24511 * passes.c (pass_manager::register_one_dump_file): Likewise.
24512 * print-tree.c: Likewise.
24513 * statistics.c: Likewise.
24514 * tree-cfg.c: Likewise.
24515 * tree-cfg.h: Likewise.
24516 * tree-dfa.c: Likewise.
24517 * tree-dfa.h: Likewise.
24518 * tree-dump.c (dump_function): Likewise.
24519 * tree-dump.h (struct dump_info): Likewise.
24520 * tree-pretty-print.c: Likewise.
24521 * tree-pretty-print.h: Likewise.
24522 * tree-ssa-live.c: Likewise.
24523 * tree-ssa-live.h: Likewise.
24524 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24525 * tree-vect-loop.c: Likewise.
24526 * tree-vect-slp.c: Likewise.
24528 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24529 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24531 PR tree-optimization/80457
24532 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24533 of all arguments to a statement as scalar_to_vec operations.
24534 (vectorizable_call): Adjust call to vect_model_simple_cost for
24536 (vectorizable_conversion): Likewise.
24537 (vectorizable_assignment): Likewise.
24538 (vectorizable_shift): Likewise.
24539 (vectorizable_operation): Likewise.
24540 (vectorizable_comparison): Likewise.
24541 (vect_is_simple_cond): Record the def types for operands.
24542 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24543 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24544 for statement argument count.
24546 2017-05-16 Carl Love <cel@us.ibm.com>
24548 * config/rs6000/rs6000-c: Add support for built-in functions
24549 vector unsigned long long vec_bperm (vector unsigned long long,
24550 vector unsigned char)
24551 vector signed long long vec_mule (vector signed int,
24553 vector unsigned long long vec_mule (vector unsigned int,
24554 vector unsigned int)
24555 vector signed long long vec_mulo (vector signed int,
24557 vector unsigned long long vec_mulo (vector unsigned int,
24558 vector unsigned int)
24559 vector signed char vec_sldw (vector signed char,
24560 vector signed char,
24562 vector unsigned char vec_sldw (vector unsigned char,
24563 vector unsigned char,
24565 vector signed short vec_sldw (vector signed short,
24566 vector signed short,
24568 vector unsigned short vec_sldw (vector unsigned short,
24569 vector unsigned short,
24571 vector signed int vec_sldw (vector signed int,
24574 vector unsigned int vec_sldw (vector unsigned int,
24575 vector unsigned int,
24577 vector signed long long vec_sldw (vector signed long long,
24578 vector signed long long,
24580 vector unsigned long long vec_sldw (vector unsigned long long,
24581 vector unsigned long long,
24583 * config/rs6000/rs6000-c: Add support for built-in functions
24584 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24585 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24586 * doc/extend.texi: Update the built-in documentation for the
24587 new built-in functions.
24589 2017-05-16 Marek Polacek <polacek@redhat.com>
24593 * tree.c (save_expr): Don't fold the expression.
24595 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24597 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24598 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24599 and (?*y,m). Update insn attributes.
24601 2017-05-16 Martin Liska <mliska@suse.cz>
24603 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24604 flags argument of print_gimple_stmt, print_gimple_expr,
24605 print_generic_stmt and print_generic_expr.
24606 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24607 * coretypes.h: Likewise.
24608 * except.c (dump_eh_tree): Likewise.
24609 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24610 * gimple-pretty-print.h: Likewise.
24611 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24612 (backprop::push_to_worklist): Likewise.
24613 (backprop::pop_from_worklist): Likewise.
24614 (backprop::process_use): Likewise.
24615 (backprop::intersect_uses): Likewise.
24616 (note_replacement): Likewise.
24617 * gimple-ssa-store-merging.c
24618 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24619 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24620 (pass_store_merging::execute): Likewise.
24621 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24622 (ssa_base_cand_dump_callback): Likewise.
24623 (dump_incr_vec): Likewise.
24624 (replace_refs): Likewise.
24625 (replace_mult_candidate): Likewise.
24626 (create_add_on_incoming_edge): Likewise.
24627 (create_phi_basis): Likewise.
24628 (insert_initializers): Likewise.
24629 (all_phi_incrs_profitable): Likewise.
24630 (introduce_cast_before_cand): Likewise.
24631 (replace_one_candidate): Likewise.
24632 * gimplify.c (gimplify_expr): Likewise.
24633 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24634 (set_rename): Likewise.
24635 (rename_uses): Likewise.
24636 (copy_loop_phi_nodes): Likewise.
24637 (add_close_phis_to_merge_points): Likewise.
24638 (copy_loop_close_phi_args): Likewise.
24639 (copy_cond_phi_args): Likewise.
24640 (graphite_copy_stmts_from_block): Likewise.
24641 (translate_pending_phi_nodes): Likewise.
24642 * graphite-poly.c (print_pdr): Likewise.
24643 (dump_gbb_cases): Likewise.
24644 (dump_gbb_conditions): Likewise.
24645 (print_scop_params): Likewise.
24646 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24647 (build_cross_bb_scalars_use): Likewise.
24648 (gather_bbs::before_dom_children): Likewise.
24649 * hsa-dump.c (dump_hsa_immed): Likewise.
24650 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24651 (get_replacement_map): Likewise.
24652 * ipa-inline-analysis.c (dump_condition): Likewise.
24653 (estimate_function_body_sizes): Likewise.
24654 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24655 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24656 * ipa-prop.c (ipa_dump_param): Likewise.
24657 (ipa_print_node_jump_functions_for_edge): Likewise.
24658 (ipa_modify_call_arguments): Likewise.
24659 (ipa_modify_expr): Likewise.
24660 (ipa_dump_param_adjustments): Likewise.
24661 (ipa_dump_agg_replacement_values): Likewise.
24662 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24663 * ipa-pure-const.c (check_stmt): Likewise.
24664 (pass_nothrow::execute): Likewise.
24665 * ipa-split.c (execute_split_functions): Likewise.
24666 * omp-offload.c (dump_oacc_loop_part): Likewise.
24667 (dump_oacc_loop): Likewise.
24668 * trans-mem.c (tm_log_emit): Likewise.
24669 (tm_memopt_accumulate_memops): Likewise.
24670 (dump_tm_memopt_set): Likewise.
24671 (dump_tm_memopt_transform): Likewise.
24672 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24673 (print_loop): Likewise.
24674 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24675 (chkp_gather_checks_info): Likewise.
24676 (chkp_get_check_result): Likewise.
24677 (chkp_remove_check_if_pass): Likewise.
24678 (chkp_use_outer_bounds_if_possible): Likewise.
24679 (chkp_reduce_bounds_lifetime): Likewise.
24680 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24681 (chkp_mark_completed_bounds): Likewise.
24682 (chkp_register_incomplete_bounds): Likewise.
24683 (chkp_mark_invalid_bounds): Likewise.
24684 (chkp_maybe_copy_and_register_bounds): Likewise.
24685 (chkp_build_returned_bound): Likewise.
24686 (chkp_get_bound_for_parm): Likewise.
24687 (chkp_build_bndldx): Likewise.
24688 (chkp_get_bounds_by_definition): Likewise.
24689 (chkp_generate_extern_var_bounds): Likewise.
24690 (chkp_get_bounds_for_decl_addr): Likewise.
24691 * tree-chrec.c (chrec_apply): Likewise.
24692 * tree-data-ref.c (dump_data_reference): Likewise.
24693 (dump_subscript): Likewise.
24694 (dump_data_dependence_relation): Likewise.
24695 (analyze_overlapping_iterations): Likewise.
24696 * tree-inline.c (expand_call_inline): Likewise.
24697 (tree_function_versioning): Likewise.
24698 * tree-into-ssa.c (dump_defs_stack): Likewise.
24699 (dump_currdefs): Likewise.
24700 (dump_names_replaced_by): Likewise.
24701 (dump_update_ssa): Likewise.
24702 (update_ssa): Likewise.
24703 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24704 * tree-parloops.c (build_new_reduction): Likewise.
24705 (try_create_reduction_list): Likewise.
24706 (ref_conflicts_with_region): Likewise.
24707 (oacc_entry_exit_ok_1): Likewise.
24708 (oacc_entry_exit_single_gang): Likewise.
24709 * tree-pretty-print.h: Likewise.
24710 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24711 (get_scalar_evolution): Likewise.
24712 (add_to_evolution): Likewise.
24713 (get_loop_exit_condition): Likewise.
24714 (analyze_evolution_in_loop): Likewise.
24715 (analyze_initial_condition): Likewise.
24716 (analyze_scalar_evolution): Likewise.
24717 (instantiate_scev): Likewise.
24718 (number_of_latch_executions): Likewise.
24719 (gather_chrec_stats): Likewise.
24720 (final_value_replacement_loop): Likewise.
24721 (scev_const_prop): Likewise.
24722 * tree-sra.c (dump_access): Likewise.
24723 (disqualify_candidate): Likewise.
24724 (create_access): Likewise.
24725 (reject): Likewise.
24726 (maybe_add_sra_candidate): Likewise.
24727 (create_access_replacement): Likewise.
24728 (analyze_access_subtree): Likewise.
24729 (analyze_all_variable_accesses): Likewise.
24730 (sra_modify_assign): Likewise.
24731 (initialize_constant_pool_replacements): Likewise.
24732 (find_param_candidates): Likewise.
24733 (decide_one_param_reduction): Likewise.
24734 (replace_removed_params_ssa_names): Likewise.
24735 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24736 * tree-ssa-copy.c (dump_copy_of): Likewise.
24737 (copy_prop_visit_cond_stmt): Likewise.
24738 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24739 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24740 (record_equivalences_from_stmt): Likewise.
24741 * tree-ssa-dse.c (compute_trims): Likewise.
24742 (delete_dead_call): Likewise.
24743 (delete_dead_assignment): Likewise.
24744 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24745 (forward_propagate_into_cond): Likewise.
24746 (pass_forwprop::execute): Likewise.
24747 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24748 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24750 (move_computations_worker): Likewise.
24751 (execute_sm): Likewise.
24752 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24753 (remove_exits_and_undefined_stmts): Likewise.
24754 (remove_redundant_iv_tests): Likewise.
24755 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24756 (adjust_iv_update_pos): Likewise.
24757 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24758 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24759 (value_replacement): Likewise.
24760 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24761 * tree-ssa-pre.c (print_pre_expr): Likewise.
24762 (get_representative_for): Likewise.
24763 (create_expression_by_pieces): Likewise.
24764 (insert_into_preds_of_block): Likewise.
24765 (eliminate_insert): Likewise.
24766 (eliminate_dom_walker::before_dom_children): Likewise.
24767 (eliminate): Likewise.
24768 (remove_dead_inserted_code): Likewise.
24769 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24770 * tree-ssa-reassoc.c (get_rank): Likewise.
24771 (eliminate_duplicate_pair): Likewise.
24772 (eliminate_plus_minus_pair): Likewise.
24773 (eliminate_not_pairs): Likewise.
24774 (undistribute_ops_list): Likewise.
24775 (eliminate_redundant_comparison): Likewise.
24776 (update_range_test): Likewise.
24777 (optimize_range_tests_var_bound): Likewise.
24778 (optimize_vec_cond_expr): Likewise.
24779 (rewrite_expr_tree): Likewise.
24780 (rewrite_expr_tree_parallel): Likewise.
24781 (linearize_expr): Likewise.
24782 (break_up_subtract): Likewise.
24783 (linearize_expr_tree): Likewise.
24784 (attempt_builtin_powi): Likewise.
24785 (attempt_builtin_copysign): Likewise.
24786 (transform_stmt_to_copy): Likewise.
24787 (transform_stmt_to_multiply): Likewise.
24788 (dump_ops_vector): Likewise.
24789 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24790 (print_scc): Likewise.
24791 (set_ssa_val_to): Likewise.
24792 (visit_reference_op_store): Likewise.
24793 (visit_use): Likewise.
24794 (sccvn_dom_walker::before_dom_children): Likewise.
24795 (run_scc_vn): Likewise.
24796 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24798 (expr_hash_elt::print): Likewise.
24799 (const_and_copies::pop_to_marker): Likewise.
24800 (const_and_copies::record_const_or_copy_raw): Likewise.
24801 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24802 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24803 (dump_predicates): Likewise.
24804 (find_uninit_use): Likewise.
24805 (warn_uninitialized_phi): Likewise.
24806 (pass_late_warn_uninitialized::execute): Likewise.
24807 * tree-ssa.c (verify_vssa): Likewise.
24808 (verify_ssa): Likewise.
24809 (maybe_optimize_var): Likewise.
24810 * tree-vrp.c (dump_value_range): Likewise.
24811 (dump_all_value_ranges): Likewise.
24812 (dump_asserts_for): Likewise.
24813 (register_edge_assert_for_2): Likewise.
24814 (vrp_visit_cond_stmt): Likewise.
24815 (vrp_visit_switch_stmt): Likewise.
24816 (vrp_visit_stmt): Likewise.
24817 (vrp_visit_phi_node): Likewise.
24818 (simplify_cond_using_ranges_1): Likewise.
24819 (fold_predicate_in): Likewise.
24820 (evrp_dom_walker::before_dom_children): Likewise.
24821 (evrp_dom_walker::push_value_range): Likewise.
24822 (evrp_dom_walker::pop_value_range): Likewise.
24823 (execute_early_vrp): Likewise.
24825 2017-05-16 Richard Biener <rguenther@suse.de>
24827 * dwarf2out.c (loc_list_from_tree_1): Do not create
24828 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24830 2017-05-16 Richard Biener <rguenther@suse.de>
24832 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24834 (note_variable_value_in_expr): If we resolved the decl ref
24835 do not push to the stack.
24837 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24839 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24840 operations in fast-math mode.
24841 (vaddq_f16): Likewise.
24842 (vmul_f16): Likewise.
24843 (vmulq_f16): Likewise.
24844 (vsub_f16): Likewise.
24845 (vsubq_f16): Likewise.
24846 * config/arm/neon.md (add<mode>3): New.
24848 (fma:<VH:mode>3): New. Also remove outdated comment.
24851 2017-05-16 Martin Liska <mliska@suse.cz>
24855 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24856 (odr_types_equivalent_p): Likewise.
24858 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24860 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24862 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24865 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24866 non-interunit SSE move alternatives with '?'.
24867 (zero-extendsidi peephole2): New peephole to skip intermediate
24868 general register in SSE zero-extend sequence.
24870 2017-05-15 Jeff Law <law@redhat.com>
24872 * reorg.c (relax_delay_slots): Create a new variable to hold
24873 the temporary target rather than clobbering TARGET_LABEL.
24875 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24876 missing argument to extract_bit_field call.
24877 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24879 2017-05-15 Martin Liska <mliska@suse.cz>
24882 * gcc.c (process_command): Do not allow empty argument of -o option.
24884 2017-05-15 Renlin Li <renlin.li@arm.com>
24886 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24887 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24888 * config/aarch64/constraints.md (Usf): Add long call check.
24889 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24890 (call_value): Likewise.
24891 (sibcall): Likewise.
24892 (sibcall_value): Likewise.
24894 (call_value_insn): New.
24895 (sibcall_insn): Update rtx pattern.
24896 (sibcall_value_insn): Likewise.
24897 (call_internal): Remove.
24898 (call_value_internal): Likewise.
24899 (sibcall_internal): Likewise.
24900 (sibcall_value_internal): Likewise.
24901 (call_reg): Likewise.
24902 (call_symbol): Likewise.
24903 (call_value_reg): Likewise.
24904 (call_value_symbol): Likewise.
24906 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24909 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24911 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24913 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24914 compatible with CCGOCmode and with CCZmode.
24916 2017-05-14 Martin Sebor <msebor@redhat.com>
24918 PR middle-end/77671
24919 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24920 (gimple_fold_builtin_snprintf): Same.
24921 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24922 (gimple_fold_builtin_snprintf): Same.
24923 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24924 of character types.
24925 (is_call_safe): New function.
24926 (try_substitute_return_value): Call it.
24927 (try_simplify_call): New function.
24928 (pass_sprintf_length::handle_gimple_call): Call it.
24930 2017-05-14 Martin Sebor <msebor@redhat.com>
24932 PR middle-end/80669
24933 * builtins.c (expand_builtin_stpncpy): Simplify.
24935 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24937 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24938 * config/i386/i386.h
24939 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24940 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24941 (struct machine_function): Add new members call_ms2sysv,
24942 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24943 (struct machine_frame_state): New fields sp_realigned and
24944 sp_realigned_offset.
24945 * config/i386/i386.c
24946 (enum xlogue_stub): New enum.
24947 (enum xlogue_stub_sets): New enum.
24948 (class xlogue_layout): New class.
24949 (struct ix86_frame): New fields stack_realign_allocate_offset,
24950 stack_realign_offset and outlined_save_offset. Modify comments to
24951 detail stack layout when using out-of-line stubs.
24952 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24953 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24954 -mcall-ms2sysv-xlogues.
24955 (stub_managed_regs): New static variable.
24956 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24957 registers managed by out-of-line stub.
24958 (disable_call_ms2sysv_xlogues): New function.
24959 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24960 m->call_ms2sysv when appropriate and compute frame layout for
24962 (sp_valid_at, fp_valid_at): New inline functions.
24963 (choose_basereg): New function.
24964 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24966 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24967 Use align parameter of choose_baseaddr to generated aligned SSE movs
24969 (pro_epilogue_adjust_stack): Modify to track
24970 machine_frame_state::sp_realigned.
24971 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24972 (ix86_nsaved_sseregs): Likewise.
24973 (ix86_emit_save_regs): Likewise.
24974 (ix86_emit_save_regs_using_mov): Likewise.
24975 (ix86_emit_save_sse_regs_using_mov): Likewise.
24976 (get_scratch_register_on_entry): Likewise.
24977 (gen_frame_set): New function.
24978 (gen_frame_load): Likewise.
24979 (gen_frame_store): Likewise.
24980 (emit_outlined_ms2sysv_save): Likewise.
24981 (emit_outlined_ms2sysv_restore): Likewise.
24982 (ix86_expand_prologue): Modify stack re-alignment code and call
24983 emit_outlined_ms2sysv_save when appropriate.
24984 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24985 parameter rtx_insn *insn, which allows the function to be used to only
24986 generate the notes.
24987 (ix86_expand_epilogue): Modify validity checks of frame and stack
24988 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24989 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24990 * config/i386/predicates.md
24991 (save_multiple): New predicate.
24992 (restore_multiple): Likewise.
24993 * config/i386/sse.md
24994 (save_multiple<mode>): New pattern.
24995 (save_multiple_realign<mode>): Likewise.
24996 (restore_multiple<mode>): Likewise.
24997 (restore_multiple_and_return<mode>): Likewise.
24998 (restore_multiple_leave_return<mode>): Likewise.
24999 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25001 2017-05-14 Julia Koval <julia.koval@intel.com>
25003 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25004 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25005 (__builtin_ia32_xsetbv): New builtins.
25006 * config/i386/i386.c (ix86_expand_special_args_builtin):
25008 (ix86_expand_builtin): Special expand for new intrinsics.
25009 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25010 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25011 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25013 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25015 * cfganal.c (inverted_post_order_compute): Change argument type
25017 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25018 * df-core.c (rest_of_handle_df_initialize): Adjust.
25019 (rest_of_handle_df_finish): Likewise.
25020 (df_analyze_1): Likewise.
25021 (df_analyze): Likewise.
25022 (loop_inverted_post_order_compute): Change argument to be a vec *.
25023 (df_analyze_loop): Adjust.
25024 (df_get_n_blocks): Likewise.
25025 (df_get_postorder): Likewise.
25026 * df.h (struct df_d): Change field to be a vec.
25027 * lcm.c (compute_laterin): Adjust.
25028 (compute_available): Likewise.
25029 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25030 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25031 * tree-ssa-pre.c (compute_antic): Likewise.
25033 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25035 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25036 (depth_first_search::depth_first_search): Change structure init
25037 function to this constructor.
25038 (depth_first_search::add_bb): Rename function to this member.
25039 (depth_first_search::execute): Likewise.
25040 (flow_dfs_compute_reverse_finish): Adjust.
25042 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25044 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25045 (longest_simple_path): Likewise.
25046 * shrink-wrap.c (spread_components): Likewise.
25047 (disqualify_problematic_components): Likewise.
25048 (emit_common_heads_for_components): Likewise.
25049 (emit_common_tails_for_components): Likewise.
25050 (insert_prologue_epilogue_for_components): Likewise.
25052 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25054 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25057 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25059 * df-core.c (df_set_blocks): Start using auto_bitmap.
25060 (df_compact_blocks): Likewise.
25061 * df-problems.c (df_rd_confluence_n): Likewise.
25062 * df-scan.c (df_insn_rescan_all): Likewise.
25063 (df_process_deferred_rescans): Likewise.
25064 (df_update_entry_block_defs): Likewise.
25065 (df_update_exit_block_uses): Likewise.
25066 (df_entry_block_bitmap_verify): Likewise.
25067 (df_exit_block_bitmap_verify): Likewise.
25068 (df_scan_verify): Likewise.
25069 * lra-constraints.c (lra_constraints): Likewise.
25070 (undo_optional_reloads): Likewise.
25071 (lra_undo_inheritance): Likewise.
25072 * lra-remat.c (calculate_gen_cands): Likewise.
25073 (do_remat): Likewise.
25074 * lra-spills.c (assign_spill_hard_regs): Likewise.
25075 (spill_pseudos): Likewise.
25076 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25077 (bitmap_set_subtract_values): Likewise.
25079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25081 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25082 management with auto_bitmap.
25083 (fix_inter_tick): Likewise.
25084 (fix_recovery_deps): Likewise.
25085 * ira.c (add_store_equivs): Likewise.
25086 (find_moveable_pseudos): Likewise.
25087 (split_live_ranges_for_shrink_wrap): Likewise.
25088 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25089 (rtx_reuse_manager::seen_def_p): Likewise.
25090 (rtx_reuse_manager::set_seen_def): Likewise.
25091 * print-rtl.h (class rtx_reuse_manager): Likewise.
25093 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25095 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25097 (migrate_btr_def): Likewise.
25098 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25099 * df-core.c (loop_post_order_compute): Likewise.
25100 (loop_inverted_post_order_compute): Likewise.
25101 * hsa-common.h: Likewise.
25102 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25103 * init-regs.c (initialize_uninitialized_regs): Likewise.
25104 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25105 (inline_small_functions): Likewise.
25106 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25107 * ira.c (combine_and_move_insns): Likewise.
25108 (build_insn_chain): Likewise.
25109 * loop-invariant.c (find_invariants): Likewise.
25110 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25111 * predict.c (tree_predict_by_opcode): Likewise.
25112 (predict_paths_leading_to): Likewise.
25113 (predict_paths_leading_to_edge): Likewise.
25114 (estimate_loops_at_level): Likewise.
25115 (estimate_loops): Likewise.
25116 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25117 (spread_components): Likewise.
25118 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25119 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25120 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25121 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25122 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25123 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25124 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25125 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25126 (mark_threaded_blocks): Likewise.
25127 (thread_through_all_blocks): Likewise.
25128 * tree-ssa.c (verify_ssa): Likewise.
25129 (execute_update_addresses_taken): Likewise.
25130 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25132 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25134 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25136 (post_order_compute): Likewise.
25137 (inverted_post_order_compute): Likewise.
25138 (pre_and_rev_post_order_compute_fn): Likewise.
25140 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25142 * genrecog.c (int_set::int_set): Explicitly construct our
25143 auto_vec base class.
25144 * vec.h (auto_vec::auto_vec): New constructor.
25146 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25148 * bitmap.h (class auto_bitmap): New constructor taking
25149 bitmap_obstack * argument.
25151 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25153 * bitmap.h (class auto_bitmap): Change type of m_bits to
25154 bitmap_head, and adjust ctor / dtor and member operators.
25156 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25158 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25159 when returned register mode doesn't match original mode.
25161 2017-05-12 Jeff Law <law@redhat.com>
25162 Jakub Jelinek <jakub@redhat.com>
25164 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25165 we look for cc setter after the compare-elim changes.
25166 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25167 within the vector to match what compare-elim now expects.
25168 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25169 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25171 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25172 after the compare-elim changes.
25173 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25174 the vector to match what compare-elim now expects.
25175 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25176 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25177 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25178 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25179 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25181 * config/visium/visium.c (single_set_and_flags): Fix where
25182 we look for cc setter after the compare-elim changes.
25183 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25184 with the vector to match what compare-elim now expects.
25185 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25186 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25187 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25188 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25189 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25190 (neg<mode>2_insn_set_overflow): Likewise.
25192 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25194 PR middle-end/79794
25195 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25196 maybe_expand_insn call, set ops[0].target. If still set after call,
25197 set alt_rtl. Add extra arg to recursive calls.
25198 (extract_bit_field): Add alt_rtl argument. Pass to
25200 * expmed.h (extract_bit_field): Fix prototype.
25201 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25202 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25203 to extract_bit_field_calls.
25204 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25205 Pass alt_rtl to extract_bit_field calls.
25206 * calls.c (store_unaligned_arguments_into_psuedos)
25207 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25208 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25210 * optabs.h (struct expand_operand): Add target bitfield.
25212 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25214 * compare-elim.c (try_eliminate_compare): Canonicalize
25215 operation with embedded compare to
25216 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25217 (set (reg) (operation)].
25219 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25221 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25224 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25225 cost of adding a carry flag for ADC instruction.
25226 [case MINUS]: Ignore the cost of subtracting a carry flag
25227 for SBB instruction.
25229 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25231 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25233 * config/rs6000/bmiintrin.h: New file.
25234 * config/rs6000/bmi2intrin.h: New file.
25235 * config/rs6000/x86intrin.h: New file.
25237 2017-05-12 Jeff Law <law@redhat.com>
25239 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25242 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25244 PR middle-end/80707
25245 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25247 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25249 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25250 early expansion of vector divide builtins.
25251 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25252 builtins identified as having unsigned arguments.
25254 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25256 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25257 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25258 expansion of vector logical operations (and, andc, or, xor,
25261 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25263 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25264 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25266 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25268 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25269 early GIMPLE expansion of vector multiplies.
25271 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25273 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25274 TARGET_HAVE_MOVT conditional.
25275 (movt splitter): Likewise.
25277 2017-05-12 Richard Biener <rguenther@suse.de>
25279 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25280 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25281 Fold all stmts not inplace.
25283 2017-05-12 Richard Biener <rguenther@suse.de>
25285 PR tree-optimization/80713
25286 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25287 inserted_exprs bit for not removed stmts.
25289 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25291 PR middle-end/69921
25292 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25293 parallelized" attribute for parallelized OpenACC kernels.
25294 * omp-offload.c (execute_oacc_device_lower): Use it.
25296 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25297 Set "oacc kernels" attribute.
25298 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25299 parameter. Adjust all users.
25300 (oacc_fn_attrib_kernels_p): Remove function.
25301 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25302 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25303 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25304 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25305 assert "oacc kernels" attribute is set.
25307 2017-05-11 Carl Love <cel@us.ibm.com>
25309 * config/rs6000/rs6000-c: Add support for built-in functions
25310 vector unsigned char vec_popcnt (vector signed char)
25311 vector unsigned char vec_popcnt (vector unsigned char)
25312 vector unsigned short vec_popcnt (vector signed short)
25313 vector unsigned short vec_popcnt (vector unsigned short)
25314 vector unsigned int vec_popcnt (vector signed int)
25315 vector unsigned int vec_popcnt (vector unsigned int)
25316 vector unsigned long long vec_popcnt (vector signed long long)
25317 vector unsigned long long vec_popcnt (vector unsigned long long)
25318 vector signed long long vec_slo (vector signed long long,
25319 vector signed char)
25320 vector signed long long vec_slo (vector signed long long,
25321 vector unsigned char)
25322 vector unsigned long long vec_slo (vector unsigned long long,
25323 vector signed char)
25324 vector unsigned long long vec_slo (vector unsigned long long,
25325 vector unsigned char)
25326 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25327 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25328 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25329 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25330 * doc/extend.texi: Update the built-in documentation file for the
25331 new built-in functions.
25333 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25335 * attribs.h (sorted_attr_string): Move machine independent
25336 functions for target clone support from the i386 port to common
25337 code. Rename ix86_function_versions to common_function_versions.
25338 Rename make_name to make_unique_name.
25339 (common_function_versions): Likewise.
25340 (make_unique_name): Likewise.
25341 (make_dispatcher_decl): Likewise.
25342 (is_function_default_version): Likewise.
25343 * attribs.c (attr_strcmp): Likewise.
25344 (sorted_attr_string): Likewise.
25345 (common_function_versions): Likewise.
25346 (make_unique_name): Likewise.
25347 (make_dispatcher_decl): Likewise.
25348 (is_function_default_version): Likewise.
25349 * config/i386/i386.c (attr_strcmp): Likewise.
25350 (sorted_attr_string): Likewise.
25351 (ix86_function_versions): Likewise.
25352 (make_name): Likewise.
25353 (make_dispatcher_decl): Likewise.
25354 (is_function_default_version): Likewise.
25355 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25357 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25360 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25361 Account for direct move costs for vec_construct of integer
25364 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25367 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25368 (UNSPEC_STX_ATOMIC): Ditto.
25369 (loaddi_via_sse): New insn.
25370 (storedi_via_sse): Ditto.
25371 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25372 Update corresponding peephole2 patterns.
25373 (atomic_storedi_fpu): Ditto.
25375 2017-05-11 Julia Koval <julia.koval@intel.com>
25377 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25378 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25380 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25381 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25382 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25384 2017-05-11 Nathan Sidwell <nathan@acm.org>
25386 * graphite-poly.c: Include dumpfile.h.
25388 * dumpfle.h (dump_function): Declare here ...
25389 * tree-dump.h (dump_function): ... not here.
25390 * dumpfile.c: #include tree-cfg.h.
25391 (dump_function): Move here from ...
25392 * tree-dump.c (dump_function): ... here.
25393 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25394 * graphite-poly.c: Don't include tree-dump.h.
25395 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25396 * print-tree.c: Likewise.
25397 * stor-layout.c: Likewise.
25398 * tree-nested.c: Likewise.
25400 * dumpfile.c (dump_start): Use TDF_FLAGS.
25401 (dump_enable_all): Fix TDF_KIND check thinko.
25403 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25405 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25406 array entries to represent two legal parameterizations of the
25407 overloaded __builtin_cmpb function, as represented by the
25408 P6_OV_BUILTIN_CMPB constant.
25409 (altivec_resolve_overloaded_builtin): Add special case handling
25410 for the __builtin_cmpb function, as represented by the
25411 P6_OV_BUILTIN_CMPB constant.
25412 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25413 (BU_P6_64BIT_2): New macro.
25414 (BU_P6_OVERLOAD_2): New macro
25415 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25416 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25417 (CMPB): Add overload support to represent both 32-bit and 64-bit
25418 compare-bytes function.
25419 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25420 support for TARGET_CMPB.
25421 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25422 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25423 documentation of the __builtin_cmpb overloaded built-in function.
25425 2017-05-11 Richard Biener <rguenther@suse.de>
25427 PR tree-optimization/80705
25428 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25429 bases are not vectorizable.
25431 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25433 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25434 when counting register pressure.
25436 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25438 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25439 (iv_ca_more_deps): Renamed to ...
25440 (iv_ca_compare_deps): ... this.
25441 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25443 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25445 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25447 (determine_group_iv_costs): ... here.
25448 (find_inv_vars_cb): Record inv var if it's not recorded before.
25450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25452 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25453 (get_shiftadd_cost): Ditto.
25455 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25457 * tree-ssa-address.c: Include header file.
25458 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25460 (add_to_parts): Refactor.
25461 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25462 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25465 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25467 PR tree-optimization/53090
25468 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25470 (extract_cond_operands): Detect condition with IV on both sides
25471 and return COMP_IV_EXPR_2.
25472 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25473 (rewrite_use_compare): Simplify by removing call to function
25474 extract_cond_operands.
25476 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25478 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25479 (extract_cond_operands): Detect condition comparing against non-
25480 invariant bound and return appropriate enum value.
25481 (find_interesting_uses_cond): Update use of extract_cond_operands.
25482 Handle its return value accordingly.
25483 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25485 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25487 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25488 nonlinear iv_use computation in loop invariant sensitive way.
25490 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25492 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25493 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25495 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25497 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25498 (dump_cand): Support iv_cand.inv_exprs.
25499 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25501 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25504 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25506 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25508 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25509 as local function. Include necessary header files.
25510 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25512 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25514 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25516 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25518 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25519 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25520 RSHIFT_EXPR and BIT_NOT_EXPR.
25522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25524 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25525 (adjust_setup_cost): New parameter supporting round up adjustment.
25526 (struct address_cost_data): Delete.
25527 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25528 (split_address_cost, ptr_difference_cost): Delete.
25529 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25530 (struct ainc_cost_data): New struct.
25531 (get_address_cost_ainc): New function.
25532 (get_address_cost, get_computation_cost): Reimplement.
25533 (determine_group_iv_cost_address): Record inv_expr for all uses of
25535 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25536 (iv_ca_has_deps): Reimplemented to ...
25537 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25539 (iv_ca_extend): Call iv_ca_more_deps.
25541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25543 * tree-ssa-address.c (struct mem_address): Move to header file.
25544 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25545 * tree-ssa-address.h (struct mem_address): Move from C file.
25546 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25548 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25550 * tree-affine.h (aff_combination_type): New interface.
25551 (aff_combination_zero_p): Remove static.
25552 (aff_combination_const_p): New interface.
25553 (aff_combination_singleton_var_p): New interfaces.
25555 2017-05-11 Richard Biener <rguenther@suse.de>
25557 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25558 Skip unreachable blocks and destinations.
25559 (eliminate): Move stmt removal and fixup ...
25560 (fini_eliminate): ... here. Skip inserted exprs.
25561 (pass_pre::execute): Move fini_pre after fini_eliminate.
25562 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25563 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25564 PRE to get rid of dead code that has invalid SSA form and
25565 split critical edges again.
25567 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25569 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25571 2017-05-11 Richard Biener <rguenther@suse.de>
25573 * passes.c (execute_function_todo): Verify loops if they are
25574 said to be up-to-date.
25575 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25576 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25578 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25581 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25582 handle calling assemble_external ourself.
25585 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25586 modes with zero size. Enhance comment.
25588 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25590 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25591 built-ins for vec_xl and vec_xst with short and char pointer
25594 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25596 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25597 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25598 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25599 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25600 (_mm_maskz_min_round_ss): New intrinsics.
25601 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25602 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25603 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25604 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25605 (__builtin_ia32_minss_mask_round): New builtins.
25606 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25607 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25608 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25610 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25611 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25613 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25616 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25618 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25619 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25620 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25621 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25622 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25623 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25624 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25625 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25626 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25627 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25628 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25629 (__builtin_ia32_mulss_mask_round): New builtins.
25630 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25631 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25632 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25634 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25635 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25637 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25640 2017-05-10 Julia Koval <julia.koval@intel.com>
25642 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25643 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25644 (_mm256_setr_m128i): New intrinsics.
25646 2017-05-10 Julia Koval <julia.koval@intel.com>
25648 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25649 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25650 (_mm_maskz_rcp14_ss): New intrinsics.
25651 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25652 (__builtin_ia32_rcp14ss_mask): New builtins.
25653 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25655 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25657 PR tree-optimization/51513
25658 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25659 (assert_unreachable_fallthru_edge_p): Use it.
25660 (group_case_labels_stmt): Likewise.
25661 * tree-cfg.h: Prototype it.
25662 * stmt.c: Include cfghooks.h and tree-cfg.h.
25663 (emit_case_dispatch_table) <gap_label>: New local variable.
25664 Use it to fill dispatch table gaps.
25665 Test for default_label before updating probabilities.
25666 (expand_case) <default_label>: Remove unneeded initialization.
25667 Test for unreachable default case statement and remove its edge.
25668 Set default_label accordingly.
25669 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25671 2017-05-10 Carl Love <cel@us.ibm.com>
25673 * config/rs6000/rs6000-c: Add support for built-in functions
25674 vector signed char vec_neg (vector signed char)
25675 vector signed short int vec_neg (vector short int)
25676 vector signed int vec_neg (vector signed int)
25677 vector signed long long vec_neg (vector signed long long)
25678 vector float vec_neg (vector float)
25679 vector double vec_neg (vector double)
25680 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25682 * config/rs6000/altivec.h: Add define for vec_neg
25683 * doc/extend.texi: Update the built-in documentation for the
25684 new built-in functions.
25686 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25688 PR tree-optimization/77644
25689 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25691 2017-05-10 Nathan Sidwell <nathan@acm.org>
25693 * dumpfile.h (TDI_lang_all): New.
25694 (TDF_KIND): New. Renumber others
25695 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25697 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25699 (get_dump_file_name): Adjust suffix generation.
