1 2017-11-19 Jeff Law <law@redhat.com>
3 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4 of degenerates resulting from ignoring an edge.
6 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
9 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
11 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
14 * profile-count.c (profile_count::to_sreal_scale): Fix return value
15 for uninitialied counts.
17 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
20 * tree-inline.c (estimate_num_insns): Set to 1 at least.
22 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
25 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
28 2017-11-19 Tom de Vries <tom@codesourcery.com>
30 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
33 2017-11-19 Tom de Vries <tom@codesourcery.com>
35 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
38 2017-11-19 Tom de Vries <tom@codesourcery.com>
40 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
43 2017-11-19 Tom de Vries <tom@codesourcery.com>
45 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
47 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
49 2017-11-19 Tom de Vries <tom@codesourcery.com>
51 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
53 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
54 semicolon after MCORE_EXPORT_NAME call.
56 2017-11-19 Tom de Vries <tom@codesourcery.com>
59 * vmsdbgout.c (vmsdbgout_early_finish): New function.
60 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
62 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
64 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
65 (cgraph_edge::clone): Cleanup updating of profile.
66 * ipa-cp.c (update_profiling_info): Likewise.
67 * ipa-inline-transform.c (inline_transform): Likewise.
68 * ipa-inline.c (inline_small_functions): Add missing space to dump.
69 * ipa-split.c (execute_split_functions): Do not split when function
71 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
72 * profile-count.c (profile_count::dump): Add global0.
73 (profile_count::to_cgraph_frequency): Do not ICE when entry is
75 (profile_count::to_sreal_scale): Likewise.
76 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
77 (profile_count::combine_with_ipa_count): New function.
78 * profile-count.h (profile_guessed_global0adjusted): New.
79 (profile_count::adjusted_zero): New.
80 (profile_count::global0adjusted): New.
81 (profile_count::combine_with_ipa_count): New.
82 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
83 correct profile of return block of split functions.
84 (copy_cfg_body): Remove unused profile_count.
85 (copy_body): Likewise.
86 (expand_call_inline): Update.
87 (tree_function_versioning): Update.
89 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
91 * hash-set.h (hash_set::empty): New.
92 * tree-ssa-threadbackward.h: Delete.
93 * tree-ssa-threadbackward.c (class thread_jumps): New.
94 Move max_threaded_paths into class.
95 (fsm_find_thread_path): Remove arguments that are now in class.
96 (profitable_jump_thread_path): Rename to...
97 (thread_jumps::profitable_jump_thread_path): ...this.
98 (convert_and_register_jump_thread_path): Rename to...
99 (thread_jumps::convert_and_register_current_path): ...this.
100 (check_subpath_and_update_thread_path): Rename to...
101 (thread_jumps::check_subpath_and_update_thread_path): ...this.
102 (register_jump_thread_path_if_profitable): Rename to...
103 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
104 (handle_phi): Rename to...
105 (thread_jumps::handle_phi): ...this.
106 (handle_assignment): Rename to...
107 (thread_jumps::handle_assignment): ...this.
108 (fsm_find_control_statement_thread_paths): Rename to...
109 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
110 (find_jump_threads_backwards): Rename to...
111 (thread_jumps::find_jump_threads_backwards): ...this.
112 Initialize path local data.
113 (pass_thread_jumps::execute): Call find_jump_threads_backwards
114 from within thread_jumps class.
115 (pass_early_thread_jumps::execute): Same.
117 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
119 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
121 2017-11-17 Steve Ellcey <sellcey@cavium.com>
123 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
124 to canonical location.
126 2017-11-17 Steve Ellcey <sellcey@cavium.com>
129 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
131 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
133 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
135 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
136 rcrt1.o%s/grcrt1.o%s for -static-pie.
138 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
140 * i386.c (ix86_multiplication_cost, ix86_division_cost,
141 ix86_shift_rotate_cost): Break out from ...
142 (ix86_rtx_costs): ... here.
143 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
146 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
148 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
149 when entry block was promoted unlikely.
150 (estimate_bb_frequencies): Increase frequency scale.
151 * profile-count.h (profile_count): Export precision info.
153 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
155 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
156 disturbin profile of entry block.
158 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
160 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
163 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
165 * ipa-cp.c (update_profiling_info): Handle conversion to local
167 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
169 2017-11-17 Jeff Law <law@redhat.com>
171 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
172 from evrp_dom_walker class. Various methods moved into new class.
173 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
174 (evrp_range_analyzer::enter): New method.
175 (evrp_range_analyzer::leave): New method.
176 (evrp_dom_walker): Remove delegators no longer needed by this class.
177 Replace vr_values data member with evrp_range_analyzer
179 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
180 method extracted from evrp_dom_walker::before_dom_children.
181 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
182 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
184 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
185 Add private copy constructor and move assignment operators.
186 Privatize methods and class data where trivially possible.
187 (evrp_dom_walker::cleanup): New function, extracted from
188 execute_early_vrp. Simplify access to class data.
190 * vr-values.h (get_output_for_vrp): Prototype.
191 * vr-values.c (get_output_for_vrp): New function extracted from
192 vrp_visit_assignment_or_call and extract_range_from_stmt.
193 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
195 2017-11-17 Luis Machado <luis.machado@linaro.org>
197 * config/aarch64/aarch64.c
198 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
199 (qdf24xx_tunings) <autoprefetcher_model>: Set to
200 tune_params::AUTOPREFETCHER_WEAK.
202 2017-11-17 Tamar Christina <tamar.christina@arm.com>
205 * config/arm/arm.c (arm_valid_target_attribute_rec):
206 Parse "arch=" and "+<ext>".
207 (arm_valid_target_attribute_tree): Re-init global options.
208 (arm_option_override): Make non-static.
209 (arm_options_perform_arch_sanity_checks): Make errors fatal.
210 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
211 (__ARM_FEATURE_CRC32): Support undef.
212 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
213 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
215 2017-11-17 David Malcolm <dmalcolm@redhat.com>
217 * gdbinit.in (break-on-diagnostic): New command.
219 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
221 * config/i386/i386.c (ix86_expand_epilogue): Change simple
222 return to indirect jump for EH return if control-flow
223 protection is enabled. Change explicit 'false' argument in
224 pro_epilogue_adjust_stack with a value of
226 * config/i386/i386.md (simple_return_indirect_internal):
227 Remove SImode restriction to support 64-bit.
229 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
231 * combine.c (added_notes_insn): New.
232 (try_combine): Handle added_notes_insn like added_links_insn.
233 Rewrite return value code.
234 (distribute_notes): Set added_notes_insn to the earliest insn we added
237 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
239 PR rtl-optimization/82621
240 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
241 dest of one of those SETs is unused.
243 2017-11-17 Richard Biener <rguenther@suse.de>
246 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
247 * tree-pretty-print.c (dump_generic_node): Handle
248 annot_expr_parallel_kind.
249 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
250 * gimplify.c (gimple_boolify): Likewise.
252 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
254 * config.gcc (extra_headers): Add cet.h for x86 targets.
255 * config/i386/cet.h: New file.
256 * doc/install.texi: Add --enable-cet/--disable-cet.
258 2017-11-17 Richard Biener <rguenther@suse.de>
260 PR tree-optimization/83017
261 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
262 (gen_parallel_loop): Properly count iterations.
263 (parallelize_loops): Handle loop->can_be_parallel independent
264 of flag_loop_parallelize_all. Make static profitability test match
266 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
267 * invoke.texi (parloops-min-per-thread): Document.
269 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
271 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
272 upstreaming review comments.
274 2017-11-17 Tamar Christina <tamar.christina@arm.com>
276 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
277 with fast unaligned access.
278 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
280 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
282 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
283 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
285 2017-11-17 Richard Biener <rguenther@suse.de>
287 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
288 folding of references.
290 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
293 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
294 of replacing call to strncmp with corresponding call to strcmp when
297 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
299 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
300 option prefer-avx256 for skylake-avx512 configuration.
301 * config/i386/i386.c (ix86_option_override_internal): Ditto.
302 (get_builtin_code_for_version): Ditto.
304 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
305 Monk Chiang <sh.chiang04@gmail.com>
307 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
308 (FIXED_REGISTERS): Reserve more register numbers.
309 (CALL_USED_REGISTERS): Likewise.
310 (REG_ALLOC_ORDER): Likewise.
311 (REG_CLASS_CONTENTS): Likewise.
312 (REGISTER_NAMES): Likewise.
314 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
315 Kito Cheng <kito.cheng@gmail.com>
317 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
319 * config/nds32/iterators.md: Add vector mode iterators and attributes.
321 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
323 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
324 parameter list for vec_splats.
326 2017-11-16 Joseph Myers <joseph@codesourcery.com>
328 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
330 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
333 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
336 * doc/install.texi: Document incompatibility of Perl >=5.6.26
337 with the required version of automake 1.11.6.
339 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
341 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
343 (DU_C2_3_power9): Correct reservation combinations.
344 (FP_DIV_power9, VEC_DIV_power9): New.
345 (power9-alu): Split out rotate/shift...
346 (power9-rot): ...to here, correct dispatch resource.
347 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
349 (power9-fp): Correct latency.
350 (power9-sdiv): Add div/sqrt resource.
351 (power9-ddiv): Correct latency, add div/sqrt resource.
352 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
353 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
355 (power9-qpdiv, power9-qpmul): Adjust resource usage.
357 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
359 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
360 switch statement mapping KF built-ins to TF built-ins if we don't
361 have the proper ISA 3.0 assembler support.
363 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
365 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
366 (lower_emutls_function_body): Do not compute it.
368 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
370 * ipa-split.c (split_bb_info): Turn time to sreal.
371 (split_point): Likewise.
372 (dump_split_point): Likewise.
373 (fine_split_points): Likewise.
374 (execute_split_functions): Only zero split_bbs; turn time to sreals.
376 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
378 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
381 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
383 * predict.c (combine_predictions_for_bb): Preserve zero predicted
385 (expensive_function_p): Remove useless assert.
386 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
388 2017-11-16 Martin Sebor <msebor@redhat.com>
390 PR tree-optimization/82588
391 PR tree-optimization/82583
392 * tree-vrp.c (check_array_ref): Handle flexible array members,
393 string literals, and inner indices.
394 (search_for_addr_array): Add detail to diagnostics.
396 2017-11-16 Nathan Sidwell <nathan@acm.org>
400 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
401 (SET_DECL_ASSEMBLER_NAME): Forward to
402 overwrite_decl_assembler_name.
403 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
404 (overwrite_decl_assembler_name): Declare.
405 * tree.c (overwrite_decl_assembler_name): New.
406 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
407 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
408 (LANG_HOOKS_INITIALIZER): Add it.
409 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
410 * langhooks.c (lhd_set_decl_assembler_name): Use
411 SET_DECL_ASSEMBLER_NAME.
412 (lhd_overwrite_decl_assembler_name): Default implementation.
414 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
415 Jackson Woodruff <jackson.woodruff@arm.com>
417 PR tree-optimization/71026
418 * match.pd: Canonicalize constant multiplies in division.
420 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
422 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
423 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
424 Remove OPT_fomit_frame_pointer entry.
425 * common/config/arc/arc-common.c: Likewise.
426 * common/config/arm/arm-common.c: Likewise.
427 * common/config/avr/avr-common.c: Likewise.
428 * common/config/c6x/c6x-common.c: Likewise.
429 * common/config/cr16/cr16-common.c: Likewise.
430 * common/config/cris/cris-common.c: Likewise.
431 * common/config/epiphany/epiphany-common.c: Likewise.
432 * common/config/fr30/fr30-common.c: Likewise.
433 * common/config/frv/frv-common.c: Likewise.
434 * common/config/ia64/ia64-common.c: Likewise.
435 * common/config/iq2000/iq2000-common.c: Likewise.
436 * common/config/lm32/lm32-common.c: Likewise.
437 * common/config/m32r/m32r-common.c: Likewise.
438 * common/config/mcore/mcore-common.c: Likewise.
439 * common/config/microblaze/microblaze-common.c: Likewise.
440 * common/config/mips/mips-common.c: Likewise.
441 * common/config/mmix/mmix-common.c: Likewise.
442 * common/config/mn10300/mn10300-common.c: Likewise.
443 * common/config/nios2/nios2-common.c: Likewise.
444 * common/config/pa/pa-common.c: Likewise.
445 * common/config/pdp11/pdp11-common.c: Likewise.
446 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
447 * common/config/riscv/riscv-common.c: Likewise.
448 * common/config/rs6000/rs6000-common.c: Likewise.
449 * common/config/rx/rx-common.c: Likewise.
450 * common/config/s390/s390-common.c: Likewise.
451 * common/config/sh/sh-common.c: Likewise.
452 * common/config/sparc/sparc-common.c: Likewise.
453 * common/config/tilegx/tilegx-common.c: Likewise.
454 * common/config/tilepro/tilepro-common.c: Likewise.
455 * common/config/v850/v850-common.c: Likewise.
456 * common/config/visium/visium-common.c: Likewise.
457 * common/config/xstormy16/xstormy16-common.c: Likewise.
458 * common/config/xtensa/xtensa-common.c: Likewise.
459 * invoke.texi (-fomit-frame-pointer): Update documentation.
461 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
463 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
465 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
467 * cfg.c (scale_bbs_frequencies_int,
468 cale_bbs_frequencies_gcov_type): Remove.
469 * cfg.h (scale_bbs_frequencies_int,
470 cale_bbs_frequencies_gcov_type): Remove.
472 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
474 * tree-ssa-loop-manip.c
475 (scale_dominated_blocks_in_loop): Update to profile counts.
476 (tree_transform_and_unroll_loop): Likewise.
478 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
480 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
481 scale_bbs_frequencies_int.
483 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
485 * final.c (compute_alignments): Use counts rather than frequencies.
487 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
489 * cfgloopanal.c: Include sreal.h
490 (average_num_loop_insns): Use counts and sreal for accounting.
492 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
494 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
497 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
498 Kito Cheng <kito.cheng@gmail.com>
500 * config/nds32/constraints.md: Provide more constraints.
501 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
502 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
503 support constraints usage.
505 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
507 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
509 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
510 Kito Cheng <kito.cheng@gmail.com>
512 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
513 * config/nds32/nds32.opt: Refine the layout.
514 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
515 TARGET_EXT_STRING): Support new options.
516 * config/nds32/nds32.h: Likewise.
517 * config/nds32/nds32.md: Likewise.
518 * config/nds32/nds32-predicates.c: Likewise.
519 * config/nds32/constraints.md: Likewise.
520 * common/config/nds32/nds32-common.c: Likewise.
522 2017-11-16 Julia Koval <julia.koval@intel.com>
525 * config/i386/gfniintrin.h: Add sse check.
526 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
528 2017-11-16 Julia Koval <julia.koval@intel.com>
530 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
531 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
532 (ix86_handle_option): Handle -mavx512vbmi2.
533 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
534 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
535 * config/i386/i386-c.c (__AVX512VBMI2__): New.
536 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
537 (ix86_valid_target_attribute_inner_p): Ditto.
538 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
539 * config/i386/i386.opt (mavx512vbmi2): New option.
540 * doc/invoke.texi: Add new option.
542 [2017-11-16 Julia Koval <julia.koval@intel.com>
544 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
545 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
546 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
547 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
548 _mm512_gf2p8mul_epi8): New intrinsics.
549 * config/i386/i386-builtin-types.def
550 (V64QI_FTYPE_V64QI_V64QI): New type.
551 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
552 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
553 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
554 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
555 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
556 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
558 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
560 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
561 explicitly as a stream of bytes.
563 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
565 * config/rs6000/altivec.h (vec_xst_be): New #define.
566 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
567 and externalize from *altivec_vperm_<mode>_internal.
568 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
570 (XL_BE_V8HI): Likewise.
571 (XL_BE_V4SI): Likewise.
572 (XL_BE_V4SI): Likewise.
573 (XL_BE_V2DI): Likewise.
574 (XL_BE_V4SF): Likewise.
575 (XL_BE_V2DF): Likewise.
576 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
577 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
578 all array entries with these keys: VSX_BUILTIN_VEC_XL,
579 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
580 VSX_BUILTIN_VEC_XST_BE.
581 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
582 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
584 (altivec_init_builtins): Replace conditional calls to def_builtin
585 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
586 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
587 with unconditional calls. Remove calls to def_builtin for
588 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
589 __builtin_vec_xst_be.
590 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
591 to define_expand, and add alternate RTL generation for P8.
592 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
594 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
595 add alternate RTL generation for P8.
596 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
597 vsx_ld_elemrev_v16qi.
598 (vsx_st_elemrev_v8hi): Convert define_insn
599 to define_expand, and add alternate RTL generation for P8.
600 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
602 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
603 add alternate RTL generation for P8.
604 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
605 vsx_st_elemrev_v16qi.
607 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
610 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
611 TARGET_AVX512ER check.
612 (ix86_option_override_internal): Set MASK_VZEROUPPER if
613 neither -mzeroupper nor -mno-zeroupper is used and
614 TARGET_EMIT_VZEROUPPER is set.
615 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
616 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
618 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
620 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
621 folding of vector compares.
622 (fold_build_vec_cmp): New helper function.
623 (fold_compare_helper): New helper function.
624 (builtin_function_type): Add compare builtins to the list of functions
625 having unsigned arguments. Cosmetic updates to comment indentation.
626 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
627 the not+eq combination.
629 2017-11-15 Bin Cheng <bin.cheng@arm.com>
631 PR tree-optimization/82726
632 PR tree-optimization/70754
633 * tree-predcom.c (order_drefs_by_pos): New function.
634 (combine_chains): Move code setting has_max_use_after to...
635 (try_combine_chains): ...here. New parameter. Sort combined chains
636 according to position information.
637 (tree_predictive_commoning_loop): Update call to above function.
638 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
640 2017-11-15 Bin Cheng <bin.cheng@arm.com>
642 PR tree-optimization/82726
644 2017-01-23 Bin Cheng <bin.cheng@arm.com>
646 PR tree-optimization/70754
647 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
648 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
649 combined stmt before it if not NULL.
650 (combine_chains): Process refs reversely and compute dominance point
654 2017-02-23 Bin Cheng <bin.cheng@arm.com>
656 PR tree-optimization/79663
657 * tree-predcom.c (combine_chains): Process refs in reverse order
658 only for ZERO length chains, and add explaining comment.
660 2017-11-15 Tamar Christina <tamar.christina@arm.com>
662 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
663 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
664 * doc/invoke.texi (ARM Options): Add armv8.3-a.
666 2017-11-15 Tamar Christina <tamar.christina@arm.com>
668 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
670 2017-11-15 Martin Liska <mliska@suse.cz>
672 * tree-cfg.c (pass_warn_function_return::execute):
673 Compare warn_return_type for greater than zero.
675 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
679 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
680 to return true on Xeon and not on Xeon Phi.
681 (ix86_check_avx256_register): Changed to ...
682 (ix86_check_avx_upper_register): ... this. Add extra check for
683 VALID_AVX512F_REG_OR_XI_MODE.
684 (ix86_avx_u128_mode_needed): Changed
685 ix86_check_avx256_register to ix86_check_avx_upper_register.
686 (ix86_check_avx256_stores): Changed to ...
687 (ix86_check_avx_upper_stores): ... this. Changed
688 ix86_check_avx256_register to ix86_check_avx_upper_register.
689 (ix86_avx_u128_mode_after): Changed
690 avx_reg256_found to avx_upper_reg_found. Changed
691 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
692 (ix86_avx_u128_mode_entry): Changed
693 ix86_check_avx256_register to ix86_check_avx_upper_register.
694 (ix86_avx_u128_mode_exit): Ditto.
695 * config/i386/i386.h: (host_detect_local_cpu): New define.
697 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
699 * config/arm/xgene1.md (xgene1): Split into automatons
700 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
701 (xgene1_f_load): Adjust reservations and/or types.
702 (xgene1_f_store): Likewise.
703 (xgene1_load_pair): Likewise.
704 (xgene1_store_pair): Likewise.
705 (xgene1_fp_load1): Likewise.
706 (xgene1_load1): Likewise.
707 (xgene1_store1): Likewise.
708 (xgene1_move): Likewise.
709 (xgene1_alu): Likewise.
710 (xgene1_simd): Likewise.
711 (xgene1_bfm): Likewise.
712 (xgene1_neon_load1): Likewise.
713 (xgene1_neon_store1): Likewise.
714 (xgene1_neon_logic): Likewise.
715 (xgene1_neon_st1): Likewise.
716 (xgene1_neon_ld1r): Likewise.
717 (xgene1_alu_cond): Added.
718 (xgene1_shift_reg): Likwise.
719 (xgene1_bfx): Likewise.
720 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
722 2017-11-15 Jakub Jelinek <jakub@redhat.com>
725 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
727 (can_widen_mult_without_libcall): New function.
728 (expand_mul_overflow): If only checking unsigned mul overflow,
729 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
730 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
731 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
732 (expand_DIVMOD): Formatting fix.
733 * expmed.h (expand_mult): Add NO_LIBCALL argument.
734 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
735 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
737 PR tree-optimization/82977
738 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
739 constructed temporary to strlen_to_stridx.put.
741 2017-11-15 Martin Liska <mliska@suse.cz>
743 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
744 * configure: Regenerate.
746 2017-11-15 Martin Liska <mliska@suse.cz>
749 * config/sh/sh-mem.cc: Use proper probability for
752 2017-11-14 Jeff Law <law@redhat.com>
754 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
755 the red zone for stack_clash_protection_final_dynamic_probe targets
756 when the total dynamic stack size is zero bytes.
758 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
759 blocks is post order.
761 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
763 * dumpfile.h (TDF_COMPARE_DEBUG): New.
764 * final.c (rest_of_clean_state): Set it for the
765 -fcompare-debug dump.
766 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
767 class when TDF_COMPARE_DEBUG is set.
769 * dwarf2out.c (gen_producer_string): Discard
772 2017-11-15 Joseph Myers <joseph@codesourcery.com>
775 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
776 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
777 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
778 (__TGMATH_REAL_2_3): Remove macros.
779 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
780 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
782 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
784 (remquo): Define using __TGMATH_REAL_3.
786 2017-11-14 Jeff Law <law@redhat.com>
788 * vr-values.c: New file with contents extracted from tree-vrp.c.
789 * Makefile.in (OBJS): Add vr-values.o
790 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
791 (set_value_range, set_and_canonicalize_value_range): Likewise.
792 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
793 (value_range_constant_singleton, symbolic_range_p): Likewise.
794 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
795 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
796 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
797 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
798 (find_case_label_range, find_case_label_index): Likewise.
799 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
800 (range_int_cst_singleton_p, value_inside_range): Likewise.
801 (get_single_symbol): Likewise.
802 (switch_update): Move structure definition here.
803 (to_remove_edges, to_update_switch_stmts): Provide externs.
804 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
805 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
806 (vrp_val_is_min, set_value_range): Likewise.
807 (set_and_canonicalize_value_range, copy_value_range): Likewise.
808 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
809 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
810 (range_is_nonnull, range_int_cst_p): Likewwise.
811 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
812 (get_single_symbol, operand_less_p): Likewise
813 (compare_values_warnv, compare_values): Likewise.
814 (value_inside_range, value_range_constant_singleton): Likewise.
815 (zero_nonzero_bitgs_from_vr): Likewise.
816 (extract_range_from_binary_expr_1): Likewise.
817 (overflow_comparison_p): Likewise.
818 (to_remove_edges, to_update_switch_stmts): Likewise.
819 (find_case_label-index, find_case_label_range): Likewise.
820 (switch_update, set_value_range_to_nonnegative): Remove.
821 (set_value_range_to_truthvalue): Likewise.
822 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
823 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
824 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
825 (find_case_label_ranges, test_for_singularity): Likewise.
826 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
827 (x_vr_values): Move to its remaining use site.
829 2017-11-10 Jeff Law <law@redhat.com>
831 * vr-values.h (VR_INITIALIZER): Move #define here.
832 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
833 * Makefile.in (OBJS): Add tree-evrp.o
834 * tree-vrp.h (assert_info): Move structure definition here.
835 (set_value_range_to_varying): Prototype.
836 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
837 (infer_value_range, register_edge_assert_for): Likewise.
838 (stmt_interesting_for_vrp): Likewise.
839 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
840 (set_value_range_to_varying): No longer static.
841 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
842 (infer_value_range, register_edge_assert_for): Likewise.
844 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
846 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
847 generate the XXBRD instruction.
849 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
850 (rs6000_builtin_type_compatible): Treat _Float128 and long double
851 as being compatible if -mabi=ieeelongdouble.
852 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
853 to setup float128 built-ins with hardware support.
854 (BU_FLOAT128_HW_2): Likewise.
855 (BU_FLOAT128_HW_3): Likewise.
856 (BU_FLOAT128_HW_VSX_1): Likewise.
857 (BU_FLOAT128_HW_VSX_2): Likewise.
858 (scalar_extract_expq): Change float128 built-in functions to
859 accommodate having both KFmode and TFmode functions. Use the
860 KFmode variant as the default.
861 (scalar_extract_sigq): Likewise.
862 (scalar_test_neg_qp): Likewise.
863 (scalar_insert_exp_q): Likewise.
864 (scalar_insert_exp_qp): Likewise.
865 (scalar_test_data_class_qp): Likewise.
866 (sqrtf128_round_to_odd): Delete processing the round to odd
867 built-in functions as special built-in functions, and define them
868 as float128 built-ins. Use the KFmode variant as the default.
869 (truncf128_round_to_odd): Likewise.
870 (addf128_round_to_odd): Likewise.
871 (subf128_round_to_odd): Likewise.
872 (mulf128_round_to_odd): Likewise.
873 (divf128_round_to_odd): Likewise.
874 (fmaf128_round_to_odd): Likewise.
875 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
876 support for KFmode and TFmode xststdcqp calls.
877 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
878 point, switch the built-in handlers for the get/set float128
879 exponent, get float128 mantissa, float128 test built-ins, and the
880 float128 round to odd built-in functions. Eliminate creating the
881 float128 round to odd built-in functions as special built-ins.
882 (rs6000_init_builtins): Eliminate special creation of the float128
883 round to odd built-in functions.
884 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
885 function insns to support both TFmode and KFmode variants.
886 (xsxsigqp_<mode>): Likewise.
887 (xsiexpqpf_<mode>): Likewise.
888 (xsiexpqp_<mode>): Likewise.
889 (xststdcqp_<mode>): Likewise.
890 (xststdcnegqp_<mode>): Likewise.
891 (xststdcqp_<mode>): Likewise.
893 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
895 * tree-ssa-threadupdate.c (compute_path_counts): Remove
896 unused path_in_freq_ptr parameter.
897 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
899 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
901 * ipa-inline.c (edge_badness): Dump sreal frequency.
902 (compute_inlined_call_time): Match natural implementaiton ...
903 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
904 forgotten division by CGRAPH_FREQ_BASE.
906 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
908 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
909 Solaris 11. Update comment.
910 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
912 * config/sol2.h (STARTFILE_SPEC): Likewise.
913 * configure: Regenerate.
915 2017-11-14 Carl Love <cel@us.ibm.com>
917 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
918 le_ and be_ prefixes to swap* variables. Remove
919 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
922 2017-11-14 Jason Merrill <jason@redhat.com>
924 Support GTY((cache)) on hash_map.
925 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
926 (ggc_cache_remove): Override it instead of ggc_mx.
927 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
928 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
929 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
930 (simple_cache_map_traits): Override maybe_mx.
931 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
932 (hash_map): Friend gt_cleare_cache.
933 (gt_cleare_cache): New.
934 * tree.h (tree_cache_traits): New hash_map traits class.
935 (tree_cache_map): New typedef.
937 2017-11-14 Richard Biener <rguenther@suse.de>
939 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
940 paramter and handling.
941 (cleanup_control_flow_bb): Likewise.
942 (cleanup_control_flow_pre): New helper performing a DFS walk
943 to call cleanup_control_flow_bb in PRE order.
944 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
945 via cleanup_control_flow_pre.
947 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
949 * config/aarch64/aarch64-simd.md
950 (aarch64_simd_bsl<mode>_internal): Remove DImode.
951 (*aarch64_simd_bsl<mode>_alt): Likewise.
952 (aarch64_simd_bsldi_internal): New.
953 (aarch64_simd_bsldi_alt): Likewise.
955 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
957 * tracer.c (better_p): Do not compare frequencies.
958 * reg-stack.c (better_edge): Likewise.
959 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
962 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
964 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
965 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
966 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
967 * lto-streamer-in.c (input_function): Use update_max_bb_count.
968 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
969 * predict.c (maybe_hot_frequency_p): Inline to ...
970 (maybe_hot_count_p): ... here; rewrite to counts.
971 (counts_to_freqs): Rename to ...
972 (update_max_bb_count): ... this one.
973 (expensive_function_p): Use counts.
974 (estimate_bb_frequencies): Update.
975 (rebuild_frequencies): Update.
976 * predict.h (counts_to_freqs): Rename to ...
977 (update_max_bb_count): ... this one.
978 * profile.c (compute_branch_probabilities): Add debug info
979 * tree-inline.c (expand_call_inline): Update debug info.
980 (optimize_inline_calls): Use update_max_bb_count..
981 (tree_function_versioning): Use update_max_bb_count..
982 * value-prof.c (gimple_value_profile_transformations):
983 Do not use update_max_bb_count.
985 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
987 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
988 always use frequencies.
990 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
992 * bb-reorder.c: Remove frequencies from comments.
993 (better_edge_p): Use profile counts.
994 (find_traces): Dump profile counts.
995 (rotate_loop): Use profile counts.
996 (find_traces_1_round): Likewise.
997 (connect_better_edge_p): Use counts instead of probabilities for
999 (copy_bb_p): Drop early check for non-0 frequency.
1000 (sanitize_hot_paths): Update comments.
1002 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1004 * ipa-split.c (struct split_point): Add count.
1005 (consider_split): Do not compute incoming frequency; compute incoming
1006 count and store it to split_point.
1007 (split_function): Set count of the call to split part correctly.
1009 2017-11-13 Carl Love <cel@us.ibm.com>
1011 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1013 2017-11-13 Tom Tromey <tom@tromey.com>
1015 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1017 2017-11-13 Carl Love <cel@us.ibm.com>
1019 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1020 Add support for builtins:
1021 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1022 vector {un,}signed {char,int,short},
1023 vector {un,}signed {char,int,short}) arguments.
1024 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1025 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1026 Add BU_P9V_AV_2 expansions for the builtins.
1027 * config/rs6000/altivec.h (vec_first_match_index,
1028 vec_first_mismatch_index, vec_first_match_or_eos_index,
1029 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1030 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1031 new extern declaration.
1032 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1033 * config/rs6000/vsx.md (first_match_index_<mode>,
1034 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1035 first_mismatch_or_eos_index_<mode>): Add define expand.
1036 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1037 * doc/extend.texi: Update the built-in documenation file for the new
1040 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1042 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1043 into the min/max operations for _Float<N> and _Float<N>X types.
1045 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1048 * dwarf2out.c (do_eh_frame): New static variable.
1049 (dwarf2out_begin_prologue): Set it.
1050 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1052 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1054 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1056 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1058 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1061 * debug.h (dwarf2out_do_eh_frame): Declare.
1062 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1063 (dwarf2out_do_frame): Use it.
1064 (dwarf2out_do_cfi_asm): Likewise.
1065 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1066 (dwarf2out_assembly_start): Likewise.
1067 (dwarf2out_begin_prologue): Fix comment.
1068 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1069 if the target needs either debug or unwind DWARF2 info.
1070 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1071 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1072 -ftrapping-math, -ftrapv and -fwrapv.
1074 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1076 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1077 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1078 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1079 (estimate_edge_size_and_time): Likewise.
1080 (ipa_merge_fn_summary_after_inlining): Likewise.
1081 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1082 (compute_uninlined_call_time): Use sreal_frequency.
1083 (compute_inlined_call_time): Likewise.
1084 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1085 * profile-count.c: Include sreal.h.
1086 (profile_count::to_sreal_scale): New.
1087 * profile-count.h: Forward declare sreal.
1088 (profile_count::to_sreal_scale): Declare.
1090 2017-11-13 Nathan Sidwell <nathan@acm.org>
1092 * diagnostic.c (maybe_line_and_column): New.
1093 (diagnostic_get_location_text): Use it.
1094 (diagnostic_report_current_module): Likewise.
1095 (test_diagnostic_get_location_text): Add tests.
1097 2017-11-13 Luis Machado <luis.machado@linaro.org>
1099 * doc/md.texi (Specifying processor pipeline description): Fix
1100 incorrect latency for the div instruction example.
1102 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1104 PR tree-optimization/78821
1105 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1106 that bit_not_p is the same.
1107 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1108 (split_group): Count precisely bit_not_p bits in each statement.
1109 (invert_op): New function.
1110 (imm_store_chain_info::output_merged_store): Use invert_op to
1111 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1112 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1114 2017-11-13 Martin Liska <mliska@suse.cz>
1116 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1117 (struct source_info): Likewise.
1118 (add_branch_counts): Likewise.
1119 (add_line_counts): Likewise.
1120 (function_summary): Likewise.
1121 (output_intermediate_line): Likewise.
1122 (generate_results): Likewise.
1124 2017-11-13 Martin Liska <mliska@suse.cz>
1126 * gcov.c (struct block_info): Remove typedef for block_t.
1127 (struct line_info): Likewise.
1128 (line_info::has_block): Likewise.
1129 (EXIT_BLOCK): Likewise.
1130 (unblock): Likewise.
1131 (circuit): Likewise.
1132 (get_cycles_count): Likewise.
1133 (process_file): Likewise.
1134 (read_graph_file): Likewise.
1135 (solve_flow_graph): Likewise.
1136 (find_exception_blocks): Likewise.
1137 (add_line_counts): Likewise.
1138 (accumulate_line_info): Likewise.
1139 (output_line_details): Likewise.
1141 2017-11-13 Martin Liska <mliska@suse.cz>
1143 * gcov.c (struct arc_info): Remove typedef for arc_t.
1144 (struct line_info): Likewise.
1145 (add_branch_counts): Likewise.
1146 (output_branch_count): Likewise.
1147 (function_info::~function_info): Likewise.
1148 (circuit): Likewise.
1149 (output_intermediate_line): Likewise.
1150 (read_graph_file): Likewise.
1151 (solve_flow_graph): Likewise.
1152 (find_exception_blocks): Likewise.
1153 (add_line_counts): Likewise.
1154 (accumulate_line_info): Likewise.
1155 (output_line_details): Likewise.
1156 (output_function_details): Likewise.
1158 2017-11-13 Martin Liska <mliska@suse.cz>
1160 * gcov.c (struct function_info): Remove typedef for function_t.
1161 (struct source_info): Likewise.
1162 (source_info::get_functions_at_location): Likewise.
1163 (solve_flow_graph): Likewise.
1164 (find_exception_blocks): Likewise.
1165 (add_line_counts): Likewise.
1166 (output_intermediate_file): Likewise.
1167 (process_file): Likewise.
1168 (generate_results): Likewise.
1169 (release_structures): Likewise.
1170 (read_graph_file): Likewise.
1171 (read_count_file): Likewise.
1172 (accumulate_line_counts): Likewise.
1173 (output_lines): Likewise.
1175 2017-11-13 Martin Liska <mliska@suse.cz>
1177 * gcov.c (function_info::function_info): Remove num_counts
1178 and add vector<gcov_type>.
1179 (function_info::~function_info): Use the vector.
1180 (process_file): Likewise.
1181 (read_graph_file): Likewise.
1182 (read_count_file): Likewise.
1183 (solve_flow_graph): Likewise.
1185 2017-11-13 Martin Liska <mliska@suse.cz>
1187 * gcov.c (function_info::is_artificial): New function.