25700 (dump_enable_all): Use TDF_KIND.
25701 * doc/invoke.texi (-fdump-lang-all): Document.
25703 * dumpfile.h: Tabify.
25705 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25708 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25709 Move member access before delete.
25711 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25713 * tree-inline.c (expand_call_inline): Split block at stmt
25716 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25719 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25720 are now unused after splitting mov{sf,sd}_hardfloat.
25721 (f32_lr2): Likewise.
25722 (f32_lm): Likewise.
25723 (f32_lm2): Likewise.
25724 (f32_li): Likewise.
25725 (f32_li2): Likewise.
25726 (f32_lv): Likewise.
25727 (f32_sr): Likewise.
25728 (f32_sr2): Likewise.
25729 (f32_sm): Likewise.
25730 (f32_sm2): Likewise.
25731 (f32_si): Likewise.
25732 (f32_si2): Likewise.
25733 (f32_sv): Likewise.
25734 (f32_dm): Likewise.
25735 (f32_vsx): Likewise.
25736 (f32_av): Likewise.
25737 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25738 For movsf, order stores so the VSX stores occur before the GPR
25739 store which encourages the register allocator to use a traditional
25740 FPR instead of a GPR. For movsd, order the stores so that the GPR
25741 store comes before the VSX stores to allow the power6 to work.
25742 This is due to the power6 not having a 32-bit integer store
25743 instruction from a FPR.
25744 (movsf_hardfloat): Likewise.
25745 (movsd_hardfloat): Likewise.
25747 2017-05-09 Martin Sebor <msebor@redhat.com>
25749 PR translation/80280
25750 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25753 PR translation/80280
25754 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25755 data member added in r247778.
25756 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25758 2017-05-09 Nathan Sidwell <nathan@acm.org>
25760 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25762 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25765 2017-05-09 Marek Polacek <polacek@redhat.com>
25767 * doc/invoke.texi: Fix typo.
25769 2017-05-09 Richard Biener <rguenther@suse.de>
25771 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25772 (vrp_val_is_min): Likewise.
25773 (set_value_range_to_value): Likewise.
25774 (set_value_range_to_nonnegative): Likewise.
25775 (gimple_assign_nonzero_p): Likewise.
25776 (gimple_stmt_nonzero_p): Likewise.
25777 (vrp_int_const_binop): Likewise. Remove unreachable case.
25778 (adjust_range_with_scev): Adjust comments.
25779 (compare_range_with_value): Likewise.
25780 (extract_range_from_phi_node): Likewise.
25781 (test_for_singularity): Likewise.
25783 2017-05-09 Richard Biener <rguenther@suse.de>
25785 * tree-vrp.c (get_single_symbol): Add assert that we don't
25786 get overflowed constants as invariant part.
25787 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25788 checks. Use wi::cmp instead of recursing for integer constants.
25789 (compare_values): Just ignore whether we assumed undefined
25790 overflow instead of failing the compare.
25791 (extract_range_for_var_from_comparison_expr): Add comment before the
25792 TREE_NO_WARNING sets.
25793 (test_for_singularity): Likewise.
25794 (extract_range_from_comparison): Do not disable optimization
25795 when we assumed undefined overflow.
25796 (extract_range_basic): Remove init of unused var.
25798 2017-05-09 Richard Biener <rguenther@suse.de>
25800 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25801 (extract_range_from_multiplicative_op_1): Adjust.
25802 (extract_range_from_binary_expr_1): Use int_const_binop.
25804 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25807 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25808 rs6000_store_data_bypass_p in seven define_bypass directives and
25809 in several comments.
25810 * config/rs6000/rs6000-protos.h: Add prototype for
25811 rs6000_store_data_bypass_p function.
25812 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25813 function implements slightly different (rs6000-specific) semantics
25814 than store_data_bypass_p, returning false rather than aborting
25815 with assertion error when arguments do not satisfy the
25816 requirements of store data bypass.
25817 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25818 rs6000_store_data_bypass_p.
25820 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25822 * config/xtensa/xtensa-protos.h
25823 (xtensa_initial_elimination_offset): New declaration.
25824 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25825 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25826 macro definition, add case for FRAME_POINTER_REGNUM when
25827 FRAME_GROWS_DOWNWARD.
25828 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25829 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25830 xtensa_initial_elimination_offset.
25832 2017-05-08 Nathan Sidwell <nathan@acm.org>
25834 * doc/invoke.texi: Alphabetize -fdump options.
25836 2017-05-08 Martin Sebor <msebor@redhat.com>
25838 PR translation/80280
25839 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25841 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25843 * target.def (compute_frame_layout): New optional target hook.
25844 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25845 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25846 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25848 * reload1.c (verify_initial_elim_offsets): Likewise.
25849 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25850 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25851 (arm_get_frame_offsets): Split up into this ...
25852 (arm_compute_frame_layout): ... and this function.
25854 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25856 * config/aarch64/constraints.md (Usa): New constraint.
25857 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25859 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25861 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25862 with_multilib_list after it has been checked.
25864 2017-05-08 Richard Biener <rguenther@suse.de>
25866 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25867 (bitmap_set_subtract_values): Likewise.
25869 2017-05-08 Richard Biener <rguenther@suse.de>
25871 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25872 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25874 (gimple_stmt_nonzero_warnv_p): Rename to ...
25875 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25877 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25878 (extract_range_basic): Adjust, do not disable propagation on
25879 strict overflow sensitive simplification.
25880 (vrp_visit_cond_stmt): Likewise.
25882 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25884 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25885 body size unconditionally.
25887 2017-05-07 Jeff Law <law@redhat.com>
25890 2017-05-06 Jeff Law <law@redhat.com>
25891 PR tree-optimization/78496
25892 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25895 PR tree-optimization/78496
25896 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25897 (simplify_stmt_using_ranges): Call it.
25898 (vrp_dom_walker::before_dom_children): Extract equivalences
25899 from an ASSERT_EXPR with an equality comparison against a
25902 2017-05-06 Jeff Law <law@redhat.com>
25904 PR tree-optimization/78496
25905 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25908 PR tree-optimization/78496
25909 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25910 (simplify_stmt_using_ranges): Call it.
25911 (vrp_dom_walker::before_dom_children): Extract equivalences
25912 from an ASSERT_EXPR with an equality comparison against a
25915 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25917 * lra-constraints.c (lra_copy_reg_equiv): New function.
25918 (split_reg): Use it to copy equivalence information from the
25919 original register to the spill register.
25921 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25923 PR rtl-optimization/75964
25924 * simplify-rtx.c (simplify_const_relational_operation): Remove
25925 invalid handling of comparisons of integer ABS.
25927 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25929 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25930 initialize to zero.
25931 (init_regs): Remove declaration.
25932 (function_arg_advance_32): Initialize error_p as boolean variable.
25934 2017-05-05 Nathan Sidwell <nathan@acm.org>
25936 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25937 lines. Use for (;;).
25939 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25941 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25942 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25943 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25944 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25945 VF=2 that require versioning.
25947 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25949 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25952 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25954 * diagnostic.h (diagnostic_override_option_index): Convert from
25955 macro to inline function.
25957 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25959 * diagnostic.c (last_module_changed_p): New function.
25960 (set_last_module): New function.
25961 (diagnostic_report_current_module): Convert macro usage to
25962 the above functions.
25963 * diagnostic.h (diagnostic_context::last_module): Strengthen
25964 from const line_map * to const line_map_ordinary *.
25965 (diagnostic_last_module_changed): Delete macro.
25966 (diagnostic_set_last_module): Delete macro.
25968 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25970 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25971 with diagnostic_report_diagnostic.
25972 (diagnostic_n_impl_richloc): Likewise.
25973 * diagnostic.h (report_diagnostic): Delete macro.
25974 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25975 with diagnostic_report_diagnostic.
25976 * substring-locations.c (format_warning_va): Likewise.
25978 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25980 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25981 save/restor of format_spec. Move option-printing code to...
25982 (print_option_information): ...this new function, and
25983 reimplement by simply printing to the pretty_printer,
25984 rather than appending to the format string.
25986 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25988 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25989 handling logic into...
25990 (update_effective_level_from_pragmas): ...this new function.
25992 2017-05-04 Andrew Waterman <andrew@sifive.com>
25994 * config/riscv/riscv.opt (mstrict-align): New option.
25995 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25996 (SLOW_UNALIGNED_ACCESS): Define.
25997 (riscv_slow_unaligned_access): Declare.
25998 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26000 (riscv_slow_unaligned_access): New variable.
26001 (rocket_tune_info): Set slow_unaligned_access to true.
26002 (optimize_size_tune_info): Set slow_unaligned_access to false.
26003 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26004 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26005 (riscv_option_override): Set riscv_slow_unaligned_access.
26006 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26008 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26010 * config/riscv/riscv.md: Unify indentation.
26012 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26017 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26019 (extendsi<mode>2): Add support for doing sign extension via
26020 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26021 don't have ISA 3.0 instructions.
26022 (extendsi<mode>2 splitter): Likewise.
26023 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26024 generate the normal insns since SImode can now go in vector
26025 registers. Disallow the special UNSPECs needed for previous
26026 machines to hide SImode being used. Add new insns
26027 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26028 (fix_trunc<mode>si2_stfiwx): Likewise.
26029 (fix_trunc<mode>si2_internal): Likewise.
26030 (fixuns_trunc<mode>si2): Likewise.
26031 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26032 (fctiw<u>z_<mode>_smallint): Likewise.
26033 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26034 of floating point to 32-bit integer from doing a direct move to
26035 the GPR registers to do a store.
26036 (fctiwz_<mode>): Break long line.
26038 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26040 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26041 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26042 (addr_list, addr_offset_valid_p): New.
26043 (split_address_groups): Check offset validity with above function.
26044 (gt-tree-ssa-loop-ivopts.h): Include header file.
26046 2017-05-05 Nathan Sidwell <nathan@acm.org>
26048 * config.gcc (arm*-*-*): Add missing 'fi'.
26050 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26052 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26053 included in -fopt-info does not matter.
26054 * doc/optinfo.texi (-fopt-info): Fix description of default
26055 behavour. Explicitly say order of options included in -fopt-info
26058 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26060 * config.gcc: Allow combinations of aprofile and rmprofile values for
26061 --with-multilib-list.
26062 * config/arm/t-multilib: New file.
26063 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26064 variables. Remove setting of ISA and floating-point ABI in
26065 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26066 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26067 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26068 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26070 * config/arm/t-rmprofile: Likewise except for the matches changes.
26071 * doc/install.texi (--with-multilib-list): Document the combination of
26072 aprofile and rmprofile values and warn about pitfalls in doing that.
26074 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26076 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26077 (movdi_aarch64): Likewise.
26079 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26081 PR tree-optimization/80632
26082 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26084 (build_arrays): Initialize it for virtual phis.
26085 (fix_phi_nodes): Use it for virtual phis.
26087 PR tree-optimization/80558
26088 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26089 [x, y] op z into [x op, y op z] for op & or | if conditions
26092 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26093 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26096 * config/arm/arm.md (use_literal_pool): Remove.
26097 (64-bit immediate split): No longer takes cost into consideration
26098 if arm_disable_literal_pool is enabled.
26099 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26100 used when arm_disable_literal_pool is enabled.
26101 (arm_max_const_double_inline_cost): Remove use of
26102 arm_disable_literal_pool.
26103 (push_minipool_fix): Add assert.
26104 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26105 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26106 (no_literal_pool_sf_immediate): New.
26108 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26110 PR tree-optimization/80613
26111 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26112 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26114 2017-05-05 Richard Biener <rguenther@suse.de>
26116 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26118 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26120 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26121 of this flag from insn conditions due to removal from r247495.
26123 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26125 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26127 (arm_early_store_addr_dep_ptr): Likewise.
26128 * config/arm/aarch-common-protos.h
26129 (arm_early_load_addr_dep_ptr): Add prototype.
26130 (arm_early_store_addr_dep_ptr): Likewise.
26131 * config/arm/cortex-a53.md: Add new bypasses.
26133 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26135 * tree.c (next_type_uid): Change type to unsigned.
26136 (type_hash_canon): Decrement back next_type_uid if
26137 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26138 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26141 2017-05-04 Martin Sebor <msebor@redhat.com>
26143 * builtins.c: Fix a trivial typo in a comment.
26145 PR middle-end/79234
26146 * builtins.c (check_sizes): Adjust to handle reading past the end.
26147 Avoid printing excessive upper bound of ranges. Use %E to print
26148 tree nodes instead of converting them to %wu.
26149 (expand_builtin_memchr): New function.
26150 (compute_dest_size): Rename...
26151 (compute_objsize): ...to this.
26152 (expand_builtin_memcpy): Adjust.
26153 (expand_builtin_mempcpy): Adjust.
26154 (expand_builtin_strcat): Adjust.
26155 (expand_builtin_strcpy): Adjust.
26156 (check_strncat_sizes): Adjust.
26157 (expand_builtin_strncat): Adjust.
26158 (expand_builtin_strncpy): Adjust and simplify.
26159 (expand_builtin_memset): Adjust.
26160 (expand_builtin_bzero): Adjust.
26161 (expand_builtin_memcmp): Adjust.
26162 (expand_builtin): Handle memcmp.
26163 (maybe_emit_chk_warning): Check strncat just once.
26165 2017-05-04 Martin Sebor <msebor@redhat.com>
26167 PR preprocessor/79214
26168 PR middle-end/79222
26169 PR middle-end/79223
26170 * builtins.c (check_sizes): Add inlining context and issue
26171 warnings even when -Wno-system-headers is set.
26172 (check_strncat_sizes): Same.
26173 (expand_builtin_strncat): Same.
26174 (expand_builtin_memmove): New function.
26175 (expand_builtin_stpncpy): Same.
26176 (expand_builtin): Handle memmove and stpncpy.
26178 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26180 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26181 which is not used any more.
26183 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26185 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26187 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26189 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26190 (cortexa53_tunings): Likewise.
26191 (cortexa57_tunings): Likewise.
26192 (cortexa72_tunings): Likewise.
26193 (cortexa73_tunings): Likewise.
26195 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26197 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26198 Set loop alignment to 8.
26200 2017-05-04 Martin Sebor <msebor@redhat.com>
26202 PR translation/80280
26203 * builtins.c (expand_builtin_object_size): Add missing quoting to
26204 %D and like directives.
26205 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26206 (hsa_type_for_tree_type): Same.
26207 (verify_function_arguments): Same.
26208 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26209 * varasm.c (get_section): Same.
26212 2017-05-04 Martin Sebor <msebor@redhat.com>
26214 PR translation/80280
26215 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26217 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26219 * config/aarch64/aarch64.c (generic_addrcost_table):
26220 Change HI/TI mode setting.
26222 2017-05-04 Martin Jambor <mjambor@suse.cz>
26224 PR tree-optimization/80622
26225 * tree-sra.c (comes_initialized_p): New function.
26226 (build_accesses_from_assign): Only set write lazily when
26227 comes_initialized_p is false.
26228 (analyze_access_subtree): Use comes_initialized_p.
26229 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26230 instead of testing for PARM_DECL.
26232 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26234 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26235 constraint on operand 0 to allow more general addressing modes.
26236 Adjust output template.
26237 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26239 * config/aarch64/aarch64-protos.h
26240 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26241 * config/aarch64/constraints.md (Dp): New address constraint.
26242 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26245 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26247 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26248 update use of estimate_ipcp_clone_size_and_time.
26249 (estimate_local_effects): Update use of
26250 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26251 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26252 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26253 Return nonspecialized time.
26255 2017-05-04 Richard Biener <rguenther@suse.de>
26257 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26258 for the last VUSE which def dominates the PHI. Directly call
26260 (get_continuation_for_phi_1): Remove.
26262 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26264 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26265 to explain the use of truncating division. Cap the number of
26266 iterations to the maximum given by nb_iterations_upper_bound,
26269 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26271 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26272 * configure: Regenerate.
26273 * config.in: Regenerate.
26274 * config/i386/driver-mingw32.c: new file.
26275 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26276 * config.host: Link driver-mingw32.o on MinGW host.
26277 * doc/install.texi: Document new --enable-mingw-wildcard configure
26280 2017-05-04 Marek Polacek <polacek@redhat.com>
26282 PR tree-optimization/80612
26283 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26285 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26286 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26288 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26289 (movt splitter): Likewise.
26290 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26291 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26292 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26293 block for Thumb-1 with MOVT.
26294 (thumb2_legitimate_address_p): Move code block ...
26295 (can_avoid_literal_pool_for_label_p): ... into this new function.
26296 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26298 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26299 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26300 "M-profile targets with the MOVT instruction".
26302 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26304 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26305 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26306 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26308 2017-05-04 Martin Liska <mliska@suse.cz>
26310 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26311 variable cond_code.
26313 2017-05-04 Richard Biener <rguenther@suse.de>
26315 * tree.c (array_at_struct_end_p): Handle arrays at struct
26316 end with flexarrays more conservatively. Refactor and treat
26317 arrays of arrays or aggregates more strict. Fix
26318 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26319 * tree.c (array_at_struct_end_p): Adjust prototype.
26320 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26321 * gimple-fold.c (get_range_strlen): Likewise.
26322 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26324 2017-05-04 Richard Biener <rguenther@suse.de>
26326 PR tree-optimization/31130
26327 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26329 (supports_overflow_infinity): Likewise.
26330 (is_negative_overflow_infinity): Likewise.
26331 (is_positive_overflow_infinity): Likewise.
26332 (is_overflow_infinity): Likewise.
26333 (stmt_overflow_infinity): Likewise.
26334 (overflow_infinity_range_p): Likewise.
26335 (usable_range_p): Remove as always returning true.
26336 (make_overflow_infinity): Remove.
26337 (negative_overflow_infinity): Likewise.
26338 (positive_overflow_infinity): Likewise.
26339 (avoid_overflow_infinity): Likewise.
26340 (set_value_range): Adjust accordingly.
26341 (set_value_range_to_nonnegative): Likewise, remove now unused
26342 overflow_infinity arg.
26343 (vrp_operand_equal_p): Adjust.
26344 (update_value_range): Likewise.
26345 (range_int_cst_singleton_p): Likewise.
26346 (operand_less_p): Likewise.
26347 (compare_values_warnv): Likewise.
26348 (extract_range_for_var_from_comparison_expr): Likewise.
26349 (vrp_int_const_binop): Likewise.
26350 (zero_nonzero_bits_from_vr): Likewise.
26351 (extract_range_from_multiplicative_op_1): Likewise.
26352 (extract_range_from_binary_expr_1): Likewise.
26353 (extract_range_from_unary_expr): Likewise.
26354 (extract_range_from_comparison): Likewise.
26355 (extract_range_basic): Likewise.
26356 (adjust_range_with_scev): Likewise.
26357 (compare_ranges): Likewise.
26358 (compare_range_with_value): Likewise.
26359 (dump_value_range): Likewise.
26360 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26362 (simplify_cond_using_ranges): Adjust.
26364 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26366 * brig-builtins.def: Added a builtin for class_f64.
26367 * builtin-types.def: Added a builtin type needed by class_f64.
26369 2017-05-03 Jason Merrill <jason@redhat.com>
26371 * timevar.def: Add TV_CONSTEXPR.
26373 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26375 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26377 2017-05-03 Martin Jambor <mjambor@suse.cz>
26379 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26380 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26381 * ipa-cp.c (ipcp_propagate_stage): Do not call
26382 ipa_update_after_lto_read.
26383 * ipa-inline.c (ipa_inline): Likewise.
26385 2017-05-03 Martin Jambor <mjambor@suse.cz>
26387 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26388 tag. Added a default constructor and a destructor.
26389 (ipa_edge_args_sum_t): New class;
26390 (ipa_edge_args_sum): Declare.
26391 (ipa_edge_args_vector): Remove declaration.
26392 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26393 (ipa_free_edge_args_substructures): Remove declaration.
26394 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26395 (ipa_edge_args_info_available_for_edge_p): Likewise.
26396 * ipa-prop.c (ipa_edge_args_vector): Removed.
26397 (edge_removal_hook_holder): Likewise.
26398 (edge_duplication_hook_holder): Likewise.
26399 (ipa_edge_args_sum): New variable.
26400 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26401 ipa_edge_args_vector.
26402 (ipa_free_edge_args_substructures): Likewise.
26403 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26404 ipa_edge_args_vector.
26405 (ipa_edge_removal_hook): Turned into method
26406 ipa_edge_args_sum_t::remove.
26407 (ipa_edge_duplication_hook): Turned into method
26408 ipa_edge_args_sum_t::duplicate.
26409 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26410 registering edge hooks.
26411 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26412 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26413 ipa_edge_args_sum instead of ipa_edge_args_vector.
26414 * ipa-profile.c (ipa_profile): Likewise.
26416 2017-05-03 Martin Jambor <mjambor@suse.cz>
26418 * symbol-summary.h (function_summary): New method exists.
26419 (function_summary::symtab_removal): Deallocate through release.
26420 (call_summary): New class.
26421 (gt_ggc_mx): New overload.
26422 (gt_pch_nx): Likewise.
26423 (gt_pch_nx): Likewise.
26425 2017-05-03 Jeff Law <law@redhat.com>
26427 PR tree-optimization/78496
26428 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26429 from simplify_cond_using_ranges. Split off code to walk
26430 backwards through casts into ...
26431 (simplify_cond_using_ranges_2): New function.
26432 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26433 (execute_vrp): After identifying jump threads, call
26434 simplify_cond_using_ranges_2.
26436 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26439 * ipa-inline.h (inline_summary): Add ctor.
26440 (create_ggc): Do not use ggc_cleared_alloc.
26442 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26443 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26445 * gcc.c (handle_braces): Support escaping in switch matching
26447 * doc/invoke.texi (Spec Files): Document it.
26448 Remove superfluous @code markup in items.
26450 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26452 * diagnostic-show-locus.c (struct column_range): New struct.
26453 (get_affected_columns): New function.
26454 (get_printed_columns): New function.
26455 (struct correction): New struct.
26456 (correction::ensure_capacity): New function.
26457 (correction::ensure_terminated): New function.
26458 (struct line_corrections): New struct.
26459 (line_corrections::~line_corrections): New dtor.
26460 (line_corrections::add_hint): New function.
26461 (layout::print_trailing_fixits): Reimplement in terms of the new
26463 (selftest::test_overlapped_fixit_printing): New function.
26464 (selftest::diagnostic_show_locus_c_tests): Call it.
26466 2017-05-03 Nathan Sidwell <nathan@acm.org>
26468 Canonicalize canonical type hashing
26469 * tree.h (type_hash_canon_hash): Declare.
26470 * tree.c (type_hash_list, attribute_hash_list): Move into
26471 type_hash_canon_hash.
26472 (build_type_attribute_qual_variant): Break out hash code calc into
26473 type_hash_canon_hash.
26474 (type_hash_canon_hash): New. Generic type hash computation.
26475 (build_range_type_1, build_array_type_1, build_function_type,
26476 build_method_type_directly, build_offset_type, build_complex_type,
26477 make_vector_type): Call it.
26479 2017-05-03 Richard Biener <rguenther@suse.de>
26481 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26482 When all DRs have unknown misaligned do not always peel
26483 when there is a store but apply the same costing model as if
26484 there were only loads.
26486 2017-05-03 Richard Biener <rguenther@suse.de>
26489 PR tree-optimization/80492
26490 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26491 compare_base_decls returning dont-know properly.
26493 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26495 * config/arm/iterators.md (CCSI): New mode iterator.
26496 (arch): New mode attribute.
26497 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26498 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26499 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26500 code iterator for success result mode.
26501 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26502 the corresponding new insn generators.
26504 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26507 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26508 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26510 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26512 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26513 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26514 (DDR_A): Wrap DDR argument in brackets.
26515 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26516 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26517 (DDR_REVERSED_P): Likewise.
26519 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26521 PR tree-optimization/79472
26522 * tree-switch-conversion.c (struct switch_conv_info): Add
26523 contiguous_range and default_case_nonstandard fields.
26524 (collect_switch_conv_info): Compute contiguous_range and
26525 default_case_nonstandard fields, don't clear final_bb if
26526 contiguous_range and only the default case doesn't have the required
26528 (check_all_empty_except_final): Set default_case_nonstandard instead
26529 of failing if contiguous_range and the default case doesn't have empty
26531 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26532 and only the default case doesn't have the required constants. Skip
26534 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26535 if default_case_nonstandard.
26536 (build_constructors): Build constant 1 just once. Assert that default
26537 values aren't inserted in between cases if contiguous_range. Skip
26539 (build_arrays): Skip virtual phis.
26540 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26541 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26542 Handle virtual phis.
26543 (gen_inbound_check): Handle default_case_nonstandard case.
26544 (process_switch): Adjust check_final_bb caller. Call
26545 gather_default_values with the first non-default case instead of
26546 default case if default_case_nonstandard.
26548 2017-05-02 Nathan Sidwell <nathan@acm.org>
26550 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26551 check. Fix formatting.
26553 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26555 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26556 errors when comparing specialized and unspecialized times.
26558 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26560 * diagnostic-show-locus.c
26561 (layout::should_print_annotation_line_p): Make private.
26562 (layout::print_annotation_line): Make private.
26563 (layout::annotation_line_showed_range_p): Make private.
26564 (layout::show_ruler): Make private.
26565 (layout::print_source_line): Make private. Pass in line and
26566 line_width, rather than calling location_get_source_line. Drop
26568 (layout::print_leading_fixits): New method.
26569 (layout::print_any_fixits): Rename to...
26570 (layout::print_trailing_fixits): ...this, and make private.
26571 Don't print newline fixits.
26572 (diagnostic_show_locus): Move logic for printing one row into...
26573 (layout::print_line): ...this new function. Move the
26574 location_get_source_line call and error-handling from
26575 print_source_line to here. Call print_leading_fixits, and rename
26576 print_any_fixits to print_trailing_fixits.
26577 (selftest::test_fixit_insert_containing_newline): Update now that
26578 newlines are partially supported.
26579 (selftest::test_fixit_insert_containing_newline_2): New test.
26580 (selftest::test_fixit_replace_containing_newline): Update comments.
26581 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26582 * edit-context.c (class added_line): New class.
26583 (class edited_line): Describe newline handling in comment.
26584 (edited_line::actually_edited_p): New method.
26585 (edited_line::print_content): Delete redundant decl.
26586 (edited_line::m_predecessors): New field.
26587 (edited_file::print_content): Call edited_line::print_content.
26588 (edited_file::print_diff): Update to support newlines.
26589 (edited_file::print_diff_hunk): Likewise.
26590 (edited_file::print_run_of_changed_lines): New function.
26591 (edited_file::print_diff_line): Convert to...
26592 (print_diff_line): ...this.
26593 (edited_file::get_effective_line_count): New function.
26594 (edited_line::edited_line): Initialize new field m_predecessors.
26595 (edited_line::~edited_line): Clean up m_predecessors.
26596 (edited_line::apply_fixit): Handle newlines.
26597 (edited_line::get_effective_line_count): New function.
26598 (edited_line::print_content): New function.
26599 (edited_line::print_diff_lines): New function.
26600 (selftest::test_applying_fixits_insert_containing_newline): New
26602 (selftest::test_applying_fixits_replace_containing_newline): New
26604 (selftest::insert_line): New function.
26605 (selftest::test_applying_fixits_multiple_lines): Add example of
26607 (selftest::edit_context_c_tests): Call the new tests.
26609 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26611 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26612 parameter cand. Update dump information.
26613 (get_computation_cost): Update uses.
26615 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26617 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26618 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26619 (get_computation_at, rewrite_use_address): Update use of
26620 get_computation_aff.
26622 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26624 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26625 (get_computation): Delete.
26626 (get_computation_cost): Implement like get_computation_cost_at.
26627 Use get_computation_at.
26628 (get_computation_cost_at): Delete.
26629 (rewrite_use_nonlinear_expr): Use get_computation_at.
26630 (rewrite_use_compare, remove_unused_ivs): Ditto.
26632 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26634 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26636 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26638 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26639 (ivopts_global_cost_for_size): Rename parameter and update uses.
26640 (iv_ca_recount_cost): Update uses.
26641 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26642 candidates seperately in n_invs and n_cands.
26643 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26645 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26647 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26648 (find_inv_vars_cb): New.
26649 (find_depends): Renamed to ...
26650 (find_inv_vars): ... this.
26651 (add_candidate_1, force_var_cost): Call find_inv_vars.
26652 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26654 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26656 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26657 inv_vars. Add inv_exprs.
26658 (struct iv_cand): Rename depends_on to inv_vars.
26659 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26660 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26661 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26662 (dump_cand): Dump inv_vars.
26663 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26664 (record_invariant, find_depends, add_candidate_1): Ditto.
26665 (set_group_iv_cost, force_var_cost): Ditto.
26666 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26667 (get_computation_cost_at, get_computation_cost): Ditto.
26668 (determine_group_iv_cost_generic): Ditto.
26669 (determine_group_iv_cost_address): Ditto.
26670 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26671 (determine_group_iv_costs): Ditto.
26672 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26673 (iv_ca_set_remove_invariants): Renamed to ...
26674 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26675 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26676 (iv_ca_set_add_invariants): Renamed to ...
26677 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26678 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26679 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26680 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26681 (create_new_ivs): Remove useless dump.
26683 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26685 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26687 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26688 (iv_ca_set_no_cp, create_new_iv): Ditto.
26690 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26692 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26694 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26696 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26697 function tree_check2.
26699 2017-05-02 Martin Liska <mliska@suse.cz>
26701 * doc/gcov.texi: Add missing preposition.
26702 * gcov.c (function_info::function_info): Properly fill up
26703 all member variables.
26705 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26707 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26709 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26711 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26713 2017-05-02 Martin Liska <mliska@suse.cz>
26716 * lto-streamer-in.c (lto_read_tree_1): Remove
26717 LTO_STREAMER_DEBUG.
26718 * lto-streamer.c (struct tree_hash_entry): Likewise.
26719 (struct tree_entry_hasher): Likewise.
26720 (tree_entry_hasher::hash): Likewise.
26721 (tree_entry_hasher::equal): Likewise.
26722 (lto_streamer_init): Likewise.
26723 (lto_orig_address_map): Likewise.
26724 (lto_orig_address_get): Likewise.
26725 (lto_orig_address_remove): Likewise.
26726 * lto-streamer.h: Likewise.
26727 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26728 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26730 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26732 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26733 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26734 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26735 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26736 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26737 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26738 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26739 (mm_maskz_sub_ss): New intrinsics.
26740 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26741 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26742 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26743 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26744 (__builtin_ia32_subss_mask_round): New builtins.
26745 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26746 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26747 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26749 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26750 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26752 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26755 2017-05-02 Martin Jambor <mjambor@suse.cz>
26757 PR tree-optimization/78687
26758 * tree-sra.c (access): New field parent.
26759 (process_subtree_disqualification): New function.
26760 (disqualify_candidate): Call it.
26761 (build_accesses_from_assign): Reset write flag if creating an
26763 (build_access_subtree): Fill in parent field and also prpagate
26764 down grp_write flag.
26765 (create_artificial_child_access): New parameter set_grp_write, set
26766 grp_write to its value.
26767 (propagate_subaccesses_across_link): Also propagate grp_write flag
26769 (propagate_all_subaccesses): Push the closest parent back to work
26770 queue if add_access_to_work_queue returned true.
26772 2017-05-02 Richard Biener <rguenther@suse.de>
26774 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26775 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26776 -fstrict-overflow documentation.
26777 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26778 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26779 flag_strict_overflow.
26780 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26781 * lto-opts.c (lto_write_options): Do not stream it.
26782 * lto-wrapper.c (merge_and_complain): Do not handle it.
26783 * opts.c (default_options_table): Do not set -fstrict-overflow.
26784 (finish_options): Likewise do not clear it when sanitizing.
26785 * simplify-rtx.c (simplify_const_relational_operation): Do not
26786 test flag_strict_overflow.
26788 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26790 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26791 using enabled attribute.
26792 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26793 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26794 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26795 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26796 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26797 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26798 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26799 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26800 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26801 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26803 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26805 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26807 2017-05-02 Richard Biener <rguenther@suse.de>
26809 PR tree-optimization/80591
26811 2017-04-10 Richard Biener <rguenther@suse.de>
26813 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26816 2017-05-02 Richard Biener <rguenther@suse.de>
26818 PR tree-optimization/80549
26819 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26820 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26821 headers if they do not have a preheader.
26823 2017-05-02 Martin Liska <mliska@suse.cz>
26826 * common.opt: Fix typo.
26827 * doc/invoke.texi: Likewise.
26829 2017-05-01 Jan Beulich <jbeulich@suse.com>
26831 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26832 swapping, add (x,x,m,x,n) alternative.
26834 2017-05-01 Nathan Sidwell <nathan@acm.org>
26836 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26837 unnecessary unadjusted_alignment check.
26839 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26842 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26844 * gimplify.c (gimplify_cilk_detach): New function.
26845 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26846 * tree-core.h: Document EXPR_CILK_SPAWN.
26847 * tree.h (EXPR_CILK_SPAWN): Define.
26849 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26851 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26852 to use new fixit_hint representation, using the "replace" logic.
26853 (get_line_span_for_fixit_hint): Likewise.
26854 (layout::print_any_fixits): Likewise.
26855 (selftest::test_one_liner_many_fixits): Rename to...
26856 (selftest::test_one_liner_many_fixits_1): ...this, and update
26857 comment and expected output to reflect that the multiple fix-it
26858 hints are now consolidated into one insertion.
26859 (selftest::test_one_liner_many_fixits_2): New test.
26860 (selftest::test_diagnostic_show_locus_one_liner): Update for
26862 (selftest::test_fixit_consolidation): Update for fix-it API
26864 * diagnostic.c (print_parseable_fixits): Likewise.
26865 * edit-context.c (edited_line::m_line_events): Convert from
26866 auto_vec <line_event *> to auto_vec <line_event>.
26867 (class line_event): Convert from abstract base class to a concrete
26868 class, taking over the role of replace_event.
26869 (class insert_event): Delete.
26870 (class replace_event): Rename to class line_event. Convert to
26872 (edit_context::add_fixits): Reimplement.
26873 (edit_context::apply_insert): Delete.
26874 (edit_context::apply_replace): Rename to...
26875 (edit_context::apply_fixit): ...this. Convert to half-open range.
26876 (edited_file::apply_insert): Delete.
26877 (edited_file::apply_replace): Rename to...
26878 (edited_file::apply_fixit): ...this.
26879 (edited_line::~edited_line): Drop deletion of events.
26880 (edited_line::apply_insert): Delete.
26881 (edited_line::apply_replace): Rename to...
26882 (edited_line::apply_fixit): ...this. Convert to half-open range.
26883 Update for change to type of m_line_events.
26884 * edit-context.h (edit_context::apply_insert): Delete.
26885 (edit_context::apply_replace): Rename to...
26886 (edit_context::apply_fixit): ...this.
26888 2017-05-01 Martin Sebor <msebor@redhat.com>
26890 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26893 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26896 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26897 __get_cpuid_max returns 0.
26898 (__get_cpuid_count): Ditto.
26900 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26902 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26903 replacement expression is another instance of one of its arguments.
26905 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26908 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26909 check for stack push/pop autoinc.
26910 * config/i386/i386.c (ix86_agi_dependent): Return false
26911 if the only reason why modified_in_p returned true is that
26912 addr is SP based and set_insn is a push or pop.
26914 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26916 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26919 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26922 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26923 (account_size_time): Use two predicates - exec_pred and
26925 (evaluate_conditions_for_known_args): Compute both clause and
26927 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26928 (inline_summary_t::duplicate): Update.
26929 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26931 (compute_inline_parameters): Likewise.
26932 (estimate_edge_size_and_time): Update caluclation of time.
26933 (estimate_node_size_and_time): Compute both time and nonspecialized
26935 (estimate_ipcp_clone_size_and_time): Update.
26936 (inline_merge_summary): Update.
26937 (do_estimate_edge_time): Update.
26938 (do_estimate_edge_size): Update.
26939 (do_estimate_edge_hints): Update.
26940 (inline_read_section, inline_write_summary): Stream both new predicates.
26941 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26943 (compute_inlined_call_time): Cleanup.
26944 (big_speedup_p): Update.
26945 (edge_badness): Update.
26946 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26947 (size_time_entry): Replace predicate by exec_predicate and
26948 nonconst_predicate.
26949 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26950 (estimate_edge_time): Return also nonspec_time.
26951 (reset_edge_growth_cache): Update.
26953 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26955 PR rtl-optimization/80491
26956 * ifcvt.c (noce_process_if_block): When looking for x setter
26957 with missing else_bb, don't check only the insn right before
26958 cond_earliest, but look for the last insn that x is modified in
26959 within the same bb.