1188 (process_file): Erase all artificial early.
1189 (generate_results): Skip as all artificial are already
1192 2017-11-13 Martin Liska <mliska@suse.cz>
1194 * gcov.c (read_graph_file): Store to global vector of functions.
1195 (read_count_file): Iterate the vector.
1196 (process_file): Likewise.
1197 (generate_results): Likewise.
1198 (release_structures): Likewise.
1200 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1202 PR tree-optimization/82954
1203 * gimple-ssa-store-merging.c
1204 (imm_store_chain_info::coalesce_immediate_stores): If
1205 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1207 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1209 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1210 Upddate call to ENDIAN_LANE_N.
1211 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1212 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1213 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1214 and use aarch64_endian_lane_rtx.
1215 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1217 2017-11-12 Tom de Vries <tom@codesourcery.com>
1219 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1221 2017-11-12 Tom de Vries <tom@codesourcery.com>
1223 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1225 2017-11-12 Tom de Vries <tom@codesourcery.com>
1227 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1229 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1230 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1231 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1233 2017-11-11 Martin Sebor <msebor@redhat.com>
1236 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1239 * prefic.c (translate_name): Replace strncpy with memcpy to
1240 avoid -Wstringop-truncation.
1242 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1244 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1246 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1248 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1250 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1252 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1255 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1257 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1258 enable generating XXBRH if the value is in a vector register.
1259 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1260 value is in a vector register.
1261 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1262 register to register bswap64's instead of doing the GPR sequence
1263 used on previous machines.
1264 (bswapdi2_xxbrd): New insn.
1265 (bswapdi2_reg): Disallow on ISA 3.0.
1266 (register to register bswap64 splitter): Do not split the insn on
1267 ISA 3.0 systems that use XXBRD.
1269 2017-11-10 Martin Sebor <msebor@redhat.com>
1272 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1274 (find_subframework_file): Same.
1276 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1278 * auto-profile.c (afdo_indirect_call): Drop frequency.
1279 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1280 (cgraph_node::create_edge): Drop frequency argument.
1281 (cgraph_node::create_indirect_edge): Drop frequency argument.
1282 (cgraph_edge::make_speculative): Drop frequency arguments.
1283 (cgraph_edge::resolve_speculation): Do not update frequencies
1284 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1285 (cgraph_node::dump): Check consistency in IPA mode.
1286 (cgraph_edge::maybe_hot_p): Use IPA counter.
1287 (cgraph_edge::verify_count_and_frequency): Rename to ...
1288 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1289 (cgraph_node::verify_node): Update.
1290 * cgraph.h (struct cgraph_edge): Drop frequency.
1291 (cgraph_edge::frequency): New function.
1292 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1294 (cgraph_edge::rebuild_edges): Likewise.
1295 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1296 (duplicate_thunk_for_node): Do not pass frequency.
1297 (cgraph_node::create_clone): Scale only counts.
1298 (cgraph_node::create_virtual_clone): Do not pass frequency.
1299 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1300 (cgraph_node::create_version_clone): Do not pass frequency.
1301 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1302 (cgraph_node::expand_thunk): Do not pass frequency.
1303 (cgraph_node::create_wrapper): Do not pass frequency.
1304 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1306 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1307 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1308 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1309 (gather_caller_stats): Use frequency function.
1310 (ipcp_cloning_candidate_p): Use frequency function.
1311 (ipcp_propagate_stage): Use frequency function.
1312 (get_info_about_necessary_edges): Use frequency function.
1313 (update_profiling_info): Update only IPA profile.
1314 (update_specialized_profile): Use frequency functoin.
1315 (perhaps_add_new_callers): Update only IPA profile.
1316 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1317 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1318 (dump_ipa_call_summary): Use frequency function.
1319 (estimate_edge_size_and_time): Use frequency function.
1320 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1321 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1322 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1323 (update_noncloned_counts): ... ths one; scale counts only.
1324 (clone_inlined_nodes): Do not scale frequency.
1325 (inline_call): Do not pass frequency.
1326 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1327 (compute_inlined_call_time): Use IPA profile.
1328 (want_inline_small_function_p): Use IPA profile.
1329 (want_inline_self_recursive_call_p): Use IPA profile.
1330 (edge_badness): Use IPA profile.
1331 (lookup_recursive_calls): Use IPA profile.
1332 (recursive_inlining): Do not pass frequency.
1333 (resolve_noninline_speculation): Do not update frequency.
1334 (inline_small_functions): Collect max of IPA profile.
1335 (dump_overall_stats): Dump IPA porfile.
1336 (dump_inline_stats): Dump IPA porfile.
1337 (ipa_inline): Collect IPA stats.
1338 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1339 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1340 (ipa_propagate_frequency): Use frequency function.
1341 (ipa_profile): Cleanup.
1342 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1343 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1344 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1345 (input_node): Do not stream frequency.
1346 (input_edge): Do not stream frequency.
1347 (merge_profile_summaries): Scale only IPA profiles.
1348 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1349 * predict.c (drop_profile): Do not recompute frequency.
1350 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1351 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1352 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1353 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1354 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1355 * tree-inline.c (copy_bb): Do not scale frequency.
1356 (expand_call_inline): Do not scale frequency.
1357 (tree_function_versioning): Do not scale frequency.
1358 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1360 2017-11-10 Julia Koval <julia.koval@intel.com>
1362 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1363 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1364 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1365 (_mm256_maskz_gf2p8affine_epi64_epi8)
1366 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1367 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1368 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1369 (__builtin_ia32_vgf2p8affineqb_v32qi)
1370 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1371 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1373 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1377 (arm_option_override): Refactor.
1378 (arm_option_reconfigure_globals): New.
1379 (arm_options_perform_arch_sanity_checks): New.
1380 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1382 (arm_options_perform_arch_sanity_checks): Likewise
1384 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1386 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1387 (power9-qpmul): New.
1388 * rs6000/rs6000.md ("type" attr): Add qmul.
1389 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1390 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1391 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1393 2017-11-10 Martin Sebor <msebor@redhat.com>
1396 * builtins.c (compute_objsize): Handle arrays that
1397 compute_builtin_object_size likes to fail for. Make extern.
1398 * builtins.h (compute_objsize): Declare.
1399 (check_strncpy_sizes): New function.
1400 (expand_builtin_strncpy): Call check_strncpy_sizes.
1401 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1402 -Wstringop-truncation.
1403 (gimple_fold_builtin_strncat): Same.
1404 * gimple.c (gimple_build_call_from_tree): Set call location.
1405 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1406 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1407 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1408 (handle_builtin_strlen): Use strlen_to_stridx.
1409 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1411 Use strlen_to_stridx.
1412 (pass_strlen::execute): Release strlen_to_stridx.
1413 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1414 (-Wstringop-truncation): Document new option.
1416 2017-11-10 Martin Liska <mliska@suse.cz>
1418 PR gcov-profile/82702
1419 * gcov.c (main): Handle intermediate files in a different
1421 (get_gcov_intermediate_filename): New function.
1422 (output_gcov_file): Remove support of intermediate files.
1423 (generate_results): Allocate intermediate file.
1424 (release_structures): Clean-up properly fn_end.
1425 (output_intermediate_file): Start iterating with line 1.
1427 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1429 PR tree-optimization/82929
1430 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1431 ops_swapped_p non-static data member.
1432 (store_immediate_info::store_immediate_info): Clear it.
1433 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1434 ops set ops_swapped_p.
1435 (count_multiple_uses): Handle ops_swapped_p.
1437 2017-11-10 Martin Liska <mliska@suse.cz>
1439 * coverage.c (coverage_init): Stream information about
1440 support of has_unexecuted_blocks.
1441 * doc/gcov.texi: Document that.
1442 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1443 * gcov.c (read_graph_file): Likewise.
1444 (output_line_beginning): Fix a small issue with
1447 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1449 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1450 reference of trivial component.
1452 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1455 * gimple-ssa-store-merging.c
1456 (pass_store_merging::terminate_all_aliasing_chains): For
1457 gimple_store_p stmts also call refs_output_dependent_p.
1459 PR rtl-optimization/82913
1460 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1463 2017-11-09 Jeff Law <law@redhat.com>
1465 * vr-values.h: New file with vr_values class.
1466 * tree-vrp.c: Include vr-values.h
1467 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1468 data objects into the vr_values class.
1469 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1470 (get_value_range): Make it a member function within vr_values class.
1471 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1472 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1473 (op_with_constant_singleton_value_range): Likewise.
1474 (extract_range_for_var_from_comparison_expr): Likewise.
1475 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1476 (extract_range_from_binary_expr): Likewise.
1477 (extract_range_from_unary_expr): Likewise.
1478 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1479 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1480 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1481 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1482 (compare_name_with_value, compare_names): Likewise.
1483 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1484 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1485 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1486 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1487 (extract_range_from_phi_node): Likewise.
1488 (simplify_truth_ops_using_ranges): Likewise.
1489 (simplify_div_or_mod_using_ranges): Likewise.
1490 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1491 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1492 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1493 (simplify_float_conversion_using_ranges): Likewise.
1494 (simplify_internal_call_using_ranges): Likewise.
1495 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1496 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1497 poitner via x_vr_values for calls into gimple folder.
1498 (vrp_initialize_lattice): Make this the vr_values ctor.
1499 (vrp_free_lattice): Make this the vr_values dtor.
1500 (set_vr_value): New function.
1501 (class vrp_prop): Add vr_values data member. Add various member
1502 functions as well as member functions that delegate to vr_values.
1503 (check_array_ref): Make a member function within vrp_prop class.
1504 (search_for_addr_array, vrp_initialize): Likewise.
1505 (vrp_finalize): Likewise. Revamp to avoid direct access to
1506 vr_value, values_propagated, etc.
1507 (check_array_bounds): Extract vrp_prop class instance pointer from
1508 walk info structure. Use it to call member functions.
1509 (check_all_array_refs): Make a member function within vrp_prop class.
1510 Smuggle class instance pointer via walk info structure.
1511 (x_vr_values): New local static.
1512 (vrp_valueize): Use x_vr_values to get class instance.
1513 (vr_valueize_1): Likewise.
1514 (class vrp_folder): Add vr_values data member. Add various member
1515 functions as well as member functions that delegate to vr_values.
1516 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1517 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1518 class instance from vr_values. Use it to call member functions.
1519 (vrp_dom_walker): Add vr_values data member.
1520 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1521 instance via x_vr_values.
1522 (identify_jump_threads): Accept vr_values as argument. Store
1523 it into the walker structure.
1524 (evrp_dom_walker): Add vr_values class data member. Add various
1526 (evrp_dom_walker::try_find_new_range): Use vr_values data
1527 member to access the memory allocator.
1528 (evrp_dom_walker::before_dom_children): Store vr_values class
1529 instance into the vrp_folder class.
1530 (evrp_dom_walker::push_value_range): Rework to avoid direct
1531 access to num_vr_values and vr_value.
1532 (evrp_dom_walker::pop_value_range): Likewise.
1533 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1534 Use vr_values to get to dump_all_value_ranges member function.
1535 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1536 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1537 Pass vr_values class instance down to identify_jump_threads.
1538 Remove call to vrp_free_lattice.
1539 (debug_all_value_ranges): Remove.
1541 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1542 (vrp_folder): Likewise.
1544 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1545 Get it from the existing bitmap instead.
1546 (vrp_intersect_ranges_1): Likewise.
1548 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1550 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1552 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1553 set bit_not_p to it.
1554 (imm_store_chain_info::coalesce_immediate_stores): Break group
1555 if bit_not_p is different.
1556 (count_multiple_uses, split_group,
1557 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1558 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1559 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1560 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1561 to store_immediate_info ctor.
1563 2017-11-09 Jim Wilson <jimw@sifive.com>
1565 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1566 (scan_prog_file): Likewise.
1568 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1570 * bb-reorder.c (max_entry_frequency): Remove.
1571 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1572 connect_traces, push_to_next_round_p): Remove prototypes.
1573 (find_traces_1_round): Use counts only.
1574 (push_to_next_round_p): Likewise.
1575 (find_traces): Likewise.
1576 (rotate_loop): Likewise.
1577 (find_traces_1_round): Likewise.
1578 (connect_traces): Likewise.
1579 (edge_order): Likewise.
1581 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1583 * config/arm/arm.c (output_return_instruction): Add comments to
1584 indicate requirement for cmse_nonsecure_entry return to account
1585 for the size of clearing instruction output here.
1586 (thumb_exit): Likewise.
1587 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1588 return in hardfloat mode.
1590 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1592 * config/rs6000/rs6000.c (machine_function): Add a bool,
1593 "toc_is_wrapped_separately".
1594 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1595 if it wasn't explicitly set or unset, we are optimizing for speed, and
1596 doing separate shrink-wrapping.
1597 (rs6000_get_separate_components): Enable the TOC component if
1598 saving the TOC register in the prologue.
1599 (rs6000_components_for_bb): Handle the TOC component.
1600 (rs6000_emit_prologue_components): Store the TOC register where needed.
1601 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1602 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1604 2017-11-09 Martin Jambor <mjambor@suse.cz>
1606 * ipa-param-manipulation.c: New file.
1607 * ipa-param-manipulation.h: Likewise.
1608 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1609 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1610 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1611 (ipa_parm_adjustment): Likewise.
1612 (ipa_parm_adjustment_vec): Likewise.
1613 (ipa_get_vector_of_formal_parms): Moved declaration to
1614 ipa-param-manipulation.h.
1615 (ipa_get_vector_of_formal_parm_types): Likewise.
1616 (ipa_modify_formal_parameters): Likewise.
1617 (ipa_modify_call_arguments): Likewise.
1618 (ipa_combine_adjustments): Likewise.
1619 (ipa_dump_param_adjustments): Likewise.
1620 (ipa_modify_expr): Likewise.
1621 (ipa_get_adjustment_candidate): Likewise.
1622 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1623 ipa-param-manipulation.c.
1624 (ipa_get_vector_of_formal_parm_types): Likewise.
1625 (ipa_modify_formal_parameters): Likewise.
1626 (ipa_modify_call_arguments): Likewise.
1627 (ipa_modify_expr): Likewise.
1628 (get_ssa_base_param): Likewise.
1629 (ipa_get_adjustment_candidate): Likewise.
1630 (index_in_adjustments_multiple_times_p): Likewise.
1631 (ipa_combine_adjustments): Likewise.
1632 (ipa_dump_param_adjustments): Likewise.
1633 * tree-sra.c: Also include ipa-param-manipulation.h
1634 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1637 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1638 Alan Hayward <alan.hayward@arm.com>
1639 David Sherwood <david.sherwood@arm.com>
1641 * doc/sourcebuild.texi (vect_masked_store): Document.
1643 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1644 Alan Hayward <alan.hayward@arm.com>
1645 David Sherwood <david.sherwood@arm.com>
1647 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1649 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1650 Alan Hayward <alan.hayward@arm.com>
1651 David Sherwood <david.sherwood@arm.com>
1653 * doc/sourcebuild.texi (vect_variable_length): Document.
1655 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1656 Alan Hayward <alan.hayward@arm.com>
1657 David Sherwood <david.sherwood@arm.com>
1659 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1661 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1662 Alan Hayward <alan.hayward@arm.com>
1663 David Sherwood <david.sherwood@arm.com>
1665 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1667 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1668 Alan Hayward <alan.hayward@arm.com>
1669 David Sherwood <david.sherwood@arm.com>
1671 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1672 previously undocumented selectors.
1673 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1675 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1677 * doc/rtl.texi (const_vector): Say that elements can be
1678 const_wide_ints too.
1679 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1680 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1681 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1682 * optabs.c (expand_vector_broadcast): Likewise.
1684 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1685 Alan Hayward <alan.hayward@arm.com>
1686 David Sherwood <david.sherwood@arm.com>
1688 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1689 scaled index even if the unscaled address was invalid.
1690 Don't increase the complexity of using a scale in that case.
1692 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1693 Alan Hayward <alan.hayward@arm.com>
1694 David Sherwood <david.sherwood@arm.com>
1696 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1697 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1698 UNITS_PER_WORD bytes.
1699 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1700 based on REGMODE_NATURAL_SIZE of the inner mode.
1701 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1702 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1703 * expmed.c (lowpart_bit_field_p): Divide the value up into
1704 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1705 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1706 whether something is likely to occupy more than one register.
1708 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1711 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1712 profile_count::adjust_for_ipa_scaling.
1713 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1714 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1716 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1718 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1720 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1721 (split_group): Add total_orig and total_new arguments, estimate the
1722 number of statements related to the store group without store merging
1723 and with store merging.
1724 (imm_store_chain_info::output_merged_store): Adjust split_group
1725 callers, punt if estimated number of statements with store merging
1726 is not smaller than estimated number of statements without it.
1728 (handled_load): Remove has_single_use checks.
1729 (pass_store_merging::process_store): Likewise.
1731 2017-11-09 Richard Biener <rguenther@suse.de>
1733 PR tree-optimization/82902
1734 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1736 2017-11-09 Martin Liska <mliska@suse.cz>
1739 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1742 2017-11-09 Martin Liska <mliska@suse.cz>
1744 PR tree-optimization/82669
1745 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1747 2017-11-09 Martin Liska <mliska@suse.cz>
1749 PR gcov-profile/48463
1750 * coverage.c (coverage_begin_function): Output also end locus
1751 of a function and information whether the function is
1753 * gcov-dump.c (tag_function): Parse and print the information.
1754 * gcov.c (INCLUDE_MAP): Add include.
1755 (INCLUDE_SET): Likewise.
1756 (struct line_info): Move earlier in the source file because
1757 of vector<line_info> in function_info structure.
1758 (line_info::line_info): Likewise.
1759 (line_info::has_block): Likewise.
1760 (struct source_info): Add new member index.
1761 (source_info::get_functions_at_location): New function.
1762 (function_info::group_line_p): New function.
1763 (output_intermediate_line): New function.
1764 (output_intermediate_file): Use the mentioned function.
1765 (struct function_start): New.