26961 PR rtl-optimization/80491
26962 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26964 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26966 PR tree-optimization/80487
26967 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26969 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26971 PR tree-optimization/79697
26972 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26973 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26974 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26976 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26977 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26979 2017-04-28 Martin Sebor <msebor@redhat.com>
26981 PR tree-optimization/80523
26982 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26983 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26985 (maybe_warn, format_directive, parse_directive): Use new functions.
26986 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26988 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26990 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26992 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26994 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26995 target_header_dir): Set correctly.
26996 * configure: Regenerated.
26997 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26998 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26999 instead of SYSTEM_HEADER_DIR.
27001 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27003 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27004 (estimate_local_effects): Likewise.
27005 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27006 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27007 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27008 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27009 do_estimate_edge_time, estimate_edge_time): Likewise.
27010 * ipa-inline-analysis.c (estimate_node_size_and_time,
27011 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27012 (estimate_time_after_inlining): Remove.
27014 2017-04-28 Martin Liska <mliska@suse.cz>
27016 * doc/gcov.texi: Enhance documentation of gcov.
27018 2017-04-28 Martin Liska <mliska@suse.cz>
27020 * doc/gcov.texi: Sort options in alphabetic order.
27021 * doc/gcov-dump.texi: Likewise.
27022 * doc/gcov-tool.texi: Likewise.
27023 * gcov.c (print_usage): Likewise.
27024 * gcov-dump.c (print_usage): Likewise.
27025 * gcov-tool.c (print_merge_usage_message): Likewise.
27026 (print_rewrite_usage_message): Likewise.
27027 (print_overlap_usage_message): Likewise.
27029 2017-04-28 Martin Liska <mliska@suse.cz>
27031 PR gcov-profile/53915
27032 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27034 2017-04-28 Martin Liska <mliska@suse.cz>
27036 PR gcov-profile/79891
27037 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27038 is marked by compiler as living on a line.
27039 (get_cycles_count): Remove usage of the union.
27040 (output_intermediate_file): Likewise.
27041 (find_source): Fix GNU coding style.
27042 (accumulate_line_counts): Remove old non-all block mode.
27043 (output_lines): Remove usage of the union.
27044 * profile.c (output_location): Include all BBs, even if
27045 belonging to a same line (and file) as a previous BB.
27047 2017-04-28 Martin Liska <mliska@suse.cz>
27049 * gcov.c (process_args): Handle new argument 'w'.
27050 (read_graph_file): Assign ID to BBs.
27051 (output_branch_count): Display BB # if verbose flag is set.
27052 (output_lines): Likewise for arcs.
27053 (print_usage): Add '--verbose' option help.
27054 * doc/gcov.texi: Document --verbose (-w) option.
27056 2017-04-28 Martin Liska <mliska@suse.cz>
27058 * gcov.c (struct block_location_info): New struct.
27059 (process_file): Fill up the new structure.
27060 (read_graph_file): Replace usage of encoding by the newly added
27062 (add_line_counts): Likewise.
27063 (accumulate_line_counts): Remove usage of the union.
27064 (function_info::function_info): New function.
27065 (function_info::~function_info): Likewise.
27066 (process_file): Call delete instead of release_function.
27067 (release_function): Release the function.
27068 (release_structures): Call delete instead of release_function.
27069 (solve_flow_graph): Replace usage of num_blocks.
27070 (find_exception_blocks): Likewise.
27071 (output_lines): Fix GNU coding style.
27073 2017-04-28 Martin Liska <mliska@suse.cz>
27076 * coverage.c (coverage_remove_note_file): New function.
27077 * coverage.h: Declare the function.
27078 * toplev.c (finalize): Clean if an error has been seen.
27080 2017-04-28 Martin Liska <mliska@suse.cz>
27082 PR gcov-profile/80031
27083 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27084 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27085 * gcov.c (read_graph_file): Read just number of blocks.
27086 * profile.c (branch_prob): Do not stream 0 flags per a basic
27089 2017-04-28 Martin Liska <mliska@suse.cz>
27091 * gcov-dump.c (tag_*): Add new argument to declarations.
27092 (dump_gcov_file): Likewise.
27093 (tag_blocks): Add and use new argument depth.
27094 (tag_arcs): Likewise.
27095 (tag_lines): Likewise.
27096 (tag_counters): Likewise.
27097 (tag_summary): Likewise.
27098 (dump_working_sets): Use depth to do a proper indentation.
27100 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27103 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27104 * symtab.c (symtab_node::debug_symtab): Move definition here.
27106 2017-04-28 Richard Biener <rguenther@suse.de>
27108 * lto-streamer.h (LTO_major_version): Bump to 7.
27110 2017-04-28 Richard Biener <rguenther@suse.de>
27112 * tree-vrp.c (assert_info): New struct.
27113 (add_assert_info): New helper.
27114 (register_edge_assert_for_2): Refactor to add asserts to a vector
27116 (register_edge_assert_for_1): Likewise.
27117 (register_edge_assert_for): Likewise.
27118 (finish_register_edge_assert_for): New helper actually registering
27119 asserts where live on edge.
27120 (find_conditional_asserts): Adjust.
27121 (find_switch_asserts): Likewise.
27122 (evrp_dom_walker::try_find_new_range): Generalize.
27123 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27125 2017-04-27 Marek Polacek <polacek@redhat.com>
27128 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27129 arg10 and arg11 to itype.
27131 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27133 * doc/extend.texi (Object Size Checking): Improve grammar.
27135 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27138 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27139 that the logic for permitting reciprocal estimates matches that
27142 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27145 * tree.c (type_cache_hasher::equal): Only compare
27146 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27147 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27148 non-aggregate element types.
27149 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27150 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27152 2017-04-27 Richard Biener <rguenther@suse.de>
27154 PR middle-end/80533
27155 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27156 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27157 keeping a reference to a trailing array.
27159 2017-04-27 Richard Biener <rguenther@suse.de>
27161 PR middle-end/80539
27162 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27163 being in loop-closed SSA form conservatively.
27164 (chrec_fold_multiply_poly_poly): Likewise.
27166 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27168 PR middle-end/79665
27169 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27170 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27172 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27175 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27176 (aarch64_function_arg_alignment): Return unsigned int again, but still
27177 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27178 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27179 Don't emit -Wpsabi note.
27180 (aarch64_function_arg_boundary): Likewise.
27181 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27184 2017-04-26 Nathan Sidwell <nathan@acm.org>
27186 * tree.h (crc32_unsigned_n): Declare.
27187 (crc32_unsigned, crc32_unsigned): Make inline.
27188 * tree.c (crc32_unsigned_bits): Replace with ...
27189 (crc32_unsigned_n): ... this.
27190 (crc32_unsigned, crc32_byte): Remove.
27191 (crc32_string): Remove unnecessary braces.
27193 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27195 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27196 * ipa-inline-analysis.c (MAX_TIME): Remove.
27197 (account_size_time): Use sreal for time.
27198 (dump_inline_summary): Update.
27199 (estimate_function_body_sizes): Update.
27200 (estimate_edge_size_and_time): Update.
27201 (estimate_calls_size_and_time): Update.
27202 (estimate_node_size_and_time): Update.
27203 (inline_merge_summary): Update.
27204 (inline_update_overall_summary): Update.
27205 (estimate_time_after_inlining): Update.
27206 (inline_read_section): Update.
27207 (inline_write_summary): Update.
27208 * ipa-inline.c (compute_uninlined_call_time): Update.
27209 (compute_inlined_call_time): Update.
27210 (recursive_inlining): Update.
27211 (inline_small_functions): Update.
27212 (dump_overall_stats): Update.
27213 * ipa-inline.h: Include sreal.h.
27214 (size_time_entry): Turn time to sreal.
27215 (inline_summary): Turn self_time nad time to sreal.
27217 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27219 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27221 (sreal::stream_out, sreal::stream_in): New.
27222 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27224 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27226 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27229 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27232 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27234 (dimode_scalar_chain::compute_convert_gain): Ditto.
27235 (dimode_scalar_chain::make_vector_copies): Ditto.
27236 (dimode_scalar_chain::convert_reg): Ditto.
27237 (dimode_scalar_chain::convert_insn): Ditto.
27238 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27239 (VI248_AVX512BW_1): New mode iterator.
27240 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27241 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27244 2017-04-25 Martin Sebor <msebor@redhat.com>
27246 PR tree-optimization/80497
27247 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27248 constants are representable in HOST_WIDE_INT.
27249 (parse_directive): Ditto.
27251 2017-04-25 Martin Sebor <msebor@redhat.com>
27254 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27255 (new_zero_array): Adjust signature.
27256 (dom_info::dom_init): Used unsigned rather that size_t.
27257 (dom_info::dom_info): Same.
27259 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27260 Jakub Jelinek <jakub@redhat.com>
27263 * config/arm/arm.c: Include gimple.h.
27264 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27265 returns negative, increment ncrn only if it returned positive.
27266 (arm_needs_doubleword_align): Return int instead of bool,
27267 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27268 members, but if there is any such non-FIELD_DECL
27269 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27270 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27271 returns negative, increment nregs only if it returned positive.
27272 (arm_setup_incoming_varargs): Likewise.
27273 (arm_function_arg_boundary): Emit -Wpsabi note if
27274 arm_needs_doubleword_align returns negative, return
27275 DOUBLEWORD_ALIGNMENT only if it returned positive.
27277 2017-04-25 Marek Polacek <polacek@redhat.com>
27280 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27281 first argument to type.
27283 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27286 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27287 type checks to test for compatibility instead of equality.
27289 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27290 Jakub Jelinek <jakub@redhat.com>
27293 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27295 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27296 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27297 the alignment computation, but return their maximum in warn_alignment.
27298 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27299 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27301 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27302 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27305 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27307 * config/arc/simdext.md (dmpyh): Fix typo.
27309 2017-04-25 Richard Biener <rguenther@suse.de>
27311 PR tree-optimization/80492
27312 * alias.c (compare_base_decls): Handle registers with asm
27313 specification conservatively.
27314 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27315 compare_base_decls returning dont-know properly.
27317 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27319 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27320 (legitimate_offset_address_p): New function.
27321 (arc_legitimate_address_p): Use above function.
27323 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27325 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27327 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27329 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27330 ACCH registers whenever they are available.
27332 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27334 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27335 double regs fix when not used.
27337 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27339 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27341 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27342 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27344 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27346 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27347 when emitting short ADD instructions.
27349 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27351 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27353 (cmpsi_cc_c_insn): Likewise.
27354 (cbranchsi4_scratch): Compute proper instruction length using
27355 compact_hreg_operand.
27356 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27358 2017-04-25 Richard Biener <rguenther@suse.de>
27360 PR middle-end/80509
27361 * passes.c (pass_manager::pass_manager): Initialize
27362 m_name_to_pass_map.
27364 2017-04-25 Richard Biener <rguenther@suse.de>
27366 PR tree-optimization/79201
27367 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27369 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27372 * config/s390/vector.md: Split MEM->GPR vector moves for
27373 non-s_operand addresses.
27375 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27378 * config/s390/predicates.md (reload_const_wide_int_operand): New
27380 * config/s390/s390.md ("movti"): Remove d/P alternative.
27381 ("movti_bigconst"): New pattern definition.
27383 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27386 * s390-protos.h (s390_expand_cs_hqi): Removed.
27387 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27388 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27389 modes as well as CCZ1mode and CCZmode.
27390 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27391 signature of s390_emit_compare_and_swap.
27392 (s390_expand_cs_hqi): Likewise, make static.
27393 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27394 compare-and-swap, in some cases.
27395 (s390_expand_cs): Wrapper function.
27396 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27398 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27399 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27400 patterns for small and large integers. Forbid symref memory operands.
27401 Move expander to s390.c. Require cc register.
27402 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27403 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27404 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27405 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27406 symref memory operands. Remove CC mode and call s390_match_ccmode
27408 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27410 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27412 * config/s390/s390.md (define_peephole2): New peephole to help
27413 combining the load-and-test pattern with volatile memory.
27415 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27417 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27418 with CCZmode for TARGET_Z196.
27420 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27422 PR rtl-optimization/80501
27423 * combine.c (make_compound_operation_int): Set subreg_code to SET
27424 even for AND with mask of the sign bit of mode.
27426 PR rtl-optimization/80500
27427 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27428 sum's initial value.
27430 2017-04-25 Julian Brown <julian@codesourcery.com>
27431 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27433 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27435 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27437 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27439 2017-04-25 Julian Brown <julian@codesourcery.com>
27440 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27442 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27443 (thunderx2t99_sha): New Reservation.
27445 2017-04-25 Julian Brown <julian@codesourcery.com>
27446 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27448 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27449 type for 1-element load.
27451 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27453 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27455 2017-04-24 Martin Jambor <mjambor@suse.cz>
27457 PR tree-optimization/80293
27458 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27459 char arrays not totally scalarizable if it is false.
27460 (analyze_all_variable_accesses): Pass correct value in the new
27461 parameter. Add a statistics counter.
27463 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27465 PR middle-end/79931
27466 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27468 2017-04-24 Richard Biener <rguenther@suse.de>
27470 PR tree-optimization/80494
27471 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27472 out for complex types.
27474 2017-04-24 Richard Biener <rguenther@suse.de>
27476 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27477 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27478 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27479 (DFS): Adjust and never fail.
27480 (sccvn_dom_walker::fail): Remove.
27481 (sccvn_dom_walker::before_dom_children): Adjust.
27482 (run_scc_vn): Likewise and never fail.
27483 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27484 (pass_fre::execute): Likewise.
27486 2017-04-24 Richard Biener <rguenther@suse.de>
27488 PR tree-optimization/79725
27489 * tree-ssa-sink.c (statement_sink_location): Return whether
27490 failure reason was zero uses. Move that check later.
27491 (sink_code_in_bb): Deal with zero uses by removing the stmt
27494 2017-04-24 Richard Biener <rguenther@suse.de>
27497 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27498 pointer-based references.
27500 2017-04-24 Richard Biener <rguenther@suse.de>
27503 * pass_manager.h (pass_manager::operator new): Remove.
27504 (pass_manager::operator delete): Likewise.
27505 * passes.c (pass_manager::operator new): Remove.
27506 (pass_manager::operator delete): Likewise.
27507 (pass_manager::pass_manager): Zero individual pass members.
27509 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27512 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27513 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27514 Check "XEXP (src, 1)" operand here.
27515 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27516 Check "XEXP (src, 1)" operand here.
27517 (dimode_scalar_chain::make_vector_copies): Detect count register
27518 of a shift instruction. Zero extend count register from QImode
27519 to DImode to satisfy vector shift pattern count operand predicate.
27520 Substitute vector shift count operand with a DImode copy.
27521 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27524 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27526 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27527 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27528 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27529 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27530 (UNSPEC_NOREX_MEM): Remove definition.
27532 2017-04-21 Richard Biener <rguenther@suse.de>
27534 PR tree-optimization/79547
27535 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27536 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27537 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27538 without any constraints.
27540 2017-04-21 Richard Biener <rguenther@suse.de>
27542 PR tree-optimization/78847
27543 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27545 2017-04-21 Richard Biener <rguenther@suse.de>
27547 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27548 (build_distinct_type_copy): Likewise.
27549 (build_variant_type_copy): Likewise.
27550 * tree.c (build_qualified_type): Pass down mem-stat info.
27551 (build_distinct_type_copy): Likewise.
27552 (build_variant_type_copy): Likewise.
27554 2017-04-21 Richard Biener <rguenther@suse.de>
27556 PR tree-optimization/80237
27557 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27559 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27560 for a simplified result.
27562 2016-04-21 Richard Biener <rguenther@suse.de>
27564 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27565 sth as strict as a simple_iv but a chrec without symbols and an
27566 operand defined in the loop we are peeling (and not some subloop).
27567 (propagate_constants_for_unrolling): Propagate all constants.
27569 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27572 * config/i386/i386.c (print_reg): Remove assert for disalowed
27573 regno values, call output_operand_lossage instead.
27575 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27578 * config/i386/constraints.md (Yc): New register constraint.
27579 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27580 Use Yc constraint for alternative 2 of operand 0. Remove
27581 preferred_for_speed attribute.
27583 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27585 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27586 lastprivate clauses in SIMT case.
27588 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27590 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27592 2017-04-20 Richard Biener <rguenther@suse.de>
27594 PR tree-optimization/57796
27595 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27596 as N scalar stores.
27597 (vect_model_load_cost): Cost gathers as N scalar loads.
27599 2017-04-20 Richard Biener <rguenther@suse.de>
27601 * ggc-page.c (ggc_allocated_p): Rename to ...
27602 (safe_lookup_page_table_entry): ... this and return the lookup
27604 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27606 2017-04-20 Richard Biener <rguenther@suse.de>
27608 PR tree-optimization/80453
27609 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27610 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27611 from the conditions.
27612 (vn_phi_eq): Pass them down.
27613 (vn_phi_lookup): Record them.
27614 (vn_phi_insert): Likewise.
27616 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27618 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27619 uninitialized variable warning to avoid buffer overrun.
27621 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27624 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27625 is suppressed for '{ 0 }' in C.
27627 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27629 * BASE-VER: Set to 8.0.0.
27631 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27633 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27634 priority .init_array and .fini_array section with SECTION_NOTYPE
27637 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27639 PR middle-end/80423
27640 * tree.h (build_array_type): Add typeless_storage default argument.
27641 * tree.c (type_cache_hasher::equal): Also compare
27642 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27643 (build_array_type): Add typeless_storage argument, set
27644 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27646 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27647 (build_array_type): Likewise. Add typeless_storage argument.
27649 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27650 Jakub Jelinek <jakub@redhat.com>
27652 PR tree-optimization/80426
27653 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27654 operation on symbolic operands, also compute the overflow for the
27655 invariant part when the operation degenerates into a negation.
27657 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27660 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27661 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27664 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27666 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27669 * config/avr/avr.c (tree.h): Include it.
27670 (cgraph.h): Include it.
27671 (avr_encode_section_info): Don't warn for uninitialized progmem
27672 variable if it's just an alias.
27674 2017-04-19 Richard Biener <rguenther@suse.de>
27677 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27678 when needed by AutoPGO.
27680 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27683 * doc/lto.texi: Remove an extra 'that'.
27685 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27687 PR rtl-optimization/80429
27688 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27689 are only used in debug insns.
27691 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27692 Vladimir Makarov <vmakarov@redhat.com>
27694 * config/sparc/predicates.md (input_operand): Add comment. Return
27695 true for any memory operand when LRA is in progress.
27696 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27698 2017-04-18 Jeff Law <law@redhat.com>
27701 * mips.md ({return,simple_return}_internal): Do not overwrite
27704 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27706 PR tree-optimization/80443
27707 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27708 instead of adding 1, subtract -1 and similarly instead of subtracting
27711 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27713 PR rtl-optimization/80357
27714 * haifa-sched.c (tmp_bitmap): New variable.
27715 (model_recompute): Handle duplicate use records.
27716 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27717 (free_global_sched_pressure_data): Free it.
27719 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27722 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27723 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27724 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27725 instead of SYSTEM_HEADER_DIR.
27727 2017-04-18 Jeff Law <law@redhat.com>
27729 PR middle-end/80422
27730 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27731 predecessors after walking up the insn chain.
27733 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27736 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27737 sizetype type into debug info.
27739 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27742 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27743 unneeded test for TARGET_UPPER_REGS_SF.
27744 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27746 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27749 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27750 instead of gsi_after_labels.
27752 2017-04-18 Jeff Law <law@redhat.com>
27754 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27758 2017-04-13 Jeff Law <law@redhat.com>
27759 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27760 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27762 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27765 * config/avr/avr.c (intl.h): Include it.
27766 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27768 2017-04-18 Martin Liska <mliska@suse.cz>
27770 PR gcov-profile/78783
27771 * gcov-tool.c (gcov_output_files): Validate that destination
27772 file is either removed by the tool or by a user.
27774 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27775 Guy Benyei <guybe@mellanox.com>
27777 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27778 block, and do not negate it, the stored id is already negative.
27780 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27782 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27784 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27787 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27788 masks of options that should be turned off if the VSX vector
27789 options are turned off.
27790 (OTHER_P8_VECTOR_MASKS): Likewise.
27791 (OTHER_VSX_VECTOR_MASKS): Likewise.
27792 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27793 rs6000_disable_incompatible_switches to validate no type switches
27795 (rs6000_incompatible_switch): New function to disallow turning on
27796 other vector options if -mno-vsx, -mno-power8-vector, or
27797 -mno-power9-vector are specified.
27799 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27801 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27803 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27805 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27806 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27807 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27808 (ARG_POINTER_CFA_OFFSET): Likewise.
27810 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27812 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27813 conditions to take advantage of various optimizations.
27815 2017-04-13 Jeff Law <law@redhat.com>
27817 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27818 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27819 (zero_extendsidi2_dext): Likewise.
27821 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27824 * fold-const.c (fold_ternary_loc): Revert
27825 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27828 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27830 PR rtl-optimization/80343
27831 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27832 new scratch pseudo.
27834 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27837 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27838 to ubsan_encode_value.
27840 2017-04-13 Jeff Law <law@redhat.com>
27842 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27843 appearing in DEBUG_INSNs.
27845 2017-04-13 Martin Liska <mliska@suse.cz>
27847 PR gcov-profile/80413
27848 * gcov-io.c (gcov_write_string): Copy to buffer just when
27849 allocated size is greater than zero.
27851 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27854 * dwarf2out.c (decls_for_scope): Ignore declarations of
27855 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27857 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27860 * ipa-visibility.c (non_local_p): Fix typos.
27861 (localize_node): When localizing symbol in same comdat group,
27862 dissolve the group only when we know external symbols are going
27864 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27866 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27868 PR tree-optimization/79390
27869 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27870 order does not result in usable sequence, retry with reversed operand
27876 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27877 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27878 op0 instead of fold_convert_loc (loc, type, arg0).
27880 2017-04-12 Jeff Law <law@redhat.com>
27882 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27883 has a delay slot in the generated code.
27885 * config/cris/cris.md (cris_preferred_reload_class): Return
27886 GENNONACR_REGS rather than GENERAL_REGS.
27888 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27891 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27892 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27893 signedness of the result type.
27895 2017-04-12 Richard Biener <rguenther@suse.de>
27896 Jeff Law <law@redhat.com>
27898 PR tree-optimization/80359
27899 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27900 trim stores to TARGET_MEM_REFs.
27902 2017-04-12 Richard Biener <rguenther@suse.de>
27904 PR tree-optimization/79390
27905 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27906 threading case even more.
27908 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27911 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27912 for quad_address_p for TImode, instead of just not indexed_address.
27914 2017-04-12 Richard Biener <rguenther@suse.de>
27915 Bernd Edlinger <bernd.edlinger@hotmail.de>
27917 PR middle-end/79671
27918 * alias.c (component_uses_parent_alias_set_from): Handle
27919 TYPE_TYPELESS_STORAGE.
27920 (get_alias_set): Likewise.
27921 * tree-core.h (tree_type_common): Add typeless_storage flag.
27922 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27923 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27924 for types containing members with TYPE_TYPELESS_STORAGE.
27925 (place_field): Likewise.
27926 (layout_type): Likewise for ARRAY_TYPE.
27927 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27928 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27929 TYPE_TYPELESS_STORAGE.
27930 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27932 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27935 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27936 first argument to type.
27938 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27942 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27943 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27944 (rs6000_expand_binop_builtin): Likewise.
27945 (rs6000_expand_ternop_builtin): Likewise; also add missing
27946 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27947 vshasigma built-ins.
27948 * doc/extend.texi: Document that vec_xxpermdi's third argument
27949 must be a constant.
27951 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27953 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27954 Use shift_const cost parameter when calculating gain of STV shifts.
27956 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27958 PR rtl-optimization/70478
27959 * lra-constraints.c (process_alt_operands): Check memory for
27960 disfavoring memory insn operand.
27962 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27964 PR middle-end/80100
27965 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27966 left shift in unsigned HOST_WIDE_INT type.
27968 PR rtl-optimization/80385
27969 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27970 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27973 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27974 if they have any depend clauses.
27976 2017-04-11 Martin Liska <mliska@suse.cz>
27979 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27980 * ipa-split.c (split_function): Create a local comdat symbol
27981 if caller is in a comdat group.
27983 2017-04-11 Martin Liska <mliska@suse.cz>
27986 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27989 2017-04-11 Martin Sebor <msebor@redhat.com>
27991 PR middle-end/80364
27992 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27993 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27995 (directive::set_width, directive::set_precision, format_character):
27997 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28000 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28003 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28004 conflict, set target->arch_name instead of target->cpu_name.
28006 2017-04-11 Richard Biener <rguenther@suse.de>
28008 PR tree-optimization/80374
28009 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28010 build_zero_cst, remove fold_convertible_p check again.
28012 2017-04-11 Martin Liska <mliska@suse.cz>
28015 * ubsan.c (instrument_object_size): Do not instrument register
28018 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28021 * config/i386/i386-builtin-types.def
28022 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28023 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28024 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28025 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28026 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28027 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28028 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28029 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28030 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28031 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28032 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28033 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28034 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28035 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28036 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28037 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28038 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28039 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28040 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28041 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28042 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28043 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28044 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28045 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28046 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28047 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28048 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28049 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28050 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28051 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28052 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28053 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28054 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28055 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28056 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28057 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28058 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28059 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28060 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28061 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28062 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28063 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28064 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28065 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28066 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28067 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28069 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28070 flag to second_arg_count, handle 4 argument function type _COUNT
28071 aliases, handle second_arg_count on second argument rather than last.
28073 2017-04-10 Jeff Law <law@redhat.com>
28075 PR tree-optimization/80374
28076 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28077 record anything if we can not convert integer_zero_node to the
28080 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28083 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28084 Enhance special handling given to the TARGET_P9_MINMAX option in
28085 relation to certain other options.
28087 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28089 PR tree-optimization/80153
28090 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28091 remove POINTER_PLUS_EXPR's base part directly, rather than through
28094 2017-04-10 Richard Biener <rguenther@suse.de>
28095 Bin Cheng <bin.cheng@arm.com>
28097 PR tree-optimization/80153
28098 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28099 the first element of pointer type aff_tree. Build result expr in
28101 (add_elt_to_tree): Convert to type unconditionally. Remove other
28102 fold_convert calls.
28103 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28104 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28106 2017-04-10 Richard Biener <rguenther@suse.de>
28108 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28111 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28113 PR rtl-optimization/70478
28114 * lra-constraints.c (curr_small_class_check): New.
28115 (update_and_check_small_class_inputs): New.
28116 (process_alt_operands): Update curr_small_class_check. Disfavor
28117 alternative insn memory operands. Check available regs for small
28120 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28123 * config/mips/mips.opt (-mvirt): Update description.
28124 * doc/invoke.texi (-mvirt): Likewise.
28126 2017-04-10 Richard Biener <rguenther@suse.de>
28128 PR middle-end/80362
28129 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28130 looking for NEGATE_EXPR in -A / -B to A / B folding.
28132 2017-04-10 Martin Liska <mliska@suse.cz>
28134 PR gcov-profile/80224
28135 * gcov.c (print_usage): Fix usage string.
28136 (get_gcov_intermediate_filename): Remove.
28137 (output_gcov_file): Use both for normal and intermediate format.
28138 (generate_results): Do not initialize special file for
28139 intermediate format.
28141 2017-04-10 Richard Biener <rguenther@suse.de>
28143 PR tree-optimization/80304
28144 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28147 2017-04-10 Nathan Sidwell <nathan@acm.org>
28150 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28151 (rs6000_init_builtins): Use it.
28153 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28155 * config/arm/arm.md (<mrc>): Add mode to SET source.
28156 (<mrrc>): Likewise.
28158 2017-04-10 Richard Biener <rguenther@suse.de>
28160 PR middle-end/80344
28161 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28163 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28166 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28167 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28168 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28169 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28170 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28171 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28172 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28173 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28174 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28175 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28176 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28177 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28178 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28179 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28180 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28181 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28182 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28183 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28184 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28185 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28186 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28187 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28188 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28190 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28192 PR rtl-optimization/70478
28193 * lra-constraints.c: Reverse the last patch.
28195 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28197 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28198 Add comment for WCHAR_T.
28200 2017-04-08 Martin Liska <mliska@suse.cz>
28203 2017-04-07 Martin Liska <mliska@suse.cz>
28206 * ipa-split.c (split_function): Add function part to a same comdat
28209 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28212 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28214 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28216 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28217 (vec_pairing): ...to this.
28218 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28219 (rs6000_sched_init): Adjust for name change.
28220 (struct rs6000_sched_context): Likewise.
28221 (rs6000_init_sched_context): Likewise.
28222 (rs6000_set_sched_context): Likewise.
28224 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28230 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28232 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28233 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28234 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28236 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28238 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28240 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28242 PR rtl-optimization/70703
28243 * ira-color.c (update_conflict_hard_regno_costs): Use
28244 int64_t instead of HOST_WIDE_INT.
28246 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28248 PR rtl-optimization/70478
28249 * lra-constraints.c (process_alt_operands): Disfavor alternative
28250 insn memory operands.
28252 2017-04-07 Jeff Law <law@redhat.com>
28254 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28255 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28257 2017-04-07 Martin Liska <mliska@suse.cz>
28260 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28261 Show error message instead of an ICE.
28263 2017-04-07 Martin Liska <mliska@suse.cz>
28266 * ipa-split.c (split_function): Add function part to a same comdat
28269 2017-04-07 Richard Biener <rguenther@suse.de>
28271 PR middle-end/80341
28272 * tree.c (get_unwidened): Also handle ! for_type case for
28274 * convert.c (do_narrow): Split out from ...
28275 (convert_to_integer_1): ... here. Do not pass final truncation
28276 type to get_unwidened for TRUNC_DIV_EXPR.
28278 2017-04-07 Richard Biener <rguenther@suse.de>
28280 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28282 (aff_combination_const): Adjust.
28283 (aff_combination_scale): Likewise.
28284 (aff_combination_add_elt): Likewise.
28285 (aff_combination_add_cst): Likewise.
28286 (aff_combination_convert): Likewise.
28287 (add_elt_to_tree): Likewise. Remove unused argument.
28288 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28290 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28292 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28294 * config/arm/arm.c (arm_default_short_enums): Use
28295 ARM_DEFAULT_SHORT_ENUMS.
28296 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28298 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28301 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28302 members with redundant out-of-class redeclaration.
28304 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28307 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28308 * config/i386/i386.md (*zero_extendsidi2):
28309 Add (?*x,*x) and (?*v,*v) alternatives.
28311 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28314 * config/i386/i386.c (ix86_expand_builtin)
28315 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28316 mode from insn data. Convert operands to insn operand mode.
28317 Copy operands that don't satisfy insn predicate to a register.
28319 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28321 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28324 2017-04-06 Richard Biener <rguenther@suse.de>
28326 PR tree-optimization/80334
28327 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28328 preserve alignment of accesses.
28330 2017-04-06 Richard Biener <rguenther@suse.de>
28332 PR tree-optimization/80262
28333 * tree-sra.c (build_ref_for_offset): Preserve address-space
28335 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28336 Drop useless address-space information on MEM_REF offsets.
28338 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28340 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28342 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28344 PR rtl-optimization/70703
28345 * ira-color.c (update_conflict_hard_regno_costs): Use
28346 HOST_WIDE_INT instead of long.
28348 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28351 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28352 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28354 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28355 for x86_64 target. Handle -m3dnowa option.
28357 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28359 PR rtl-optimization/70703
28360 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28361 (update_conflict_hard_regno_costs): Use long instead of unsigned
28362 arithmetic for cost calculation.
28364 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28365 Bernd Edlinger <bernd.edlinger@hotmail.de>
28368 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28371 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28374 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28375 ptr_mode with Pmode throughout.
28376 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28377 into probe_stack_range and use DImode.
28379 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28382 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28383 call_eh_return is true.
28385 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28387 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28388 Initialize last_match_fntype_index.
28390 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28393 * tree-nvr.c: Include internal-fn.h.
28394 (pass_return_slot::execute): Ignore internal calls without
28397 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28398 Richard Biener <rguenther@suse.de>
28401 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28402 captures used multiple times, except for the last use.
28403 * generic-match-head.c: Include gimplify.h.
28405 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28407 PR tree-optimization/79390
28408 * target.h (struct noce_if_info): Declare.
28409 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28410 * target.def (noce_conversion_profitable_p): New target hook.
28411 * ifcvt.h (struct noce_if_info): New type, moved from ...
28412 * ifcvt.c (struct noce_if_info): ... here.
28413 (noce_conversion_profitable_p): Renamed to ...
28414 (default_noce_conversion_profitable_p): ... this. No longer
28416 (noce_try_store_flag_constants, noce_try_addcc,
28417 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28418 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28419 instead of noce_conversion_profitable_p.
28420 * config/i386/i386.c: Include ifcvt.h.
28421 (ix86_option_override_internal): Don't override
28422 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28423 (ix86_noce_conversion_profitable_p): New function.
28424 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28425 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28426 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28427 * doc/tm.texi: Regenerated.
28429 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28431 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28434 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28437 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28438 instructions for small multiply cores.
28440 2017-04-04 Jeff Law <law@redhat.com>
28442 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28444 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28445 that are not set by the loop over the elements.
28447 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28450 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28451 int mode, convert_modes it to mode as unsigned, otherwise use
28452 lowpart_subreg to mode rather than SImode.
28453 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28454 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28455 Use DImode instead of SImode for the shift count operand.
28456 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28459 2017-04-04 Richard Biener <rguenther@suse.de>
28461 PR middle-end/80281
28462 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28463 arithmetic done for the negate or the plus. Simplify.
28464 (A - (-B) -> A + B): Likewise.
28465 * fold-const.c (split_tree): Make sure to not negate pointers.
28467 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28469 PR rtl-optimization/60818
28470 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28471 a compare of comparisons with the thing compared if this results
28472 in a different machine mode.
28474 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28476 * alias.c (base_alias_check): Fix typo in comment.
28477 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28478 * cgraphunit.c (symbol_table::compile): Likewise.
28479 * collect2.c (maybe_run_lto_and_relink): Likewise.
28480 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28481 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28482 * config/avr/avr.c (avr_map_op_t): Likewise.
28483 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28484 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28485 * config/epiphany/epiphany.md (movcc): Likewise.
28486 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28487 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28489 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28490 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28491 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28492 * config/sh/sh.c (sh_rtx_costs): Likewise.
28493 * fold-const.c (fold_truth_andor): Likewise.
28494 * genautomata.c (collapse_flag): Likewise.
28495 * gengtype.h (struct type::u::s): Likewise.
28496 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28497 * input.c (FORMAT_AMOUNT): Likewise.
28498 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28499 (known_aggs_to_agg_replacement_list): Likewise.
28500 * ipa-inline-analysis.c: Likewise.
28501 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28502 * ipa-polymorphic-call.c
28503 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28504 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28505 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28507 * modulo-sched.c (apply_reg_moves): Likewise.
28508 * omp-expand.c (build_omp_regions_1): Likewise.
28509 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28510 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28511 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28512 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28513 * value-prof.c: Likewise.
28514 * var-tracking.c (val_reset): Likewise.
28516 2017-04-03 Richard Biener <rguenther@suse.de>
28518 PR tree-optimization/80275
28519 * fold-const.c (split_address_to_core_and_offset): Handle
28522 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28524 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28525 descriptors is at least equal to that of functions.
28527 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28529 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28531 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28534 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28535 (mov<IMOD4:mode>): New expander.
28536 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28538 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28540 PR rtl-optimization/79405
28541 * fwprop.c (propagations_left): New variable.
28542 (forward_propagate_into): Decrement it.
28543 (fwprop_init): Initialize it.
28544 (fw_prop): If the variable has reached zero, stop propagating.
28545 (fwprop_addr): Ditto.
28547 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28550 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28551 a FUNCTION_DECL, pass it as decl instead of origin to
28554 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28556 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28559 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28562 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28563 TARGET_VSX_SMALL_INTEGER.
28565 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28567 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28568 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28570 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28572 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28573 extraction from odd-numbered MSA register.
28575 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28577 PR middle-end/80173
28578 * expmed.c (store_bit_field_1): Don't attempt to create
28579 a word subreg out of hard registers wider than word if they
28580 have HARD_REGNO_NREGS of 1 for their mode.
28582 PR middle-end/80163
28583 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28584 conversions to integer types wider than word and pointer.
28587 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28588 (rtx_equal_for_cselib_p): Pass 0 to it.
28589 * cselib.c (cselib_hasher::equal): Likewise.
28590 (rtx_equal_for_cselib_1): Add depth argument. If depth
28591 is 128, don't look up VALUE locs and punt. Increment
28592 depth in recursive calls when walking VALUE locs.
28594 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28596 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28597 (make_gcov_file_name): Use the canonical path name for generating
28599 (read_line): Fix handling of files with ascii null bytes.
28601 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28603 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28604 to initialise a vector register instead
28605 of using a const_int.
28607 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28609 PR translation/80189
28610 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28611 diagnostic messages.
28613 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28616 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28617 (dfp_diex_<mode>): Update mode of operand 1.
28618 * doc/extend.texi (dxex, dxexq): Document change to return type.
28619 (diex, diexq): Document change to argument type.
28621 2017-03-30 Martin Jambor <mjambor@suse.cz>
28624 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28625 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28626 it reflects the signature changes performed at the callee side.
28627 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28628 to cgraph_build_function_type_skip_args.
28629 (build_function_decl_skip_args): Adjust call to the above function.
28631 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28634 * config/i386/sse.md
28635 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28636 register as dest whenever it is a MEM not rtx_equal_p to the
28637 corresponding dup operand, and when forcing into reg move the
28638 reg into the memory afterwards.
28639 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28640 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28641 for the force_reg mode.
28642 (avx512vl_vextractf128<mode>): Use register as dest either
28643 always when a MEM, or when it is a MEM not rtx_equal_p to the
28644 corresponding dup operand, or even not when it is a CONST_VECTOR
28645 depending on the mode and lo vs. hi.
28646 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28648 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28649 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28650 Likewise. Require that operands[2] is even.
28651 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28652 Remove extraneous parens. Require that operands[2] is a multiple
28654 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28655 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28656 disallow memory then.
28658 2017-03-30 Richard Biener <rguenther@suse.de>
28660 PR tree-optimization/77498
28661 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28662 to non-constants over backedges.
28664 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28666 PR rtl-optimization/80233
28667 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28668 as last_combined_insn. Do not test for BARRIER_P separately.
28670 2017-03-29 Andreas Schwab <schwab@suse.de>
28673 * calls.c (prepare_call_address): Convert funexp to Pmode before
28674 copying to temp reg.
28676 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28678 PR tree-optimization/80158
28679 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28680 Handle possible future case of more than one alternate
28682 (replace_rhs_if_not_dup): Likewise.
28683 (replace_one_candidate): Likewise.
28685 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28687 PR rtl-optimization/80193
28688 * ira.c (ira): Do not check allocation for LRA.
28690 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28692 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28693 (nvptx_output_simt_exit): Declare.
28694 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28695 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28696 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28697 (nvptx_declare_function_name): ...here. Emit declaration of local
28698 memory space buffer for omp_simt_enter insn.
28699 (nvptx_output_unisimt_switch): New.
28700 (nvptx_output_softstack_switch): New.
28701 (nvptx_output_simt_enter): New.
28702 (nvptx_output_simt_exit): New.
28703 * config/nvptx/nvptx.h (struct machine_function): New fields
28704 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28705 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28706 (UNSPECV_SIMT_EXIT): Ditto.
28707 (omp_simt_enter_insn): New insn.
28708 (omp_simt_enter): New expansion.
28709 (omp_simt_exit): New insn.
28710 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28712 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28713 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28714 (expand_GOMP_SIMT_EXIT): New.
28715 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28716 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28717 (GOMP_SIMT_EXIT): Ditto.
28718 * target-insns.def (omp_simt_enter): New insn.
28719 (omp_simt_exit): Ditto.
28720 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28722 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28723 (lower_rec_input_clauses): Likewise.
28724 (lower_lastprivate_clauses): Handle SIMT privatization.
28726 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28727 (ompdevlow_adjust_simt_enter): New.
28728 (find_simtpriv_var_op): New.
28729 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28730 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28732 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28733 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28734 (copy_decl_for_dup_finish): Ditto.
28736 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28738 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28741 * config/i386/i386.c (ix86_option_override_internal): Always
28742 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28744 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28746 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28748 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28750 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28751 mark new edge's irreducible flag accordign to it.
28752 (vect_do_peeling): Check loop preheader edge's irreducible flag
28753 and pass it to function slpeel_add_loop_guard.
28755 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28757 PR tree-optimization/80218
28758 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28759 Update block frequencies and counts.
28761 2017-03-28 Richard Biener <rguenther@suse.de>
28763 PR tree-optimization/78644
28764 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28765 of a simplification result we may not use it at all.
28767 2017-03-28 Richard Biener <rguenther@suse.de>
28770 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28771 without arguments, generate default definition of a SSA name.
28773 2017-03-28 Richard Biener <rguenther@suse.de>
28775 PR middle-end/80222
28776 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28777 TYPE_REF_CAN_ALIAS_ALL references.
28778 * fold-const.c (fold_indirect_ref_1): Likewise.
28780 2017-03-28 Martin Liska <mliska@suse.cz>
28783 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28784 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28786 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28787 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28789 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28790 (EXTRA_SPECS): Define.
28791 (SUBTARGET_EXTRA_SPECS): Likewise.
28792 (SUBTARGET_CPP_SPEC): Likewise.
28793 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28794 SUBTARGET_EXTRA_SPECS.
28795 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28797 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28799 * config/arc/simdext.md (vst64_insn): Update pattern.
28800 (vld32wh_insn): Likewise.
28801 (vld32wl_insn): Likewise.
28802 (vld64_insn): Likewise.
28803 (vld32_insn): Likewise.
28805 2017-03-28 Marek Polacek <polacek@redhat.com>
28808 * fold-const.c (fold_comparison): Use protected_set_expr_location
28809 instead of SET_EXPR_LOCATION.
28811 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28813 * tree.c (add_expr): Avoid name lookup warning.
28815 2017-03-27 Jeff Law <law@redhat.com>
28817 PR tree-optimization/80216
28818 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28819 function name. Limit recursion depth.
28820 (record_temporary_equivalences): Corresponding changes.
28822 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28824 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28827 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28830 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28832 * cfgcleanup.c (reg_note_cfa_p): New array.
28833 (insns_have_identical_cfa_notes): New function.
28834 (old_insns_match_p): Don't cross-jump in between /f
28835 and non-/f instructions. If both i1 and i2 are frame related,
28836 verify all CFA notes, their order and content.
28838 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28841 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28842 HImode and SImode with zero extend to DImode to one insn.
28843 (bswap<mode>2_extenddi): Likewise.
28844 (bswapsi2_extenddi): Likewise.
28845 (bswaphi2_extendsi): Likewise.
28846 (bswaphi2): Combine bswap HImode and SImode into one insn.
28847 Separate memory insns from swapping register.
28848 (bswapsi2): Likewise.
28849 (bswap<mode>2): Likewise.
28850 (bswaphi2_internal): Delete, no longer used.
28851 (bswapsi2_internal): Likewise.
28852 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28853 store, and gpr<-gpr swap insns.
28854 (bswap<mode>2_store): Likewise.
28855 (bswaphi2_reg): Register only splitter, combine with the splitter.
28856 (bswaphi2 splitter): Likewise.
28857 (bswapsi2_reg): Likewise.
28858 (bswapsi2 splitter): Likewise.
28859 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28860 the insns into load, store, and register/register insns.
28861 (bswapdi2_ldbrx): Likewise.
28862 (bswapdi2_load): Likewise.
28863 (bswapdi2_store): Likewise.
28864 (bswapdi2_reg): Likewise.
28866 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28868 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28869 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28871 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28874 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28876 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28877 special handling for target option conflicts between dform
28878 options (-mpower9-dform, -mpower9-dform-vector,
28879 -mpower9-dform-scalar) and -mno-direct-move.
28881 2017-03-27 Richard Biener <rguenther@suse.de>
28883 PR tree-optimization/80181
28884 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28886 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28888 * config/arc/predicates.md (move_double_src_operand): Replace the
28889 call to move_double_src_operand with a call to address_operand.
28891 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28893 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28894 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28895 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28897 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28899 * config/arc/predicates.md (long_immediate_loadstore_operand):
28900 Consider scaled addresses cases.
28902 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28904 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28905 restored when in interrupt.
28906 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28907 doesn't have delay slot.
28909 2017-03-27 Richard Biener <rguenther@suse.de>
28912 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28913 inlined thunk clones.
28915 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28918 * asan.c (instrument_derefs): Copy over last operand from
28919 original COMPONENT_REF to the new COMPONENT_REF with
28920 DECL_BIT_FIELD_REPRESENTATIVE.
28921 * ubsan.c (instrument_object_size): Likewise.
28923 2017-03-27 Richard Biener <rguenther@suse.de>
28925 PR tree-optimization/80170
28926 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28927 sure DR/SCEV didnt fold in constants we do not see when looking
28928 at the reference base alignment.
28930 2017-03-27 Richard Biener <rguenther@suse.de>
28932 PR middle-end/80171
28933 * gimple-fold.c (fold_ctor_reference): Properly guard against
28934 NULL return value from canonicalize_constructor_val.
28936 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28939 * config/i386/i386.c (ix86_expand_builtin)
28940 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28941 flags reg setting and flags reg using instructions.
28942 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28943 clobbering instructions to zero extend op2.
28945 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28947 * doc/install.texi (Configuration) <--with-aix-soname>:
28948 Update link to AIX ld.
28950 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28952 PR rtl-optimization/80160
28953 PR rtl-optimization/80159
28954 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28955 reg_alternate_class into account.
28957 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28960 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28961 to consider in curr_insn_transform.
28963 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28965 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28966 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28967 and emit_mode_inner.
28969 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28971 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28972 argument to the overloaded builtin variants. Use the new flag to
28973 deprecate certain builtin variants.
28974 * config/s390/s390-builtin-types.def: Add new builtin types.
28975 * config/s390/s390-builtins.h: Support new flags field for
28976 overloaded builtins.
28977 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28978 (s390_macro_to_expand): Enable vector float data type.
28979 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28980 builtins by incrementing the __VEC__ version number.
28981 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28983 (s390_resolve_overloaded_builtin): Emit error messages depending
28984 on the builtin flags.
28985 * config/s390/s390.c (s390_expand_builtin): Support additional
28986 flags argument. Change error message to match the messages
28987 emitted in s390-c.c.
28988 * config/s390/s390.md: New UNSPEC_* constants.
28989 (op_type): Add new instruction types.
28990 * config/s390/vecintrin.h: Add new builtins and test data class
28992 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28993 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28994 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28995 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28996 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28997 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28999 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29000 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29001 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29002 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29004 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29005 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29006 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29007 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29009 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29010 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29011 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29012 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29013 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29014 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29015 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29017 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29018 ("vec_scatter_element<V_HW_4:mode>_DI")
29019 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29020 ("vec_fpint<mode>", "vflls")
29021 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29022 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29023 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29024 ("*vec_cmphe<mode>_cc"): ... these.
29026 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29027 mode constant instead of magic value.
29029 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29031 * config/s390/s390.c (s390_expand_vec_compare): Support other
29032 vector floating point modes than just V2DF.
29033 (s390_expand_vcond): Likewise.
29034 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29035 (s390_cannot_change_mode_class): Prevent mode changes between TF
29036 and V1TF in vector registers.
29037 * config/s390/s390.md (DF, SF): New mode attributes.
29038 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29039 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29040 SFmode support for VRs.
29041 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29043 (VFT, VF_HW): New mode iterators.
29044 (vw, sdx): New mode attributes.
29045 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29046 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29047 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29048 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29049 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29050 also the new vector floating point modes. Renaming to ...
29052 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29053 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29054 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29055 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29056 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29057 ("vec_unordered<mode>"): ... these.
29059 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29060 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29061 ("*vec_extendv2df"): New insn definitions.
29063 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29065 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29066 ("mulditi3_2", "*muldi3_sign"): New patterns.
29067 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29068 rename the pattern definition.
29070 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29072 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29074 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29076 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29078 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29079 instruction if possible.
29080 * config/s390/vector.md (vec_halfnumelts): New mode
29082 ("*vec_vllezlf<mode>"): New pattern.
29084 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29086 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29087 ("popcountv4si2", "popcountv2di2"): Rename to ...
29088 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29089 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29091 ("popcount<mode>2_vxe"): New pattern.
29093 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29095 * common/config/s390/s390-common.c (processor_flags_table): Add
29097 * config.gcc: Add arch12.
29098 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29099 Default to arch12 for unknown CPU model numbers.
29100 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29101 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29102 PROCESSOR_max sanity check.
29103 * config/s390/s390-opts.h (enum processor_type): Add
29105 * config/s390/s390.c (processor_table): Add arch12.
29106 (s390_expand_builtin): Add check for B_VXE flag.
29107 (s390_issue_rate): Add PROCESSOR_ARCH12.
29108 (s390_get_sched_attrmask): Likewise.
29109 (s390_get_unit_mask): Likewise.
29110 (s390_sched_score): Enable z13 scheduling for arch12.
29111 (s390_sched_reorder): Likewise.
29112 (s390_sched_variable_issue): Likewise.
29113 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29115 (s390_tune_attr): Use z13 scheduling also for arch12.
29116 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29117 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29118 (TARGET_VXE_P): New macros.
29119 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29120 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29121 * config/s390/s390.opt: Add arch12 as processor_type.
29123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29125 * config/s390/s390.md
29126 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29127 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29128 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29130 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29131 Rename expanders to ...
29133 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29134 ("fixuns_truncdddi2_emu"): ... these.
29136 ("fixuns_trunc<mode>si2_emu"): New expander.
29138 ("*fixuns_truncdfdi2_z13"): Rename to ...
29139 ("*fixuns_truncdfdi2_vx"): ... this.
29141 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29143 * config/s390/2964.md: Remove the single element vector compare
29144 instructions which are no longer used.
29145 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29147 (s390_canonicalize_comparison): Remove handling of DFmode
29149 (s390_expand_vec_compare_scalar): Remove function.
29150 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29151 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29153 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29155 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29157 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29159 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29160 will anyway by matched by mov<mode>_64dfp.
29162 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29164 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29165 vlef/vstef. Add missing operand to vleif.
29167 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29169 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29170 pair for all vector types with 64 bit elements.
29171 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29172 * config/s390/vector.md (V_HW_64): ... here.
29173 (V_128_NOSINGLE): New mode iterator.
29174 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29175 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29176 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29177 ("*vec_load_pairv2di"): Change to ...
29178 ("*vec_load_pair<mode>"): ... this one.
29180 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29182 * config/s390/constraints.md: Add comments.
29183 (jKK): Reject element sizes > 8 bytes.
29184 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29186 * config/s390/s390.md: Add the s_operand checks formerly in
29187 s390_split_ok_p to various splitters where they are still
29189 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29190 for 128 bit vectors. Plus two splitters.
29192 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29194 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29197 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29200 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29201 error if the boundary argument is not constant.
29203 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29205 PR rtl-optimization/80112
29206 * loop-doloop.c (doloop_condition_get): Don't check condition
29207 if cmp isn't SET with IF_THEN_ELSE src.
29209 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29211 PR tree-optimization/80158
29212 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29213 replacing a candidate statement, also replace it for the
29214 candidate's alternate interpretation.
29215 (replace_rhs_if_not_dup): Likewise.
29216 (replace_one_candidate): Likewise.
29218 2017-03-24 Richard Biener <rguenther@suse.de>
29220 PR tree-optimization/80167
29221 * graphite-isl-ast-to-gimple.c
29222 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29224 (translate_isl_ast_to_gimple::get_rename): Likewise.
29226 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29228 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29229 handling of certain combinations of target options, including the
29230 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29231 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29233 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29236 * config/arm/arm.md (*load_multiple): Add reload_completed to
29237 matching condition.
29239 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29240 Richard Biener <rguenth@suse.de>
29242 PR tree-optimization/79908
29243 PR tree-optimization/80136
29244 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29245 been cast away, gimplify_and_add suffices.
29247 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29249 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29251 2017-03-23 Richard Biener <rguenther@suse.de>
29253 PR tree-optimization/80032
29254 * gimplify.c (gimple_push_cleanup): Forced unconditional
29255 cleanups still have to go to the conditional_cleanups
29258 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29260 PR tree-optimization/80072
29261 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29263 (next_operand_entry_id): Change type to unsigned int.
29264 (sort_by_operand_rank): Make sure to return the right return value
29265 even if unsigned fields are bigger than INT_MAX.
29266 (struct oecount): Change cnt and id type to unsigned int.
29267 (oecount_hasher::equal): Formatting fix.
29268 (oecount_cmp): Make sure to return the right return value
29269 even if unsigned fields are bigger than INT_MAX.
29270 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29273 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29274 TREE_READONLY on result if writing it more than once.
29277 * doc/invoke.texi (-fsanitize=thread): Document that with
29278 -fnon-call-exceptions atomics are not able to throw
29282 * tsan.c: Include tree-eh.h.
29283 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29284 maybe_clean_or_replace_eh_stmt where needed.
29285 (instrument_memory_accesses): Add cfg_changed argument.
29286 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29287 if it returned true.
29288 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29290 PR rtl-optimization/63191
29291 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29292 wrapper function, moved the whole old content into ...
29293 (ix86_delegitimize_address_1): ... this. New inline function.
29294 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29295 true as last argument instead of ix86_delegitimize_address.
29297 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29299 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29300 cortexa57_branch_cost.
29302 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29304 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29306 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29309 * doc/md.texi (Constraints): Document wA constraint.
29310 * config/rs6000/constraints.md (wA): New.
29311 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29312 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29313 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29314 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29316 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29319 * gimplify.c (is_oacc_declared): New function.
29320 (oacc_default_clause): Use it to set default flags for acc declared
29321 variables inside parallel regions.
29322 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29323 declared variables.
29324 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29325 declare attribute to any decl as necessary.
29327 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29330 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29331 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29332 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29333 (arm_arch_lpae): This.
29334 * config/arm/arm.c (arm_arch7ve): Rename into ...
29335 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29336 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29339 2017-03-22 Martin Liska <mliska@suse.cz>
29342 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29343 error message instead of an ICE.
29345 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29347 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29349 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29351 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29353 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29356 2017-03-21 Martin Sebor <msebor@redhat.com>
29358 * doc/extend.texi: Use "cannot" instead of "can't."
29359 * doc/hostconfig.texi: Same.
29360 * doc/install.texi: Same.
29361 * doc/invoke.texi: Same.
29362 * doc/loop.texi: Same.
29363 * doc/md.texi: Same.
29364 * doc/objc.texi: Same.
29365 * doc/rtl.texi: Same.
29366 * doc/tm.texi: Same.
29367 * doc/tm.texi.in: Same.
29368 * doc/trouble.texi: Same.
29370 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29373 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29374 (collect_checksum_attributes): Set it.
29375 (die_checksum_ordered): Use it.
29377 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29379 PR tree-optimization/79908
29380 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29381 change: For a VA_ARG whose LHS has been cast away, use
29382 force_gimple_operand to construct the side effects.
29384 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29386 PR translation/80001
29387 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29388 more amenable to translation.
29389 (oacc_loop_auto_partitions): Likewise.
29391 2017-03-21 Marek Polacek <polacek@redhat.com>
29392 Martin Sebor <msebor@redhat.com>
29394 PR tree-optimization/80109
29395 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29396 on INTEGRAL_TYPE_P.
29398 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29399 Segher Boessenkool <segher@kernel.crashing.org>
29402 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29403 check reg_used_between_p between insn and one of succ or succ2
29404 depending on if succ is artificial insn not inserted into insn
29407 2017-03-21 Martin Liska <mliska@suse.cz>
29409 PR gcov-profile/80081
29410 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29411 * doc/gcc.texi: Include gcov-dump stuff.
29412 * doc/gcov-dump.texi: New file.
29414 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29416 PR rtl-optimization/79150
29417 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29418 conditional jump, if the jump is the last insn of the loop.
29420 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29421 Richard Biener <rguenth@suse.de>
29423 PR tree-optimization/79908
29424 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29425 been cast away, use force_gimple_operand to construct the side
29428 2017-03-21 Martin Liska <mliska@suse.cz>
29430 PR libfortran/79956
29431 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29434 2017-03-21 Brad Spengler <spender@grsecurity.net>
29437 * plugin.c (htab_hash_plugin): New function.
29438 (add_new_plugin): Use it and adjust.
29439 (parse_plugin_arg_opt): Adjust.
29440 (init_one_plugin): Likewise.
29442 2017-03-21 Richard Biener <rguenther@suse.de>
29444 PR tree-optimization/80032
29445 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29446 if set force the cleanup to happen unconditionally.
29447 (gimplify_target_expr): Push inserted clobbers with force_uncond
29448 to avoid them being removed by control-dependent DCE.
29450 2017-03-21 Richard Biener <rguenther@suse.de>
29452 PR tree-optimization/80122
29453 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29454 va_arg_pack_len when the inlined call stmt requires pack
29456 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29458 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29461 * tsan.c (instrument_builtin_call): If the memory model argument
29462 is not a constant, assume it is valid.
29465 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29468 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29470 PR rtl-optimization/79910
29471 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29472 if its dest is used by an insn before I2 (other than the combined
29473 insns themselves, which are properly handled already).
29475 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29478 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29480 * combine.c (record_used_regs): New static function.
29481 (try_combine): Handle situations where there is an additional
29482 instruction between I2 and I3 which needs to have a LOG_LINK
29486 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29488 * combine.c (try_combine): Delete redundant i1 test. Call
29489 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29491 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29494 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29495 alternatives 13/14.
29497 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29499 PR tree-optimization/80054
29500 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29501 the optimization if a PHI or any of its arguments is not dominated
29502 by the candidate's basis. Use gphi* rather than gimple* as
29504 (replace_profitable_candidates): Clean up a gimple* variable that
29505 should be a gphi* variable.
29507 2017-03-20 Martin Sebor <msebor@redhat.com>
29510 * doc/extend.texi (attribute constructor): Document present limitation.
29512 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29515 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29516 __POWER9_VECTOR__ #ifdef control, change template definition to
29517 use Power9-specific built-in function.
29518 (vec_any_eq): Likewise.
29519 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29520 to control outcomes from this test.
29521 (vector_ae_<mode>p): For VEC_F modes, likewise.
29523 2017-03-20 Ian Lance Taylor <iant@google.com>
29525 * config/i386/i386.c (ix86_function_regparm): Save an extra
29526 register for -fsplit-stack with DECL_STATIC_CHAIN.
29528 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29531 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29532 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29534 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29536 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29538 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29541 2017-03-20 Marek Polacek <polacek@redhat.com>
29544 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29546 2017-03-20 Richard Biener <rguenther@suse.de>
29548 PR tree-optimization/80113
29549 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29550 allocate extra SSA name for PHI def.
29551 (add_close_phis_to_outer_loops): Likewise.
29552 (add_close_phis_to_merge_points): Likewise.
29553 (copy_loop_close_phi_args): Likewise.
29554 (copy_cond_phi_nodes): Likewise.
29556 2017-03-20 Martin Liska <mliska@suse.cz>
29558 PR middle-end/79753
29559 * tree-chkp.c (chkp_build_returned_bound): Do not build
29560 returned bounds for a LHS that's not a BOUNDED_P type.
29562 2017-03-20 Martin Liska <mliska@suse.cz>
29566 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29567 COMPLEX_CST and VECTOR_CST.
29569 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29572 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29573 target operand. A new splitter adds the clobber statement in case
29574 the target operand is dead anyway.
29576 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29578 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29579 to age-old versions of binutils and glibc.
29581 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29583 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29585 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29587 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29589 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29591 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29592 requirement for binutils 2.13.
29594 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29596 * combine.c (try_combine): Delete redundant i1 test. Call
29597 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29599 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29601 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29602 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29604 <riscv64-*-elf>: Re-arrange section
29605 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29606 <riscv32-*-linux>: Likewise.
29607 <riscv64-*-elf>: Likewise
29608 <riscv64-*-linux>: Likewise.
29610 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29613 * aarch64.opt(verbose-cost-dump): Fix typo.
29615 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29618 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29619 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29621 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29623 * reload.c (find_reloads): When reloading a nonoffsettable address,
29624 use RELOAD_OTHER for it and its address reloads.
29626 PR rtl-optimization/79910
29627 * combine.c (record_used_regs): New static function.
29628 (try_combine): Handle situations where there is an additional
29629 instruction between I2 and I3 which needs to have a LOG_LINK
29632 2017-03-17 Jeff Law <law@redhat.com>
29634 PR tree-optimization/71437
29635 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29636 conditional in the hash table first.
29637 (vrp_dom_walker::before_dom_children): Extract condition from
29638 ASSERT_EXPR. Record condition, its inverion and any implied
29639 conditions as well.
29641 2017-03-17 Marek Polacek <polacek@redhat.com>
29642 Markus Trippelsdorf <markus@trippelsdorf.de>
29644 PR tree-optimization/80079
29645 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29648 2017-03-17 Richard Biener <rguenther@suse.de>
29650 PR middle-end/80075
29651 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29652 Properly verify the LHS before the RHS possibly claims to be
29654 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29657 2017-03-17 Martin Jambor <mjambor@suse.cz>
29659 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29660 (List of -O2 options): Likewise.
29661 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29664 2017-03-17 Tom de Vries <tom@codesourcery.com>
29666 * gcov-dump.c (print_usage): Print bug_report_url.
29668 2017-03-17 Richard Biener <rguenther@suse.de>
29670 PR middle-end/80050
29671 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29672 (parser::peek): Likewise.
29674 2017-03-17 Richard Biener <rguenther@suse.de>
29676 PR tree-optimization/80048
29677 * sese.c (free_sese_info): Properly release rename_map and
29678 copied_bb_map elements.
29680 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29682 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29683 Add linked-list forward and backlinks. Insert on
29684 construction, remove on destruction.
29685 (class pass_store_merging): Add m_stores_head field.
29686 (pass_store_merging::terminate_and_process_all_chains):
29687 Iterate over m_stores_head list.
29688 (pass_store_merging::terminate_all_aliasing_chains):
29690 (pass_store_merging::execute): Check for debug stmts first.
29691 Push new chains onto the m_stores_head stack.
29693 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29696 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29697 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29698 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29700 2017-03-16 Jeff Law <law@redhat.com>
29702 PR tree-optimization/71437
29703 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29704 member function. Implementation moved into after_dom_children
29705 member function and into the threader's thread_outgoing_edges
29707 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29708 some code into new thread_outgoing_edges.
29709 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29710 definition. Simplify marker handling (do it here). Assume we always
29711 have the available expression and the const/copies tables.
29712 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29714 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29715 * tree-vrp.c (equiv_stack): No longer file scoped.
29716 (vrp_dom_walker): New class.
29717 (vrp_dom_walker::before_dom_children): New member function.
29718 (vrp_dom_walker::after_dom_children): Likewise.
29719 (identify_jump_threads): Setup domwalker. Use it rather than
29720 walking edges in a random order by hand. Simplify setup/finalization.
29721 (finalize_jump_threads): Remove.
29722 (vrp_finalize): Do not call identify_jump_threads here.
29723 (execute_vrp): Do it here instead and call thread_through_all_blocks
29726 PR tree-optimization/71437
29727 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29729 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29731 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29732 (dom_opt_dom_walker::thread_across_edge): Remove
29733 handle_dominating_asserts argument. All callers changed.
29734 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29735 changes. Remove calls to lhs_of_dominating_assert. Other
29736 uses of handle_dominating_asserts turn into unconditional code
29737 (simplify_control_stmt_condition_1): Likewise.
29738 (simplify_control_stmt_condition): Likewise.
29739 (thread_through_normal_block, thread_across_edge): Likewise.
29740 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29741 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29742 object if it is not an SSA_NAME.
29743 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29744 before calling into the VRP specific simplifiers.
29745 (identify_jump_threads): Remove handle_dominating_asserts
29748 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29751 * tree-diagnostic.c (default_tree_printer): No longer static.
29752 * tree-diagnostic.h (default_tree_printer): New prototype.
29754 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29756 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29757 Change ins into fmov.
29759 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29761 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29762 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29763 Use h_con constraint for operand 1.
29764 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29765 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29767 2017-03-15 Jeff Law <law@redhat.com>
29769 PR tree-optimization/71437
29770 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29771 (record_temporary_equivalences): Use it.
29773 PR tree-optimization/71437
29774 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29775 tree-ssa-scopedtables.
29776 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29777 (record_conditions, record_cond, vuse_eq): Likewise.
29778 (record_edge_info): Adjust to API tweak of record_conditions.
29779 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29780 (record_temporary_equivalences, optimize_stmt): Likewise.
29781 (eliminate_redundant_computations): Likewise.
29782 (record_equivalences_from_stmt): Likewise.
29783 * tree-ssa-scopedtables.c: Include options.h and params.h.
29784 (vuse_eq): New function, moved from tree-ssa-dom.c
29785 (build_and_record_new_cond): Likewise.
29786 (record_conditions): Likewise. Accept vector of conditions rather
29787 than edge_equivalence structure for first argument.
29788 for the first argument.
29789 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29790 from tree-ssa-dom.c.
29791 (avail_exprs_stack::record_cond): Likewise.
29792 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29793 from tree-ssa-dom.c.
29794 (avail_exprs_stack): Add new member functions lookup_avail_expr
29796 (record_conditions): Declare.
29798 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29801 * lra-constraints.c (process_alt_operands): Increase reject for
29802 reloading an input/output operand.
29804 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29807 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29808 insns to convert from signed/unsigned char/short to IEEE 128-bit
29810 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29812 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29815 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29816 subreg of inner mode for values already in registers.
29818 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29820 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29821 iteration reg is used after the loop.
29823 2017-03-14 Martin Sebor <msebor@redhat.com>
29825 PR tree-optimization/79800
29826 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29827 precision in negative-positive range.
29828 (format_floating): Call non-const overload with adjusted precision.
29830 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29833 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29836 2017-03-14 Martin Sebor <msebor@redhat.com>
29838 PR middle-end/80020
29839 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29840 * builtins.def (aligned_alloc): Use it.
29843 * Makefile.in (GTFILES): Add calls.c.
29844 * calls.c: Include "gt-calls.h".
29846 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29848 PR rtl-optimization/79728
29849 * regs.h (struct target_regs): New field
29850 x_contains_allocatable_regs_of_mode.
29851 (contains_allocatable_regs_of_mode): New macro.
29852 * reginfo.c (init_reg_sets_1): Initialize it, and change
29853 contains_reg_of_mode so it includes global regs as well.
29854 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29855 rather than contains_regs_of_mode.
29857 2017-03-14 Martin Liska <mliska@suse.cz>
29859 * doc/invoke.texi: Document options that can't be combined with
29860 -fcheck-pointer-bounds.
29862 2017-03-14 Martin Liska <mliska@suse.cz>
29864 PR middle-end/79831
29865 * doc/invoke.texi (-Wchkp): Document the option.
29867 2017-03-14 Martin Liska <mliska@suse.cz>
29869 * Makefile.in: Install gcov-dump.
29871 2017-03-14 Martin Liska <mliska@suse.cz>
29873 * multiple_target.c (expand_target_clones): Bail out for
29874 an invalid attribute.
29876 2017-03-14 Richard Biener <rguenther@suse.de>
29878 * alias.c (struct alias_set_entry): Pack properly.
29879 * cfgloop.h (struct loop): Likewise.
29880 * cse.c (struct set): Likewise.
29881 * ipa-utils.c (struct searchc_env): Likewise.
29882 * loop-invariant.c (struct invariant): Likewise.
29883 * lra-remat.c (struct cand): Likewise.
29884 * recog.c (struct change_t): Likewise.
29885 * rtl.h (struct address_info): Likewise.
29886 * symbol-summary.h (function_summary): Likewise.
29887 * tree-loop-distribution.c (struct partition): Likewise.
29888 * tree-object-size.c (struct object_size_info): Likewise.
29889 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29890 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29891 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29892 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29893 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29894 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29895 (struct _stmt_vec_info): Likewise.
29897 2017-03-14 Martin Liska <mliska@suse.cz>
29900 * multiple_target.c (create_dispatcher_calls): Check that
29901 a target can create a function dispatcher.
29903 2017-03-14 Martin Liska <mliska@suse.cz>
29906 * multiple_target.c (expand_target_clones): Drop local.local
29907 flag for default implementation.
29909 2017-03-14 Richard Biener <rguenther@suse.de>
29911 PR tree-optimization/80030
29912 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29914 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29916 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29917 gcc_fallthrough() instead of __attribute__((fallthrough));
29919 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29921 * doc/gcc.texi: Remove "up" link to (DIR).
29922 * doc/gccint.texi: Ditto.
29924 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29926 * doc/install.texi (Specific) <avr>: Remove reference to
29929 2017-03-13 Jeff Law <law@redhat.com>
29931 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29932 attribute rather than comments.
29934 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29935 match_scratch operand is highest.
29937 2017-03-13 Martin Liska <mliska@suse.cz>
29939 PR middle-end/78339
29940 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29941 is a CHKP clone, use original declaration.
29943 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29945 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29946 (arc_conditional_register_usage): Use a different allocation order
29947 when optimizing for size.
29948 * common/config/arc/arc-common.c (arc_option_optimization_table):
29949 Section anchors default on when optimizing for size.
29951 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29953 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29955 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29957 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29958 * config/arc/arc.md (cpu_facility): Add cd variant.
29959 (*movqi_insn): Add code density variant.
29960 (*movhi_insn): Likewise.
29961 (*movqi_insn): Likewise.
29962 (*addsi3_mixed): Likewise.
29963 (subsi3_insn): Likewise.
29965 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29967 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29969 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29971 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29972 expressions with MINUS and UNARY ops.
29974 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29977 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29979 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29980 between vec_select and vector argument.
29981 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29982 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29983 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29984 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29985 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29986 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29988 2017-03-13 Richard Biener <rguenther@suse.de>
29991 * params.def (vect-max-peeling-for-alignment): Fix typo.
29993 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29995 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29996 issue that only occurred with binutils below 2.18.
29998 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30000 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30001 refer to binutils 2.11/2.12 minimum.
30003 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30005 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30006 ftp.kernel.org and simplify binutils requirement.
30008 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30010 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30012 (Optimize Options): Ditto. Also remove redundancy.
30014 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30016 PR translation/79848
30017 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30019 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30020 to G_ to avoid double translation.
30022 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30024 PR translation/79923
30025 * auto-profile.c (get_combined_location): Convert leading
30026 character of diagnostics to lower case and remove trailing period.
30027 (read_profile): Likewise for various diagnostics.
30028 * config/arm/arm.c (arm_option_override): Remove trailing period
30029 from various diagnostics.
30030 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30031 (msp430_expand_delay_cycles): Likewise.
30033 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30036 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30037 full command-line argument, rather than just "str".
30038 (aarch64_validate_march): Likewise.
30039 (aarch64_validate_mtune): Likewise.
30041 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30043 PR rtl-optimization/78911
30044 * lra-assigns.c (must_not_spill_p): New function.
30045 (spill_for): Use it.
30047 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30049 PR tree-optimization/79981
30050 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30051 ATOMIC_COMPARE_EXCHANGE ifn result.
30052 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30053 IFN_ATOMIC_COMPARE_EXCHANGE.
30055 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30058 * opts.c (parse_sanitizer_options): Add missing question mark to
30059 "did you mean" message.
30061 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30063 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30065 (VMULEUH_UNS): Likewise.
30066 (VMULOUB_UNS): Likewise.
30067 (VMULOUH_UNS): Likewise.
30068 * config/rs6000/rs6000.c (builtin_function_type): Remove
30069 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30071 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30074 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30075 x with result of extra_parsing_for_operand_code_0.
30076 (function_reader::extra_parsing_for_operand_code_0): Convert
30077 return type from void to rtx, returning x. When reading
30078 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30079 larger size containing struct block_symbol.
30081 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30083 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30084 -mfloat128-hardware without -m64.
30086 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30089 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30090 entries to the case statement that marks unsigned arguments to
30091 overloaded functions.
30093 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30095 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30096 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30098 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30101 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30102 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30104 2017-03-10 Martin Liska <mliska@suse.cz>
30108 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30109 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30112 2017-03-10 Olivier Hainque <hainque@adacore.com>
30114 * tree-switch-conversion (array_value_type): Start by resetting
30115 candidate type to it's main variant.
30117 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30119 PR rtl-optimization/79909
30120 * combine.c (try_combine): Use simplify_replace_rtx on individual
30121 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30122 of the whole CALL_INSN_FUNCTION_USAGE.
30124 PR tree-optimization/79972
30125 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30126 get_range_info on SSA_NAMEs. Formatting fixes.
30128 2017-03-10 Richard Biener <rguenther@suse.de>
30129 Jakub Jelinek <jakub@redhat.com>
30131 PR tree-optimization/77975
30132 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30133 edge to be constant.
30134 (get_val_for): For constant x return it. Formatting fix.
30135 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30136 would use the same bases as the current one.
30138 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30140 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30141 instead of vec_select for V1TImode.
30142 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30144 (VSX_LE_128): Add V1TI to this mode iterator.
30145 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30146 (*vsx_le_perm_store_<mode>): Likewise.