1766 (struct function_start_pair_hash): Likewise.
1767 (process_file): Add code that identifies group functions.
1768 Assign lines either to global or function scope.
1769 (generate_results): Skip artificial functions.
1770 (find_source): Assign index for each source file.
1771 (read_graph_file): Read new flag artificial and end_line.
1772 (add_line_counts): Assign it either to global of function scope.
1773 (accumulate_line_counts): Isolate core of the function to
1774 accumulate_line_info and call it for both function and global
1776 (accumulate_line_info): New function.
1777 (output_line_beginning): Fix GNU coding style.
1778 (print_source_line): New function.
1779 (output_line_details): Likewise.
1780 (output_function_details): Likewise.
1781 (output_lines): Iterate both source (global) scope and function
1783 (struct function_line_start_cmp): New class.
1784 * doc/gcov.texi: Reflect changes in documentation.
1786 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1789 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1790 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1791 and similarly for not instead of neg.
1793 2017-11-08 Andi Kleen <ak@linux.intel.com>
1795 * config/i386/i386.opt: Add -mforce-indirect-call.
1796 * config/i386/predicates.md: Check for flag_force_indirect_call.
1797 * doc/invoke.texi: Document -mforce-indirect-call
1799 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1801 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1803 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1804 config/riscv/riscv.c (predict.h): New include.
1805 (riscv_slow_unaligned_access_p): No longer static.
1806 (riscv_block_move_straight): Add require.
1807 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1809 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1812 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1813 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1814 nonimmediate_operand predicate for operand 1 instead of
1817 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1819 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1821 * config/aarch64/constraints.md (Uml): New constraint.
1822 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1825 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1827 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1828 of two vec_duplicates into a vec_concat.
1830 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1832 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1833 Simplify vec_merge of vec_duplicate and vec_concat.
1834 * config/aarch64/constraints.md (Utq): New constraint.
1835 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1838 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1840 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1841 Simplify vec_merge of vec_duplicate and const_vector.
1842 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1844 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1845 mode iterator. Update predicate on operand 1 to
1846 handle non-const_vec constants. Delete constraints.
1847 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1849 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1851 PR tree-optimization/78821
1852 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1854 (store_operand_info::store_operand_info): Initialize it to false.
1855 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1856 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1857 store in the group, and if chain_info is non-NULL, to ignore altogether
1859 (compatible_load_p): Fail if bit_not_p does not match.
1860 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1861 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1862 (pass_store_merging::process_store): Adjust
1863 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1864 call terminate_all_aliasing_chains newly when adding a store into
1865 a chain with non-NULL chain_info.
1867 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1869 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1871 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1873 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1875 (aarch64_layout_frame): Initialise emit_frame_chain.
1876 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1877 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1879 2017-11-08 Martin Liska <mliska@suse.cz>
1881 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1882 of gimple_call_internal_p.
1884 2017-11-07 Tom de Vries <tom@codesourcery.com>
1886 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1888 2017-11-07 Tom de Vries <tom@codesourcery.com>
1890 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1893 2017-11-08 Martin Liska <mliska@suse.cz>
1896 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1898 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1900 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1902 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1905 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1906 (*cmp<mode>_ccz_1): New insn with $k alternative.
1909 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1910 RTX_COMM_COMPARE as commutative as well.
1911 (ix86_binary_operator_ok): Formatting fix.
1912 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1913 *<code><mode>3<mask_name><round_saeonly_name>,
1914 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1915 *<s>mul<mode>3_highpart<mask_name>,
1916 *vec_widen_umult_even_v16si<mask_name>,
1917 *vec_widen_umult_even_v8si<mask_name>,
1918 *vec_widen_umult_even_v4si<mask_name>,
1919 *vec_widen_smult_even_v16si<mask_name>,
1920 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1921 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1922 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1923 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1924 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1925 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1926 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1927 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1928 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1929 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1930 ix86_binary_operator_ok. Formatting fixes.
1931 (*<plusminus_insn><mode>3<mask_name><round_name>,
1932 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1935 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1937 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1938 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1939 eq and ne if TARGET_ISEL.
1940 (cmp): New code_iterator.
1941 (UNS, UNSU_, UNSIK): New code_attrs.
1942 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1943 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1945 ("eq<mode>3"): ... this.
1946 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1948 ("ne<mode>3"): ... this.
1950 2017-11-07 Julia Koval <julia.koval@intel.com>
1953 * common/config/i386/i386-common.c
1954 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1955 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1956 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1957 * config/i386/i386.opt: Ditto.
1958 * config/i386/i386.c (ix86_target_string): Ditto.
1959 (ix86_option_override_internal): Ditto.
1960 (ix86_init_mpx_builtins): Move MPX to args2.
1961 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1962 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1963 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1964 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1965 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1966 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1967 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1969 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1972 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1973 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1974 (zero-extendsidi peephole2): Remove peephole.
1976 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1979 * stor-layout.c: Include attribs.h.
1980 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1981 explicit lookup of "aligned" attribute.
1983 2017-11-07 Andrew Waterman <andrew@sifive.com>
1985 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1986 (riscv_expand_block_move): Likewise.
1987 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1989 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1990 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1991 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1992 (riscv_adjust_block_mem): Likewise.
1993 (riscv_block_move_loop): Likewise.
1994 (riscv_expand_block_move): Likewise.
1995 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1997 2017-11-07 Michael Clark <michaeljclark@mac.com>
1999 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2000 (MUSL_DYNAMIC_LINKER): Likewise.
2002 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2004 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2005 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2008 2017-11-07 Richard Biener <rguenther@suse.de>
2010 * match.pd: Fix build.
2012 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2013 Jackson Woodruff <jackson.woodruff@arm.com>
2015 PR tree-optimization/71026
2016 * match.pd: Canonicalize negate in division.
2018 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2021 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2023 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2025 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2026 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2028 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2030 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2031 non-scalar integral types.
2032 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2033 (-(A-B), -(~A)): New transformations.
2035 2017-11-07 Tom de Vries <tom@codesourcery.com>
2037 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2038 semicolon after "do {} while (0)".
2039 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2040 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2041 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2042 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2043 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2045 2017-11-07 Tom de Vries <tom@codesourcery.com>
2047 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2048 after "do {} while (0)".
2049 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2050 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2051 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2052 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2053 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2055 2017-11-07 Tom de Vries <tom@codesourcery.com>
2058 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2060 (arm_rtx_costs_internal): Add missing semicolon after
2061 HANDLE_NARROW_SHIFT_ARITH call.
2063 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2066 disable isel if it was not set explicitly.
2068 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2070 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2072 (add_type_attribute) likewise.
2074 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2076 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2077 of struct ix86_frame.
2078 (ix86_initial_elimination_offset): Likewise.
2079 (ix86_expand_split_stack_prologue): Likewise.
2081 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2083 * tree-vrp.h (enum value_range_type): Update stale comment.
2085 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2090 (aarch64_expand_vec_perm_const): Take the number of units too.
2091 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2092 (aarch64_expand_vec_perm_const): Likewise.
2093 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2094 (vec_perm<mode>): Update accordingly.
2096 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2097 Alan Hayward <alan.hayward@arm.com>
2098 David Sherwood <david.sherwood@arm.com>
2100 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2101 Take the number of units too.
2102 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2103 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2104 but check for a vector mode before rather than after the call.
2105 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2106 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2107 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2108 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2109 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2110 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2111 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2112 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2113 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2114 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2115 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2116 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2117 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2118 (aarch64_sqdmull2_n<mode>): Update accordingly.
2120 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2121 Alan Hayward <alan.hayward@arm.com>
2122 David Sherwood <david.sherwood@arm.com>
2124 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2125 the number of units too.
2126 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2127 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2128 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2129 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2130 (vec_store_lanesxi<mode>): Update accordingly.
2132 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2133 Alan Hayward <alan.hayward@arm.com>
2134 David Sherwood <david.sherwood@arm.com>
2136 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2137 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2138 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2139 of units rather than the mode.
2140 * config/aarch64/iterators.md (nunits): New mode attribute.
2141 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2142 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2143 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2144 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2145 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2146 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2147 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2148 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2149 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2150 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2151 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2152 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2153 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2154 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2155 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2156 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2157 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2158 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2159 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2160 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2161 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2162 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2163 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2164 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2165 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2166 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2167 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2168 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2169 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2170 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2171 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2172 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2173 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2174 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2175 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2176 (aarch64_simd_vec_setv2di): Likewise.
2178 2017-11-06 Carl Love <cel@us.ibm.com>
2180 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2182 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2183 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2185 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2187 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2188 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2189 Add power 8 macro expansions.
2190 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2191 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2192 power 8 instructions. (VSX_XXBR): Add iterator.
2194 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2196 * config/arm/arm.md (predicable_short_it): Change default to "no",
2197 improve documentation, remove uses that are identical to the default.
2198 (enabled_for_depr_it): Rename to enabled_for_short_it.
2199 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2200 Remove default uses.
2201 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2202 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2203 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2204 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2206 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2209 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2210 float128 helper macros, which are no longer used after deleting
2211 the old 'q' built-in functions, and moving the round to odd
2212 built-in functions to being special built-in functions.
2213 (BU_FLOAT128_2): Likewise.
2214 (BU_FLOAT128_1_HW): Likewise.
2215 (BU_FLOAT128_2_HW): Likewise.
2216 (BU_FLOAT128_3_HW): Likewise.
2217 (FABSQ): Delete old 'q' built-in functions.
2218 (COPYSIGNQ): Likewise.
2219 (SQRTF128_ODD): Move round to odd built-in functions to be
2220 special built-in functions, so that we can handle
2221 -mabi=ieeelongdouble.
2222 (TRUNCF128_ODD): Likewise.
2223 (ADDF128_ODD): Likewise.
2224 (SUBF128_ODD): Likewise.
2225 (MULF128_ODD): Likewise.
2226 (DIVF128_ODD): Likewise.
2227 (FMAF128_ODD): Likewise.
2228 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2229 built-in names to 'f128'.
2230 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2231 old 'q' built-in functions, as the machine independent code for
2232 'f128' built-in functions handles this.
2233 (rs6000_expand_builtin): Add expansion for float128 round to odd
2234 functions, keying off on -mabi=ieeelongdouble of whether to use
2235 the KFmode or TFmode variant.
2236 (rs6000_init_builtins): Initialize the _Float128 round to odd
2238 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2239 _Float128 'q' built-in functions are now mapped into the new
2240 'f128' built-in functions.
2242 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2244 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2245 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2247 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2250 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2251 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2252 * toplev.c: Include "ipa-fnsummary.h".
2253 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2255 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2257 PR tree-optimization/82838
2258 * gimple-ssa-store-merging.c
2259 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2260 on a separate gimple_seq which is then appended to seq.
2262 2017-11-06 Jeff Law <law@redhat.com>
2265 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2266 (get_probe_interval): New functions.
2267 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2268 (ix86_adjust_stack_and_probe): Likewise.
2269 (output_adjust_stack_and_probe): Likewise.
2270 (ix86_emit_probe_stack_range): Likewise.
2271 (ix86_expand_prologue): Likewise.
2273 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2275 PR tree-optimization/82816
2276 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2277 if the modes of the two types are the same.
2278 (convert_plusminus_to_widen): Likewise.
2280 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2282 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2284 (usadv16qi): New define_expand.
2285 (usadv8hi): New define_expand.
2287 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2290 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2292 (inline_transform): Likewise.
2293 * predict.c (counts_to_freqs): Remove useless conditional.
2294 * profile-count.h (profile_count::apply_scale): Move sanity check.
2295 * tree-inline.c (copy_bb): Always scale.
2296 (copy_cfg_body): Likewise.
2298 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2301 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2303 (*cmp_ite0): Add enabled_for_depr_it attribute.
2304 (*cmp_ite1): Likewise.
2306 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2308 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2311 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2313 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2314 return the value by pointer.
2315 (extract_range_from_multiplicative_op_1): Update accordingly.
2316 Return as soon as an operation fails.
2318 2017-11-05 Tom de Vries <tom@codesourcery.com>
2321 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2322 (DEF_SANITIZER_BUILTIN): ... here.
2323 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2324 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2326 2017-11-05 Tom de Vries <tom@codesourcery.com>
2329 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2331 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2332 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2333 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2335 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2336 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2337 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2339 2017-11-05 Tom de Vries <tom@codesourcery.com>
2342 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2345 2017-11-04 Michael Clark <michaeljclark@mac.com>
2347 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2348 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2350 (*addsi3_extended): Likewise.
2351 (*addsi3_extended2): Likewise.
2352 (<optab>si3): Likewise.
2353 (<optab>di3): Likewise.
2354 (<optab><mode>3): Likewise.
2355 (<*optabe>si3_internal): Likewise.
2356 (zero_extendqi<SUPERQI:mode>2): Likewise.
2357 (*add<mode>hi3): Likewise.
2358 (*xor<mode>hi3): Likewise.
2359 (<optab>di3): Likewise.
2360 (*<optab>si3_extend): Likewise.
2361 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2362 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2363 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2365 2017-11-04 Andrew Waterman <andrew@sifive.com>
2367 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2368 TARGET_STRICT_ALIGN based upon -mtune argument.
2370 2017-11-04 Andrew Waterman <andrew@sifive.com>
2372 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2374 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2376 config/i386/i386.c (choose_basereg): Use optional scratch
2377 register and add assertion.
2378 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2379 needed, and don't allocate stack.
2380 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2381 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2382 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2384 2017-11-03 Jeff Law <law@redhat.com>
2386 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2387 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2388 to probe at the start of a noreturn function.
2390 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2392 PR tree-optimization/78821
2393 * gimple-ssa-store-merging.c: Update the file comment.
2394 (MAX_STORE_ALIAS_CHECKS): Define.
2395 (struct store_operand_info): New type.
2396 (store_operand_info::store_operand_info): New constructor.
2397 (struct store_immediate_info): Add rhs_code and ops data members.
2398 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2399 and op1r arguments to the ctor, initialize corresponding data members.
2400 (struct merged_store_group): Add load_align_base and load_align
2402 (merged_store_group::merged_store_group): Initialize them.
2403 (merged_store_group::do_merge): Update them.
2404 (merged_store_group::apply_stores): Pick the constant for
2405 encode_tree_to_bitpos from one of the two operands, or skip
2406 encode_tree_to_bitpos if neither operand is a constant.
2407 (class pass_store_merging): Add process_store method decl. Remove
2408 bool argument from terminate_all_aliasing_chains method decl.
2409 (pass_store_merging::terminate_all_aliasing_chains): Remove
2410 var_offset_p argument and corresponding handling.
2411 (stmts_may_clobber_ref_p): New function.
2412 (compatible_load_p): New function.
2413 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2414 if there is overlap and rhs_code is not INTEGER_CST. For
2415 non-overlapping stores terminate group if rhs is not mergeable.
2416 (get_alias_type_for_stmts): Change first argument from
2417 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2418 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2419 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2421 (get_location_for_stmts): Change first argument from
2422 auto_vec<gimple *> & to vec<gimple *> &.
2423 (struct split_store): Remove orig_stmts data member, add orig_stores.
2424 (split_store::split_store): Create orig_stores rather than orig_stmts.
2425 (find_constituent_stmts): Renamed to ...
2426 (find_constituent_stores): ... this. Change second argument from
2427 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2428 to info structures rather than the statements.
2429 (split_group): Rename ALLOW_UNALIGNED argument to
2430 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2431 it. Adjust find_constituent_stores caller.
2432 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2433 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2434 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2435 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2436 (mem_valid_for_store_merging): New function.
2437 (handled_load): New function.
2438 (pass_store_merging::process_store): New method.
2439 (pass_store_merging::execute): Use process_store method. Adjust
2440 terminate_all_aliasing_chains caller.
2442 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2444 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2445 Return true for more constants, symbols and label references.
2446 (aarch64_valid_floating_const): Remove unused function.
2448 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2451 * config/aarch64/aarch64.c (aarch64_layout_frame):
2452 Undo forcing of LR at bottom of frame.
2454 2017-11-03 Jeff Law <law@redhat.com>
2457 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2458 for int_registers_saved.
2460 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2461 extracted from tree-ssa-dom.c.
2462 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2463 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2464 (record_equivalences_from_incoming_edge): Add additional argument
2465 to single_pred_edge_ignoring_loop_edges call.
2466 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2467 (uncprop_dom_walker::before_dom_children): Add additional argument
2468 to single_pred_edge_ignoring_loop_edges call.
2469 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2470 single_pred_edge_ignoring_loop_edges rather than open coding.
2471 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2473 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2475 * match.pd (-(-A)): Rewrite.
2477 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2479 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2480 (rs6000_emit_int_cmove): New declaration.
2481 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2482 (rs6000_emit_sISEL): Delete.
2483 (rs6000_emit_int_cmove): Make non-static.
2484 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2485 instead of rs6000_emit_sISEL.
2487 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2489 * asan.c (create_cond_insert_point): Maintain profile.
2490 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2492 * basic-block.h (struct basic_block_def): Remove frequency.
2493 (EDGE_FREQUENCY): Use to_frequency
2494 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2496 (find_traces): Update to use to_frequency.
2497 (find_traces_1_round): Likewise; use only IPA counts.
2498 (bb_to_key): Likewise.
2499 (connect_traces): Use IPA counts only.
2500 (copy_bb_p): Update to use to_frequency.
2501 (fix_up_crossing_landing_pad): Likewise.
2502 (sanitize_hot_paths): Likewise.
2503 * bt-load.c (basic_block_freq): Likewise.
2504 * cfg.c (init_flow): Set count_max to uninitialized.
2505 (check_bb_profile): Remove frequencies; check counts.
2506 (dump_bb_info): Do not dump frequencies.
2507 (update_bb_profile_for_threading): Update counts only.
2508 (scale_bbs_frequencies_int): Likewise.
2509 (MAX_SAFE_MULTIPLIER): Remove.
2510 (scale_bbs_frequencies_gcov_type): Update counts only.
2511 (scale_bbs_frequencies_profile_count): Update counts only.
2512 (scale_bbs_frequencies): Update counts only.
2513 * cfg.h (struct control_flow_graph): Add count-max.
2514 (update_bb_profile_for_threading): Update prototype.
2515 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2516 (find_many_sub_basic_blocks): Likewise.
2517 * cfgcleanup.c (try_forward_edges): Likewise.
2518 (try_crossjump_to_edge): Likewise.
2519 * cfgexpand.c (expand_gimple_cond): Likewise.
2520 (expand_gimple_tailcall): Likewise.
2521 (construct_init_block): Likewise.
2522 (construct_exit_block): Likewise.
2523 * cfghooks.c (verify_flow_info): Check consistency of counts.
2524 (dump_bb_for_graph): Do not dump frequencies.
2525 (split_block_1): Do not update frequencies.
2526 (split_edge): Do not update frequencies.
2527 (make_forwarder_block): Do not update frequencies.
2528 (duplicate_block): Do not update frequencies.
2529 (account_profile_record): Do not update frequencies.
2530 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2531 for global heuristics.
2532 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2533 (expected_loop_iterations_unbounded): Use counts only.
2534 * cfgloopmanip.c (scale_loop_profile): Simplify.
2535 (create_empty_loop_on_edge): Simplify
2537 (duplicate_loop_to_header_edge): Simplify
2538 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2539 (update_br_prob_note): Take care of removing note when profile
2541 (relink_block_chain): Do not dump frequency.
2542 (rtl_account_profile_record): Use to_frequency.
2543 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2544 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2545 (cgraph_update_edges_for_call_stmt_node): Likewise.
2546 (cgraph_edge::verify_count_and_frequency): Update.
2547 (cgraph_node::verify_node): Temporarily disable frequency verification.
2548 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2549 to_cgraph_frequency.
2550 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2551 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2553 (cgraph_node::expand_thunk): Update profile.
2554 * except.c (dw2_build_landing_pads): Do not update frequency.
2555 * final.c (compute_alignments): Use to_frequency.
2556 (dump_basic_block_info): Do not dump frequency.
2557 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2558 (dump_gimple_bb_header): Do not dump frequency.
2559 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2561 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2562 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2563 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2564 (init_before_recovery): Do not update frequency.
2565 (sched_create_recovery_edges): Do not update frequency.
2566 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2567 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2568 (ipa_cp_c_finalize): Set max_count to uninitialized.
2569 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2570 (param_change_prob): Use counts.
2571 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2573 * ipa-split.c (consider_split): Use to_frequency.
2574 (split_function): Use to_frequency.
2575 * ira-build.c (loop_compare_func): Likewise.
2576 (mark_loops_for_removal): Likewise.
2577 (mark_all_loops_for_removal): Likewise.
2578 * loop-doloop.c (doloop_modify): Do not update frequency.
2579 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2581 * lto-streamer-in.c (input_function): Update count_max.
2582 * omp-expand.c (expand_omp_taskreg): Update count_max.
2583 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2584 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2585 (maybe_hot_count_p): Use ipa counts only.
2586 (maybe_hot_bb_p): Simplify.
2587 (maybe_hot_edge_p): Simplify.
2588 (probably_never_executed): Do not take frequency argument.
2589 (probably_never_executed_bb_p): Do not pass frequency.
2590 (probably_never_executed_edge_p): Likewise.
2591 (combine_predictions_for_bb): Check that profile is nonzero.
2592 (propagate_freq): Do not set frequency.
2593 (drop_profile): Simplify.
2594 (counts_to_freqs): Simplify.
2595 (expensive_function_p): Use to_frequency.
2596 (propagate_unlikely_bbs_forward): Simplify.
2597 (determine_unlikely_bbs): Simplify.
2598 (estimate_bb_frequencies): Add hack to silence graphite issues.
2599 (compute_function_frequency): Use ipa counts.
2600 (pass_profile::execute): Update.
2601 (rebuild_frequencies): Use counts only.
2602 (force_edge_cold): Use counts only.
2603 * profile-count.c (profile_count::dump): Dump new count types.
2604 (profile_count::differs_from_p): Check compatiblity.
2605 (profile_count::to_frequency): New function.
2606 (profile_count::to_cgraph_frequency): New function.
2607 * profile-count.h (struct function): Declare.
2608 (enum profile_quality): Add profile_guessed_local and
2609 profile_guessed_global0.
2610 (class profile_proability): Decrease number of bits to 29;
2611 update from_reg_br_prob_note and to_reg_br_prob_note.
2612 (class profile_count: Update comment; decrease number of bits
2613 to 61. Check compatibility.
2614 (profile_count::compatible_p): New private member function.
2615 (profile_count::ipa_p): New member function.
2616 (profile_count::operator<): Handle global zero correctly.
2617 (profile_count::operator>): Handle global zero correctly.
2618 (profile_count::operator<=): Handle global zero correctly.
2619 (profile_count::operator>=): Handle global zero correctly.
2620 (profile_count::nonzero_p): New member function.
2621 (profile_count::force_nonzero): New member function.
2622 (profile_count::max): New member function.
2623 (profile_count::apply_scale): Handle IPA scalling.
2624 (profile_count::guessed_local): New member function.
2625 (profile_count::global0): New member function.
2626 (profile_count::ipa): New member function.
2627 (profile_count::to_frequency): Declare.
2628 (profile_count::to_cgraph_frequency): Declare.
2629 * profile.c (OVERLAP_BASE): Delete.
2630 (compute_frequency_overlap): Delete.
2631 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2632 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2633 * sched-ebb.c (rank): Use counts only.
2634 * shrink-wrap.c (handle_simple_exit): Use counts only.
2635 (try_shrink_wrapping): Use counts only.
2636 (place_prologue_for_one_component): Use counts only.
2637 * tracer.c (find_best_predecessor): Use to_frequency.
2638 (find_trace): Use to_frequency.
2639 (tail_duplicate): Use to_frequency.
2640 * trans-mem.c (expand_transaction): Do not update frequency.
2641 * tree-call-cdce.c: Do not update frequency.
2642 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2643 (gimple_merge_blocks): Likewise.
2644 (gimple_split_edge): Likewise.
2645 (gimple_duplicate_sese_region): Likewise.
2646 (gimple_duplicate_sese_tail): Likewise.
2647 (move_sese_region_to_fn): Likewise.
2648 (gimple_account_profile_record): Likewise.
2649 (insert_cond_bb): Likewise.
2650 * tree-complex.c (expand_complex_div_wide): Likewise.
2651 * tree-eh.c (lower_resx): Update profile.
2652 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2654 (initialize_cfun): Do not initialize frequencies
2655 (freqs_to_counts): Delete.
2656 (copy_cfg_body): Ignore count parameter.
2657 (copy_body): Update.
2658 (expand_call_inline): Update count_max.
2659 (optimize_inline_calls): Update count_max.
2660 (tree_function_versioning): Update count_max.
2661 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2662 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2664 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2665 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2666 (try_peel_loop): Likewise.
2667 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2669 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2670 (tree_transform_and_unroll_loop): Do not use frequencies
2671 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2672 Use reliable prediction only.
2673 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2674 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2675 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2676 probability scaling.
2677 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2678 not update frequency
2679 (any_remaining_duplicated_blocks): Likewise.
2680 (update_profile): Likewise.
2681 (estimated_freqs_path): Delete.
2682 (freqs_to_counts_path): Delete.
2683 (clear_counts_path): Delete.
2684 (ssa_fix_duplicate_block_edges): Likewise.
2685 (duplicate_thread_path): Likewise.
2686 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2687 * tree-tailcall.c (decrease_profile): Do not update frequency.
2688 (eliminate_tail_call): Likewise.
2689 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2690 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2691 (optimize_mask_stores): Likewise.
2692 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2693 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2694 (ubsan_expand_ptr_ifn): Update profile.
2695 * value-prof.c (gimple_ic): Simplify.
2696 * value-prof.h (gimple_ic): Update prototype.
2697 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2698 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2700 (want_inline_self_recursive_call_p): Likewise.
2701 (resolve_noninline_speculation): Only cummulate defined counts.
2702 (inline_small_functions): Use nonzero_p.
2703 (ipa_inline): Do not access freed node.
2705 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2707 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2708 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2710 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2712 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2713 non-legitimate address.
2715 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2717 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2718 (*lt0_<mode>di, *lt0_<mode>si): New.
2720 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2722 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2723 TARGET_PAIRED_FLOAT.
2725 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2726 Jim Wilson <jim.wilson@linaro.org>
2728 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2729 * config/aarch64/aarch64-tune.md: Regenerated.
2730 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2731 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2733 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2735 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2736 (arc_expand_prologue): Restore blink for millicode.
2737 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2739 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2742 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2743 gen_vec_duplicate after forcing the scalar into a register.
2745 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2747 * combine (try_combine): Print the insns input to try_combine to the
2750 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2753 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2754 Remove second argument from aarch64_process_target_attr call.
2755 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2757 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2759 (aarch64_handle_attr_arch): Remove second argument.
2760 (aarch64_handle_attr_cpu): Ditto.
2761 (aarch64_handle_attr_tune): Ditto.
2762 (aarch64_handle_attr_isa_flags): Ditto.
2763 (aarch64_process_one_target_attr): Ditto.
2764 (aarch64_process_target_attr): Ditto.
2765 (aarch64_option_valid_attribute_p): Remove second argument.
2766 on aarch64_process_target_attr call.
2768 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2770 * diagnostic.c: Include "selftest-diagnostic.h".
2771 (selftest::assert_location_text): New function.
2772 (selftest::test_diagnostic_get_location_text): New function.
2773 (selftest::diagnostic_c_tests): Call it.
2775 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2777 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2778 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2779 (class selftest::test_diagnostic_context): Move to...
2780 * selftest-diagnostic.c: New file.
2781 * selftest-diagnostic.h: New file.
2783 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2785 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2786 offset range for FT32B.
2787 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2788 * config/ft32/ft32.md: Add TARGET_NOPM.
2789 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2790 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2792 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2794 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2796 2017-11-02 Jeff Law <law@redhat.com>
2798 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2799 virtual keyword on FINAL OVERRIDE members.
2801 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2802 virtuals together. Add virtual destructor.
2803 (substitute_and_fold_engine): Similarly.
2805 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2807 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2809 2017-11-02 Richard Biener <rguenther@suse.de>
2811 PR tree-optimization/82795
2812 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2814 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2816 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2818 (gcc_GAS_CHECK_FEATURE): Remove.
2819 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2820 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2821 * configure: Regenerate.
2823 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2825 * config/arc/arc.c (hwloop_optimize): Account for empty
2828 2017-11-02 Richard Biener <rguenther@suse.de>
2831 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2832 Truncate ARRAY_REF index and element size.
2834 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2836 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2838 2017-11-01 Jeff Law <law@redhat.com>
2840 * tree-ssa-ccp.c (ccp_folder): New class derived from
2841 substitute_and_fold_engine.
2842 (ccp_folder::get_value): New member function.
2843 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2844 (ccp_fold_stmt): Remove prototype.
2845 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2846 * tree-ssa-copy.c (copy_folder): New class derived from
2847 substitute_and_fold_engine.
2848 (copy_folder::get_value): Renamed from get_value.
2849 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2850 * tree-vrp.c (vrp_folder): New class derived from
2851 substitute_and_fold_engine.
2852 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2853 (vrp_folder::get_value): New member function.
2854 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2855 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2856 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2857 provide a class interface to folder/substitute routines.
2858 (ssa_prop_fold_stmt_fn): Remove typedef.
2859 (ssa_prop_get_value_fn): Likewise.
2860 (subsitute_and_fold): Remove prototype.
2861 (replace_uses_in): Likewise.
2862 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2863 Renamed from replace_uses_in. Call the virtual member function
2864 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2865 (substitute_and_fold_dom_walker): Remove initialization of
2866 data member entries for calbacks. Add substitute_and_fold_engine
2867 member and initialize it.
2868 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2869 member functions for get_value, replace_phi_args_in c
2870 replace_uses_in, and fold_stmt calls.
2871 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2872 substitute_and_fold. Remove assert. Update ctor call.
2874 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2875 (ssa_prop_visit_phi_fn): Likewise.
2876 (class ssa_propagation_engine): New class to provide an interface
2878 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2880 (ssa_prop_visit_phi): Likewise.
2881 (ssa_propagation_engine::simulate_stmt): Moved into class.
2882 Call visit_phi/visit_stmt from the class rather than via
2883 file scoped static variables.
2884 (ssa_propagation_engine::simulate_block): Moved into class.
2885 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2886 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2887 set file scoped statics for the visit_stmt/visit_phi callbacks.
2888 * tree-complex.c (complex_propagate): New class derived from
2889 ssa_propagation_engine.
2890 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2891 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2892 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2894 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2895 ssa_propagation_engine.
2896 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2897 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2898 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2899 * tree-ssa-copy.c (copy_prop): New class derived from
2900 ssa_propagation_engine.
2901 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2902 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2903 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2904 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2905 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2906 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2907 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2909 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2911 PR rtl-optimization/82778
2912 PR rtl-optimization/82597
2913 * compare-elim.c (struct comparison): Add in_a_setter field.
2914 (find_comparison_dom_walker::before_dom_children): Remove killed
2915 bitmap and df_simulate_find_defs call, instead walk the defs.
2916 Compute last_setter and initialize in_a_setter. Merge definitions
2917 with first initialization for a few variables.
2918 (try_validate_parallel): Use insn_invalid_p instead of
2919 recog_memoized. Return insn rather than just the pattern.
2920 (try_merge_compare): Fix up comment. Don't uselessly test if
2921 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2923 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2926 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2927 Alan Hayward <alan.hayward@arm.com>
2928 David Sherwood <david.sherwood@arm.com>
2930 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2931 aarch64_hard_regno_nregs to get the number of registers
2934 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2935 Alan Hayward <alan.hayward@arm.com>
2936 David Sherwood <david.sherwood@arm.com>
2938 * config/aarch64/constraints.md (Upl): Rename to...
2940 * config/aarch64/aarch64.md
2941 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2944 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2945 Alan Hayward <alan.hayward@arm.com>
2946 David Sherwood <david.sherwood@arm.com>
2948 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2949 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2952 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2953 Alan Hayward <alan.hayward@arm.com>
2954 David Sherwood <david.sherwood@arm.com>
2956 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2957 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2958 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2960 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2961 of a single element.
2962 * config/aarch64/iterators.md: Add a comment above the permute
2963 unspecs to say that they are generated directly by
2964 aarch64_expand_vec_perm_const.
2965 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2967 2017-11-01 Nathan Sidwell <nathan@acm.org>
2969 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2971 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2973 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2974 and medany code models, and describe what they do.
2976 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2978 Revert accidental duplicate:
2980 * combine.c (can_change_dest_mode): Reject changes in
2981 REGMODE_NATURAL_SIZE.
2983 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2985 PR rtl-optimization/64682
2986 PR rtl-optimization/69567
2987 PR rtl-optimization/69737
2988 PR rtl-optimization/82683
2989 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2990 register mentioned in the note, drop the note, unless it came from I3,
2991 in which case it should go to I3 again.
2993 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2995 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2996 and return false if not.
2997 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2999 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3001 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3002 range is known to be empty.
3004 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3005 Alan Hayward <alan.hayward@arm.com>
3006 David Sherwood <david.sherwood@arm.com>
3008 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3009 and CONST_VECTOR_NUNITS instead of computing the number of units from
3010 the byte sizes of the vector and element.
3011 (simplify_binary_operation_1): Likewise.
3012 (simplify_const_binary_operation): Likewise.
3013 (simplify_ternary_operation): Likewise.
3015 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3016 Alan Hayward <alan.hayward@arm.com>
3017 David Sherwood <david.sherwood@arm.com>
3019 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3020 (int_mem_offset): ...this new function.
3021 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3022 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3023 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3026 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3027 Alan Hayward <alan.hayward@arm.com>
3028 David Sherwood <david.sherwood@arm.com>
3030 * lower-subreg.c (interesting_mode_p): New function.
3031 (compute_costs, find_decomposable_subregs, decompose_register)
3032 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3033 (resolve_clobber, dump_choices): Use it.
3035 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3036 Alan Hayward <alan.hayward@arm.com>
3037 David Sherwood <david.sherwood@arm.com>
3039 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3041 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3042 Alan Hayward <alan.hayward@arm.com>
3043 David Sherwood <david.sherwood@arm.com>
3045 * alias.c (find_base_value, find_base_term): Only process integer
3046 truncations. Check the precision rather than the size.
3048 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3049 Alan Hayward <alan.hayward@arm.com>
3050 David Sherwood <david.sherwood@arm.com>
3052 * machmode.h (is_narrower_int_mode): New function
3053 * optabs.c (expand_float, expand_fix): Use it.
3054 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3056 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3057 Alan Hayward <alan.hayward@arm.com>
3058 David Sherwood <david.sherwood@arm.com>
3060 * rtl.h (narrower_subreg_mode): New function.
3061 * ira-color.c (update_costs_from_allocno): Use it.
3063 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3064 Alan Hayward <alan.hayward@arm.com>
3065 David Sherwood <david.sherwood@arm.com>
3067 * optabs-query.h (convert_optab_p): New function, split out from...
3068 (convert_optab_handler): ...here.
3069 (widening_optab_handler): Delete.
3070 (find_widening_optab_handler): Remove permit_non_widening parameter.
3071 (find_widening_optab_handler_and_mode): Likewise. Provide an
3072 override that operates on mode class wrappers.
3073 * optabs-query.c (widening_optab_handler): Delete.
3074 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3075 parameter. Assert that the two modes are the same class and that
3076 the "from" mode is narrower than the "to" mode. Use
3077 convert_optab_handler instead of widening_optab_handler.
3078 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3079 instead of widening_optab_handler.
3080 * expr.c (expand_expr_real_2): Update calls to
3081 find_widening_optab_handler.
3082 * optabs.c (expand_widen_pattern_expr): Likewise.