30147 (pre-reload splitter for VSX stores): Likewise.
30148 (post-reload splitter for VSX stores): Likewise.
30149 (*vsx_xxpermdi2_le_<mode>): Likewise.
30150 (*vsx_lxvd2x2_le_<mode>): Likewise.
30151 (*vsx_stxvd2x2_le_<mode>): Likewise.
30153 2017-03-09 Michael Eager <eager@eagercon.com>
30155 Correct failures with --enable-checking=yes,rtl.
30157 * config/microblaze/microblaze.c (microblaze_expand_shift):
30158 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30159 test for const0_rtx.
30160 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30161 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30163 2017-03-09 Richard Biener <rguenther@suse.de>
30165 PR tree-optimization/79977
30166 * graphite-scop-detection.c (scop_detection::merge_sese):
30167 Handle the case of extra exits to blocks dominating the entry.
30169 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30171 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30174 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30176 PR rtl-optimization/79949
30177 * lra-constraints.c (process_alt_operands): Check memory when
30178 trying to predict a cycle. Print about the overall increase.
30180 2017-03-09 Richard Biener <rguenther@suse.de>
30182 PR middle-end/79971
30183 * gimple-expr.c (useless_type_conversion_p): Preserve
30184 TYPE_SATURATING for fixed-point types.
30186 2017-03-09 Richard Biener <rguenther@suse.de>
30189 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30190 alignment of BLKmode params.
30192 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30195 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30196 (VALL_NO_V2Q): Likewise.
30198 * config/aarch64/aarch64-simd.md
30199 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30201 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30202 VALL_NO_V2Q mode iterator.
30203 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30205 2017-03-09 Martin Liska <mliska@suse.cz>
30207 PR tree-optimization/79631
30208 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30209 tree_int_cst_sign_bit just for INTEGER constants.
30211 2017-03-09 Martin Liska <mliska@suse.cz>
30215 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30218 2017-03-09 Marek Polacek <polacek@redhat.com>
30221 * tree.c (inchash::add_expr): Handle TREE_VEC.
30223 2017-03-09 Martin Liska <mliska@suse.cz>
30226 (chkp_narrow_size_and_offset): New function.
30227 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30228 (void chkp_parse_bit_field_ref): New function.
30229 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30230 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30232 2017-03-09 Martin Liska <mliska@suse.cz>
30235 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30236 (chkp_find_bounds_1): Remove gcc_unreachable.
30238 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30241 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30242 BUILT_IN_SYNC*, determine the access type from the size suffix and
30243 always build a MEM_REF with that type. Handle forgotten
30244 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30247 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30248 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30249 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30250 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30251 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30252 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30253 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30254 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30255 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30256 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30257 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30258 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30259 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30260 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30261 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30262 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30263 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30264 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30265 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30266 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30267 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30268 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30269 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30270 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30271 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30272 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30273 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30274 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30275 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30276 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30277 definitions outside of __OPTIMIZE__ guarded section.
30280 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30281 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30282 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30283 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30286 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30288 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30289 ("vfenez<mode>"): Add missing constraints.
30291 2017-03-08 Martin Sebor <msebor@redhat.com>
30294 * config/nds32/nds32.c (nds32_option_override):
30295 Fix misspelled diagnostic.
30297 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30300 * gimplify.c (gimplify_omp_for): Replace index var in outer
30301 taskloop statement with an artificial variable and add
30302 OMP_CLAUSE_PRIVATE clause for it.
30304 2017-03-08 Richard Biener <rguenther@suse.de>
30306 PR tree-optimization/79955
30307 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30308 for accesses that are completely outside of the variable.
30310 2017-03-08 Andrew Haley <aph@redhat.com>
30312 PR tree-optimization/79943
30313 * tree-ssa-loop-split.c (compute_new_first_bound): When
30314 calculating the new upper bound, (END-BEG) should be added, not
30317 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30319 * config/avr/avr.md (setmemhi): Make sure match_dup
30320 operand number comes before match_scratch.
30322 2017-03-08 Richard Biener <rguenther@suse.de>
30324 PR tree-optimization/79920
30325 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30326 with ncopies == 1 to ...
30327 (vect_transform_slp_perm_load): ... here. Properly compute
30328 all element loads by iterating VF times over the group. Do
30329 not handle ncopies (computed in a broken way) in
30330 vect_create_mask_and_perm.
30332 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30335 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30336 is a uniform vector, use uniform_vector_p return value instead of
30337 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30339 2017-03-07 Marek Polacek <polacek@redhat.com>
30341 PR middle-end/79809
30342 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30343 (alloca_call_type): Likewise.
30345 2017-03-07 Martin Liska <mliska@suse.cz>
30347 * gcov.c (process_args): Put comment to correct location.
30349 2017-03-07 Martin Liska <mliska@suse.cz>
30351 PR middle-end/68270
30352 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30353 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30354 (chkp_narrow_bounds_for_field): Likewise.
30355 (chkp_parse_array_and_component_ref): Pass one more argument to
30358 2017-03-07 Richard Biener <rguenther@suse.de>
30360 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30363 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30365 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30366 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30368 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30371 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30372 to end of description.
30373 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30375 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30377 PR rtl-optimization/79901
30378 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30380 (*avx512f_<code><mode>3<mask_name>): ... this.
30381 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30382 iterator instead of VI8_AVX2_AVX512BW.
30384 PR rtl-optimization/79901
30385 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30386 min/max expander, expand it using expand_vec_cond_expr.
30389 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30392 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30395 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30396 to void * for PCH reasons.
30397 * dwarf2out.c (output_loc_operands, output_die): Cast
30398 v.val_vec.array to unsigned char *.
30400 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30403 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30406 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30408 PR rtl-optimization/79571
30409 * lra-constraints.c (process_alt_operands): Calculate static
30410 reject and subtract it from overall when only addresses will be
30413 2017-03-06 Julia Koval <julia.koval@intel.com>
30416 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30417 incoming stack boundary to 128 for 64-bit targets.
30419 2017-03-06 Richard Biener <rguenther@suse.de>
30421 PR tree-optimization/79894
30422 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30423 to NULL after folding it.
30425 2017-03-06 Richard Biener <rguenther@suse.de>
30427 PR tree-optimization/79824
30428 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30429 check disabling peeling for gaps.
30431 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30433 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30434 attributes): Document gettimeofday.
30436 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30438 * config/s390/s390.c (s390_option_override_internal): Set
30439 PARAM_MIN_VECT_LOOP_BOUND
30441 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30443 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30444 * config/s390/s390.md: Likewise.
30446 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30449 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30450 (<avx2_avx512>_perm<mode>): Rename to ...
30451 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30453 (<avx512>_perm<mode>_mask): Rename to ...
30454 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30456 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30457 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30458 instead of VI8F_256_512.
30459 (avx512f_perm<mode>): New define_expand.
30460 (avx512f_perm<mode>_mask): Likewise.
30461 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30462 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30464 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30466 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30467 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30469 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30471 2017-03-06 Martin Liska <mliska@suse.cz>
30474 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30475 when having a SSA NAME w/o VAR_DECL assigned to it.
30477 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30479 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30480 msa_dpsub_<su>_d): Fix MODE for vec_select.
30482 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30484 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30486 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30488 2017-03-06 Richard Biener <rguenther@suse.de>
30490 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30491 * plugin.c (register_plugin_info): Likewise.
30492 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30494 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30496 * config/i386/sse.md (sse_storehps, sse_storelps,
30497 avx_<castmode><avxsizesuffix>_<castmode>,
30498 avx512f_<castmode><avxsizesuffix>_<castmode>,
30499 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30500 in condition that at least one operand is not a MEM.
30502 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30504 PR middle-end/79805
30505 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30506 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30508 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30509 gimple_call_nothrow_p flag based on whether original builtin can throw.
30510 If it can, emit following stmts on the fallthrough edge.
30511 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30512 don't create new bb if inserting just debug stmts on the edge, try to
30513 insert them on the fallthru bb or just reset debug stmts.
30515 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30518 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30519 restore recog_data (including the operand rtxes inside it) around
30520 the call to get_insn_template.
30522 2017-03-03 Martin Sebor <msebor@redhat.com>
30524 PR tree-optimization/79699
30525 * context.c (context::~context): Free MPFR caches to avoid
30526 a memory leak on program exit.
30528 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30530 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30531 Use wide_int::ulow () instead of .elt (0).
30533 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30535 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30536 (*pushxf): Limit oF constraint to 32bit targets and add oC
30537 constraint for 64bit targets.
30538 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30539 (*pushdf): Change rmF constraint to rmC.
30541 2017-03-03 Martin Liska <mliska@suse.cz>
30543 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30544 Remove unused variable.
30546 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30549 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30550 is a memory operand, increase num_memory.
30551 (ix86_expand_args_builtin): Likewise.
30553 2017-03-03 Jan Hubicka <jh@suse.cz>
30556 * ipa-devirt.c (maybe_record_node): Properly handle
30557 __cxa_pure_virtual visibility.
30559 2017-03-03 Martin Liska <mliska@suse.cz>
30561 PR tree-optimization/79803
30562 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30564 (pass_loop_prefetch::execute): Disabled optimization if an
30565 assumption about L1 cache size is not met.
30567 2017-03-03 Martin Liska <mliska@suse.cz>
30569 PR rtl-optimization/79574
30570 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30571 (hash_scan_set): Likewise.
30572 (dump_hash_table): Likewise.
30573 (hoist_code): Likewise.
30575 2017-03-03 Richard Biener <rguenther@suse.de>
30577 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30578 in place of uhigh ().
30579 (fixed_convert_from_real): Likewise.
30581 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30584 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30586 2017-03-03 Richard Biener <rguenther@suse.de>
30588 PR middle-end/79818
30589 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30590 TYPE_OVERFLOW_UNDEFINED check.
30592 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30594 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30596 (vector_ae_<mode>_p): Likewise.
30597 (vector_nez_<mode>_p): Likewise.
30598 (vector_ne_v2di_p): Likewise.
30599 (vector_ae_v2di_p): Likewise.
30600 (vector_ne_<mode>_p): Likewise.
30601 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30603 (vsx_tsqrt<mode>2_fe): Likewise.
30605 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30608 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30610 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30612 PR rtl-optimization/79780
30613 * cprop.c (one_cprop_pass): When second and further conditional trap
30614 in a single basic block is turned into an unconditional trap, turn it
30615 into a deleted note to avoid RTL verification failures.
30617 2017-03-02 Richard Biener <rguenther@suse.de>
30619 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30621 2017-03-02 Richard Biener <rguenther@suse.de>
30623 PR tree-optimization/79345
30625 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30626 param and abort the walk, returning -1 if it is hit.
30627 (walk_aliased_vdefs): Take a limit param and pass it on.
30628 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30629 defaulting to 0 and return a signed int.
30630 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30631 (check_defs): New helper.
30632 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30633 about uninitialized memory.
30634 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30635 bogus uninitialized warning.
30636 (fixed_convert_from_real): Likewise.
30638 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30640 PR tree-optimization/66768
30641 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30642 iv_use if base object can't be determined.
30644 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30646 PR tree-optimization/79345
30647 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30648 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30649 (get_pattern_stats): Initialize it.
30650 * genemit.c (gen_expand): Verify match_scratch numbers come after
30651 match_operand/match_dup numbers.
30652 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30653 match_scratch numbers.
30654 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30656 * config/s390/s390.md (trunctdsd2): Likewise.
30658 2017-03-02 Richard Biener <rguenther@suse.de>
30660 * wide-int.h (wide_int_storage::operator=): Implement in terms
30663 2017-03-02 Richard Biener <rguenther@suse.de>
30665 PR tree-optimization/79777
30666 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30667 the to insert expression to sth existing.
30669 2017-03-01 Martin Sebor <msebor@redhat.com>
30671 PR middle-end/79692
30672 * gimple-ssa-sprintf.c
30673 (directive::known_width_and_precision): New function.
30674 (format_integer): Use it.
30675 (get_mpfr_format_length): Consider the full range of precision
30676 when computing %g output with the # flag. Set the likely byte
30677 count to 3 rather than 1 when precision is indeterminate.
30678 (format_floating): Correct the lower bound of precision.
30680 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30682 * doc/invoke.texi: Document default code model for 64-bit Linux.
30684 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30687 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30688 udiv rather than div since input pattern is unsigned.
30690 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30692 * config/i386/i386.c (print_reg): Warn for values of
30693 unsupported size in integer register.
30695 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30698 * config/rs6000/predicates.md (current_file_function_operand): Do
30699 not allow self calls to be local if the function is replaceable.
30701 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30704 * config/rs6000/altivec.h (vec_ctz and others): Change the
30705 preprocessor macro that controls conditional compilation from
30706 _ARCH_PWR9 to __POWER9_VECTOR__.
30707 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30708 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30709 control (instead of _ARCH_PWR9 control) so that template
30710 definition uses power9-specific function.
30711 (vec_any_eq): Likewise.
30712 (vec_all_ne): Change macro definition to use a power9-specific
30713 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30714 _ARCH_PWR9 control).
30715 (vec_any_eq) Likewise.
30716 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30717 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30718 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30719 support for xvcmpnedp instruction.
30720 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30721 macro expansion so that Power9 implementation of vec_all_ne does
30722 not use the AltiVec predicate framework.
30723 (VCMPNEH_P): Likewise.
30724 (VCMPNEW_P): Likewise.
30725 (VCMPNED_P): Likewise.
30726 (VCMPNEFP_P): Likewise.
30727 (VCMPNEDP_P): Likewise.
30728 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30729 implementation of vec_any_eq to not use AltiVec predicate
30731 (VCMPAEH_P): Likewise.
30732 (VCMPAEW_P): Likewise.
30733 (VCMPAED_P): Likewise.
30734 (VCMPAEFP_P): Likewise.
30735 (VCMPAEDP_P): Likewise.
30736 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30737 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30738 not use the AltiVec predicate framework.
30739 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30740 of vec_any_eq to not use AltiVec predicate framework.
30741 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30742 support for predefined __POWER9_VECTOR__ macro to indicate that
30743 Power9 instruction selection is enabled.
30744 (altivec_overloaded_builtins): Remove extraneous
30745 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30746 function argument types RS6000_BTI_bool_V16QI and
30747 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30748 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30749 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30750 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30751 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30752 Power9 for implementations of vec_cmpne. Change the signature for
30753 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30754 (representing vec_all_ne) to remove the previously described first
30755 argument of type RS6000_BTI_INTSI, as this was an artifact of
30756 reliance on the AltiVec predicate framework, which is no longer
30757 used in the implementation of these functions. Add
30758 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30759 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30760 since, unlike the AltiVec predicate framework implementation, we
30761 do not share function descriptors between vec_alle and vec_anyeq.
30762 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30763 set of modes that receive special treatment even when
30764 TARGET_P9_VECTOR is true. The special treatment emits code that
30765 does not depend on Power9 instructions.
30766 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30767 define_expand to not rely on AltiVec predicate framework.
30768 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30770 (vector_ne_v2di_p): Change this define_expand to not rely on
30771 AltiVec predicate framework.
30772 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30774 (vector_ne_<mode>_p): Change this define_expand to not rely on
30775 AltiVec predicate framework.
30776 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30778 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30779 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30780 define_insn pattern.
30781 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30782 define_insn pattern because the xvcmpne<VSs>. instruction is not
30784 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30785 instruction is not supported.
30787 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30789 * config/nvptx/nvptx.c: Include intl.h.
30791 2017-03-01 Martin Jambor <mjambor@suse.cz>
30794 * ipa-prop.h (ipa_bits): Removed field known.
30795 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30796 to pointers. Adjusted their comments to warn about their sharing.
30797 (ipcp_transformation_summary): Change bits to a vector of pointers.
30798 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30799 (ipa_get_ipa_bits_for_value): Declare.
30800 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30801 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30802 (ipa_bits_hash_table): Likewise.
30803 (ipa_vr_ggc_hash_traits): Likewise.
30804 (ipa_vr_hash_table): Likewise.
30805 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30806 being pointers and vr_known being removed.
30807 (ipa_set_jf_unknown): Likewise.
30808 (ipa_get_ipa_bits_for_value): New function.
30809 (ipa_set_jfunc_bits): Likewise.
30810 (ipa_get_value_range): New overloaded functions.
30811 (ipa_set_jfunc_vr): Likewise.
30812 (ipa_compute_jump_functions_for_edge): Use the above functions to
30813 construct bits and vr parts of jump functions.
30814 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30815 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30817 (ipcp_grow_transformations_if_necessary): Also allocate
30818 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30820 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30821 them. Fix too long lines.
30822 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30823 vr_known being removed.
30824 (ipa_read_jump_function): Use new setter functions to construct bits
30825 and vr parts of jump functions or set them to NULL.
30826 (write_ipcp_transformation_info): Adjust for bits being pointers.
30827 (read_ipcp_transformation_info): Likewise.
30828 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30830 Include gt-ipa-prop.h.
30831 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30833 (ipcp_store_bits_results): Likewise.
30834 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30835 Do not write to existing jump functions but use a temporary instead.
30837 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30840 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30841 attempt to use its first operand as BIT_FIELD_REF base.
30843 2017-03-01 Richard Biener <rguenther@suse.de>
30845 PR middle-end/79721
30846 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30847 interpolating formula in wrapping arithmetic.
30848 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30850 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30852 PR tree-optimization/79734
30853 * tree-vect-generic.c (expand_vector_condition): Optimize
30854 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30855 Handle VEC_COND_EXPR where comparison has different inner width from
30856 type's inner width.
30858 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30860 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30861 markup, and similar issues. Remove @opindex entries for things
30862 that aren't options. Add missing -mmpy-option entries.
30864 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30866 PR tree-optimization/79737
30867 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30868 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30869 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30870 instead of byte_size. Formatting fix.
30871 (shift_bytes_in_array_right): Formatting fix.
30873 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30876 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30877 condition on optimize for the leaf function test.
30879 2017-02-28 Martin Liska <mliska@suse.cz>
30882 * read-rtl-function.c (function_reader::handle_unknown_directive):
30883 Bail out when one uses -flto.
30885 2017-02-28 Martin Liska <mliska@suse.cz>
30887 * common.opt: Replace space with tabular for options of <number>
30889 * config/i386/i386.opt: Show <number> value for
30890 -mlarge-data-threshold.
30891 * opts.c (print_filtered_help): Do not display number in hexadecimal
30894 2017-02-28 Martin Liska <mliska@suse.cz>
30896 * common.opt: Fix --help=option -Q for options which are of
30899 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30901 * config/i386/i386.c (print_reg): Error out for values
30902 of 8-bit size in invalid integer register.
30904 2017-02-28 Martin Sebor <msebor@redhat.com>
30906 PR tree-optimization/79691
30907 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30909 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30912 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30913 gcc_unreachable with output_operand_lossage.
30915 2017-02-28 Richard Biener <rguenther@suse.de>
30917 PR tree-optimization/79740
30918 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30920 (visit_nary_op): Insert the nary into the hashtable if we
30921 pattern-matched sth.
30922 * tree-ssa-pre.c (eliminate_insert): Robustify.
30924 2017-02-28 Richard Biener <rguenther@suse.de>
30926 PR middle-end/79731
30927 * fold-const.c (decode_field_reference): Reject out-of-bound
30930 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30932 * config/i386/i386.c: Include intl.h.
30933 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30934 instead of just cond ? "..." : "...".
30935 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30936 * coverage.c (read_counts_file): Likewise.
30937 * omp-offload.c: Include intl.h.
30938 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30939 of just cond ? "..." : "...".
30940 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30941 of just cond ? "..." : "...".
30943 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30946 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30948 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30949 'tune for' CPU name.
30950 * config/arm/arm-cpu-data.h: Regenerated.
30952 2017-02-28 Richard Biener <rguenther@suse.de>
30954 PR tree-optimization/79732
30955 * tree-inline.c (expand_call_inline): Do not shadow var.
30957 2017-02-28 Richard Biener <rguenther@suse.de>
30959 PR tree-optimization/79723
30960 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30961 address-space properly.
30963 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30965 * doc/optinfo.texi (Optimization groups): Fix option used for
30967 * doc/invoke.texi (-fopt-info): Document "omp".
30968 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30969 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30970 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30971 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30972 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30974 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30976 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30977 OPTGROUP_OMP with "omp".
30979 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30982 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30983 for arithmetic shift of unsigned V2DI.
30985 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30987 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30988 arc/linux.h headers.
30989 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30990 (LINK_SPEC): Likewise.
30991 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30992 (EXTRA_SPECS): Likewise.
30993 (STARTFILE_SPEC): Likewise.
30994 (ENDFILE_SPEC): Likewise.
30995 (LIB_SPEC): Likewise.
30996 (TARGET_SDATA_DEFAULT): Likewise.
30997 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30998 (MULTILIB_DEFAULTS): Likewise.
30999 (DWARF2_UNWIND_INFO): Likewise.
31000 * config/arc/big.h: New file.
31001 * config/arc/elf.h: Likewise.
31002 * config/arc/linux.h: Likewise.
31003 * config/arc/t-uClibc: Remove.
31005 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31007 PR tree-optimization/77536
31008 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31009 (tree_transform_and_unroll_loop): Use above function to compute the
31010 estimated niter of unrolled loop and use it when scaling profile.
31011 Also use count info rather than frequency if it's non-zero.
31012 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31013 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31014 (vect_transform_loop): Call above function.
31016 2017-02-27 Richard Biener <rguenther@suse.de>
31018 PR tree-optimization/45397
31019 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31020 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31021 (visit_nary_op): Add pattern matching for CSEing sign-changed
31022 or truncated operations with wider ones.
31024 2017-02-27 Richard Biener <rguenther@suse.de>
31026 PR tree-optimization/79690
31027 * tree-vect-stmts.c (vectorizable_store): Use vector type
31028 built from the DR with address-space.
31030 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31032 * doc/invoke.texi (Optimize Options): Refine the description
31033 of asan-use-after-return.
31035 2017-02-25 Alan Modra <amodra@gmail.com>
31037 PR rtl-optimization/79584
31038 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31039 base, not ad->base_term, the reg within base. Remove assertion
31040 that ad->base == ad->base_term. Replace gen_int_mode using
31041 bogus mode with const0_rtx.
31043 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31045 PR middle-end/79396
31046 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31047 FMA_EXPR like tcc_binary or tcc_unary.
31049 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31052 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31054 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31055 (output_loc_operands): Handle DW_OP_call_ref and
31056 DW_OP_GNU_variable_value.
31057 (struct variable_value_struct): New type.
31058 (struct variable_value_hasher): Likewise.
31059 (variable_value_hash): New variable.
31060 (string_types): Remove.
31061 (copy_loc_descr): New function.
31062 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31063 (prepend_loc_descr_to_each): New function.
31064 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31065 instead of add_loc_descr_to_each if the first argument is single
31066 location list and the second has multiple.
31067 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31068 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31069 when looking for variable value which doesn't have other location info.
31070 (loc_list_from_tree): Formatting fix.
31071 (gen_array_type_die): Simplify DW_AT_string_length handling.
31072 (adjust_string_types): Remove.
31073 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31074 string_types. Call resolve_variable_values.
31075 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31076 (resolve_addr_in_expr): Likewise. Add A argument.
31077 (copy_deref_exprloc): Remove deref argument. Adjust for the
31078 original expression being DW_OP_GNU_variable_value with optionally
31079 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31080 optionally after it.
31081 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31082 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31083 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31084 (variable_value_hasher::hash, variable_value_hasher::equal): New
31086 (resolve_variable_value_in_expr, resolve_variable_value,
31087 resolve_variable_values, note_variable_value_in_expr,
31088 note_variable_value): New functions.
31089 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31092 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31095 * opts.h (handle_generated_option): Add GENERATED_P argument.
31096 * opts-common.c (handle_option): Adjust function comment.
31097 (handle_generated_option): Add GENERATED_P argument, pass it to
31099 (control_warning_option): Pass false to handle_generated_option
31101 * opts.c (maybe_default_option): Pass true to handle_generated_option
31103 * optc-gen.awk: Likewise.
31105 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31107 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31108 a REG, look at the REG it is a SUBREG of.
31109 (splitter for cmpeqsi_t): Ditto.
31111 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31113 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31114 the special USEs with the pattern of the insn, not the insn itself.
31116 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31119 * doc/invoke.texi: Document -mload-store-pairs.
31121 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31122 Sandra Loosemore <sandra@codesourcery.com>
31124 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31125 argument isn't a CONST_INT.
31126 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31127 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31128 (nios2_validate_compare): Bypass alternate compare logic if *op2
31129 is not a CONST_INT.
31130 (ldstwm_operation_p): Return false if first_base is not a REG or
31131 if first_offset is not a CONST_INT.
31133 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31135 * config/cris/cris.md: Use correct operand in a define_peephole2.
31137 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31139 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31141 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31143 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31144 this_insn if it is an INSN or JUMP_INSN.
31145 (force_offsettable): Look at base, not at addr.
31146 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31147 on things that aren't necessarily CONST_INTs.
31149 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31151 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31152 -mfpmath=sse is the default also for x86-32 targets with SSE2
31153 instruction set when @option{-ffast-math} is enabled
31155 2017-02-24 Jeff Law <law@redhat.com>
31157 PR rtl-optimizatoin/79286
31158 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31161 2017-02-24 Richard Biener <rguenther@suse.de>
31163 PR tree-optimization/79389
31164 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31167 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31169 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31170 function comment to reflect reality.
31171 (loop_exits_before_overflow): Fix typo in function description.
31173 2017-02-24 Richard Biener <rguenther@suse.de>
31175 PR tree-optimization/79389
31176 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31177 properly that a threading opportunity exists. Detect conditional
31178 copy/constant propagation opportunities.
31180 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31182 * config/visium/visium.md (type): Add trap.
31183 (b): New mode attribute.
31184 (*btst): Rename into...
31185 (*btst<mode>): ...this and adjust.
31186 (*cbranchsi4_btst_insn): Rename into...
31187 (*cbranch<mode>4_btst_insn): ...this and adjust.
31188 (trap): New define_insn.
31190 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31192 PR tree-optimization/79389
31193 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31194 (noce_reversed_cond_code): New function.
31195 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31196 reversed_comparison_code. Formatting fix.
31197 (noce_try_store_flag): Test rev_cond != NULL in addition to
31198 reversed_comparison_code.
31199 (noce_try_store_flag_constants): Likewise.
31200 (noce_try_store_flag_mask): Likewise.
31201 (noce_try_addcc): Use rev_cond if non-NULL instead of
31202 reversed_comparison_code.
31203 (noce_try_cmove_arith): Likewise. Formatting fixes.
31204 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31205 (noce_find_if_block): Initialize rev_cond.
31206 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31207 instead of false as last argument never attempt to reverse it
31210 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31212 PR tree-optimization/79663
31213 * tree-predcom.c (combine_chains): Process refs in reverse order
31214 only for ZERO length chains, and add explaining comment.
31216 2017-02-23 Jeff Law <law@redhat.com>
31218 PR tree-optimization/79578
31219 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31220 in call to operand_equal_p.
31222 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31225 * config/i386/cpuid.h: Fix another undefined behavior.
31227 2017-02-23 Richard Biener <rguenther@suse.de>
31229 PR tree-optimization/79683
31230 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31231 vector types for data-refs.
31233 2017-02-23 Martin Liska <mliska@suse.cz>
31235 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31237 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31239 PR middle-end/79665
31240 * internal-fn.c (get_range_pos_neg): Moved to ...
31241 * tree.c (get_range_pos_neg): ... here. No longer static.
31242 * tree.h (get_range_pos_neg): New prototype.
31243 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31244 are known to be in between 0 and signed maximum inclusive, try to
31245 expand both unsigned and signed divmod and use the cheaper one from
31248 2017-02-22 Jeff Law <law@redhat.com>
31250 PR tree-optimization/79578
31251 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31252 to compare base operands.
31254 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31257 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31258 gpc_reg_operand instead of fpr_reg_operand.
31260 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31262 * config/mips/mips.c (mips_return_in_memory): Force FP
31263 vector types to be returned in memory for o32 ABI.
31265 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31267 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31268 instead of DW_TAG_member for static data member declarations and don't
31269 set no_linkage_name for static inline data members.
31270 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31273 2017-02-22 Martin Liska <mliska@suse.cz>
31275 * doc/invoke.texi: Replace inequality signs with square brackets
31278 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31281 * lra-constraints.c (simplify_operand_subreg): Handle
31282 WORD_REGISTER_OPERATIONS targets.
31284 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31287 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31288 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31289 elimination by swapping fld*.
31291 2017-02-22 Richard Biener <rguenther@suse.de>
31293 PR tree-optimization/79673
31294 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31295 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31296 irrelevant address-space qualifiers and avoiding a
31297 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31299 2017-02-22 Richard Biener <rguenther@suse.de>
31301 PR tree-optimization/79666
31302 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31303 to not symbolically negate if that may introduce undefined
31306 2017-02-22 Martin Liska <mliska@suse.cz>
31309 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31310 * data-streamer-out.c (streamer_write_gcov_count_stream):
31312 * value-prof.c (stream_out_histogram_value): Make assert more
31313 precise based on type of counter.
31315 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31318 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31319 Use nonimmediate_operand instead of memory_operand for operand 1.
31320 (float-extend standard_x87sse_constant_load splitter): Ditto.
31322 2017-02-21 Jeff Law <law@redhat.com>
31324 PR tree-optimization/79621
31325 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31326 blocks with edges to themselves.
31328 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31331 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31332 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31333 Use gimple_call_builtin_p.
31336 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31337 on temporarily removed DEBUG_INSNs.
31339 PR tree-optimization/79649
31340 * tree-loop-distribution.c (classify_partition): Give up on
31341 non-generic address space loads/stores.
31343 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31345 * doc/loop.texi (Loop manipulation): Remove nonexistent
31346 tree_ssa_loop_version from the documentation.
31347 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31349 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31352 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31353 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31354 * config/rs6000/rs6000.c: Include except.h.
31355 (rs6000_expand_split_stack_prologue): Call
31356 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31358 2017-02-21 Martin Jambor <mjambor@suse.cz>
31361 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31362 have been analyzed.
31364 2017-02-21 Martin Jambor <mjambor@suse.cz>
31366 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31367 for backward compatibility only.
31368 * doc/invoke.texi (Option Summary): Remove all references to
31369 -fipa-cp-alignment.
31371 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31375 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31377 * lra-constraints.c (curr_insn_transform): Handle
31378 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31380 2017-02-21 Martin Liska <mliska@suse.cz>
31382 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31384 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31386 PR translation/79638
31387 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31389 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31392 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31393 (arm_function_ok_for_sibcall): Return false for an indirect call by
31394 descriptor if all the argument registers are used.
31395 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31396 alignment of the function.
31398 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31400 PR tree-optimization/61441
31401 * simplify-rtx.c (simplify_const_unary_operation): For
31402 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31403 the sNaN unmodified.
31405 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31407 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31408 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31409 instead of SYSTEM_HEADER_DIR.
31411 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31412 Martin Liška <mliska@suse.cz>
31414 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31415 Fix typos and grammar, use active voice, and clarify.
31417 2017-02-20 Marek Polacek <polacek@redhat.com>
31419 PR middle-end/79537
31420 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31423 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31425 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31428 * config/i386/i386.c (ix86_expand_builtin): Handle
31429 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31430 ix86_builtins_isa[fcode].isa as a requirement of those
31431 flags and any other flag in the bitmask.
31432 (ix86_init_mmx_sse_builtins): Use 0 instead of
31433 ~OPTION_MASK_ISA_64BIT as mask.
31434 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31435 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31436 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31437 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31439 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31442 * lra-constraints.c (split_reg): Check requested split mode
31443 is supported by the register.
31445 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31447 * lra-constraints.c (simplify_operand_subreg): Remove early
31450 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31453 * lra-constraints.c (curr_insn_transform): Tighten condition
31454 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31456 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31459 * lra-constraints.c (curr_insn_transform): Handle
31460 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31462 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31465 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31467 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31469 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31472 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31475 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31477 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31480 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31482 * final.c (last_columnnum, override_columnnum): New variables.
31483 (final_start_function): Set last_columnnum, pass it to begin_prologue
31484 hook and pass 0 to dwarf2out_begin_prologue.
31485 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31486 to source_line debug hook.
31487 (notice_source_line): Compute last_columnnum and for debug_column_info
31488 return true on column changes.
31489 * debug.h (struct gcc_debug_hooks): Add column argument to
31490 source_line and begin_prologue hooks.
31491 (debug_nothing_int_charstar_int_bool): Remove prototype.
31492 (debug_nothing_int_int_charstar,
31493 debug_nothing_int_int_charstar_int_bool): New prototypes.
31494 (dwarf2out_begin_prologue): Add column argument.
31495 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31496 begin_prologue hooks.
31497 (debug_nothing_int_charstar_int_bool): Remove.
31498 (debug_nothing_int_int_charstar,
31499 debug_nothing_int_int_charstar_int_bool): New functions.
31500 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31501 through to dwarf2out_source_line.
31502 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31503 (dwarf2out_source_line): Add column argument, emit it if requested.
31504 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31506 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31507 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31508 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31509 through to dwarf2out_begin_prologue.
31510 (vmsdbgout_source_line): Add column argument, pass it through to
31511 dwarf2out_source_line.
31512 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31513 dbxout_source_line caller.
31514 (dbxout_source_line): Add column argument.
31516 * common.opt (gno-column-info, gcolumn-info): New options.
31517 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31518 (check_die): Also test for multiple DW_AT_decl_column attributes.
31519 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31520 DW_AT_decl_column if requested.
31521 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31523 (gen_variable_die): Likewise.
31524 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31525 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31528 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31529 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31530 (ix86_handle_option): Handle OPT_m3dnowa.
31531 * doc/invoke.texi (-m3dnowa): Document.
31532 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31533 -m3dnowa instead of -m3dnow -march=athlon.
31536 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31537 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31539 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31542 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31543 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31544 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31545 generator for vsx_xxpermdi_<mode>_be.
31546 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31547 force big-endian semantics.
31548 (vsx_xxpermdi_<mode>_be): New define_expand with same
31549 implementation as previous version of vsx_xxpermdi_<mode>.
31551 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31553 PR tree-optimization/79327
31554 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31555 variable, its initialization and use.
31557 2017-02-17 Julia Koval <julia.koval@intel.com>
31559 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31560 (OPTION_MASK_ISA_PKU_UNSET): New.
31561 (ix86_handle_option): Handle -mrdpid.
31562 * config/i386/cpuid.h (bit_RDPID): New.
31563 * config/i386/driver-i386.c (host_detect_local_cpu):
31564 Detect RDPID feature.
31565 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31566 * config/i386/i386-c.c (ix86_target_macros_internal):
31568 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31569 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31570 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31571 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31572 * config/i386/i386.md (define_insn "rdpid"): New.
31573 * config/i386/i386.opt Add -mrdpid.
31574 * config/i386/immintrin.h (_rdpid_u32): New.
31576 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31578 PR rtl-optimization/79541
31579 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31580 instead of transforming it into USE.
31582 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31584 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31585 If HONOR_SNANS (SFmode) force the input to a register.
31586 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31587 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31588 an frsp or similar insn.
31590 2017-02-17 Martin Liska <mliska@suse.cz>
31592 PR rtl-optimization/79577
31593 * params.def (selsched-max-sched-times): Increase minimum to 1.
31595 2017-02-17 Martin Liska <mliska@suse.cz>
31597 PR rtl-optimization/79574
31598 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31600 2017-02-17 Martin Liska <mliska@suse.cz>
31602 PR tree-optimization/79529
31603 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31604 ssa_defined_default_def_p to handle cases which are implicitly
31606 * tree-ssa.c (ssa_defined_default_def_p): New function.
31607 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31608 which are implicitly defined.
31609 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31611 2017-02-17 Richard Biener <rguenther@suse.de>
31613 PR middle-end/79576
31614 * params.def (max-ssa-name-query-depth): Limit to 10.
31616 2017-02-17 Richard Biener <rguenther@suse.de>
31618 PR tree-optimization/79552
31619 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31622 2017-02-17 Richard Biener <rguenther@suse.de>
31625 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31627 2017-02-17 Marek Polacek <polacek@redhat.com>
31629 PR middle-end/79536
31630 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31631 (fold_negate_expr): New wrapper.
31633 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31635 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31636 Correct terminology and de-emphasize pre-standard behavior.
31638 2017-02-16 Alan Modra <amodra@gmail.com>
31640 PR rtl-optimization/79286
31641 * ira.c (def_dominates_uses): New function.
31642 (update_equiv_regs): Don't create an equivalence for insns that
31643 may trap where the register def does not dominate the use.
31645 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31647 PR rtl-optimization/78127
31648 * lra.c (lra): Call lra_eliminate before finish the loop after
31651 2017-02-16 Richard Biener <rguenther@suse.de>
31653 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31655 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31656 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31657 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31658 (isl_val_int_from_wi): New function.