3083 (expand_binop_directly): Take the insn_code as a parameter.
3084 (expand_binop): Only call find_widening_optab_handler for
3085 conversion optabs; use optab_handler otherwise. Update calls
3086 to find_widening_optab_handler and expand_binop_directly.
3087 Use convert_optab_handler instead of widening_optab_handler.
3088 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3089 find_widening_optab_handler and use scalar_mode rather than
3091 (convert_plusminus_to_widen): Likewise.
3093 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3094 Alan Hayward <alan.hayward@arm.com>
3095 David Sherwood <david.sherwood@arm.com>
3097 * machmode.h (fixed_size_mode): New class.
3098 * rtl.h (get_pool_mode): Return fixed_size_mode.
3099 * gengtype.c (main): Add fixed_size_mode.
3100 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3101 (get_raw_arg_mode): Likewise.
3102 * doc/tm.texi: Regenerate.
3103 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3104 * targhooks.c (default_get_reg_raw_mode): Likewise.
3105 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3106 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3107 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3108 (msp430_get_raw_result_mode): Likewise.
3109 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3110 * dbxout.c (dbxout_parms): Require fixed-size modes.
3111 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3112 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3113 * omp-low.c (lower_oacc_reductions): Likewise.
3114 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3115 (simplify_subreg): Update accordingly.
3116 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3117 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3118 that aren't fixed-size.
3119 (get_pool_mode): Return a fixed_size_mode.
3120 (output_constant_pool_2): Take a fixed_size_mode.
3122 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3123 Alan Hayward <alan.hayward@arm.com>
3124 David Sherwood <david.sherwood@arm.com>
3126 * doc/rtl.texi (vec_series): Document.
3127 (const): Say that the operand can be a vec_series.
3128 * rtl.def (VEC_SERIES): New rtx code.
3129 * rtl.h (const_vec_series_p_1): Declare.
3130 (const_vec_series_p): New function.
3131 * emit-rtl.h (gen_const_vec_series): Declare.
3132 (gen_vec_series): Likewise.
3133 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3134 (gen_vec_series): Likewise.
3135 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3136 * simplify-rtx.c (simplify_unary_operation): Handle negations
3138 (simplify_binary_operation_series): New function.
3139 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3140 (test_vector_ops_series): New function.
3141 (test_vector_ops): Call it.
3142 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3143 gen_const_vec_series.
3144 (altivec_lvsr): Likewise.
3145 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3147 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3148 Alan Hayward <alan.hayward@arm.com>
3149 David Sherwood <david.sherwood@arm.com>
3151 * doc/rtl.texi (const): Update description of address constants.
3152 Say that vector constants are allowed too.
3153 * common.md (E, F): Use CONSTANT_P instead of checking for
3155 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3156 checking for CONST_VECTOR.
3157 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3159 * expr.c (expand_expr_real_2): Check for vector modes instead
3160 of checking for CONST_VECTOR.
3161 * rtl.h (const_vec_p): New function.
3162 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3163 (unwrap_const_vec_duplicate): Handle them here too.
3165 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3166 David Malcolm <dmalcolm@redhat.com>
3167 Alan Hayward <alan.hayward@arm.com>
3168 David Sherwood <david.sherwood@arm.com>
3170 * rtl.h (vec_duplicate_p): New function.
3171 * selftest-rtl.c (assert_rtx_eq_at): New function.
3172 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3173 (assert_rtx_eq_at): Declare.
3174 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3175 * selftest-run-tests.c (selftest::run_tests): Call it.
3176 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3177 (simplify_unary_operation_1): Recursively handle vector duplicates.
3178 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3180 (simplify_subreg): Handle subregs of vector duplicates.
3181 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3182 (selftest::simplify_rtx_c_tests): New functions.
3184 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3185 Alan Hayward <alan.hayward@arm.com>
3186 David Sherwood <david.sherwood@arm.com>
3188 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3189 (gen_vec_duplicate): Likewise.
3190 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3192 (gen_const_vector): ...here.
3193 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3194 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3195 whose elements are all equal.
3196 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3197 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3198 (simplify_relational_operation): Likewise.
3199 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3201 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3202 (aarch64_expand_vector_init): Likewise.
3203 * config/arm/arm.c (neon_vdup_constant): Likewise.
3204 (neon_expand_vector_init): Likewise.
3205 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3206 (arm_block_set_unaligned_vect): Likewise.
3207 (arm_block_set_aligned_vect): Likewise.
3208 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3209 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3210 (expand_vec_perm_even_odd_pack): Likewise.
3211 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3212 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3213 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3214 gen_const_vec_duplicate.
3215 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3216 * config/mips/mips.c (mips_gen_const_int_vector): Use
3217 gen_const_vec_duplicate.
3218 (mips_expand_vector_init): Use CONST0_RTX.
3219 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3220 (define_split): Use gen_const_vec_duplicate.
3221 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3222 (define_split): Use gen_const_vec_duplicate.
3223 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3224 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3225 * config/spu/spu.c (spu_const): Likewise.
3227 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3228 Alan Hayward <alan.hayward@arm.com>
3229 David Sherwood <david.sherwood@arm.com>
3231 * combine.c (can_change_dest_mode): Reject changes in
3232 REGMODE_NATURAL_SIZE.
3234 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3236 * configure.ac (--enable-libssp): New.
3237 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3238 trying to determine target-specific default. Adjust indentation.
3239 * configure: Regenerated.
3240 * doc/install.texi (Configuration): Expand --disable-libssp
3243 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3245 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3248 2017-10-31 Martin Jambor <mjambor@suse.cz>
3251 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3253 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3255 * auto-profile.c (autofdo_source_profile::read): Use
3256 UNKNOWN_LOCATION rather than 0.
3257 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3258 (warning_at): ...this overload.
3259 (warning_at_rich_loc_n): Rename to...
3260 (warning_n): ...this overload.
3261 (error_at_rich_loc): Rename to...
3262 (error_at): ...this overload.
3263 (pedwarn_at_rich_loc): Rename to...
3264 (pedwarn): ...this overload.
3265 (permerror_at_rich_loc): Rename to...
3266 (permerror): ...this overload.
3267 (inform_at_rich_loc): Rename to...
3268 (inform): ...this overload.
3269 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3270 (diagnostic_n_impl_richloc): Rename to...
3271 (diagnostic_n_impl): ...this rich_location *-based decl.
3272 (inform_at_rich_loc): Rename to...
3273 (inform): ...this, and add an assertion.
3274 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3275 (warning_at_rich_loc): Rename to...
3276 (warning_at): ...this, and add an assertion.
3277 (warning_at_rich_loc_n): Rename to...
3278 (warning_n): ...this, and add an assertion.
3279 (warning_n): Update location_t-based implementation for removal of
3280 location_t-based diagnostic_n_impl.
3281 (pedwarn_at_rich_loc): Rename to...
3282 (pedwarn): ...this, and add an assertion.
3283 (permerror_at_rich_loc): Rename to...
3284 (permerror): ...this, and add an assertion.
3285 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3286 (error_at_rich_loc): Rename to...
3287 (error_at): ...this, and add an assertion.
3288 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3289 (driver::do_spec_on_infiles): Likewise.
3290 * substring-locations.c (format_warning_va): Update for renaming
3291 of inform_at_rich_loc.
3293 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3295 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3296 _Float<N>X built-in functions so that the variant without the
3297 "__builtin_" prefix is only enabled for the GNU C and Objective C
3298 languages when they are in non-strict ANSI/ISO mode.
3299 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3300 * target.def (floatn_builtin_p): Add a target hook to control
3301 whether _Float<N> and _Float<N>X built-in functions without the
3302 "__builtin_" prefix are enabled, and return true for C and
3303 Objective C in the default hook. Include langhooks.h in
3305 * targhooks.h (default_floatn_builtin_p): Likewise.
3306 * targhooks.c (default_floatn_builtin_p): Likewise.
3307 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3308 floatn_builtin_p target hook.
3309 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3311 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3312 Eric Botcazou <ebotcazou@adacore.com>
3314 PR rtl-optimization/81803
3315 * lra-constraints.c (curr_insn_transform): Also reload the whole
3316 register for a strict subreg no wider than a word if this is for
3317 a WORD_REGISTER_OPERATIONS target.
3319 2017-10-31 Jason Merrill <jason@redhat.com>
3321 * gdbinit.in: Skip over inlines from timevar.h.
3323 2017-10-31 Martin Liska <mliska@suse.cz>
3325 * doc/gcov.texi: Document new option.
3326 * gcov.c (print_usage): Likewise print it.
3327 (process_args): Support the argument.
3328 (format_count): New function.
3329 (format_gcov): Use the function.
3331 2017-10-31 Martin Liska <mliska@suse.cz>
3333 * gcov.c (struct name_map): do not use typedef.
3334 Define operator== and operator<.
3335 (name_search): Remove.
3336 (name_sort): Remove.
3337 (main): Do not allocate names.
3338 (process_file): Add vertical space.
3339 (generate_results): Use std::find.
3340 (release_structures): Do not release memory.
3341 (find_source): Use std::find.
3343 2017-10-31 Martin Liska <mliska@suse.cz>
3345 * gcov.c (struct line_info): Remove it's typedef.
3346 (line_info::line_info): Add proper ctor.
3347 (line_info::has_block): Do not use a typedef.
3348 (struct source_info): Do not use typedef.
3349 (circuit): Likewise.
3350 (get_cycles_count): Likewise.
3351 (output_intermediate_file): Iterate via vector iterator.
3352 (add_line_counts): Use std::vector methods.
3353 (accumulate_line_counts): Likewise.
3354 (output_lines): Likewise.
3356 2017-10-31 Martin Liska <mliska@suse.cz>
3358 * gcov.c (struct source_info): Remove typedef.
3359 (source_info::source_info): Add proper ctor.
3360 (accumulate_line_counts): Use struct, not it's typedef.
3361 (output_gcov_file): Likewise.
3362 (output_lines): Likewise.
3363 (main): Do not allocate an array.
3364 (output_intermediate_file): Use size of vector container.
3365 (process_file): Resize the vector.
3366 (generate_results): Do not preallocate, use newly added vector
3368 (release_structures): Do not release sources.
3369 (find_source): Use vector methods.
3370 (add_line_counts): Do not use typedef.
3372 2017-10-31 Martin Liska <mliska@suse.cz>
3374 * doc/gcov.texi: Document that.
3375 * gcov.c (add_line_counts): Mark lines with a non-executed
3377 (output_line_beginning): Handle such lines.
3378 (output_lines): Pass new argument.
3379 (output_intermediate_file): Print it in intermediate format.
3381 2017-10-31 Martin Liska <mliska@suse.cz>
3383 * color-macros.h: New file.
3384 * diagnostic-color.c: Factor out color related to macros to
3386 * doc/gcov.texi: Document -k option.
3387 * gcov.c (INCLUDE_STRING): Include string.h.
3388 (print_usage): Add -k option.
3389 (process_args): Parse it.
3390 (pad_count_string): New function.
3391 (output_line_beginning): Likewise.
3392 (DEFAULT_LINE_START): New macro.
3393 (output_lines): Support color output.
3395 2017-10-31 Martin Liska <mliska@suse.cz>
3397 PR gcov-profile/82633
3398 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3399 their interaction with GCOV infrastructure.
3400 * configure.ac: Add -fkeep-{inline,static}-functions to
3402 * configure: Regenerate.
3404 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3407 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3409 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3412 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3413 into a register if it does not fit into an immediate offset field.
3415 2017-10-31 Olivier Hainque <hainque@adacore.com>
3417 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3419 2017-10-31 Julia Koval <julia.koval@intel.com>
3421 * config.gcc: Add gfniintrin.h.
3422 * config/i386/gfniintrin.h: New.
3423 * config/i386/i386-builtin-types.def
3424 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3425 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3426 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3427 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3428 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3429 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3430 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3431 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3432 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3433 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3434 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3435 * config/i386/immintrin.h: Include gfniintrin.h.
3436 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3438 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3440 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3442 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3444 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3445 (arm_ashldi3_1bit): Remove pattern.
3446 (ashrdi3): Remove shift by 1 expansion.
3447 (arm_ashrdi3_1bit): Remove pattern.
3448 (lshrdi3): Remove shift by 1 expansion.
3449 (arm_lshrdi3_1bit): Remove pattern.
3450 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3451 cost of ashldi3 by 1.
3452 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3453 (<shift>di3_neon): Likewise.
3455 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3457 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3458 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3459 and (*aarch64_simd_mov<VQ:mode>).
3460 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3461 pattern alternative.
3462 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3463 attributes to match pattern alternative.
3465 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3467 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3468 * config/rs6000/emmintrin.h: New file.
3469 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3471 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3473 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3474 * (movdi_vfp_cortexa8): Remove pattern.
3476 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3478 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3480 (Specific, alpha*-dec-osf5.1): Remove.
3481 (Specific, mips-sgi-irix5): Remove.
3482 (Specific, mips-sgi-irix6): Remove.
3484 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3487 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3488 arguments to clear_bit_region_be.
3490 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3492 * gimplify.c: Include memmodel.h.
3494 2017-10-30 Martin Jambor <mjambor@suse.cz>
3496 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3497 condition whether loop should be executed at all.
3499 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3501 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3502 gimple folding of vec_madd() intrinsics.
3503 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3504 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3505 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3507 2017-10-30 Richard Biener <rguenther@suse.de>
3509 PR tree-optimization/82762
3511 2017-10-23 Richard Biener <rguenther@suse.de>
3513 PR tree-optimization/82129
3515 2017-08-01 Richard Biener <rguenther@suse.de>
3517 PR tree-optimization/81181
3518 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3519 (compute_antic): ... end of iteration here.
3521 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3523 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3524 -std=iso9899:2017 and -std=gnu17.
3525 * doc/standards.texi (C Language): Document C17 support.
3526 * doc/cpp.texi (Overview): Mention -std=c17.
3527 (Standard Predefined Macros): Document C11 and C17 values of
3528 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3529 * doc/extend.texi (Inline): Do not list individual options for
3530 standards newer than C99.
3531 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3533 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3536 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3538 * asan.c (asan_finish_file): Align asan globals array by shadow
3541 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3544 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3545 (struct store_immediate_info): Add bitregion_start and bitregion_end
3547 (store_immediate_info::store_immediate_info): Add brs and bre
3548 arguments and initialize bitregion_{start,end} from those.
3549 (struct merged_store_group): Add bitregion_start, bitregion_end,
3550 align_base and mask fields. Drop unnecessary struct keyword from
3551 struct store_immediate_info. Add do_merge method.
3552 (clear_bit_region_be): Use memset instead of loop storing zeros.
3553 (merged_store_group::do_merge): New method.
3554 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3555 stores as long as the surrounding bitregions have no gaps.
3556 (merged_store_group::merge_overlapping): Use do_merge.
3557 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3558 is byte aligned, rather than requiring that start and width are
3559 byte aligned. Drop unnecessary struct keyword from
3560 struct store_immediate_info. Allocate and populate also mask array.
3561 Make start of the arrays relative to bitregion_start rather than
3562 start and size them according to bitregion_{end,start} difference.
3563 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3564 struct store_immediate_info.
3565 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3566 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3567 struct keyword from struct store_immediate_info.
3568 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3569 between stores as long as the surrounding bitregions have no gaps.
3571 (struct split_store): Add orig non-static data member.
3572 (split_store::split_store): Initialize orig to false.
3573 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3574 if there is exactly a single original stmt. Change stmts argument
3575 to pointer from reference, if NULL, don't push anything to it. Add
3576 first argument, use it to optimize skipping over orig stmts that
3577 are known to be before bitpos already. Simplify.
3578 (split_group): Return unsigned int count how many stores are or
3579 would be needed rather than a bool. Add allow_unaligned argument.
3580 Change split_stores argument from reference to pointer, if NULL,
3581 only do a dry run computing how many stores would be produced.
3582 Rewritten algorithm to use both alignment and misalign if
3583 !allow_unaligned and handle bitfield stores with gaps.
3584 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3585 from bitregion_start instead of start. Compute allow_unaligned
3586 here, if true, do 2 split_group dry runs to compute which one
3587 produces fewer stores and prefer aligned if equal. Punt if
3588 new count is bigger or equal than original before emitting any
3589 statements, rather than during that. Remove no longer needed
3590 new_ssa_names tracking. Replace num_stmts with
3591 split_stores.length (). Use 32-bit stack allocated entries
3592 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3593 if possible. Handle bitfields with gaps.
3594 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3595 Compute bitregion_{start,end} for the stores and construct
3596 store_immediate_info with that. Formatting fixes.
3598 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3601 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3602 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3604 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3606 * gimplify.c: Include tm_p.h.
3608 * common.opt (gcoff): Re-add as ignored option.
3609 (gcoff1, gcoff2, gcoff3): Likewise.
3611 * Makefile.in (OBJS): Delete sdbout.o.
3612 (GTFILES): Delete $(srcdir)/sdbout.c.
3613 * debug.h: Delete sdb_debug_hooks.
3614 * final.c: Delete sdbout.h include.
3615 (final_scan_insn): Delete SDB_DEBUG check.
3616 (rest_of_clean_state): Likewise.
3617 * output.h: Delete sdb_begin_function_line.
3620 * toplev.c: Delete sdbout.h include.
3621 (process_options): Delete SDB_DEBUG check.
3622 * tree-core.h (tree_type_common): Delete pointer field of
3624 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3625 TYPE_SYMTAB_POINTER.
3626 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3627 (TYPE_SYMTAB_IS_POINTER): Delete.
3628 (TYPE_SYMTAB_IS_DIE): Renumber.
3629 * xcoffout.c: Refer to former sdbout.c file.
3630 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3632 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3633 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3634 (Debugging Options): Delete -gcoff.
3635 (-gstabs): Delete SDB reference.
3637 (-gcoff@var{level}): Delete.
3638 * doc/passes.texi (Debugging information output): Delete SDB and
3639 sdbout.c references.
3640 * doc/tm.texi: Regenerate.
3641 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3642 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3644 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3645 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3646 (SDB_DEBUGGING_INFO): Delete.
3647 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3648 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3649 * target.def (output_source_filename): Delete COFF reference.
3651 * common.opt (gcoff): Delete.
3652 (gxcoff+): Update Negative chain.
3653 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3655 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3656 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3657 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3659 (expand_function_start): Change sdb reference to past tense.
3660 (expand_function_end): Change sdb reference to past tense.
3661 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3662 * opts.c (debug_type_names): Delete coff entry.
3663 (common_handle_option): Delete OPT_gcoff case.
3664 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3666 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3667 * config/cris/cris.h: Delete SDB reference in comment.
3668 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3669 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3670 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3671 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3673 (ix86_expand_prologue): Likewise.
3674 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3675 * config/ia64/ia64.h: Likewise.
3676 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3677 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3679 * config/mmix/mmix.h: Likewise.
3680 * config/nds32/nds32.c: Likewise.
3681 * config/stormy/storym16.h: Likewise.
3682 * config/visium/visium.h: Likewise.
3683 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3685 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3687 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3688 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3689 FRAME_POINTER_REGNUM point at high end of local var area.
3691 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3693 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3694 Move comment around. Do not reset best_edge for a copiable
3695 destination if the copy would cause a partition change.
3696 (better_edge_p): Remove redundant check.
3698 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3700 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3702 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3704 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3705 for math functions that have _Float<N> and _Float<N>X variants.
3706 (mathfn_built_in_2): Add support for math functions that have
3707 _Float<N> and _Float<N>X variants.
3708 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3709 (expand_builtin_mathfn_ternary): Add support for fma with
3710 _Float<N> and _Float<N>X variants.
3711 (expand_builtin): Likewise.
3712 (fold_builtin_3): Likewise.
3713 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3714 create math function _Float<N> and _Float<N>X variants as external
3716 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3717 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3718 the __builtin_ prefix and if not strict ansi, without the prefix.
3719 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3720 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3721 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3722 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3723 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3724 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3725 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3726 function signatures for fma _Float<N> and _Float<N>X variants.
3727 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3728 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3729 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3730 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3731 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3732 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3733 * gencfn-macros.c (print_case_cfn): Add support for math functions
3734 that have _Float<N> and _Float<N>X variants.
3735 (print_define_operator_list): Likewise.
3736 (fltfn_suffixes): Likewise.
3738 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3739 for math functions that have _Float<N> and _Float<N>X variants.
3740 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3741 and _Float<N>X variants.
3742 (COPYSIGN): Likewise.
3745 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3746 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3748 (integer_valued_read_call_p): Likewise.
3749 * fold-const-call.c (fold_const_call_ss): Likewise.
3750 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3751 _Float<N> and _Float<N>X variants.
3752 (fold_const_call_ssss): Add support for fma _Float<N> and
3753 _Float<N>X variants.
3754 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3755 support for copysign and fma _Float<N> and _Float<N>X variants.
3756 (backprop::process_builtin_call_use): Likewise.
3757 * tree-call-cdce.c (can_test_argument_range); Add support for
3758 sqrt _Float<N> and _Float<N>X variants.
3759 (edom_only_function): Likewise.
3760 (get_no_error_domain): Likewise.
3761 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3762 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3763 copysign _Float<N> and _Float<N>X variants.
3764 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3765 handled by machine independent code.
3766 (FMAF128): Likewise.
3767 * doc/cpp.texi (Common Predefined Macros): Document defining
3768 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3769 fma _Float<N> and _Float<N>X variants.
3771 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3774 * config/i386/i386-modes.def (CCFPU): Remove definition.
3775 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3776 (ix86_cc_modes_compatible): Ditto.
3777 (ix86_expand_carry_flag_compare): Ditto.
3778 (ix86_expand_int_movcc): Ditto.
3779 (ix86_expand_int_addcc): Ditto.
3780 (ix86_reverse_condition): Ditto.
3781 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3782 Return true/false for unordered/ordered fp comparisons.
3783 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3784 (ix86_prepare_fp_compare_args): Update for rename.
3785 (ix86_expand_fp_compare): Update for rename. Generate unordered
3786 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3787 (ix86_expand_sse_compare_and_jump): Ditto.
3788 * config/i386/predicates.md (fcmov_comparison_operator):
3789 Remove CCFPU mode handling.
3790 (ix86_comparison_operator): Ditto.
3791 (ix86_carry_flag_operator): Ditto.
3792 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3793 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3794 (*cmpu<mode>_cc_i387): Ditto.
3795 (FPCMP): Remove mode iterator.
3796 (unord): Remove mode attribute.
3797 (unord_subst): New define_subst transformation
3798 (unord): New define_subst attribute.
3800 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3801 (*cmpi<unord>xf_i387): Ditto.
3802 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3803 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3804 using unord_subst transformation.
3805 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3806 (round_saeonly): Also handle CCFP mode.
3807 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3808 Remove UNSPEC_SAHF unspec handling.
3810 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3812 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3814 2017-10-27 Jeff Law <law@redhat.com>
3816 * gimple-ssa-sprintf.c: Include domwalk.h.
3817 (class sprintf_dom_walker): New class, derived from dom_walker.
3818 (sprintf_dom_walker::before_dom_children): New function.
3819 (struct call_info): Moved into sprintf_dom_walker class
3820 (compute_formath_length, handle_gimple_call): Likewise.
3821 (sprintf_length::execute): Call the dominator walker rather
3822 than walking the statements.
3824 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3825 gimple statement locations.
3826 (check_array_bounds): Corresponding changes. Get the statement's
3827 location directly from wi->stmt.
3829 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3832 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3834 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3836 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3837 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3839 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3842 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3843 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3844 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3845 maybe_get_pool_constant.
3846 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3849 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3851 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3852 2.26 caveat. Update gas and gld versions.
3853 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3856 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3858 * cgraph.h (set_malloc_flag): Declare.
3859 * cgraph.c (set_malloc_flag_1): New function.
3860 (set_malloc_flag): Likewise.
3861 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3862 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3864 (read_ipa_call_summary): Add support for reading is_return_callee.
3865 (write_ipa_call_summary): Stream is_return_callee.
3866 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3867 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3868 ipa-prop.h, ipa-fnsummary.h.
3869 (pure_const_names): Change to static.
3870 (malloc_state_e): Define.
3871 (malloc_state_names): Define.
3872 (funct_state_d): Add field malloc_state.
3873 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3874 (check_retval_uses): New function.
3875 (malloc_candidate_p): Likewise.
3876 (analyze_function): Add support for malloc attribute.
3877 (pure_const_write_summary): Stream malloc_state.
3878 (pure_const_read_summary): Add support for reading malloc_state.
3879 (dump_malloc_lattice): New function.
3880 (propagate_malloc): New function.
3881 (warn_function_malloc): New function.
3882 (ipa_pure_const::execute): Call propagate_malloc and
3883 ipa_free_fn_summary.
3884 (pass_local_pure_const::execute): Add support for malloc attribute.
3885 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3886 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3888 2017-10-27 Martin Liska <mliska@suse.cz>
3890 PR gcov-profile/82457
3891 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3892 for fork-like functions to be properly instrumented.
3894 2017-10-27 Richard Biener <rguenther@suse.de>
3897 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3898 info when we redirected EH.
3900 2017-10-26 Michael Collison <michael.collison@arm.com>
3902 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3904 (<optab>_trunchf<GPI:mode>2: New pattern.
3905 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3906 * config/aarch64/iterators.md (wv): New mode attribute.
3907 (vf, VF): New mode attributes.
3908 (vgp, VGP): New mode attributes.
3909 (s): Update attribute with SImode and DImode prefixes.
3911 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3913 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3914 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3915 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3916 (nios2_option_overide): Initialize it. Don't allow R0-relative
3917 addressing with PIC.
3918 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3919 (nios2_symbolic_constant_p): Likewise.
3920 (nios2_legitimate_address_p): Likewise.
3921 (nios2_r0rel_section_name_p): New.
3922 (nios2_symbol_ref_in_r0rel_data_p): New.
3923 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3924 (r0rel_constant_p): New.
3925 (nios2_print_operand_address): Handle r0rel_constant_p.
3926 (nios2_cdx_narrow_form_p): Likewise.
3927 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3928 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3929 (Nios II Options): Document -mr0rel-sec.
3931 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3933 * config/nios2/nios2.c: Include xregex.h.
3934 (nios2_gprel_sec_regex): New.
3935 (nios2_option_overide): Initialize it. Don't allow GP-relative
3936 addressing with PIC.
3937 (nios2_small_section_name_p): Check for regex match.
3938 * config/nios2/nios2.opt (mgprel-sec=): New option.
3939 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3940 (Nios II Options): Document -mgprel-sec.
3942 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3944 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3946 2017-10-26 Tom de Vries <tom@codesourcery.com>
3948 PR tree-optimization/82707
3949 * gimple.c (gimple_copy): Fix unsharing of
3950 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3952 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3954 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3955 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3956 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3957 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3958 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3959 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3960 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3961 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3962 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3963 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3964 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3965 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3966 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3967 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3968 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3969 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3970 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3972 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3974 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3976 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3977 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3978 warning to rs6000.c. Remove the Undocumented flag, since it has
3980 (-mabi=ibmlongdouble): Likewise.
3981 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3982 already set, set the default format for long double.
3983 (rs6000_debug_reg_global): Print whether long double is IBM or
3985 (rs6000_option_override_internal): Rework setting long double
3986 format. Only warn if the user is changing the long double default
3987 and they did not use -Wno-psabi.
3988 * doc/invoke.texi (PowerPC options): Update the documentation for
3989 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3991 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3992 Alan Hayward <alan.hayward@arm.com>
3993 David Sherwood <david.sherwood@arm.com>
3995 * rtl.h (wider_subreg_mode): New function.
3996 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3997 rather than an unsigned int *.
3998 * ira-color.c (regno_max_ref_width): Replace with...
3999 (regno_max_ref_mode): ...this new variable.
4000 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4001 Use wider_subreg_mode.
4002 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4003 rather than an unsigned int *.
4004 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4005 (process_alt_operands): Likewise.
4006 (invariant_p): Likewise.
4007 * lra-spills.c (assign_mem_slot): Likewise.
4008 (add_pseudo_to_slot): Likewise.
4009 * lra.c (collect_non_operand_hard_regs): Likewise.
4010 (add_regs_to_insn_regno_info): Likewise.
4011 * reload1.c (regno_max_ref_width): Replace with...
4012 (regno_max_ref_mode): ...this new variable.
4013 (reload): Update accordingly. Update call to
4014 ira_sort_regnos_for_alter_reg.
4015 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4016 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4017 (scan_paradoxical_subregs): Likewise.
4019 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4021 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4022 (aarch64_frame): Add emit_frame_chain boolean.
4023 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4024 Move eh_return case to aarch64_layout_frame.
4025 (aarch64_layout_frame): Initialize emit_frame_chain.
4026 (aarch64_expand_prologue): Use emit_frame_chain.
4028 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4030 * config/aarch64/aarch64.c (aarch64_layout_frame):
4031 Ensure LR is always stored at the bottom of the callee-saves.
4032 Remove rarely used frame layout which saves callee-saves at top of
4033 frame, so the store of LR can be used as a valid probe in all cases.
4035 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4037 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4038 Improve unaligned TImode/TFmode base/offset split.
4040 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4041 Alan Hayward <alan.hayward@arm.com>
4042 David Sherwood <david.sherwood@arm.com>
4044 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4045 * combine.c (find_single_use_1): Likewise.
4046 (expand_field_assignment): Likewise.
4047 (move_deaths): Likewise.
4048 * lra-constraints.c (simplify_operand_subreg): Likewise.
4049 (curr_insn_transform): Likewise.
4050 * lra.c (collect_non_operand_hard_regs): Likewise.
4051 (add_regs_to_insn_regno_info): Likewise.
4052 * rtlanal.c (reg_referenced_p): Likewise.
4053 (covers_regno_no_parallel_p): Likewise.
4055 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4057 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4058 Don't print any bits outside the precision of the value.
4059 * wide-int.cc (test_printing): Add some new tests.
4061 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4063 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4064 supports -xbrace_comment option.
4065 * configure: Regenerate.
4066 * config.in: Regenerate.
4067 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4068 (ASM_CPU_SPEC): Use it.
4070 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4072 * target.def (static_rtx_alignment): New hook.
4073 * targhooks.h (default_static_rtx_alignment): Declare.
4074 * targhooks.c (default_static_rtx_alignment): New function.
4075 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4076 * doc/tm.texi: Regenerate.
4077 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4078 instead of targetm.constant_alignment. Remove call to
4080 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4081 (cris_preferred_mininum_alignment): New function, split out from...
4082 (cris_constant_alignment): ...here.
4083 (cris_static_rtx_alignment): New function.
4084 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4086 (ix86_constant_alignment): ...here.
4087 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4088 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4089 (mmix_static_rtx_alignment): New function.
4090 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4091 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4093 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4096 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4097 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4099 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4101 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4102 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4103 mark as a sign-extending load.
4104 (local_pic_load_u): Define.
4106 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4109 * fold-const.c (operand_equal_for_comparison_p): Also return true
4110 if ARG0 is a simple variant of ARG1 with narrower precision.
4111 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4113 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4115 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4117 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4118 scatter_static, scatter_per_elt.
4119 * x86-tune-costs.h: Add new cost entries.
4121 2017-10-25 Richard Biener <rguenther@suse.de>
4123 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4124 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4125 class pass_fre): Move to ...
4126 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4127 class pass_fre): ... here and adjust for statistics.
4129 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4132 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4133 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4134 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4135 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4138 2017-10-25 Richard Biener <rguenther@suse.de>
4140 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4141 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4142 eliminate_push_avail, eliminate_insert): Move inside...
4143 (class eliminate_dom_walker): ... this class in preparation
4145 (fini_eliminate): Remove by merging with ...
4146 (eliminate): ... this function. Adjust for class changes.
4147 (pass_pre::execute): Remove fini_eliminate call.
4148 (pass_fre::execute): Likewise.
4150 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4153 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4154 (VPERMI2, VPERMI2I): New mode iterators.
4155 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4156 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4158 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4159 mode iterator. Remove 3 old define_insn patterns.
4160 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4161 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4162 VPERMI2 mode iterator, remove the other two expanders.
4163 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4164 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4165 instructions, remove the other two patterns.
4166 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4167 mode iterator, remove the other two patterns.
4168 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4169 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4170 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4171 and adjust argument order accordingly.
4172 (ix86_expand_vec_perm): Adjust caller.
4173 (expand_vec_perm_1): Likewise.
4174 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4175 (expand_vec_perm_vpermt2_vpshub2): ... this.
4176 (ix86_expand_vec_perm_const_1): Adjust caller.
4177 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4180 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4181 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4182 (vec_shl_<mode>): Remove unused expander.
4183 (avx512bw_<shift_insn><mode>3): New define_insn.
4184 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4185 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4187 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4190 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4193 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4195 PR rtl-optimization/82396
4196 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4197 (autopref_multipass_init): Simplify initialization.
4198 (autopref_rank_data): Simplify sort order.
4199 * gcc/sched-int.h (autopref_multipass_data_): Remove
4200 multi_mem_insn_p, min_offset and max_offset.
4202 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4205 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4206 Check special value of flag_omit_frame_pointer.
4207 (aarch64_can_eliminate): Likewise.
4208 (aarch64_override_options_after_change_1): Simplify handling of
4209 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4211 2017-10-24 Richard Biener <rguenther@suse.de>
4213 PR tree-optimization/82697
4214 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4215 zero for conditional load and unconditional store.
4217 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4219 * doc/install.texi: Document bootstrap-cet.
4221 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4224 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4225 ENDBR instruction at function entrance if function is only
4228 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4231 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4232 patterns to better describe from which operation the CF is computed.
4233 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4234 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4235 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4236 is 0, use _0 suffixed expanders instead of emitting a comparison
4239 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4241 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4242 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4244 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4247 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4248 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4249 * loop-iv.c (iv_get_reaching_def): Likewise.
4250 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4251 variable is promoted and the partition contains undefined values.
4253 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4255 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4257 (nios2_address_cost): Define.
4258 (nios2_legitimize_address): Recognize (exp + constant) directly.
4259 (TARGET_ADDRESS_COST): Define.
4261 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4263 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4264 (nios2_symbolic_memory_operand_p): Declare.
4265 (nios2_split_large_constant): Declare.
4266 (nios2_split_symbolic_memory_operand): Declare.
4267 * config/nios2/nios2.c: Adjust includes.
4268 (nios2_symbolic_constant_allowed): New.
4269 (nios2_symbolic_constant_p): New.
4270 (nios2_plus_symbolic_constant_p): New.
4271 (nios2_valid_addr_expr_p): Recognize addresses involving
4273 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4274 (nios2_symbolic_memory_operand_p): New.
4275 (nios2_large_constant_p): New.
4276 (nios2_split_large_constant): New.
4277 (nios2_split_plus_large_constant): New.
4278 (nios2_split_symbolic_memory_operand): New.
4279 (nios2_legitimize_address): Code refactoring. Handle addresses
4280 involving symbolic constants.
4281 (nios2_emit_move_sequence): Likewise.
4282 (nios2_print_operand): Improve error output.
4283 (nios2_print_operand_address): Handle LO_SUM.
4284 (nios2_cdx_narrow_form_p): Likewise.
4285 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4286 operands involving symbolic constants.
4287 (movhi_internal, movsi_internal): Likewise.
4288 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4289 (extendhisi2, extendqi<mode>2): Likewise.
4291 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4293 * tree-pass.h (PROP_rtl_split_insns): Define.
4294 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4296 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4298 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4300 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4303 * target.def (const_not_ok_for_debug_p): Default to
4304 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4305 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4306 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4307 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4308 which targetm.const_not_ok_for_debug_p returned true.
4309 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4311 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4313 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4314 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4315 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4317 (ix86_const_not_ok_for_debug_p): New function.