31659 (extract_affine_gmp): Rename to ...
31660 (extract_affine_wi): ... this, take a widest_int.
31661 (extract_affine_int): Just wrap extract_affine_wi.
31662 (add_param_constraints): Use isl_val_int_from_wi.
31663 (add_loop_constraints): Likewise, and extract_affine_wi.
31665 2017-02-15 Jeff Law <law@redhat.com>
31667 PR middle-end/79521
31668 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31669 ira_init_register_move_cost_if_necessary.
31671 2017-02-15 Martin Sebor <msebor@redhat.com>
31673 PR middle-end/32003
31674 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31675 removed in a prior commit.
31677 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31679 PR tree-optimization/79347
31680 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31681 counters during peeling.
31683 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31685 * Makefile.in (site.exp): Remove "set ISLVER".
31687 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31690 * real.c (real_from_integer): Call real_convert even for decimal.
31692 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31695 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31697 2017-02-14 Andrew Pinski <apinski@cavium.com>
31699 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31700 cores and change the partno/implementer to be correct.
31701 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31702 the 'B" as the implementer.
31703 * config/aarch64/aarch64-tune.md: Regenerate.
31705 2017-02-14 Carl Love <cel@us.ibm.com>
31707 * config/rs6000/rs6000.c: Add case statement entry to make the
31708 xvcvuxdsp built-in argument unsigned.
31709 * config/rs6000/vsx.md: Fix the source and return operand types so they
31710 match the instruction definitions from the ISA document. Fix typo
31711 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31714 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31717 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31718 member early_clobber_alts.
31719 * lra-lives.c (reg_early_clobber_p): New.
31720 (process_bb_lives): Use it.
31721 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31722 (debug_operand_data): Initialize early_clobber_alts.
31723 (setup_operand_alternative): Set up early_clobber_alts.
31724 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31725 alternatives to new_insn_reg.
31726 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31728 (lra_update_insn_regno_info): Pass the new arg.
31730 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31732 PR middle-end/79505
31733 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31734 (new_oacc_loop_raw): Don't clear already cleared fields.
31737 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31738 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31739 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31741 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31744 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31746 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31749 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31750 the extra instruction to the right place to store 128-bit constant
31753 2017-02-14 Martin Sebor <msebor@redhat.com>
31755 PR middle-end/79448
31756 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31757 warning for strings of unknown length.
31759 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31761 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31763 2017-02-14 Jeff Law <law@redhat.com>
31766 * ira-costs.c (scan_one_insn): Initialize register move costs
31767 for pseudos seen in USE/CLOBBER insns.
31769 PR tree-optimization/79095
31770 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31771 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31772 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31773 if the operands are known to be not equal, then the resulting range
31775 (intersect_ranges): If the new range is ~[0,0] and the old range is
31776 wide, then prefer ~[0,0].
31777 * tree-vrp.c (overflow_comparison_p_1): New function.
31778 (overflow_comparison_p): New function.
31779 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31780 if NAME is used in an overflow test.
31781 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31782 overflow check that can be expressed as an equality test, then adjust
31783 ops to be that equality test.
31785 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31787 * config/s390/s390-builtin-types.def: Remove flags argument.
31788 * config/s390/s390.c (s390_init_builtins): Likewise.
31790 2017-02-14 Martin Liska <mliska@suse.cz>
31792 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31793 vector. Fix trailing white spaces.
31795 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31797 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31800 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31802 PR rtl-optimization/68664
31803 * config/arm/arm.c (arm_sched_can_speculate_insn):
31804 New function. Declare prototype.
31805 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31807 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31809 PR rtl-optimization/68664
31810 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31812 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31814 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31816 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31817 max skip bytes for function, loop and jump.
31819 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31821 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31822 ABS_EXPR for gimple dump.
31824 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31827 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31829 PR tree-optimization/79408
31830 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31831 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31832 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31833 also if rhs1 is INTEGER_CST.
31835 2017-02-14 Richard Biener <rguenther@suse.de>
31837 PR middle-end/79432
31838 * tree-into-ssa.c (insert_phi_nodes): When the function can
31839 have abnormal edges rewrite SSA names with broken use-def
31840 dominance out of SSA and register them for PHI insertion.
31842 2017-02-13 Martin Sebor <msebor@redhat.com>
31844 PR middle-end/79496
31845 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31846 clearing info.nowrite flag when snprintf size argument is a range.
31848 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31850 * cprop.c (cprop_jump): Add missing space in string literal.
31851 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31852 (get_constraint_for_component_ref): Likewise.
31853 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31854 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31855 * lra-constraints.c (process_alt_operands): Likewise.
31856 * ipa-inline.c (inline_small_functions): Likewise.
31857 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31858 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31859 * trans-mem.c (diagnose_tm_1_op): Likewise.
31860 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31861 (grid_parallel_clauses_gridifiable): Likewise.
31863 * config/nvptx/mkoffload.c (process): Add space in between
31866 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31867 "MOD4_SSE_REGS" and "ALL_REGS".
31869 * spellcheck.c (test_data): Add , in between "foo" and "food".
31871 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31874 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31875 boundary crossing check and subsequent code generation agree.
31877 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31879 * config/aarch64/aarch64.c (has_memory_op): Delete.
31880 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31883 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31885 PR rtl-optimization/79388
31886 PR rtl-optimization/79450
31887 * combine.c (distribute_notes): When removing TEM_INSN for which
31888 corresponding dest has last value recorded, invalidate that last
31891 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31893 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31894 of explicit '@'. Add missing assembly comment marker on branch costs
31897 2017-02-13 Nathan Sidwell <nathan@acm.org>
31899 * gengtype-lex.l (<in_struct>): Add '/'.
31901 2017-02-13 Martin Liska <mliska@suse.cz>
31904 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31906 2017-02-13 Richard Biener <rguenther@suse.de>
31908 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31910 * configure: Re-generate.
31911 * config.in: Likewise.
31912 * graphite-dependences.c: Simplify as if
31913 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31914 * graphite-isl-ast-to-gimple.c: Likewise.
31915 * graphite-optimize-isl.c: Likewise.
31916 * graphite-poly.c: Likewise.
31917 * graphite-sese-to-poly.c: Likewise.
31918 * graphite.h: Likewise.
31919 * toplev.c: Include isl/version.h and use isl_version () for
31920 printing the ISL version.
31921 * doc/install.texi: Update ISL requirement.
31923 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31925 * doc/standards.texi (Standards): Update reference to
31928 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31930 * doc/extend.texi (Named Address Spaces): sourceware.org now
31932 * doc/install.texi (Binaries): Ditto.
31935 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31937 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31938 standard terminology "stringize"/"stringizing" throughout.
31939 * doc/cppinternals.texi: Likewise.
31941 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31943 * doc/extend.texi: Fix some spelling mistakes and typos.
31944 * doc/invoke.texi: Likewise.
31946 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31949 * params.def (inline-min-speedup) Change from 10 to 8.
31951 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31953 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31956 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31959 * ipa-inline-analysis.c (get_minimal_bb): New function.
31960 (record_modified): Use it.
31961 (remap_edge_change_prob): Handle also ancestor functions.
31963 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31965 * doc/contrib.texi (Contributors): Remove broken link into
31966 the Mauve CVS repository.
31968 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31970 PR middle-end/79454
31971 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31972 result computation whenever lhs doesn't have vector mode, not
31973 just when it has BLKmode.
31975 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31977 * doc/makefile.texi (profiledbootstrap): Refer to the
31978 installation instructions only in textual form.
31980 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31983 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31985 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31987 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31988 (Specific): Update mingw-w64 reference.
31990 (Specific): Remove broken link to Renesas RX processor.
31992 2017-02-10 Richard Biener <rguenther@suse.de>
31994 * toplev.c (process_options): Do not mention obsolete graphite
31995 options when printing sorry message about missing graphite support.
31996 Mention -floop-nest-optimize.
31998 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32000 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32001 (vtst_p16): Likewise.
32002 (vtstq_p8): Likewise.
32003 (vtstq_p16): Likewise.
32005 (vtstq_p64): Likewise.
32006 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32007 (vset_lane_p64): New.
32008 (vsetq_lane_p64): New.
32010 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32012 PR tree-optimization/79411
32013 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32014 stmt operands are SSA_NAMEs used in abnormal phis.
32015 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32018 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32021 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32024 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32026 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32028 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32030 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32034 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32035 not arbitrary TREE_CONSTANT.
32038 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32039 "omp declare target link" attribute unless is_global_var.
32040 * omp-offload.c (find_link_var_op): Likewise.
32042 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32043 Chung-Lin Tang <cltang@codesourcery.com>
32045 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32047 (gimplify_adjust_omp_clauses): Don't delete TILE.
32048 (gimplify_omp_for): Deal with TILE.
32049 * internal-fn.c (expand_GOACC_TILE): New function.
32050 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32052 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32053 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32055 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32056 avoid DIV for outermost collapse var.
32057 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32058 Remove out of date comments, fix whitespace.
32059 * omp-general.c (omp_extract_for_data): Deal with tiling.
32060 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32061 adjust OLF_DIM_BASE value.
32062 (struct omp_for_data): Add tiling field.
32063 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32064 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32065 for auto loops. Remove default auto determining, moved to
32066 oacc_loop_fixed_partitions.
32067 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32068 stmts, add e_mask field.
32069 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32070 (oacc_thread_numbers): Use oacc_dim_call.
32071 (oacc_xform_tile): New.
32072 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32073 (finish_oacc_loop): Adjust for ifns vector.
32074 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32075 add case for GOACC_TILE fns.
32076 (oacc_loop_xform_loop): Delete.
32077 (oacc_loop_process): Iterate over call list directly, and add
32078 handling for GOACC_TILE fns.
32079 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32081 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32082 vector partitioning to outer loops. Assign 2 partitions to loops
32083 when available. Add TILE handling.
32084 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32085 (execite_oacc_device_lower): Process GOACC_TILE fns,
32086 ignore unknown specs.
32087 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32088 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32089 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32091 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32093 * configure.ac (ACX_BUGURL): Update.
32094 * configure: Regenerate.
32096 2017-02-09 Richard Biener <rguenther@suse.de>
32098 PR tree-optimization/69823
32099 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32100 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32102 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32104 * config/arc/arc-c.def: Add __NPS400__ definition.
32105 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32106 (TARGET_NPS400): Define.
32108 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32110 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32112 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32113 pointer, arch_info.
32114 (arc_cpu_types): Fill the arch_info field with a pointer into the
32115 arc_arch_types table.
32116 (arc_selected_cpu): Declare.
32117 * config/arc/arc.c (arc_selected_cpu): Make global.
32118 (arc_selected_arch): Delete.
32119 (arc_base_cpu): Delete.
32120 (arc_override_options): Remove references to deleted variables,
32121 update access to arch information.
32122 (ARC_OPT): Update access to arch information.
32123 (ARC_OPTX): Likewise.
32124 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32125 (TARGET_ARC600): Update access to arch information.
32126 (TARGET_ARC601): Likewise.
32127 (TARGET_ARC700): Likewise.
32128 (TARGET_EM): Likewise.
32129 (TARGET_HS): Likewise.
32130 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32133 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32136 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32137 condition/operands for integer GE/LE/GEU/LEU operations.
32139 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32141 PR translation/79397
32142 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32145 2017-02-08 Martin Jambor <mjambor@suse.cz>
32148 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32149 whether allocation happened.
32150 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32151 nothing was allocated.
32153 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32155 PR tree-optimization/79408
32156 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32157 constant, but SSA_NAME with a known integer range, use the minimum
32158 of that range instead of op1 to determine if modulo can be replaced
32159 with its first operand.
32161 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32163 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32165 2017-02-08 Richard Biener <rguenther@suse.de>
32167 PR tree-optimization/71824
32168 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32169 Check all loops contained in the merged region.
32171 2017-02-07 Andrew Pinski <apinski@cavium.com>
32173 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32175 2017-02-07 Andrew Pinski <apinski@cavium.com>
32177 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32178 (thunderxt88): Likewise.
32179 (thunderxt81): Disable LSE and change v8.1 to v8.
32180 (thunderxt83): Likewise.
32182 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32183 Richard Biener <rguenther@suse.de>
32185 PR middle-end/79399
32186 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32187 type from int to size_t.
32188 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32190 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32192 PR rtl-optimization/79386
32193 * cprop.c (bypass_conditional_jumps): Initialize
32194 bypass_last_basic_block already before splitting bbs after
32195 unconditional traps...
32196 (bypass_conditional_jumps): ... rather than here.
32199 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32200 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32201 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32202 fix -masm=intel patterns.
32204 2017-02-07 Richard Biener <rguenther@suse.de>
32206 PR tree-optimization/79256
32207 PR middle-end/79278
32208 * builtins.c (get_object_alignment_2): Use min_align_of_type
32209 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32210 and ADJUST_FIELD_ALIGN.
32212 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32214 * doc/tm.texi: Regenerate.
32215 * stor-layout.c (layout_decl): Adjust.
32216 (update_alignment_for_field): Likewise.
32217 (place_field): Likewise.
32218 (min_align_of_type): Likewise.
32219 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32220 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32221 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32222 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32223 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32224 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32225 * config/i386/i386.c (x86_field_alignment): Likewise.
32226 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32227 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32228 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32229 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32230 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32231 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32235 2017-01-30 Richard Biener <rguenther@suse.de>
32237 PR tree-optimization/79256
32238 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32239 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32242 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32244 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32245 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32246 builtins to SImode and emit a zero-extend, if necessary.
32248 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32250 * docs/invoke.texi (RISC-V Options): Alphabetize.
32252 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32254 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32257 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32259 * config/riscv/riscv.c: New file.
32260 * common/config/riscv/riscv-common.c: Likewise.
32261 * config.gcc: Likewise.
32262 * config/riscv/constraints.md: Likewise.
32263 * config/riscv/elf.h: Likewise.
32264 * config/riscv/generic.md: Likewise.
32265 * config/riscv/linux.h: Likewise.
32266 * config/riscv/multilib-generator: Likewise.
32267 * config/riscv/peephole.md: Likewise.
32268 * config/riscv/pic.md: Likewise.
32269 * config/riscv/predicates.md: Likewise.
32270 * config/riscv/riscv-builtins.c: Likewise.
32271 * config/riscv/riscv-c.c: Likewise.
32272 * config/riscv/riscv-ftypes.def: Likewise.
32273 * config/riscv/riscv-modes.def: Likewise.
32274 * config/riscv/riscv-opts.h: Likewise.
32275 * config/riscv/riscv-protos.h: Likewise.
32276 * config/riscv/riscv.h: Likewise.
32277 * config/riscv/riscv.md: Likewise.
32278 * config/riscv/riscv.opt: Likewise.
32279 * config/riscv/sync.md: Likewise.
32280 * config/riscv/t-elf-multilib: Likewise.
32281 * config/riscv/t-linux: Likewise.
32282 * config/riscv/t-linux-multilib: Likewise.
32283 * config/riscv/t-riscv: Likewise.
32284 * configure.ac: Likewise.
32285 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32286 Waterman as RISC-V maintainers.
32287 * doc/install.texi: Add RISC-V entries.
32288 * doc/invoke.texi: Add RISC-V options section.
32289 * doc/md.texi: Add RISC-V constraints section.
32290 * configure: Regenerated.
32292 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32295 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32296 false values to be constant vectors with all 0 or all 1 bits set.
32297 (vcondu<mode><mode>): Likewise.
32298 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32300 (fpmask_comparison_operator): Update comment.
32301 (vecint_comparison_operator): New predicate.
32302 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32303 vector conditionals when the true and false values are constant
32304 vectors with all 0 bits or all 1 bits set.
32306 2017-02-06 Martin Sebor <msebor@redhat.com>
32308 PR tree-optimization/79376
32309 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32311 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32313 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32314 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32315 to simplify split condition.
32317 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32319 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32320 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32323 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32325 PR rtl-optimization/68664
32326 * target.def (can_speculate_insn): New hook.
32327 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32328 * doc/tm.texi: Regenerate.
32329 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32330 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32331 (rs6000_sched_can_speculate_insn): New function.
32333 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32335 PR tree-optimization/79284
32336 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32337 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32338 vectorizable_mask_load_store, vectorizable_operation,
32339 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32340 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32341 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32342 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32343 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32344 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32345 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32346 is_gimple_assign (stmt). Replace another such test with
32347 is_gimple_assign (stmt).
32349 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32352 * config/avr/avr.c (rtl-iter.h): Include it.
32353 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32354 (avr_legitimate_combined_insn): ...and implementation.
32356 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32358 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32359 * config/s390/s390.c (s390_const_operand_ok)
32360 (s390_canonicalize_comparison, s390_extract_part)
32361 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32362 (s390_contiguous_bitmask_p, s390_rtx_costs)
32363 (legitimize_pic_address): Likewise.
32364 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32365 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32366 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32367 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32368 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32370 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32372 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32373 REGNO($0) == REGNO($1).
32375 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32377 * config/s390/linux.h(SIZE_TYPE): Add comment.
32379 2017-02-06 Julian Brown <julian@codesourcery.com>
32380 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32381 Virendra Pathak <virendra.pathak@broadcom.com>
32383 * config/aarch64/aarch64-cores.def: Change the scheduler
32385 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32386 * config/aarch64/thunderx2t99.md: New file.
32388 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32390 * doc/standards.texi (Go Language): Update link to language
32393 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32395 * tree-eh.c (lower_resx): Sanitize profile.
32396 (cleanup_empty_eh_move_lp): Likewise.
32398 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32401 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32403 * cfgloopmanip.h (loop_version): Update prototype.
32404 * modulo-sched.c (sms_schedule): Update call of loop_version.
32405 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32406 * tree-parloops.c (gen_parallel_loop): Likewise.
32407 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32408 * tree-ssa-loop-split.c (split_loop): Likewise.
32409 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32410 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32412 2017-02-05 Martin Liska <mliska@suse.cz>
32415 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32417 (print_operand_address): Initialize a struct to zero.
32419 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32421 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32422 garbage collector only in textual form.
32424 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32426 * doc/extend.texi (x86 specific memory model extensions for
32427 transactional memory): Simplify a phrase.
32429 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32432 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32433 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32434 (atomic_storedi_1): Likewise.
32436 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32438 PR tree-optimization/79338
32439 * tree-parloops.c (gather_scalar_reductions): Don't call
32440 vect_analyze_loop_form for loop->inner before destroying loop's
32443 2017-02-03 Martin Sebor <msebor@redhat.com>
32445 PR tree-optimization/79327
32446 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32447 when precision has resulted in leading zeros.
32448 (format_integer): Adjust the likely counter to assume an unknown
32449 argument that may be zero is non-zero.
32451 2017-02-03 Jason Merrill <jason@redhat.com>
32454 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32455 avoid copying non-taken branch.
32457 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32459 PR tree-optimization/79340
32460 * tree-vect-loop.c (vectorizable_reduction): Release
32461 vec_defs elements after safe_splicing them into other vectors.
32464 PR tree-optimization/79327
32465 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32466 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32468 (format_integer): Use wide_int_to_tree instead of build_int_cst
32469 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32470 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32471 of shortest and longest sequence.
32473 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32475 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32476 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32478 2017-02-03 Walter Lee <walt@tilera.com>
32481 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32482 after initial stackframe link reg save.
32483 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32485 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32488 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32489 wu for stxssp alternative.
32491 2017-02-03 Martin Sebor <msebor@redhat.com>
32493 PR tree-optimization/79352
32494 * gimple-fold.c (get_range_strlen): Add argument.
32495 (get_range_strlen): Change return type to bool.
32496 (get_maxval_strlen): Pass in a dummy argument.
32497 * gimple-fold.h (get_range_strlen): Change return type to bool.
32498 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32499 * tree.h (array_at_struct_end_p): Add argument.
32500 * tree.c (array_at_struct_end_p): Handle it.
32502 2017-02-03 Martin Liska <mliska@suse.cz>
32505 * multiple_target.c (create_dispatcher_calls): Redirect edge
32506 from a caller of a dispatcher.
32507 (expand_target_clones): Make the clones local.
32508 (ipa_target_clone): Do both target clones and resolvers.
32509 (ipa_dispatcher_calls): Remove the pass.
32510 (pass_dispatcher_calls::gate): Likewise.
32511 (make_pass_dispatcher_calls): Likewise.
32512 * passes.def (pass_target_clone): Put as very first IPA early
32515 2017-02-03 Martin Liska <mliska@suse.cz>
32517 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32518 in case of a function with ifunc attribute.
32520 2017-02-03 Martin Liska <mliska@suse.cz>
32522 * cgraph.c (cgraph_node::dump): Dump function version info.
32523 * symtab.c (symtab_node::dump_base): Add missing new line.
32525 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32527 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32528 (ifcombine_ifandif): Use it.
32530 2017-02-03 Martin Liska <mliska@suse.cz>
32532 * doc/invoke.texi: Document default value for
32533 use-after-scope-direct-emission-threshold.
32535 2017-02-03 Martin Liska <mliska@suse.cz>
32537 PR tree-optimization/79339
32538 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32539 (format_floating): Likewise.
32541 2017-02-03 Martin Liska <mliska@suse.cz>
32544 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32546 (ipa_node_params_t::remove): Likewise.
32547 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32548 initialization from removed ipa_node_params_t::insert.
32549 (ipa_node_params::~ipa_node_params): Move from removed
32550 ipa_node_params_t::release.
32551 * symbol-summary.h (symbol_summary::m_released): New member.
32552 Do not release a summary twice. Do not allow to call finalizer
32553 for types of a summary that live in GGC memory.
32555 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32557 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32560 2017-02-02 Martin Sebor <msebor@redhat.com>
32562 PR middle-end/79275
32563 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32564 (format_string): Tighten up the range of output for non-constant
32565 strings and correct the expected range for wide non-constant strings.
32567 2017-02-02 Martin Sebor <msebor@redhat.com>
32569 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32571 PR middle-end/32003
32572 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32574 (-fdump-tree-@var): Add to index and document how to come up
32575 with pass-specific option and dump file names.
32576 (-fdump-passes): Clarify where to look for output.
32578 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32580 PR middle-end/77445
32581 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32582 statistics of the analyzed path; allow threading for speed when
32583 any of BBs along the path are optimized for speed.
32585 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32587 PR middle-end/78468
32588 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32589 settings of the virtual registers.
32592 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32594 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32595 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32598 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32600 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32601 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32603 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32605 * config/s390/s390.md: Add missing comments with the expanded
32607 * config/s390/vector.md: Likewise.
32608 * config/s390/vx-builtins.md: Likewise.
32610 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32613 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32614 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32615 conditions on a single line.
32617 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32619 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32620 __S390_VX__ to __VX__.
32622 2017-02-01 Andrew Pinski <apinski@cavium.com>
32624 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32625 stmt_info to record_stmt_cost.
32626 (vect_get_known_peeling_cost): Pass stmt_info if known to
32628 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32629 cpu_vector_cost field into
32630 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32631 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32632 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32633 splitting of scalar_stmt_cost and vec_stmt_cost.
32634 (thunderx_vector_cost): Likewise.
32635 (cortexa57_vector_cost): LIkewise.
32636 (exynosm1_vector_cost): Likewise.
32637 (xgene1_vector_cost): Likewise.
32638 (thunderx2t99_vector_cost): Improve after the splitting of the two
32640 (aarch64_builtin_vectorization_cost): Update for the splitting of
32641 scalar_stmt_cost and vec_stmt_cost.
32643 2017-02-01 Torvald Riegel <triegel@redhat.com>
32644 Richard Henderson <rth@redhat.com>
32646 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32647 conditional on existance of a fast atomic load.
32648 * optabs-query.c (can_atomic_load_p): New function.
32649 * optabs-query.h (can_atomic_load_p): Declare it.
32650 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32651 no fast atomic load is available for the particular size of access.
32652 (expand_atomic_compare_and_swap): Likewise.
32653 (expand_atomic_load): Likewise.
32654 (expand_atomic_store): Likewise.
32655 (expand_atomic_fetch_op): Likewise.
32656 * testsuite/lib/target-supports.exp
32657 (check_effective_target_sync_int_128): Remove x86 because it provides
32658 no fast atomic load.
32659 (check_effective_target_sync_int_128_runtime): Likewise.
32661 2017-02-01 Richard Biener <rguenther@suse.de>
32663 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32664 (graphite_transform_loops): Provide opt-info for optimized nests.
32665 * tree-parloop.c (parallelize_loops): Provide opt-info for
32666 parallelized loops.
32668 2017-02-01 Richard Biener <rguenther@suse.de>
32670 PR middle-end/79315
32671 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32672 was not set before.
32674 2017-02-01 Richard Biener <rguenther@suse.de>
32676 PR tree-optimization/71824
32677 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32678 Verify the loops are valid in the merged SESE region.
32679 (scop_detection::can_represent_loop_1): Check analyzing the
32680 evolution of the number of iterations in the region succeeds.
32682 2017-01-31 Ian Lance Taylor <iant@golang.org>
32684 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32685 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32687 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32689 PR preprocessor/79210
32690 * input.c (get_substring_ranges_for_loc): Replace line_width
32691 assertion with error-handling.
32693 2017-01-31 Richard Biener <rguenther@suse.de>
32695 PR tree-optimization/77318
32696 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32697 (create_pw_aff_from_tree): Take loop parameter.
32698 (add_condition_to_pbb): Pass loop of the condition to
32699 create_pw_aff_from_tree.
32701 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32703 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32704 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32706 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32710 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32712 (convert_int_to_float128): Likewise.
32713 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32714 (convert_int_to_float128): Likewise.
32715 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32716 (UNSPEC_IEEE128_CONVERT): Likewise.
32717 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32718 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32719 Use local variables for IBM extended format.
32720 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32721 (fix_trunc<mode>si2_fprs): Likewise.
32722 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32723 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32724 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32725 to know that we can now have integers of all sizes in vector
32727 (fix<uns>_<mode>di2_hw): Likewise.
32728 (float<uns>_<mode>si2_hw): Likewise.
32729 (fix_<mode>si2_hw): Likewise.
32730 (fixuns_<mode>si2_hw): Likewise.
32731 (float<uns>_<mode>di2_hw): Likewise.
32732 (float_<mode>di2_hw): Likewise.
32733 (float_<mode>si2_hw): Likewise.
32734 (floatuns_<mode>di2_hw): Likewise.
32735 (floatuns_<mode>si2_hw): Likewise.
32736 (xscvqp<su>wz_<mode>): Delete, no longer used.
32737 (xscvqp<su>dz_<mode>): Likewise.
32738 (xscv<su>dqp_<mode>): Likewise.
32739 (ieee128_mfvsrd_64bit): Likewise.
32740 (ieee128_mfvsrd_32bit): Likewise.
32741 (ieee128_mfvsrwz): Likewise.
32742 (ieee128_mtvsrw): Likewise.
32743 (ieee128_mtvsrd_64bit): Likewise.
32744 (ieee128_mtvsrd_32bit): Likewise.
32746 2017-01-31 Martin Liska <mliska@suse.cz>
32749 * ipa-prop.c (ipa_free_all_node_params): Call release method
32750 instead of ~sumbol_summary to not to trigger double times
32753 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32755 PR tree-optimization/71691
32756 * bitmap.h (class auto_bitmap): New.
32757 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32758 is_maybe_undefined instead of ssa_undefined_value_p.
32760 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32762 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32763 __S390_ARCH_LEVEL__ to __ARCH__.
32765 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32767 PR tree-optimization/79267
32768 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32769 if should_remove_lhs_p is true.
32771 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32774 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32775 (add_alignment_attribute): New.
32776 (base_type_die): Add alignment attribute.
32777 (subrange_type_die): Likewise.
32778 (modified_type_die): Likewise.
32779 (gen_array_type_die): Likewise.
32780 (gen_descr_array_type_die: Likewise.
32781 (gen_enumeration_type_die): Likewise.
32782 (gen_subprogram_die): Likewise.
32783 (gen_variable_die): Likewise.
32784 (gen_field_die): Likewise.
32785 (gen_ptr_to_mbr_type_die): Likewise.
32786 (gen_struct_or_union_type_die): Likewise.
32787 (gen_subroutine_type_die): Likewise.
32788 (gen_typedef_die): Likewise.
32789 (base_type_cmp): Compare alignment attribute.
32791 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32794 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32795 (setb_unsigned) New pattern for setb with CCUNS.
32796 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32797 subfc./subfe sequence to avoid overflow problems. Generate a
32798 shorter sequence with cmpld/setb for power9.
32799 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32800 for generating subfc. instruction.
32801 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32802 now uses this instruction.
32804 2017-01-30 Ian Lance Taylor <iant@google.com>
32807 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32808 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32810 2017-01-30 Martin Sebor <msebor@redhat.com>
32812 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32813 Move constant to the right of a relational operator.
32814 (get_mpfr_format_length, format_character, format_string): Ditto.
32815 (should_warn_p, maybe_warn): Same.
32817 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32819 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32822 * asan.c (get_translation_unit_decl): Remove function.
32823 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32825 2017-01-30 Martin Liska <mliska@suse.cz>
32827 PR gcov-profile/79259
32828 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32829 -fprofile-generate.
32831 2017-01-30 Martin Liska <mliska@suse.cz>
32834 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32835 Initialize variables with NULL value.
32837 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32840 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32842 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32844 2017-01-30 Richard Biener <rguenther@suse.de>
32846 PR tree-optimization/79276
32847 * tree-vrp.c (process_assert_insertions): Properly adjust common
32848 when removing a duplicate.
32850 2017-01-30 Richard Biener <rguenther@suse.de>
32852 PR tree-optimization/79256
32853 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32854 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32856 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32858 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32861 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32862 ("*r<noxa>sbg_<mode>_sll_bitmask")
32863 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32864 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32865 Use contiguous_bitmask_nowrap_operand.
32867 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32870 * config/rs6000/altivec.h (vec_xl): Revise #define.
32871 (vec_xst): Likewise.
32873 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32875 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32877 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32879 PR rtl-optimization/79194
32880 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32881 traps before call to bypass_conditional_jumps.
32883 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32885 PR tree-optimization/71374
32886 * lra-constraints.c (check_conflict_input_operands): New.
32887 (match_reload): Use it.
32889 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32892 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32893 account to calculate conflict_set.
32895 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32897 PR rtl-optimization/78559
32898 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32899 other_insn in combine.
32901 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32903 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32904 uint16_type_node for BT_UINT16.
32906 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32908 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32909 "RTL Tests" to menu.
32910 (GIMPLE Tests): New node.
32911 (RTL Tests): New node.
32913 2017-01-27 Richard Biener <rguenther@suse.de>
32915 PR tree-optimization/79245
32916 * tree-loop-distribution.c (distribute_loop): Apply cost
32917 modeling also to detected patterns.
32919 2017-01-27 Richard Biener <rguenther@suse.de>
32921 PR tree-optimization/71433
32922 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32923 (compare_assert_loc): New function.
32924 (process_assert_insertions): Sort and optimize assert locations
32925 to remove duplicates and push down identical assertions on
32926 edges to their destination block.
32928 2017-01-27 Richard Biener <rguenther@suse.de>
32930 PR tree-optimization/79244
32931 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32932 out SSA names even if abnormal.
32934 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32936 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32937 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32938 instead of MPFR_RNDN.
32940 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32943 * arm.c (arm_option_override): Don't call build_target_option_node
32944 until after doing all option overrides.
32945 (arm_valid_target_attribute_tree): Likewise.
32947 2017-01-27 Martin Liska <mliska@suse.cz>
32949 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32950 for {cd}tors and C++ {cd}tors.
32952 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32954 * config/s390/s390.md ("*setmem_long_and")
32955 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32957 2017-01-26 Martin Sebor <msebor@redhat.com>
32959 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32962 2017-01-26 Martin Sebor <msebor@redhat.com>
32964 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32965 HAVE_DFmode before using XFmode or DFmode.
32966 (parse_directive): Avoid using the z length modifier to avoid
32967 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32969 PR middle-end/78703
32970 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32971 to accept adjustment as an array.
32972 (get_int_range): New function.
32973 (struct directive): Make width and prec arrays.
32974 (directive::set_width, directive::set_precision): Call get_int_range.
32975 (format_integer, format_floating): Handle width and precision ranges.
32976 (format_string, parse_directive): Same.
32978 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32981 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32982 comdat_type_p set, just clone them, but keep the children in the
32986 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32987 which have direct callers with -fvar-tracking-assignments enabled
32989 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32990 inside of type units.
32992 2017-01-26 Martin Sebor <msebor@redhat.com>
32994 PR middle-end/78703
32995 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32997 (struct format_result): Replace number_chars, number_chars_min,
32998 and number_chars_max with a single member of struct result_range.
33000 (format_result::operator+=): Adjust.
33001 (struct fmtresult): Remove bounded. Handle likely and unlikely
33003 (fmtresult::adjust_for_width_or_precision): New function.
33004 (fmtresult:type_max_digits): New function.
33005 (bytes_remaining): Handle likely and unlikely counters.
33006 (min_bytes_remaining): Remove.
33007 (format_percent): Simplify.
33008 (format_integer, format_floating): Set likely and unlikely counters.
33009 (get_string_length, format_character, format_string): Same.
33010 (format_plain, should_warn_p): New function.
33011 (maybe_warn): Call should_warn_p. Update diagnostic messages
33012 and handle those for all directives, including plain strings.
33013 (format_directive): Handle likely and unlikely counters.
33014 Remove unnecessary quoting from diagnostics. Add an informational
33016 (add_bytes): Remove.
33017 (pass_sprintf_length::compute_format_length): Simplify.
33018 (try_substitute_return_value): Handle likely and unlikely counters.
33020 2017-01-26 Carl Love <cel@us.ibm.com>
33022 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33023 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33025 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33028 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33029 endianess for subregs into account.
33030 * lra-constraints.c (lra_constraints): Do risky transformations
33031 always on the first iteration.
33032 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33033 last_call_used_reg_set.
33034 (process_bb_lives): Define and use last_call_used_reg_set.
33035 * lra.c (lra): Always continue after lra_constraints on the first
33038 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33040 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33042 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33044 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33046 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33047 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33048 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33049 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33050 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33051 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33052 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33053 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33054 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33056 2017-01-26 Marek Polacek <polacek@redhat.com>
33059 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33060 for the third operand.
33062 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33064 PR middle-end/79236
33065 * omp-low.c (struct omp_context): Add simt_stmt field.
33066 (scan_omp_for): Return omp_context *.
33067 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33068 context to the _simt_ SIMD stmt.
33069 (lower_omp_for): For combined SIMD with sibling _simt_
33070 SIMD, make sure to use the same decls in _looptemp_
33071 clauses as in the sibling.
33073 2017-01-26 David Sherwood <david.sherwood@arm.com>
33075 PR middle-end/79212
33076 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33079 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33082 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33083 emit fld b; fld a; if possible.
33085 * brig-builtins.def: Update copyright years.
33086 * config/arm/arm_acle_builtins.def: Update copyright years.
33088 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33091 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33092 constraint instead of o for the stxsd instruction.
33094 2017-01-25 Carl Love <cel@us.ibm.com>
33096 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33097 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33099 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33101 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33103 2017-01-25 Richard Biener <rguenther@suse.de>
33105 PR tree-optimization/69264
33106 * target.def (vector_alignment_reachable): Improve documentation.
33107 * doc/tm.texi: Regenerate.
33108 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33110 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33111 earlier changes with respect to TYPE_USER_ALIGN.
33112 (vector_alignment_reachable_p): Likewise. Improve dumping.
33114 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33117 * config/arm/arm.md (xordi3): Force constant operand into a register
33120 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33122 * doc/invoke.texi (-fstore-merging): Correct default optimization
33123 levels at which it is enabled.
33124 (-O): Move -fstore-merging from list to...
33127 2017-01-25 Richard Biener <rguenther@suse.de>
33130 * omp-expand.c: Include debug.h.
33131 (expand_omp_taskreg): Make sure to generate early debug before
33132 outlining anything from a function.
33133 (expand_omp_target): Likewise.
33134 (grid_expand_target_grid_body): Likewise.
33136 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33139 * asan.c (get_translation_unit_decl): New function.
33140 (asan_add_global): Extract modules file name from globals
33141 TRANSLATION_UNIT_DECL name.
33143 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33146 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33147 for long calls with APCS frame and VFP.
33149 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33151 * cfg.c (original_copy_tables_initialized_p): New function.
33152 * cfg.h (original_copy_tables_initialized_p): New decl.
33153 * cfgrtl.c (relink_block_chain): Guard the call to
33154 free_original_copy_tables with a call to
33155 original_copy_tables_initialized_p.
33156 * cgraph.h (symtab_node::native_rtl_p): New decl.
33157 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33158 (symtab_node::needed_p): Don't assert for early assembly output
33159 for __RTL functions.
33160 (cgraph_node::finalize_function): Set "force_output" for __RTL
33162 (cgraph_node::analyze): Bail out early for __RTL functions.
33163 (analyze_functions): Update assertion to support __RTL functions.
33164 (cgraph_node::expand): Bail out early for __RTL functions.
33165 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33167 * function.h (struct function): Update comment for field
33169 * gimple-expr.c: Include "tree-pass.h".
33170 (gimple_has_body_p): Return false for __RTL functions.
33171 * Makefile.in (OBJS): Add run-rtl-passes.o.
33172 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33174 (gcc::pass_manager::get_clean_slate): New accessor.
33175 * passes.c: Include "insn-addr.h".
33176 (should_skip_pass_p): Add logging. Update logic for running
33177 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33178 property-provider override so it is only done for gimple passes.
33180 (skip_pass): New function.
33181 (execute_one_pass): Call skip_pass when skipping passes.
33182 * read-md.c (md_reader::read_char): Support filtering
33183 the input to a subset of line numbers.
33184 (md_reader::md_reader): Initialize fields
33185 m_first_line and m_last_line.
33186 (md_reader::read_file_fragment): New function.
33187 * read-md.h (md_reader::read_file_fragment): New decl.
33188 (md_reader::m_first_line): New field.
33189 (md_reader::m_last_line): New field.
33190 * read-rtl-function.c (function_reader::create_function): Only
33191 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33192 curr_properties. Set DECL_INITIAL to a dummy block.
33193 (read_rtl_function_body_from_file_range): New function.
33194 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33196 * run-rtl-passes.c: New file.
33197 * run-rtl-passes.h: New file.
33199 2017-01-24 Jeff Law <law@redhat.com>
33201 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33204 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33206 PR tree-optimization/79159
33207 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33208 (record_nonwrapping_iv): Improve boundary using above function if no
33209 value range information.
33211 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33212 Martin Jambor <mjambor@suse.cz>
33214 * brig-builtins.def: New file.
33215 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33216 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33217 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33218 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33219 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33220 * builtin-types.def (BT_INT8): New.
33221 (BT_INT16): Likewise.
33222 (BT_UINT8): Likewise.
33223 (BT_UINT16): Likewise.
33224 (BT_FN_ULONG): Likewise.
33225 (BT_FN_UINT_INT): Likewise.
33226 (BT_FN_UINT_ULONG): Likewise.
33227 (BT_FN_UINT_LONG): Likewise.
33228 (BT_FN_UINT_PTR): Likewise.
33229 (BT_FN_ULONG_PTR): Likewise.
33230 (BT_FN_INT8_FLOAT): Likewise.
33231 (BT_FN_INT16_FLOAT): Likewise.
33232 (BT_FN_UINT32_FLOAT): Likewise.
33233 (BT_FN_UINT16_FLOAT): Likewise.
33234 (BT_FN_UINT8_FLOAT): Likewise.
33235 (BT_FN_UINT64_FLOAT): Likewise.
33236 (BT_FN_UINT16_UINT32): Likewise.
33237 (BT_FN_UINT32_UINT16): Likewise.
33238 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33239 (BT_FN_INT_PTR_INT): Likewise.
33240 (BT_FN_UINT_PTR_UINT): Likewise.
33241 (BT_FN_LONG_PTR_LONG): Likewise.
33242 (BT_FN_ULONG_PTR_ULONG): Likewise.
33243 (BT_FN_VOID_UINT64_UINT64): Likewise.
33244 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33245 (BT_FN_INT8_INT8_INT8): Likewise.
33246 (BT_FN_INT16_INT16_INT16): Likewise.
33247 (BT_FN_INT_INT_INT): Likewise.
33248 (BT_FN_UINT_FLOAT_UINT): Likewise.
33249 (BT_FN_FLOAT_UINT_UINT): Likewise.
33250 (BT_FN_ULONG_UINT_UINT): Likewise.
33251 (BT_FN_ULONG_UINT_PTR): Likewise.
33252 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33253 (BT_FN_UINT_UINT_UINT): Likewise.
33254 (BT_FN_VOID_UINT_PTR): Likewise.
33255 (BT_FN_UINT_UINT_PTR: Likewise.
33256 (BT_FN_UINT32_UINT64_PTR): Likewise.
33257 (BT_FN_INT_INT_UINT_UINT): Likewise.
33258 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33259 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33260 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33261 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33262 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33263 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33264 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33265 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33266 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33267 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33268 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33269 * doc/frontends.texi: List BRIG FE.
33270 * doc/install.texi (Testing): Add BRIG tesring requirements.
33271 * doc/invoke.texi (Overall Options): Mention BRIG.
33272 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33274 2017-01-24 Richard Biener <rguenther@suse.de>
33276 PR translation/79208
33277 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33279 2017-01-24 Martin Jambor <mjambor@suse.cz>
33282 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33283 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33284 and known_contexts.
33286 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33288 PR middle-end/79123
33289 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33290 casts from signed to unsigned really don't have a range.
33292 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33294 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33295 GMP_RNDx for compatiblity.
33297 2017-01-24 Martin Liska <mliska@suse.cz>
33300 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33301 that would prevent us to call alloca with -1 as argument.
33303 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33305 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33306 Avoid -Wformat-security warning.
33308 2017-01-23 Andrew Pinski <apinski@cavium.com>
33310 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33313 2017-01-23 Martin Sebor <msebor@redhat.com>
33315 PR middle-end/78703
33316 * gimple-ssa-sprintf.c (warn_level): New global.
33317 (format_integer): Use it here and throughout the rest of the file.
33318 Use the same switch to compute sign as base.
33319 (maybe_warn): New function.
33320 (format_directive): Factor out warnings into maybe_warn.
33321 Add debugging output. Use warn_level.
33322 (add_bytes): Use warn_level.
33323 (pass_sprintf_length::compute_format_length): Add debugging output.
33324 (try_substitute_return_value): Same.
33325 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33327 PR middle-end/78703
33328 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33329 (struct fmtresult, format_integer, format_floating): Adjust.
33330 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33331 (get_string_length, format_string,format_directive): Same.
33332 (pass_sprintf_length::compute_format_length): Same.
33333 (try_substitute_return_value): Simplify slightly.
33335 PR middle-end/78703
33336 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33337 (fmtresult::operator+=): Outlined.
33338 (struct fmtresult): Add ctors.
33339 (struct conversion_spec): Rename...
33340 (struct directive): ...to this. Add and remove data members.
33341 (directive::set_width, directive::set_precision): New functions.
33342 (format_percent): Use fmtresult ctor.
33343 (get_width_and_precision): Remove.
33344 (format_integer): Make naming changes. Avoid computing width and
33346 (format_floating): Same. Adjust indentation.
33347 (format_character, format_none): New functions.
33348 (format_string): Moved character handling to format_character.
33349 (format_directive): Remove arguments, change return type.
33350 (parse_directive): New function.
33351 (pass_sprintf_length::compute_format_length): Move directive
33352 parsing to parse_directive.
33354 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33356 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33357 (assign_assembler_name_if_needed): ... this.
33358 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33359 (assign_assembler_name_if_needed): ... this.
33360 (free_lang_data_in_cgraph): Adjust callers.
33361 * cgraphunit.c (cgraph_node::analyze): Likewise.
33362 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33365 2017-01-23 Richard Biener <rguenther@suse.de>
33367 PR tree-optimization/79088
33368 PR tree-optimization/79188
33369 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33370 resetting loop bounds after last path deletion. Reset loop
33371 bounds of the target loop, make code match the comments.
33372 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33373 Make sure loops need no fixups.
33375 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33377 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33378 exponent support with double type for first argument.
33379 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33380 type returned by __builtin_vec_extract_sig,
33381 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33382 functions from "vector int" to "vector unsigned int" or from
33383 "vector long long int" to "vector unsigned long long int".
33384 Changed type returned by __builtin_vec_extract_exp,
33385 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33386 functions from "vector int" to "vector unsigned int" or from
33387 "vector long long int" to "vector unsigned long long int".
33388 Changed return type of __builtin_vec_test_data_class,
33389 __builtin_vec_test_data_class_sp, and
33390 __builtin_vec_test_data_class_dp from "vector int" to
33391 "vector bool int" or from "vector long long int" to "vector bool
33392 long long int" and changed second argument type from "unsigned
33393 int" to "int". Added new overloaded function forms "vector float
33394 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33395 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33396 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33397 double, vector unsigned long long int)" and "vector double
33398 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33399 long int)". Changed return type of
33400 __builtin_scalar_test_data_class and
33401 __builtin_scalar_test_data_class_sp and
33402 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33403 int" and changed second argument from "unsigned int" to "int".
33404 Changed type returned by __builtin_scalar_test_neg,
33405 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33406 from "int" to "bool int". Added new overloaded function form
33407 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33408 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33409 exponent double-precision with floating point first argument.
33410 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33411 documentation of scalar_test_data_class, scalar_test_neg,
33412 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33413 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33414 vec_test_data_class built-in functions to reflect refinements in
33415 their type signatures.
33417 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33419 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33421 (aarch64_elf_asm_destructor): Likewise.
33423 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33425 PR rtl-optimization/78634
33426 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33427 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33428 * ifcvt.c (noce_try_cmove): Add missing cost check.
33430 PR rtl-optimization/71724
33431 * combine.c (if_then_else_cond): Look for situations where it is
33432 beneficial to undo the work of one of the recursive calls.
33434 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33436 PR tree-optimization/70754
33437 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33438 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33439 combined stmt before it if not NULL.
33440 (combine_chains): Process refs reversely and compute dominance point
33443 2017-01-23 Martin Liska <mliska@suse.cz>
33445 PR tree-optimization/79196
33446 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33447 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33449 (strlen_optimize_stmt): Call the renamed function.
33451 2017-01-23 Michael Matz <matz@suse.de>
33453 PR tree-optimization/78384
33454 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33456 2017-01-23 Richard Biener <rguenther@suse.de>
33458 PR tree-optimization/79186
33459 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33460 both incoming edges before moving an assert.
33462 2017-01-23 Martin Jambor <mjambor@suse.cz>
33464 * ipa-prop.c (load_from_param_1): Removed.
33465 (load_from_unmodified_param): Bits from load_from_param_1 put back
33467 (load_from_param): Removed.
33468 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33469 with stmt. Reverted back to use of load_from_unmodified_param.
33471 2017-01-23 Martin Jambor <mjambor@suse.cz>
33474 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33475 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33476 field a pointer to garbage collected vector, mark lattices and
33477 ipcp_orig_node with GTY((skip)).
33478 (ipa_get_param_count): Adjust to descriptors being a pointer.
33479 (ipa_get_param): Likewise.
33480 (ipa_get_type): Likewise.
33481 (ipa_get_param_move_cost): Likewise.
33482 (ipa_set_param_used): Likewise.
33483 (ipa_get_controlled_uses): Likewise.
33484 (ipa_set_controlled_uses): Likewise.
33485 (ipa_is_param_used): Likewise.
33486 (ipa_node_params_t): Move into garbage collector. New methods insert
33488 (ipa_node_params_sum): Annotate wth GTY(()).
33489 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33491 (ipa_load_from_parm_agg): Adjust declaration.
33492 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33493 * ipa-profile.c (ipa_profile): Likewise.
33494 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33495 (ipa_populate_param_decls): Make descriptors parameter garbage
33497 (ipa_dump_param): Adjust to descriptors being a pointer.
33498 (ipa_alloc_node_params): Likewise.
33499 (ipa_initialize_node_params): Likewise.
33500 (load_from_param_1): Make descriptors parameter garbage collected.
33501 (load_from_unmodified_param): Likewise.
33502 (load_from_param): Likewise.
33503 (ipa_load_from_parm_agg): Likewise.
33504 (ipa_node_params::~ipa_node_params): Removed.
33505 (ipa_free_all_node_params): Remove call to delete operator.
33506 (ipa_node_params_t::insert): New.
33507 (ipa_node_params_t::remove): Likewise.
33508 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33509 copy known_csts and known_contexts vectors.
33510 (ipa_read_node_info): Adjust to descriptors being a pointer.
33511 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33513 (ipcp_transform_function): Make descriptors variable garbage
33516 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33518 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33519 * config/i386/avx512dqintrin.h: Ditto.
33520 * config/i386/avx512fintrin.h: Ditto.
33521 * config/i386/i386.c: Handle new builtins.
33522 * config/i386/i386-builtin.def: Add new builtins.
33523 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33524 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33526 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33527 Martin Liska <mliska@suse.cz>
33529 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33530 * asan.c (asan_expand_poison_ifn): Support stores and use
33531 appropriate ASAN report function.
33532 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33533 * internal-fn.def (ASAN_POISON_USE): Declare.
33534 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33535 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33536 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33537 ASAN_POISON calls w/o LHS.
33538 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33539 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33540 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33541 * gimplify.c (asan_poison_variables): Add attribute
33542 use_after_scope_memory to variables that really needs to live
33544 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33545 having the attribute.
33547 2017-01-23 Martin Liska <mliska@suse.cz>
33549 * asan.c (create_asan_shadow_var): New function.
33550 (asan_expand_poison_ifn): Likewise.
33551 * asan.h (asan_expand_poison_ifn): New declaration.
33552 * internal-fn.c (expand_ASAN_POISON): Likewise.
33553 * internal-fn.def (ASAN_POISON): New builtin.
33554 * sanopt.c (pass_sanopt::execute): Expand
33555 asan_expand_poison_ifn.
33556 * tree-inline.c (copy_decl_for_dup_finish): Make function
33558 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33559 * tree-ssa.c (is_asan_mark_p): New function.
33560 (execute_update_addresses_taken): Rewrite local variables
33561 (identified just by use-after-scope as addressable) into SSA.
33563 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33565 * doc/install.texi (Specific): opensource.apple.com uses https
33566 now. Remove trailing slash.
33568 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33570 * README.Portability: Remove note on an Irix compatibility issue.
33572 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33574 * gcov.c (INCLUDE_ALGORITHM): Define.
33575 (INCLUDE_VECTOR): Define.
33576 No longer include <vector> and <algorithm> directly.
33578 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33580 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33582 * doc/invoke.texi (Code Gen Options): Ditto.
33584 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33587 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33589 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33591 rtl-optimization/79125
33592 * cprop.c (local_cprop_pass): Handle cases where we make an
33593 unconditional trap.
33595 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33599 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33600 read from, for big endian.
33602 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33604 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33605 register pauth builtins for LP64 only.
33607 2017-01-20 Marek Polacek <polacek@redhat.com>
33610 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33613 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33615 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33617 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33618 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33619 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33621 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33624 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33625 in big-endian mode when they are not a single duplicated value.
33627 2017-01-20 Richard Biener <rguenther@suse.de>
33629 * BASE-VER: Bump to 7.0.1.
33631 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33633 * omp-low.c (omplow_simd_context): New struct. Use it...
33634 (lower_rec_simd_input_clauses): ...here and...
33635 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33636 references to idx, lane, max_vf, is_simt.
33638 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33640 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33643 2017-01-20 Martin Jambor <mjambor@suse.cz>
33645 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33646 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33648 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33649 (GTFILES): Rename hsa.c to hsa-common.c.
33650 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33651 * hsa-dump.c: Likewise.
33652 * hsa-gen.c: Likewise.
33653 * hsa-regalloc.c: Likewise.
33654 * ipa-hsa.c: Likewise.
33655 * omp-expand.c: Likewise.
33656 * omp-low.c: Likewise.
33657 * toplev.c: Likewise.
33659 2017-01-20 Marek Polacek <polacek@redhat.com>
33662 * doc/invoke.texi: Document -Wduplicated-branches.
33663 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33664 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33665 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33666 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33667 return 0 only when not OEP_LEXICOGRAPHIC.
33668 (fold_build_cleanup_point_expr): Use the expression
33669 location when building CLEANUP_POINT_EXPR.
33670 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33671 * tree.c (add_expr): Handle error_mark_node.
33673 2017-01-20 Martin Liska <mliska@suse.cz>
33676 * tree-profile.c (init_ic_make_global_vars): Do not call
33678 (gimple_init_gcov_profiler): Likewise.
33680 2017-01-20 Martin Liska <mliska@suse.cz>
33683 * cgraph.h (maybe_create_reference): Remove argument and
33685 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33687 * ipa-cp.c (create_specialized_node): Likewise.
33688 * symtab.c (symtab_node::maybe_create_reference): Handle
33689 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33691 2017-01-20 Martin Liska <mliska@suse.cz>
33693 * read-rtl-function.c (function_reader::create_function): Use
33694 build_decl instread of build_decl_stat.
33696 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33698 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33699 * config/i386/avx512dqintrin.h: Ditto.
33700 * config/i386/avx512fintrin.h: Ditto.
33701 * config/i386/i386-builtin-types.def: Add new types.
33702 * config/i386/i386.c: Handle new types.
33703 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33704 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33705 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33706 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33707 (__builtin_ia32_kshiftridi): New.
33708 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33710 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33714 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33715 define to rs6000_init_stack_protect_guard.
33716 (rs6000_init_stack_protect_guard): New function.
33718 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33719 Yunqiang Su <yunqiang.su@imgtec.com>
33721 * config.gcc (supported_defaults): Add madd4.
33722 (with_madd4): Add validation.
33723 (all_defaults): Add madd4.
33724 * config/mips/mips.opt (mmadd4): New option.
33725 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33727 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33729 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33730 (ISA_HAS_FUSED_MADD4): Likewise.
33731 * doc/invoke.texi (-mmadd4): Document the new option.
33732 * doc/install.texi (--with-madd4): Document the new option.
33734 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33736 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33737 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33738 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33739 (aarch64_init_pauth_hint_builtins): New.
33740 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33741 (aarch64_expand_builtin): Expand new builtins.
33743 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33745 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33746 * combine-stack-adj.c (no_unhandled_cfa): Handle
33747 REG_CFA_TOGGLE_RA_MANGLE.
33748 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33749 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33750 info for return address signing.
33751 (aarch64_expand_epilogue): Likewise.
33753 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33755 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33756 * config/aarch64/aarch64-protos.h
33757 (aarch64_return_address_signing_enabled): New declaration.
33758 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33760 (aarch64_expand_prologue): Sign return address before it's pushed onto
33762 (aarch64_expand_epilogue): Authenticate return address fetched from
33764 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33765 (aarch64_attributes): New function attributes for "sign-return-address".
33766 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33767 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33768 ("*do_return"): Generate combined instructions according to key index.
33769 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33770 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33772 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33773 * config/aarch64/aarch64.opt (msign-return-address=): New.
33774 * doc/extend.texi (AArch64 Function Attributes): Documents
33775 "sign-return-address=".
33776 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33778 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33780 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33781 overall option summary.
33783 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33785 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33786 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33787 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33788 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33790 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33792 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33793 -mpower9-minmax by default for -mcpu=power9.
33794 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33795 128-bit floating point.
33797 2017-01-20 Alan Modra <amodra@gmail.com>
33799 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33800 optimizing for size.
33802 2017-01-20 Alan Modra <amodra@gmail.com>
33805 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33806 for strcmp and strncmp from corresponding builtin decl.
33808 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33810 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33811 instead of i386/rtems-64.h.
33812 * config/i386/rtems-64.h: Remove.
33814 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33818 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33820 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33822 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33824 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33825 Change int to HOST_WIDE_INT.
33826 * config/aarch64/aarch64-protos.h
33827 (aarch64_simd_gen_const_vector_dup): Likewise.
33828 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33830 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33832 * langhooks-def.h (lhd_type_for_size): New decl.
33833 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33834 * langhooks.c (lhd_type_for_size): New function, taken from
33837 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33839 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33840 define_bypass for CR latency.
33841 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33842 (power9-alu2): Add define_bypass for CR latency.
33844 (power9-mul): Update insn latency.
33845 (power9-mul-compare): Update insn latency, bypass latency and remove
33848 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33850 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33852 * config/aarch64/aarch64.md
33853 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33854 aarch64_nopcrelative_literal_loads.
33855 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33857 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33859 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33860 TARGET_LOONGSON_3A.
33861 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33863 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33866 * config.gcc (supported_defaults): Add lxc1-sxc1.
33867 (with_lxc1_sxc1): Add validation.
33868 (all_defaults): Add lxc1-sxc1.
33869 * config/mips/mips.opt (mlxc1-sxc1): New option.
33870 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33872 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33873 __mips_no_lxc1_sxc1.
33874 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33875 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33876 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33878 2017-01-19 Richard Biener <rguenther@suse.de>
33880 PR tree-optimization/72488
33881 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33882 sure to restore SSA info.
33883 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33885 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33887 PR rtl-optimization/79121
33888 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33889 of the inner type when shifting an extended value.
33891 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33894 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33895 interposable aliases.
33897 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33900 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33901 Use the evmergelohi instruction.
33902 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33903 (mov_sitf_e500_subreg8_2_be): Likewise.
33904 (mov_sitf_e500_subreg12_2_le): Likewise.
33905 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33906 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33907 (mov_sitf_e500_subreg8_2_le): Likewise.
33908 (mov_sitf_e500_subreg12_2_be): Likewise.
33910 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33912 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33913 attribute from vecsimple to vecperm.
33914 (altivec_vbpermq2): Likewise.
33916 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33919 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33921 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33922 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33923 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33924 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33925 case where N arg is SIZE_MAX.
33926 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33927 (cmpstrsi): Add pattern.
33929 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33931 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33932 __builtin_vec_revb builtins.
33933 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33934 built-in functions to support generation of the ISA 3.0 XXBR<x>
33935 vector byte reverse instructions.
33936 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33937 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33938 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33939 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33940 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33941 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33942 (P9V_BUILTIN_VEC_REVB): Likewise.
33943 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33944 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33945 (p9_xxbrq_v16qi): Likewise.
33946 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33947 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33948 (p9_xxbrh_v8hi): Likewise.
33949 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33950 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33951 vec_revb built-in functions.
33953 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33955 PR rtl-optimization/78952
33956 * config/i386/i386.md (any_extract): New code iterator.
33957 (*insvqi_2): Use any_extract for source operand.
33958 (*insvqi_3): Use any_shiftrt for source operand.
33960 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33962 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33964 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33966 2017-01-18 Matthias Klose <doko@ubuntu.com>
33968 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33970 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33972 * config/rs6000/altivec.h (vec_bperm): Change #define.
33973 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33974 (altivec_vbpermq2): New define_insn.
33975 (altivec_vbpermd): Likewise.
33976 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33977 function interface.
33978 (VBPERMD): Likewise.
33979 (VBPERM): New polymorphic function interface.
33980 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33981 Add entries for P9V_BUILTIN_VEC_VBPERM.
33982 * doc/extend.texi: Add interfaces for vec_bperm.
33984 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33986 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33987 first letter of error messages.
33988 (s390_resolve_overloaded_builtin): Likewise.
33989 * config/s390/s390.c (s390_expand_builtin): Likewise.
33990 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33991 (s390_valid_target_attribute_inner_p): Likewise.
33992 * config/s390/s390.md ("tabort"): Likewise.
33994 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33996 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33997 (ISA_AVOID_DIV_HILO): New macro.
33998 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33999 (ISA_HAS_DDIV): Likewise.
34001 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34003 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34005 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34007 * doc/invoke.texi (fabi-version): Spelling fix.
34009 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34012 * doc/invoke.texi (fabi-version): Mention mangling fix for
34015 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34018 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34020 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34023 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34024 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34026 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34028 (stack_protect_test): Ditto.
34029 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34030 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34032 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34033 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34034 -mstack-protector-guard-offset=.
34035 (RS/6000 and PowerPC Options): Ditto.
34037 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34039 * config/i386/i386.h (MASK_CLASS_P): New define.
34040 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34041 there are no registers from different register sets also when
34042 mask registers are used. Update function comment.
34043 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34044 to (*k/*r) and (*k/*km) alternatives.
34046 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34048 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34049 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34050 (EH_RETURN_HANDLER_RTX): New define.
34051 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34052 Force frame pointer in EH return functions.
34053 (aarch64_expand_epilogue): Add barrier for eh_return.
34054 (aarch64_final_eh_return_addr): Remove.
34055 (aarch64_eh_return_handler_rtx): New function.
34056 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34058 (aarch64_eh_return_handler_rtx): New prototype.
34060 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34062 * config/rs6000/altivec.h (vec_rlmi): New #define.
34063 (vec_vrlnm): Likewise.
34064 (vec_rlnm): Likewise.
34065 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34066 (UNSPEC_VRLNM): Likewise.
34067 (VIlong): New mode iterator.
34068 (altivec_vrl<VI_char>mi): New define_insn.
34069 (altivec_vrl<VI_char>nm): Likewise.
34070 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34072 (VRLDNM): Likewise.
34073 (RLNM): New polymorphic function entry.
34074 (VRLWMI): New monomorphic function entry.
34075 (VRLDMI): Likewise.
34076 (RLMI): New polymorphic function entry.
34077 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34078 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34079 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34082 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34085 * dwarf2out.c (field_byte_offset): Restore the
34086 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34087 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34088 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34091 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34094 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34096 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34099 * dwarf2out.c (add_data_member_location_attribute): For constant
34100 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34101 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34102 DW_AT_byte_size attributes.
34104 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34106 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34107 after forcing to constant memory when the code model is medium.
34109 2017-01-17 Julia Koval <julia.koval@intel.com>
34112 * config/i386/avx512fintrin.h
34113 (_mm512_i32gather_ps): Change __addr type to void const*.
34114 (_mm512_mask_i32gather_ps): Ditto.
34115 (_mm512_i32gather_pd): Ditto.
34116 (_mm512_mask_i32gather_pd): Ditto.
34117 (_mm512_i64gather_ps): Ditto.
34118 (_mm512_mask_i64gather_ps): Ditto.
34119 (_mm512_i64gather_pd): Ditto.
34120 (_mm512_mask_i64gather_pd): Ditto.
34121 (_mm512_i32gather_epi32): Ditto.
34122 (_mm512_mask_i32gather_epi32): Ditto.
34123 (_mm512_i32gather_epi64): Ditto.
34124 (_mm512_mask_i32gather_epi64): Ditto.
34125 (_mm512_i64gather_epi32): Ditto.
34126 (_mm512_mask_i64gather_epi32): Ditto.
34127 (_mm512_i64gather_epi64): Ditto.
34128 (_mm512_mask_i64gather_epi64): Ditto.
34129 (_mm512_i32scatter_ps): Change __addr type to void*.
34130 (_mm512_mask_i32scatter_ps): Ditto.
34131 (_mm512_i32scatter_pd): Ditto.
34132 (_mm512_mask_i32scatter_pd): Ditto.
34133 (_mm512_i64scatter_ps): Ditto.
34134 (_mm512_mask_i64scatter_ps): Ditto.
34135 (_mm512_i64scatter_pd): Ditto.
34136 (_mm512_mask_i64scatter_pd): Ditto.
34137 (_mm512_i32scatter_epi32): Ditto.
34138 (_mm512_mask_i32scatter_epi32): Ditto.
34139 (_mm512_i32scatter_epi64): Ditto.
34140 (_mm512_mask_i32scatter_epi64): Ditto.
34141 (_mm512_i64scatter_epi32): Ditto.
34142 (_mm512_mask_i64scatter_epi32): Ditto.
34143 (_mm512_i64scatter_epi64): Ditto.
34144 (_mm512_mask_i64scatter_epi64): Ditto.
34145 * config/i386/avx512pfintrin.h
34146 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34147 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34148 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34149 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34150 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34151 (_mm512_prefetch_i32scatter_ps): Ditto.
34152 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34153 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34154 (_mm512_prefetch_i64scatter_pd): Ditto.
34155 (_mm512_prefetch_i64scatter_ps): Ditto.
34156 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34157 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34158 * config/i386/avx512vlintrin.h
34159 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34160 (_mm_mmask_i32gather_ps): Ditto.
34161 (_mm256_mmask_i32gather_pd): Ditto.
34162 (_mm_mmask_i32gather_pd): Ditto.
34163 (_mm256_mmask_i64gather_ps): Ditto.
34164 (_mm_mmask_i64gather_ps): Ditto.
34165 (_mm256_mmask_i64gather_pd): Ditto.
34166 (_mm_mmask_i64gather_pd): Ditto.
34167 (_mm256_mmask_i32gather_epi32): Ditto.
34168 (_mm_mmask_i32gather_epi32): Ditto.
34169 (_mm256_mmask_i32gather_epi64): Ditto.
34170 (_mm_mmask_i32gather_epi64): Ditto.
34171 (_mm256_mmask_i64gather_epi32): Ditto.
34172 (_mm_mmask_i64gather_epi32): Ditto.
34173 (_mm256_mmask_i64gather_epi64): Ditto.
34174 (_mm_mmask_i64gather_epi64): Ditto.
34175 (_mm256_i32scatter_ps): Change __addr type to void*.
34176 (_mm256_mask_i32scatter_ps): Ditto.
34177 (_mm_i32scatter_ps): Ditto.
34178 (_mm_mask_i32scatter_ps): Ditto.
34179 (_mm256_i32scatter_pd): Ditto.
34180 (_mm256_mask_i32scatter_pd): Ditto.
34181 (_mm_i32scatter_pd): Ditto.
34182 (_mm_mask_i32scatter_pd): Ditto.
34183 (_mm256_i64scatter_ps): Ditto.
34184 (_mm256_mask_i64scatter_ps): Ditto.
34185 (_mm_i64scatter_ps): Ditto.
34186 (_mm_mask_i64scatter_ps): Ditto.
34187 (_mm256_i64scatter_pd): Ditto.
34188 (_mm256_mask_i64scatter_pd): Ditto.
34189 (_mm_i64scatter_pd): Ditto.
34190 (_mm_mask_i64scatter_pd): Ditto.
34191 (_mm256_i32scatter_epi32): Ditto.
34192 (_mm256_mask_i32scatter_epi32): Ditto.
34193 (_mm_i32scatter_epi32): Ditto.
34194 (_mm_mask_i32scatter_epi32): Ditto.
34195 (_mm256_i32scatter_epi64): Ditto.
34196 (_mm256_mask_i32scatter_epi64): Ditto.
34197 (_mm_i32scatter_epi64): Ditto.
34198 (_mm_mask_i32scatter_epi64): Ditto.
34199 (_mm256_i64scatter_epi32): Ditto.
34200 (_mm256_mask_i64scatter_epi32): Ditto.
34201 (_mm_i64scatter_epi32): Ditto.
34202 (_mm_mask_i64scatter_epi32): Ditto.
34203 (_mm256_i64scatter_epi64): Ditto.
34204 (_mm256_mask_i64scatter_epi64): Ditto.
34205 (_mm_i64scatter_epi64): Ditto.
34206 (_mm_mask_i64scatter_epi64): Ditto.
34207 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34208 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34209 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34210 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34211 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34212 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34213 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34214 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34215 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34216 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34217 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34218 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34219 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34220 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34221 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34222 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34223 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34224 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34225 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34226 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34227 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34228 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34229 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34230 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34231 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34232 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34233 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34234 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34235 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34236 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34237 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34238 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34239 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34240 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34241 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34242 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34243 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34244 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34245 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34246 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34247 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34248 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34249 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34250 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34251 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34252 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34253 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34254 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34255 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34256 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34257 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34258 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34259 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34260 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34261 definitions accordingly.
34263 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34264 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34267 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34270 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34273 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34274 hard regs for inner regno.
34276 2017-01-17 Martin Liska <mliska@suse.cz>
34279 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34280 assumption and add comment.
34282 2017-01-17 Nathan Sidwell <nathan@acm.org>
34284 * ipa-visibility.c (localize_node): New function, broken out of ...
34285 (function_and_variable_visibility): ... here. Call it.
34287 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34289 PR middle-end/77445
34290 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34291 correctly set frequency of oudgoing edge.
34292 (duplicate_thread_path): Fix profile updating.
34294 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34297 * configure.ac: Add GCC_BASE_VER.
34298 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34299 version from BASE-VER file.
34300 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34301 (gcc.o): Depend on $(BASEVER).
34302 * common.opt (dumpfullversion): New option.
34303 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34304 * doc/invoke.texi: Document -dumpfullversion.
34305 * doc/install.texi: Document --with-gcc-major-version-only.
34306 * configure: Regenerated.
34308 2017-01-17 Richard Biener <rguenther@suse.de>
34310 PR tree-optimization/71433
34311 * tree-vrp.c (register_new_assert_for): Merge same asserts
34312 on all incoming edges.
34313 (process_assert_insertions_for): Handle insertions at the
34316 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34318 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34319 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34321 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34324 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34327 2017-01-17 Alan Modra <amodra@gmail.com>
34330 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34331 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34332 symbolic stack limit when pic.
34334 2017-01-16 Martin Sebor <msebor@redhat.com>
34336 PR tree-optimization/78608
34337 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34339 2017-01-16 Jeff Law <law@redhat.com>
34342 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34343 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34344 for several include directories that may be relative to sysroot.
34345 * config/i386/x-mingw32 (gplus_includedir): Define.
34346 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34347 (native_system_includedir): Likewise.
34348 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34349 override if TARGET_SYSTEM_ROOT is defined.
34350 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34352 PR tree-optimization/79090
34353 PR tree-optimization/33562
34354 PR tree-optimization/61912
34355 PR tree-optimization/77485
34356 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34357 and computed trims into the dump file.
34359 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34361 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34363 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34366 * gimplify.c (gimplify_init_constructor): If want_value and
34367 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34371 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34372 sequence. Formatting fixes.
34373 (doloop_optimize): Formatting fixes.
34376 * gcc.c (debug_level_greater_than_spec_func): New function.
34377 (static_spec_functions): Add debug-level-gt spec function.
34378 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34380 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34381 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34382 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34383 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34384 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34385 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34387 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34389 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34390 QImode fixups to general and mask registers only.
34392 2017-01-16 Carl Love <cel@us.ibm.com>
34394 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34395 for built-in functions
34396 vector signed char vec_nabs (vector signed char)
34397 vector signed short vec_nabs (vector signed short)
34398 vector signed int vec_nabs (vector signed int)
34399 vector signed long long vec_nabs (vector signed long long)
34400 vector float vec_nabs (vector float)
34401 vector double vec_nabs (vector double)
34402 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34404 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34405 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34406 * doc/extend.texi: Update the documentation file for the new built-in
34409 2017-01-16 Martin Sebor <msebor@redhat.com>
34411 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34414 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34416 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34417 UNSPEC_VSX__XXSPLTD to require special splat handling.
34419 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34422 * system.h: Poison strndup.
34424 2017-01-16 Alan Modra <amodra@gmail.com>
34427 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34430 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34432 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34434 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34436 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34437 call recog here. Assert that INSN_CODE (insn) is non-negative.
34439 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34442 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34444 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34445 in the currently scheduled RTL fragment.
34447 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34449 PR rtl-optimization/78751
34450 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34453 2017-01-14 Jeff Law <law@redhat.com>
34455 PR tree-optimization/79090
34456 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34457 variable length stores.
34458 (compute_trims): Delete dead assignment to *trim_tail.
34459 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34462 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34464 PR rtl-optimization/78626
34465 PR rtl-optimization/78727
34466 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34467 of a block, and split such blocks after everything else is finished.
34469 2017-01-14 Alan Modra <amodra@gmail.com>
34472 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34473 target legitimate_combined_insn.
34474 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34475 (rs6000_legitimate_combined_insn): New function.
34476 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34478 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34479 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34480 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34482 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34484 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34486 2017-01-13 Jeff Law <law@redhat.com>
34488 PR tree-optimization/33562
34489 PR tree-optimization/61912
34490 PR tree-optimization/77485
34491 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34493 (delete_dead_assignment): Likewise.
34494 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34495 statement to delete_dead_call and delete_dead_assignment.
34497 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34500 * substring-locations.c (format_warning_va): Strengthen case 1 so
34501 that both endpoints of the substring must be within the format
34502 range for just the substring to be printed.
34504 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34506 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34507 * config/i386/i386.c (ix86_target_string): Add missing options
34508 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34509 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34510 flags_other and ix86_target_other to flags2_other. Display unknown
34512 (ix86_valid_target_attribute_inner_p): Add missing options and
34513 reorder options by implied ISAs, as in ix86_target_string.
34515 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34517 * hash-table.h (hash_table::too_empty_p): New function.
34518 (hash_table::expand): Use it.
34519 (hash_table::traverse): Likewise.
34520 (hash_table::empty_slot): Use sizeof (value_type) instead of
34521 sizeof (PTR) to convert bytes to elements. Shrink the table
34522 if the current size is excessive for the current number of
34525 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34527 * ira-costs.c (record_reg_classes): Break from the inner loop
34528 early once alt_fail is known to be true. Update outer loop
34529 handling accordingly.
34531 2017-01-13 Jeff Law <law@redhat.com>
34533 * tree-ssa-dse.c (decrement_count): New function.
34534 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34535 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34536 when we know the partially dead statement is a mem* function.
34538 PR tree-optimization/61912
34539 PR tree-optimization/77485
34540 * tree-ssa-dse.c: Include expr.h.
34541 (maybe_trim_constructor_store): New function.
34542 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34544 PR tree-optimization/33562
34545 PR tree-optimization/61912
34546 PR tree-optimization/77485
34547 * doc/invoke.texi: Document new dse-max-object-size param.
34548 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34549 * tree-ssa-dse.c: Include params.h.
34550 (dse_store_status): New enum.
34551 (initialize_ao_ref_for_dse): New, partially extracted from
34553 (valid_ao_ref_for_dse, normalize_ref): New.
34554 (setup_live_bytes_from_ref, compute_trims): Likewise.
34555 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34556 (maybe_trim_partially_dead_store): Likewise.
34557 (maybe_trim_complex_store): Likewise.
34558 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34559 Track what bytes live from the original store. Return tri-state
34560 for dead, partially dead or live.
34561 (dse_dom_walker): Add constructor, destructor and new private members.
34562 (delete_dead_call, delete_dead_assignment): New extracted from
34564 (dse_optimize_stmt): Make a member of dse_dom_walker.
34565 Use initialize_ao_ref_for_dse.
34567 PR tree-optimization/33562
34568 PR tree-optimization/61912
34569 PR tree-optimization/77485
34570 * sbitmap.h (bitmap_count_bits): Prototype.
34571 (bitmap_clear_range, bitmap_set_range): Likewise.
34572 * sbitmap.c (bitmap_clear_range): New function.
34573 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34575 2017-01-13 Martin Liska <mliska@suse.cz>
34578 * function.c (set_cfun): Add new argument force.
34579 * function.h (set_cfun): Likewise.
34580 * ipa-inline-transform.c (inline_call): Use the function when
34581 strict alising from is dropped for function we inline to.
34583 2017-01-13 Richard Biener <rguenther@suse.de>
34585 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34586 for dumping GIMPLE INTEGER_CSTs.
34588 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34590 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34591 to 201112L since C++17.
34593 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34596 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34597 if -fsanitize=kernel-address is present.
34599 2017-01-13 Richard Biener <rguenther@suse.de>
34601 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34602 as _Literal ( type ) number in case usual suffixes do not
34603 preserve all information.
34605 2017-01-13 Richard Biener <rguenther@suse.de>
34607 PR tree-optimization/77283
34608 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34609 and ssa-iterators.h.
34610 (is_feasible_trace): Implement a cost model based on joiner
34613 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34616 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34617 char or short to __float128/_Float128 directly.
34619 2017-01-12 Martin Sebor <msebor@redhat.com>
34621 to -Wformat-overflow.
34622 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34623 (min_bytes_remaining): Same.
34624 (get_string_length): Same.
34625 (format_string): Same.
34626 (format_directive): Same.
34628 (pass_sprintf_length::handle_gimple_call): Same.
34630 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34632 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34633 info.nowrite calls with no lhs that can't throw. Return bool
34634 whether gsi_remove has been called or not.
34635 (pass_sprintf_length::handle_gimple_call): Return bool whether
34636 try_substitute_return_value called gsi_remove. Formatting fix.
34637 (pass_sprintf_length::execute): Don't use gsi_remove if
34638 handle_gimple_call returned true.
34641 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34642 be removed due to side-effects, don't remove following barrier nor
34643 turn the successor edge into fallthru edge.
34645 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34648 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34649 element-reversing loads and stores as not swappable.
34651 2017-01-12 Nathan Sidwell <nathan@acm.org>
34652 Nicolai Stange <nicstange@gmail.com>
34654 * combine.c (try_combine): Don't ignore result of overlap checking
34655 loop. Combine overlap & asm check into single loop.
34657 2017-01-12 Richard Biener <rguenther@suse.de>
34659 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34660 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34662 2017-01-12 Richard Biener <rguenther@suse.de>
34664 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34665 and TS_TARGET_OPTION directly derive from TS_BASE.
34666 * tree-core.h (tree_optimization_option): Derive from tree_base.
34667 (tree_target_option): Likewise.
34669 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34671 * config/i386/i386.c (memory_address_length): Increase len
34672 only when rip_relative_addr_p returns false.
34674 2017-01-11 Julia Koval <julia.koval@intel.com>
34676 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34677 (OPTION_MASK_ISA_SGX_SET): New.
34678 (ix86_handle_option): Handle OPT_msgx.
34679 * config.gcc: Added sgxintrin.h.
34680 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34681 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34682 * config/i386/i386.c (ix86_target_string): Add -msgx.
34684 (ix86_option_override_internal): Handle new options.
34685 (ix86_valid_target_attribute_inner_p): Add sgx.
34686 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34687 * config/i386/i386.opt: Add msgx.
34688 * config/i386/sgxintrin.h: New file.
34689 * config/i386/x86intrin.h: Add sgxintrin.h.
34691 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34694 * fold-const.c (maybe_nonzero_address): Return 1 for function
34696 (tree_single_nonzero_warnv_p): Don't handle function local objects
34700 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34703 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34706 * opts.c: Include "spellcheck.h"
34707 (struct string_fragment): New struct.
34708 (struct edit_distance_traits<const string_fragment &>): New
34710 (get_closest_sanitizer_option): New function.
34711 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34713 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34715 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34717 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34718 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34719 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34720 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34721 for initial die_offset if dwarf_split_debug_info.
34722 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34723 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34725 (output_skeleton_debug_sections): Formatting fix. Use
34726 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34727 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34729 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34731 * config/arm/cortex-a53.md: Add bypasses for
34732 cortex_a53_r2f_cvt.
34733 (cortex_a53_r2f): Only use for transfers.
34734 (cortex_a53_f2r): Likewise.
34735 (cortex_a53_r2f_cvt): Add reservation for conversions.
34736 (cortex_a53_f2r_cvt): Likewise.
34738 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34740 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34741 to all inlined functions, change static to extern.
34743 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34746 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34748 (arm_assemble_integer): Likewise.
34750 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34752 * config.gcc: Use new awk script to check CPU, FPU and architecture
34753 parameters for --with-... options.
34754 * config/arm/parsecpu.awk: New file
34755 * config/arm/arm-cpus.in: New file.
34756 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34758 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34760 * config/arm/t-arm: Update dependency rules.
34761 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34762 of processing .def files.
34763 * config/arm/genopt.sh: Deleted.
34764 * config/arm/gentune.sh: Deleted.
34765 * config/arm/arm-cores.def: Deleted.
34766 * config/arm/arm-arches.def: Deleted.
34767 * config/arm/arm-fpus.def: Deleted.
34768 * config/arm/arm-tune.md: Regenerated.
34769 * config/arm/arm-tables.opt: Regenerated.
34770 * config/arm/arm-cpu.h: New generated file.
34771 * config/arm/arm-cpu-data.h: New generated file.
34772 * config/arm/arm-cpu-cdata.h: New generated file.
34774 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34777 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34779 (input_varpool_node): Unpack dynamically_initialized bit.
34781 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34783 PR rtl-optimization/79032
34784 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34785 the alignment of the adjusted memory reference against that of MODE,
34786 instead of the alignment of the original memory reference.
34788 2017-01-11 Martin Jambor <mjambor@suse.cz>
34790 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34792 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34793 decorated functions.
34795 2017-01-11 Richard Biener <rguenther@suse.de>
34797 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34798 set range/nonnull info for PHI results. Do not set it on
34799 stmts marked for removal.
34801 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34803 * expr.c (store_field): In the bitfield case, fetch the return value
34804 from the registers before applying a single big-endian adjustment.
34805 Always do a final load for a BLKmode value not larger than a word.
34807 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34810 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34811 that we correctly handle column numbers greater than
34812 LINE_MAP_MAX_COLUMN_NUMBER.
34814 2017-01-10 Martin Sebor <msebor@redhat.com>
34816 PR middle-end/78245
34817 * gimple-ssa-sprintf.c (get_destination_size): Call
34818 {init,fini}object_sizes.
34819 * tree-object-size.c (addr_object_size): Adjust.
34820 (pass_through_call): Adjust.
34821 (pass_object_sizes::execute): Adjust.
34822 * tree-object-size.h (fini_object_sizes): Declare.
34824 2017-01-10 Martin Sebor <msebor@redhat.com>
34826 PR tree-optimization/78775
34827 * builtins.c (get_size_range): Move...
34828 * calls.c: ...to here.
34829 (alloc_max_size): Accept zero argument.
34830 (operand_signed_p): Remove.
34831 (maybe_warn_alloc_args_overflow): Call get_size_range.
34832 * calls.h (get_size_range): Declare.
34834 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34836 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34837 from TI's devices.csv file as of September 2016.
34838 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34840 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34842 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34843 * doc/invoke.texi: Likewise.
34844 * doc/md.texi: Likewise.
34845 * doc/objc.texi: Likewise.
34847 2017-01-10 Joshua Conner <joshconner@google.com>
34849 * config/arm/fuchsia-elf.h: New file.
34850 * config/fuchsia.h: New file.
34851 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34852 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34854 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34856 2016-01-10 Richard Biener <rguenther@suse.de>
34858 PR tree-optimization/79034
34859 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34860 Propagate out degenerate PHIs in the joiner.
34862 2017-01-10 Martin Liska <mliska@suse.cz>
34864 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34865 (sort_congruence_classes_by_decl_uid): Likewise.
34866 (sort_congruence_class_groups_by_decl_uid): Likewise.
34867 (sem_item_optimizer::merge_classes): Sort class, groups in these
34868 classes and members in the groups by DECL_UID of declarations.
34869 This would make merge operations stable.
34871 2017-01-10 Martin Liska <mliska@suse.cz>
34873 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34874 usage of m_classes_vec.
34875 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34876 (sem_item_optimizer::get_group_by_hash): Likewise.
34877 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34878 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34879 (sem_item_optimizer::verify_classes): Likewise.
34880 (sem_item_optimizer::process_cong_reduction): Likewise.
34881 (sem_item_optimizer::dump_cong_classes): Likewise.
34882 (sem_item_optimizer::merge_classes): Likewise.
34883 * ipa-icf.h (congruence_class_hash): Rename from
34884 congruence_class_group_hash. Remove declaration of m_classes_vec.
34886 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34888 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34889 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34890 * config.gcc: Add avx512vpopcntdqintrin.h.
34891 * config/i386/avx512vpopcntdqintrin.h: New.
34892 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34893 * config/i386/i386-builtin-types.def: Add new types.
34894 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34895 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34896 __builtin_ia32_vpopcountq_v8di_mask): New.
34897 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34898 __AVX512VPOPCNTDQ__.
34899 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34900 (PTA_AVX512VPOPCNTDQ): Define.
34901 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34902 TARGET_AVX512VPOPCNTDQ_P): Define.
34903 * config/i386/i386.opt: Add mavx512vpopcntdq.
34904 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34905 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34907 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34909 PR middle-end/77484
34910 * predict.def (PRED_CALL): Set to 67.
34912 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34914 * expr.c (store_field): In the bitfield case, if the value comes from
34915 a function call and is of an aggregate type returned in registers, do
34916 not modify the field mode; extract the value in all cases if the mode
34917 is BLKmode and the size is not larger than a word.
34919 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34922 * config/i386/cpuid.h: Fix undefined behavior.
34924 2017-01-04 Jeff Law <law@redhat.com>
34926 PR tree-optimization/79007
34927 PR tree-optimization/67955
34928 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34929 conservative for pt.null when flag_non_call_exceptions is on.
34931 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34933 PR translation/79019
34934 PR translation/79020
34935 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34936 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34937 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34939 * config/avr/avr.opt (maccumulate-args): Likewise.
34940 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34941 * common.opt (freport-bug): Likewise.
34942 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34943 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34944 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34945 translatable string.
34946 * config/i386/i386.c (function_value_32): Likewise.
34947 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34948 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34950 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34951 * common/config/msp430/msp430-common.c (msp430_handle_option):
34953 * symtab.c (symtab_node::verify_base): Likewise.
34954 * opts.c (set_debug_level): Likewise.
34955 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34956 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34957 missing whitespace to translatable strings.
34958 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34959 * config/sh/superh.h: Likewise.
34960 * config/i386/xopintrin.h: Likewise.
34961 * config/i386/znver1.md: Likewise.
34962 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34963 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34964 * double-int.h (struct double_int): Likewise.
34965 * double-int.c (div_and_round_double): Likewise.
34966 * wide-int.cc: Likewise.
34967 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34968 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34969 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34970 (crossjumps_occurred): ... this.
34971 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34974 PR tree-optimization/78899
34975 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34976 returning bool return struct loop *, NULL for failure and the new
34978 (versionable_outer_loop_p): Don't version outer loop if it has
34979 dont_vectorized bit set.
34980 (tree_if_conversion): When versioning outer loop, ensure
34981 tree_if_conversion is performed also on the inner loop of the
34982 non-vectorizable outer loop copy.
34983 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34984 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34985 prevent vectorization of it.
34986 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34987 the outer loop vectorization of the non-scalar version is attempted
34988 before vectorization of the inner loop in scalar version. If
34989 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34990 vectorization of its inner loop.
34991 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34992 has 2 inner loops, rename also on edges from bb whose single pred
34993 is outer_loop->header. Fix typo in function comment.
34995 2017-01-09 Martin Sebor <msebor@redhat.com>
34998 * asan.c (asan_emit_stack_protection): Increase local buffer size
34999 to avoid snprintf truncation warning.
35001 2017-01-09 Andrew Pinski <apinski@cavium.com>
35003 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35004 to reference thunderx2t99 for the tuning structure
35005 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35007 (thunderx2t99_extra_costs): This.
35008 * config/aarch64/aarch64-tune.md: Regenerate.
35009 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35010 (vulcan_addrcost_table): This.
35011 (vulcan_regmove_cost): Rename to ...
35012 (thunderx2t99_regmove_cost): This.
35013 (vulcan_vector_cost): Rename to ...
35014 (thunderx2t99_vector_cost): this.
35015 (vulcan_branch_cost): Rename to ...
35016 (thunderx2t99_branch_cost): This.
35017 (vulcan_tunings): Rename to ...
35018 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35019 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35021 2017-01-09 Martin Jambor <mjambor@suse.cz>
35025 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35026 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35027 (propagate_vr_accross_jump_function): Use the above function for all
35028 value range computations for pass-through jump functions and type
35029 converasion from explicit value range values.
35030 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35031 parameters from TYPE_ARG_TYPES.
35032 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35033 (ipa_write_node_info): Stream type of the actual argument.
35034 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35036 2017-01-09 Martin Liska <mliska@suse.cz>
35039 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35040 (lookup_compiler): Do not show error message with have_E.
35042 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35044 PR tree-optimization/78938
35045 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35046 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35047 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35048 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35051 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35053 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35056 2017-01-09 Richard Biener <rguenther@suse.de>
35058 PR tree-optimization/78997
35059 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35060 name condition properly.
35062 2017-01-09 Richard Biener <rguenther@suse.de>
35065 * dwarf2out.c (is_cxx): New overload with context.
35066 (is_naming_typedef_decl): Use it.
35068 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35070 * invoke.texi (Option Summary): Correct spacing in option lists
35071 and add line breaks to fix over-long lines.
35073 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35075 PR middle-end/17660
35077 * extend.texi (Common Variable Attributes): Add xref to GCC
35078 Internals manual to explain mode attribute keywords.
35080 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35083 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35084 and Preprocessor Options.
35085 (Options for Linking): Document -pthread here....
35086 (RS/6000 and PowerPC Options): ...not here.
35087 (Solaris 2 Options): ...or here.
35088 * doc/cppopts.texi: Document -pthread.
35090 2017-01-08 Martin Sebor <msebor@redhat.com>
35092 PR middle-end/77708
35093 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35094 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35095 New member functions.
35096 (format_directive): Used them.
35098 (pass_sprintf_length::handle_gimple_call): Same.
35099 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35100 to avoid truncation for any argument.
35101 (extract_affine_mul): Same.
35102 * tree.c (get_file_function_name): Same.
35104 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35106 PR middle-end/77484
35107 * predict.def (PRED_INDIR_CALL): Set to 86.
35109 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35111 PR preprocessor/54124
35112 * doc/cppopts.texi: Reformat -d subtable to list the full name
35113 of the options. Add cross-reference to the docs for the general
35114 compiler -d options.
35115 * doc/invoke.texi (Developer Options): Add cross-reference to the
35116 preprocessor-specific -d option documentation.
35118 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35120 PR preprocessor/13498
35121 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35122 redudant material, and reflect new command-line options.
35123 (System Headers): Likewise.
35125 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35127 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35128 -isystem, and -idirafter. Copy-edit.
35129 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35130 default for -ftrack-macro-expansion. Delete obsolete and
35131 badly-formatted implementation details about -fdebug-cpp output.
35132 * doc/cppwarnopts.texi: Copy-edit.
35134 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35137 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35138 that the transition from a max line width >= 1<<10 to narrower
35139 lines works correctly.
35141 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35143 * doc/options.texi (PerFunction): New.
35144 * opt-functions.awk (switch_flags): Map both Optimization and
35145 PerFunction to CL_OPTIMIZATION.
35146 * opth-gen.awk: Test for PerFunction flag along with
35148 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35149 it only when the latter is present. Skip those that don't in
35150 the hash function generator.
35151 * common.opt (fvar-tracking): Mark as PerFunction instead of
35153 (fvar-tracking-assignments): Likewise.
35154 (fvar-tracking-assignments-toggle): Likewise.
35155 (fvar-tracking-uninit): Likewise.
35157 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35159 PR translation/79018
35160 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35163 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35166 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35167 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35168 TARGET_LONG_JUMP_TABLE_OFFSETS.
35169 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35170 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35171 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35172 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35173 * config/m68k/m68k.md (tablejump expander): Likewise.
35174 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35175 TARGET_LONG_JUMP_TABLE_OFFSETS.
35176 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35177 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35179 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35180 David Holsgrove <david.holsgrove@xilinx.com>
35182 * common/config/microblaze/microblaze-common.c
35183 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35184 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35186 * config/microblaze/microblaze.c (microblaze_must_save_register)
35187 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35189 (microblaze_eh_return): New function.
35190 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35191 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35192 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35193 * config/microblaze/microblaze.md (eh_return): New pattern.
35195 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35197 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35198 GCC_DIAGNOSTIC_STRINGIFY): Define.
35200 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35202 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35204 * config/arm/arm.md (<mcrr>): New.
35206 * config/arm/arm.c (arm_arch5te): New.
35207 (arm_option_override): Set arm_arch5te.
35208 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35210 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35211 (arm_mcrr_qualifiers): ... this. New.
35212 (MRRC_QUALIFIERS): Define to...
35213 (arm_mrrc_qualifiers): ... this. New.
35214 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35216 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35217 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35218 (MRRCI, mrrc, MRRC): New.
35219 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35220 VUNSPEC_MRRC2): New.
35222 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35224 * config/arm/arm.md (<mcr>): New.
35226 * config/arm/arm.c (arm_coproc_builtin_available): Add
35227 support for mcr, mrc, mcr2 and mrc2.
35228 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35229 (arm_mcr_qualifiers): ... this. New.
35230 (MRC_QUALIFIERS): Define to ...
35231 (arm_mrc_qualifiers): ... this. New.
35232 (MCR_QUALIFIERS): Define to ...
35233 (arm_mcr_qualifiers): ... this. New.
35234 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35236 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35237 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35238 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35239 VUNSPEC_MRC2): New.
35241 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35243 * config/arm/arm.md (*ldc): New.
35247 * config/arm/arm.c (arm_coproc_builtin_available): Add
35248 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35249 (arm_coproc_ldc_stc_legitimate_address): New.
35250 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35251 'qualifier_const_pointer'.
35252 (LDC_QUALIFIERS): Define to...
35253 (arm_ldc_qualifiers): ... this. New.
35254 (STC_QUALIFIERS): Define to...
35255 (arm_stc_qualifiers): ... this. New.
35256 * config/arm/arm-protos.h
35257 (arm_coproc_ldc_stc_legitimate_address): New.
35258 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35259 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35260 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35261 stc2, stcl, stc2l): New.
35262 * config/arm/constraints.md (Uz): New.
35263 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35264 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35265 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35266 VUNSPEC_STC2L): New.
35268 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35270 * config/arm/arm.md (<cdp>): New.
35271 * config/arm/arm.c (neon_const_bounds): Rename this ...
35272 (arm_const_bounds): ... this.
35273 (arm_coproc_builtin_available): New.
35274 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35275 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35276 (CDP_QUALIFIERS): Define to...
35277 (arm_cdp_qualifiers): ... this. New.
35279 (arm_expand_builtin_args): Add case for 6 arguments.
35280 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35281 (arm_const_bounds): ... this.
35282 (arm_coproc_builtin_available): New.
35283 * config/arm/arm_acle.h (__arm_cdp): New.
35285 * config/arm/arm_acle_builtins.def (cdp): New.
35287 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35288 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35289 'arm_const_bounds'.
35290 * config/arm/types.md (coproc): New.
35291 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35292 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35293 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35294 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35296 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35298 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35299 (UBINOP_QUALIFIERS): New.
35301 (acle_builtin_data): New. Change comment.
35302 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35303 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35304 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35305 arm_acle_builtins.def.
35306 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35307 (arm_init_acle_builtins): New.
35308 (CRC32_BUILTIN): Remove.
35309 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35310 crc32cb, crc32ch and crc32cw.
35311 (arm_init_crc32_builtins): Remove.
35312 (arm_init_builtins): Use arm_init_acle_builtins rather
35313 than arm_init_crc32_builtins.
35314 (arm_expand_acle_builtin): New.
35315 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35316 * config/arm/arm_acle_builtins.def: New.
35318 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35320 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35321 (arm_builtin_datum): ... this.
35322 (arm_init_neon_builtin): Rename to ...
35323 (arm_init_builtin): ... this. Add a new parameters PREFIX
35324 and USE_SIG_IN_NAME.
35325 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35326 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35327 'arm_builtin_datum'.
35328 (arm_init_vfp_builtins): Likewise.
35329 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35330 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35331 (arm_expand_neon_args): Rename to ...
35332 (arm_expand_builtin_args): ... this. Rename builtin_arg
35333 enum values and differentiate between ARG_BUILTIN_MEMORY
35334 and ARG_BUILTIN_NEON_MEMORY.
35335 (arm_expand_neon_builtin_1): Rename to ...
35336 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35337 values, arm_expand_builtin_args and add bool parameter NEON.
35338 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35339 (arm_expand_vfp_builtin): Likewise.
35340 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35342 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35344 PR middle-end/77484
35345 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35346 * predict.c (tree_estimate_probability_bb): Reverse direction of
35347 polymorphic call predictor.
35349 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35351 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35352 (determine_pass_name_match): ...this new function and...
35353 (should_skip_pass_p): ...this new function.
35355 2017-01-06 Nathan Sidwell <nathan@acm.org>
35357 * ipa-visibility.c (function_and_variable_visibility): Reformat
35358 comments and long lines. Remove extrneous if.
35359 * symtab.c (symtab_node::make_decl_local): Fix code format.
35360 (symtab_node::set_section_for_node): Fix comment typo.
35362 2017-01-06 Martin Liska <mliska@suse.cz>
35365 * lra-constraints.c: Rename invariant to lra_invariant.
35366 * predict.c (set_even_probabilities): Initialize e to NULL.
35368 2017-01-05 Martin Sebor <msebor@redhat.com>
35370 PR tree-optimization/78910
35371 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35372 (format_integer): Correct off-by-one error in the handling
35373 of precision with negative numbers in signed conversions..
35375 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35377 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35379 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35381 PR tree-optimization/71016
35382 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35383 factor_out_conditional_conversion. Formatting fix.
35384 (factor_out_conditional_conversion): Add cond_stmt argument.
35385 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35386 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35389 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35391 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35392 read-rtl-function.o, and selftest-rtl.o.
35393 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35394 (selftest::aarch64_test_loading_full_dump): New function.
35395 (selftest::aarch64_run_selftests): New function.
35396 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35397 selftest::aarch64_run_selftests.
35398 * config/i386/i386.c
35399 (selftest::ix86_test_loading_dump_fragment_1): New function.
35400 (selftest::ix86_test_loading_call_insn): New function.
35401 (selftest::ix86_test_loading_full_dump): New function.
35402 (selftest::ix86_test_loading_unspec): New function.
35403 (selftest::ix86_run_selftests): Call the new functions.
35404 * emit-rtl.c (maybe_set_max_label_num): New function.
35405 * emit-rtl.h (maybe_set_max_label_num): New decl.
35406 * function.c (instantiate_decls): Guard call to
35407 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35408 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35410 * gensupport.c (gen_reader::gen_reader): Pass "false"
35411 for new "compact" param of rtx_reader.
35412 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35413 rather than an empty string for NULL strings.
35414 * read-md.c: Potentially include config.h rather than bconfig.h.
35415 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35416 (have_error): New global, copied from errors.c.
35417 (md_reader::read_name): Rename to...
35418 (md_reader::read_name_1): ...this, adding "out_loc" param,
35419 and converting "missing name or number" to returning false, rather
35421 (md_reader::read_name): Reimplement in terms of read_name_1.
35422 (md_reader::read_name_or_nil): New function.
35423 (md_reader::read_string): Handle "(nil)" by returning NULL.
35424 (md_reader::md_reader): Add new param "compact".
35425 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35426 (md_reader::read_file): New method.
35427 * read-md.h (md_reader::md_reader): Add new param "compact".
35428 (md_reader::read_file): New method.
35429 (md_reader::is_compact): New accessor.
35430 (md_reader::read_name): Convert return type from void to file_location.
35431 (md_reader::read_name_or_nil): New decl.
35432 (md_reader::read_name_1): New decl.
35433 (md_reader::m_compact): New field.
35434 (noop_reader::noop_reader): Pass "false" for new "compact" param
35436 (rtx_reader::rtx_reader): Add new "compact" param.
35437 (rtx_reader::read_rtx_operand): Make virtual and convert return
35438 type from void to rtx.
35439 (rtx_reader::read_until): New decl.
35440 (rtx_reader::handle_any_trailing_information): New virtual function.
35441 (rtx_reader::postprocess): New virtual function.
35442 (rtx_reader::finalize_string): New virtual function.
35443 (rtx_reader::m_in_call_function_usage): New field.
35444 (rtx_reader::m_reuse_rtx_by_id): New field.
35445 * read-rtl-function.c: New file.
35446 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35447 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35448 (selftest::verify_three_block_rtl_cfg): New decl.
35449 * read-rtl-function.h: New file.
35450 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35451 For host, include function.h, memmodel.h, and emit-rtl.h.
35452 (one_time_initialization): New function.
35453 (struct compact_insn_name): New struct.
35454 (compact_insn_names): New array.
35455 (find_code): Handle insn codes in compact dumps.
35456 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35457 (bind_subst_iter_and_attr): Likewise.
35458 (add_condition_to_string): Likewise.
35459 (add_condition_to_rtx): Likewise.
35460 (apply_attribute_uses): Likewise.
35461 (add_current_iterators): Likewise.
35462 (apply_iterators): Likewise.
35463 (initialize_iterators): Guard usage of apply_subst_iterator with
35464 #ifdef GENERATOR_FILE.
35465 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35466 (md_reader::read_mapping): Likewise.
35467 (add_define_attr_for_define_subst): Likewise.
35468 (add_define_subst_attr): Likewise.
35469 (read_subst_mapping): Likewise.
35470 (check_code_iterator): Likewise.
35471 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35473 (one_time_initialization): New function.
35474 (rtx_reader::read_until): New method.
35475 (read_flags): New function.
35476 (parse_reg_note_name): New function.
35477 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35478 Handle reuse_rtx ids.
35479 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35480 Add parsing support for RTL dumps, mirroring the special-cases in
35481 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35482 values, and calling handle_any_trailing_information.
35483 (rtx_reader::read_rtx_operand): Convert return type from void
35484 to rtx, returning return_rtx. Handle case 'e'. Call
35485 finalize_string on XSTR and XTMPL fields.
35486 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35487 "(nil)" values were omitted. Call the postprocess vfunc on the
35489 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35490 class ctor. Initialize m_in_call_function_usage. Call
35491 one_time_initialization.
35492 * rtl-tests.c (selftest::test_uncond_jump): Call
35493 set_new_first_and_last_insn.
35494 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35495 * selftest-rtl.c: New file.
35496 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35497 (selftest::get_insn_by_uid): New decl.
35498 * selftest-run-tests.c (selftest::run_tests): Call
35499 read_rtl_function_c_tests.
35500 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35501 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35504 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35506 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35507 operands in a special way. Assert that pos+len <= mode precision.
35509 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35511 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35512 3 argument Alias with unlimited for the negative form.
35513 (fno-vect-cost-model): Removed.
35515 2017-01-05 Martin Liska <mliska@suse.cz>
35517 * hsa-gen.c (gen_hsa_divmod): New function.
35518 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35520 2017-01-05 Martin Liska <mliska@suse.cz>
35523 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35526 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35528 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35529 small constant length operands.
35531 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35533 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35534 between loop iterations.
35536 2017-01-05 Martin Liska <mliska@suse.cz>
35539 * gimplify.c (gimplify_decl_expr): Compare to
35540 asan_poisoned_variables instread of checking flags.
35541 (gimplify_target_expr): Likewise.
35542 (gimplify_expr): Likewise.
35543 (gimplify_function_tree): Conditionally initialize
35544 asan_poisoned_variables.
35546 2017-01-04 Jeff Law <law@redhat.com>
35548 PR tree-optimizatin/78812
35549 * rtl.h (contains_mem_rtx_p): Prototype.
35550 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35551 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35552 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35553 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35554 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35556 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35558 * input.c (assert_char_at_range): Default-initialize actual_range.
35560 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35562 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35563 to match the caller.
35565 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35567 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35568 insns after final jump in test to emit dummy move.
35570 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35572 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35573 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35575 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35577 * multiple_target.c (create_dispatcher_calls): Init e_next.
35578 * tree-ssa-loop-split.c (split_loop): Init border.
35579 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35582 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35587 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35588 (altivec_register_operand): Do not return true if the operand
35589 contains a SUBREG mixing SImode and SFmode.
35590 (vsx_register_operand): Likewise.
35591 (vsx_reg_sfsubreg_ok): New predicate.
35592 (vfloat_operand): Do not return true if the operand contains a
35593 SUBREG mixing SImode and SFmode.
35594 (vint_operand): Likewise.
35595 (vlogical_operand): Likewise.
35596 (gpc_reg_operand): Likewise.
35597 (int_reg_operand): Likewise.
35598 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35599 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35600 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35602 (rs6000_emit_move_si_sf_subreg): New helper function.
35603 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35604 fixup SUBREGs involving SImode and SFmode.
35605 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35606 numbers for the new peephole2 optimization.
35607 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35608 the GLIBC math library that do AND/IOR/XOR operations on single
35609 precision floating point.
35610 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35611 target macros to say whether we need to avoid SUBREGs mixing
35613 (TARGET_ALLOW_SF_SUBREG): Likewise.
35614 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35615 (UNSPEC_SI_FROM_SF): Likewise.
35616 (iorxor): Change spacing.
35617 (and_ior_xor): New iterator for AND, IOR, and XOR.
35618 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35619 (movdi_from_sf_zero_ext): Likewise.
35620 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35621 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35622 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35623 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35624 (fms<mode>4): Likewise.
35625 (fnma<mode>4): Likewise.
35626 (fnms<mode>4): Likewise.
35627 (nfma<mode>4): Likewise.
35628 (nfms<mode>4): Likewise.
35630 2017-01-04 Marek Polacek <polacek@redhat.com>
35633 * doc/invoke.texi: Document -Wpointer-compare.
35635 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35637 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35640 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35641 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35644 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35647 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35648 documentation of the powerpc_popcntb_ok attribute.
35649 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35650 code to issue warning messages if a requested CPU configuration is
35651 not supported by the binary (assembler and loader) toolchain.
35652 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35653 made to define a built-in function that has been disabled.
35654 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35655 made to define a built-in function that has been disabled.
35656 (altivec_init_builtins): Add comment explaining why definition
35657 of the DST built-in functions is not preceded by an assertion
35658 check. Add assertions to prevent ICE if attempts are made to
35659 define an altivec predicate or an abs* built-in function that has
35661 (htm_init_builtins): Add comment explaining why definition of the
35662 htm built-in functions is not preceded by an assertion check.
35664 2017-01-04 Jeff Law <law@redhat.com>
35666 PR tree-optimizatin/67955
35667 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35668 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35669 the points-to solution does not include pt_null. Use DECL_PT_UID
35672 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35674 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35675 Use gen_int_mode instead of gen_lopwart for const_int operands.
35677 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35679 PR tree-optimization/71563
35680 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35681 out of range value - has low bits known to be zero.
35683 2017-01-04 Alan Modra <amodra@gmail.com>
35685 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35686 * configure: Regenerate.
35687 * config.in: Regenerate.
35689 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35692 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35693 a substring of the message, but strcmp with the whole message. Ifdef
35694 ENABLE_NLS, translate the message first using dgettext.
35696 2017-01-03 Jeff Law <law@redhat.com>
35698 PR tree-optimizatin/78856
35699 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35700 (mark_threaded_blocks): Remove code to truncate thread paths that
35701 cross multiple loop headers. Instead invalidate the cached loop
35702 iteration information and handle case of a thread path walking
35703 into an irreducible region.
35705 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35708 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35709 assertions. Add support for doing the signbit if the IEEE 128-bit
35710 floating point value is in a GPR.
35711 * config/rs6000/rs6000.md (Fsignbit): Delete.
35712 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35713 Update the length attribute if the value is in a GPR.
35714 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35715 the sign or zero extension instruction, since the value is always 0/1.
35716 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35719 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35720 extracting SImode to a GPR register so that we can generate a
35721 store, limit the vector to be in a traditional Altivec register
35722 for the vextuwrx instruction.
35724 2017-01-03 Ian Lance Taylor <iant@google.com>
35726 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35728 2017-01-03 Martin Sebor <msebor@redhat.com>
35730 PR tree-optimization/78696
35731 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35732 precision. Use MPFR for %f for greater fidelity. Correct handling
35734 (pass_sprintf_length::compute_format_length): Set width and precision
35735 specified by asrerisk to void_node for vararg functions.
35736 (try_substitute_return_value): Adjust dump output.
35738 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35740 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35742 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35744 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35745 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35746 -mlra/-mno-lra was passed to the compiler.
35748 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35750 PR rtl-optimization/65618
35751 * emit-rtl.c (try_split): Move initialization of "before" and
35752 "after" to just before the call to emit_insn_after_setloc.
35754 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35756 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35758 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35760 * dwarf2out.c (gen_enumeration_type_die): When
35761 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35763 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35765 PR tree-optimization/78965
35766 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35767 Change first argument from const call_info & to call_info &. For %n
35768 set info.nowrite to false.
35770 PR middle-end/78901
35771 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35772 possibly throwing calls.
35774 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35775 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35776 and fns handling, rather than in a separate case SSA_NAME.
35778 2017-01-02 Jeff Law <law@redhat.com>
35780 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35781 fixes for first_period and second_period variables.
35783 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35786 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35787 (*insvqi_1): New insn pattern.
35788 (*insvqi_1_mem_rex64): Ditto.
35789 (*insvqi_2): Ditto.
35790 (*insvqi_3): Rename from *insvqi.
35792 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35794 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35796 * doc/cfg.texi (Edges): Remove reference to Java.
35797 (Maintaining the CFG): Ditto.
35799 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35801 PR middle-end/77674
35802 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35803 transparent aliases.
35805 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35807 PR middle-end/77484
35808 * predict.def (PRED_CALL): Update hitrate.
35809 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35810 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35811 into direct/indirect/polymorphic variants.
35813 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35815 Update copyright years.
35817 * gcc.c (process_command): Update copyright notice dates.
35818 * gcov-dump.c (print_version): Ditto.
35819 * gcov.c (print_version): Ditto.
35820 * gcov-tool.c (print_version): Ditto.
35821 * gengtype.c (create_file): Ditto.
35822 * doc/cpp.texi: Bump @copying's copyright year.
35823 * doc/cppinternals.texi: Ditto.
35824 * doc/gcc.texi: Ditto.
35825 * doc/gccint.texi: Ditto.
35826 * doc/gcov.texi: Ditto.
35827 * doc/install.texi: Ditto.
35828 * doc/invoke.texi: Ditto.
35830 Copyright (C) 2017 Free Software Foundation, Inc.
35832 Copying and distribution of this file, with or without modification,
35833 are permitted in any medium without royalty provided the copyright
35834 notice and this notice are preserved.