4318 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4319 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4321 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4324 * system.h: Conditionally include "unique-ptr.h" if
4325 INCLUDE_UNIQUE_PTR is defined.
4326 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4327 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4329 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4331 * config/rl78/rl78.md: New define_expand "subdi3".
4333 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4336 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4337 DF_REF_INSN if DF_REF_INSN_INFO is false.
4339 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4341 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4342 xmm_move instead of sse_move.
4343 (sse_store_index): New function.
4344 (ix86_register_move_cost): Be more sensible about mismatch stall;
4345 model AVX moves correctly; make difference between sse->integer and
4347 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4348 moves; make difference between SSE and AVX.
4349 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4350 and zmm_move. Increase size of sse load and store tables;
4351 add unaligned load and store tables; add ssemmx_to_integer.
4352 * x86-tune-costs.h: Update all entries according to real
4353 move latencies from Agner Fog's manual and chip documentation.
4355 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4358 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4359 * config/i386/constraints.md (Wf): New constraint.
4360 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4361 (cmp<dwi>_doubleword): Removed.
4362 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4363 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4364 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4365 expand with cmp<dwi>_doubleword. For LTU and GEU use
4366 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4368 * common.opt (gcolumn-info): Enable by default.
4369 * doc/invoke.texi (gcolumn-info): Document new default.
4371 2017-10-23 Richard Biener <rguenther@suse.de>
4373 PR tree-optimization/82672
4374 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4375 Fold the stmt if we propagated into it.
4377 2017-10-23 Richard Biener <rguenther@suse.de>
4379 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4380 (bitmap_remove_expr_from_set): ... this. All callers call this
4381 for non-constant values.
4382 (bitmap_set_subtract): Rename to...
4383 (bitmap_set_subtract_expressions): ... this. Adjust and
4385 (bitmap_set_contains_value): Remove superfluous check.
4386 (bitmap_set_replace_value): Inline into single caller ...
4387 (bitmap_value_replace_in_set): ... here and simplify.
4388 (dependent_clean): Merge into ...
4389 (clean): ... this using an overload. Adjust.
4390 (prune_clobbered_mems): Adjust.
4391 (compute_antic_aux): Likewise.
4392 (compute_partial_antic_aux): Likewise.
4394 2017-10-23 Richard Biener <rguenther@suse.de>
4396 PR tree-optimization/82129
4398 2017-08-01 Richard Biener <rguenther@suse.de>
4400 PR tree-optimization/81181
4401 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4402 (compute_antic): ... end of iteration here.
4404 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4406 * target.def (starting_frame_offset): New hook.
4407 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4408 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4409 * doc/tm.texi.in: Regenerate.
4410 * hooks.h (hook_hwi_void_0): Declare.
4411 * hooks.c (hook_hwi_void_0): New function.
4412 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4413 STARTING_FRAME_OFFSET.
4414 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4415 * reload1.c (reload): Likewise.
4416 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4417 instead of STARTING_FRAME_OFFSET.
4418 * function.c (try_fit_stack_local): Likewise.
4419 (assign_stack_local_1): Likewise
4420 (instantiate_virtual_regs): Likewise.
4421 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4422 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4423 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4424 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4425 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4426 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4427 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4428 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4429 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4430 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4431 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4432 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4433 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4434 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4435 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4436 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4437 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4438 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4439 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4440 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4441 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4442 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4443 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4444 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4445 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4446 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4447 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4448 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4449 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4450 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4451 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4452 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4453 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4454 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4455 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4456 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4457 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4458 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4459 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4460 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4461 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4462 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4463 return a HOST_WIDE_INT.
4464 (avr_builtin_setjmp_frame_value): Use it instead of
4465 STARTING_FRAME_OFFSET.
4466 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4467 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4468 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4470 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4471 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4472 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4473 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4474 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4475 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4476 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4477 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4478 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4479 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4480 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4481 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4483 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4484 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4485 * config/mips/mips.c (mips_compute_frame_info): Refer to
4486 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4487 (mips_starting_frame_offset): New function.
4488 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4489 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4490 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4491 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4492 and return a HOST_WIDE_INT.
4493 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4494 (mmix_initial_elimination_offset): Refer to
4495 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4496 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4497 * config/pa/pa.c (pa_starting_frame_offset): New function.
4498 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4499 (pa_expand_prologue): Likewise.
4500 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4501 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4502 !FRAME_GROWS_DOWNWARD handling to...
4503 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4504 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4505 !FRAME_GROWS_DOWNWARD handling to...
4506 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4507 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4508 !FRAME_GROWS_DOWNWARD handling to...
4509 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4510 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4512 (rs6000_starting_frame_offset): New function.
4513 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4514 !FRAME_GROWS_DOWNWARD handling to...
4515 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4516 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4517 !FRAME_GROWS_DOWNWARD handling to...
4518 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4519 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4520 !FRAME_GROWS_DOWNWARD handling to...
4521 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4522 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4523 (rs6000_starting_frame_offset): New function.
4524 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4525 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4526 * config/vax/vax.c (vax_starting_frame_offset): New function.
4527 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4528 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4529 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4530 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4531 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4532 * system.h (STARTING_FRAME_OFFSET): Poison.
4534 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4536 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4537 SCALAR_TYPE_MODE instead of TYPE_MODE.
4539 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4540 Alan Hayward <alan.hayward@arm.com>
4541 David Sherwood <david.sherwood@arm.com>
4543 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4545 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4546 Alan Hayward <alan.hayward@arm.com>
4547 David Sherwood <david.sherwood@arm.com>
4549 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4551 2017-10-23 Richard Biener <rguenther@suse.de>
4553 PR tree-optimization/82129
4554 * tree-ssa-pre.c (bitmap_set_and): Remove.
4555 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4556 canonicalizing expressions in the set to those with lowest
4557 ID rather than taking that from the first edge.
4559 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4561 * combine.c (rtx_equal_for_field_assignment_p): Use
4562 byte_lowpart_offset.
4564 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4565 Alan Hayward <alan.hayward@arm.com>
4566 David Sherwood <david.sherwood@arm.com>
4568 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4569 to a SUBREG_PROMOTED_VAR.
4571 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4572 Alan Hayward <alan.hayward@arm.com>
4573 David Sherwood <david.sherwood@arm.com>
4575 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4576 (expand_debug_source_expr): Likewise.
4577 * combine.c (combine_simplify_rtx): Likewise.
4578 * cse.c (fold_rtx): Likewise.
4579 * optabs.c (expand_float): Likewise.
4580 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4581 (simplify_binary_operation_1): Likewise.
4583 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4584 Alan Hayward <alan.hayward@arm.com>
4585 David Sherwood <david.sherwood@arm.com>
4587 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4588 (record_promoted_value): Likewise.
4589 * expr.c (expand_expr_real_2): Likewise.
4590 * ree.c (update_reg_equal_equiv_notes): Likewise.
4591 (combine_set_extension): Likewise.
4592 * rtlanal.c (low_bitmask_len): Likewise.
4593 * simplify-rtx.c (neg_const_int): Likewise.
4594 (simplify_binary_operation_1): Likewise.
4596 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4597 Alan Hayward <alan.hayward@arm.com>
4598 David Sherwood <david.sherwood@arm.com>
4600 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4601 * regcprop.c (maybe_mode_change): Likewise.
4602 * reload1.c (alter_reg): Likewise.
4604 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4606 * inchash.h (inchash::hash::add_wide_int): New function.
4607 * lto-streamer-out.c (hash_tree): Use it.
4609 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4611 * inchash.h (inchash::hash::add_wide_int): Rename to...
4612 (inchash::hash::add_hwi): ...this.
4613 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4614 (polymorphic_call_target_hasher::hash): Likewise.
4615 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4616 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4617 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4618 * lto-streamer-out.c (hash_tree): Likewise.
4619 * optc-save-gen.awk: Likewise.
4620 * tree.c (add_expr): Likewise.
4622 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4625 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4626 for ordered inequality comparisons even with TARGET_IEEE_FP.
4628 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4631 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4632 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4633 Expand with cmp<dwi>_doubleword.
4635 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4637 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4638 List CET intrinsics.
4639 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4640 specific to -fcf-protection option.
4642 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4644 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4645 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4646 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4647 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4648 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4649 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4650 (extra_objs): Add cet.o for Linux/x86 targets.
4651 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4652 * config/i386/cet.c: New file.
4653 * config/i386/cetintrin.h: Likewise.
4654 * config/i386/t-cet: Likewise.
4655 * config/i386/cpuid.h (bit_SHSTK): New.
4656 (bit_IBT): Likewise.
4657 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4658 pass IBT and SHSTK bits.
4659 * config/i386/i386-builtin-types.def
4660 (VOID_FTYPE_UNSIGNED_PVOID): New.
4661 (VOID_FTYPE_UINT64_PVOID): Likewise.
4662 * config/i386/i386-builtin.def: Add CET intrinsics.
4663 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4664 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4665 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4666 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4668 * config/i386/i386.c (rest_of_insert_endbranch): New.
4669 (pass_data_insert_endbranch): Likewise.
4670 (pass_insert_endbranch): Likewise.
4671 (make_pass_insert_endbranch): Likewise.
4672 (ix86_notrack_prefixed_insn_p): Likewise.
4673 (ix86_target_string): Add -mibt, -mshstk flags.
4674 (ix86_option_override_internal): Add flag_cf_protection
4676 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4677 (ix86_print_operand): Add 'notrack' prefix output.
4678 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4679 (ix86_expand_builtin): Expand CET intrinsics.
4680 (x86_output_mi_thunk): Add 'endbranch' instruction.
4681 * config/i386/i386.h (TARGET_IBT): New.
4682 (TARGET_IBT_P): Likewise.
4683 (TARGET_SHSTK): Likewise.
4684 (TARGET_SHSTK_P): Likewise.
4685 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4686 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4687 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4688 (builtin_setjmp_setup): New pattern.
4689 (builtin_longjmp): Likewise.
4690 (rdssp<mode>): Likewise.
4691 (incssp<mode>): Likewise.
4692 (saveprevssp): Likewise.
4693 (rstorssp): Likewise.
4694 (wrss<mode>): Likewise.
4695 (wruss<mode>): Likewise.
4696 (setssbsy): Likewise.
4697 (clrssbsy): Likewise.
4698 (nop_endbr): Likewise.
4699 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4701 * config/i386/immintrin.h: Include <cetintrin.h>.
4702 * config/i386/linux-common.h
4703 (file_end_indicate_exec_stack_and_cet): New prototype.
4704 (TARGET_ASM_FILE_END): New.
4706 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4708 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4709 latencies instead of having separate table; make difference between
4710 integer and float costs.
4711 * i386.h (processor_costs): Remove scalar_stmt_cost,
4712 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4713 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4715 * x86-tune-costs.h: Remove entries which has been removed in
4716 procesor_costs from all tables; make cond_taken_branch_cost
4717 and cond_not_taken_branch_cost COST_N_INSNS based.
4719 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4721 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4723 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4725 * config/i386/i386.md (isa): Remove fma_avx512f.
4726 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4727 <avx512>_fmadd_<mode>_mask3<round_name>,
4728 <avx512>_fmsub_<mode>_mask<round_name>,
4729 <avx512>_fmsub_<mode>_mask3<round_name>,
4730 <avx512>_fnmadd_<mode>_mask<round_name>,
4731 <avx512>_fnmadd_<mode>_mask3<round_name>,
4732 <avx512>_fnmsub_<mode>_mask<round_name>,
4733 <avx512>_fnmsub_<mode>_mask3<round_name>,
4734 <avx512>_fmaddsub_<mode>_mask<round_name>,
4735 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4736 <avx512>_fmsubadd_<mode>_mask<round_name>,
4737 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4738 (*vec_widen_umult_even_v16si<mask_name>,
4739 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4740 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4742 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4744 * extend.texi: Add 'nocf_check' documentation.
4745 * gimple.texi: Add second parameter to
4746 gimple_build_call_from_tree.
4747 * invoke.texi: Add -fcf-protection documentation.
4748 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4750 2017-10-20 Richard Biener <rguenther@suse.de>
4752 PR tree-optimization/82473
4753 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4754 the largest input type.
4756 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4758 * c-attribs.c (handle_nocf_check_attribute): New function.
4759 (c_common_attribute_table): Add 'nocf_check' handling.
4760 * gimple-parser.c: Add second argument NULL to
4761 gimple_build_call_from_tree.
4762 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4763 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4765 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4766 * common.opt: Add fcf-protection flag.
4767 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4768 * flag-types.h: Add enum cf_protection_level.
4769 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4770 Add 'nocf_check' attribute propagation to gimple call.
4771 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4772 (gimple_build_call_from_tree): Update prototype.
4773 (gimple_call_nocf_check_p): New function.
4774 (gimple_call_set_nocf_check): Likewise.
4775 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4776 * ipa-icf.c: Add nocf_check attribute in statement hash.
4777 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4778 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4779 * toplev.c (process_options): Add flag_cf_protection handling.
4781 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4783 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4785 2017-10-20 Richard Biener <rguenther@suse.de>
4787 PR tree-optimization/82603
4788 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4789 remove false predicated stores.
4791 2017-10-20 Richard Biener <rguenther@suse.de>
4793 * graphite-isl-ast-to-gimple.c
4794 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4795 Remove return value and simplify, dump copied stmt after lhs
4797 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4798 Reduce dump verbosity.
4799 (gsi_insert_earliest): Likewise.
4800 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4801 * graphite.c (print_global_statistics): Adjust dumping.
4802 (print_graphite_scop_statistics): Likewise.
4803 (print_graphite_statistics): Do not dump loops here.
4804 (graphite_transform_loops): But here.
4806 2017-10-20 Nicolas Roche <roche@adacore.com>
4808 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4809 * configure: Regenerate.
4811 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4814 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4815 functions when optimizing replace GIMPLE_RETURN stmts with
4816 calls to __builtin_unreachable ().
4819 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4820 for -fsanitize=thread link of executables.
4821 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4822 link of executables.
4825 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4827 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4828 define_insns for logical vector shifts to use VI248_AVX512BW
4829 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4830 condition, useless isa and prefix attributes. Change the first
4831 2 of these define_insns to ...
4832 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4833 define_insn for avx512vl.
4834 (<shift_insn><mode>3): ... and this, new define_insn without
4835 masking for non-avx512vl.
4838 * config/i386/sse.md (*andnot<mode>3,
4839 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4840 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4841 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4842 not applied use empty suffix even for TARGET_AVX512VL.
4843 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4844 is applied, supply evex,evex or evex,evex,evex instead of just
4847 2017-10-20 Julia Koval <julia.koval@intel.com>
4849 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4850 (OPTION_MASK_ISA_GFNI_UNSET): New.
4851 (ix86_handle_option): Handle OPT_mgfni.
4852 * config/i386/cpuid.h (bit_GFNI): New.
4853 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4854 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4855 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4856 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4857 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4858 * config/i386/i386.opt: Add mgfni.
4860 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4862 * config/msp430/msp430.c (msp430_option_override): Disable
4863 -fdelete-null-pointer-checks.
4864 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4866 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4868 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4869 of x87 and SSE instructions.
4871 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4873 * asan.c (create_cond_insert_point): Do not update edge count.
4874 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4875 (afdo_propagate_circuit): Likewise.
4876 (afdo_calculate_branch_prob): Likewise.
4877 (afdo_annotate_cfg): Likewise.
4878 * basic-block.h (struct edge_def): Remove count.
4879 (edge_def::count): New accessor.
4880 * bb-reorder.c (rotate_loop): Update.
4881 (find_traces_1_round): Update.
4882 (connect_traces): Update.
4883 (sanitize_hot_paths): Update.
4884 * cfg.c (unchecked_make_edge): Update.
4885 (make_single_succ_edge): Update.
4886 (check_bb_profile): Update.
4887 (dump_edge_info): Update.
4888 (update_bb_profile_for_threading): Update.
4889 (scale_bbs_frequencies_int): Update.
4890 (scale_bbs_frequencies_gcov_type): Update.
4891 (scale_bbs_frequencies_profile_count): Update.
4892 (scale_bbs_frequencies): Update.
4893 * cfganal.c (connect_infinite_loops_to_exit): Update.
4894 * cfgbuild.c (compute_outgoing_frequencies): Update.
4895 (find_many_sub_basic_blocks): Update.
4896 * cfgcleanup.c (try_forward_edges): Update.
4897 (try_crossjump_to_edge): Update
4898 * cfgexpand.c (expand_gimple_cond): Update
4899 (expand_gimple_tailcall): Update
4900 (construct_exit_block): Update
4901 * cfghooks.c (verify_flow_info): Update
4902 (redirect_edge_succ_nodup): Update
4903 (split_edge): Update
4904 (make_forwarder_block): Update
4905 (duplicate_block): Update
4906 (account_profile_record): Update
4907 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4908 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4909 * cfgloopmanip.c (scale_loop_profile): Update.
4911 (lv_adjust_loop_entry_edge): Update.
4912 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4913 (force_nonfallthru_and_redirect): Update.
4914 (purge_dead_edges): Update.
4915 (rtl_flow_call_edges_add): Update.
4916 * cgraphunit.c (init_lowered_empty_function): Update.
4917 (cgraph_node::expand_thunk): Update.
4918 * gimple-pretty-print.c (dump_probability): Update.
4919 (dump_edge_probability): Update.
4920 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4921 * haifa-sched.c (sched_create_recovery_edges): Update.
4922 * hsa-gen.c (convert_switch_statements): Update.
4923 * ifcvt.c (dead_or_predicable): Update.
4924 * ipa-inline-transform.c (inline_transform): Update.
4925 * ipa-split.c (split_function): Update.
4926 * ipa-utils.c (ipa_merge_profiles): Update.
4927 * loop-doloop.c (add_test): Update.
4928 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4929 * lto-streamer-in.c (input_cfg): Update.
4930 (input_function): Update.
4931 * lto-streamer-out.c (output_cfg): Update.
4932 * modulo-sched.c (sms_schedule): Update.
4933 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4934 * predict.c (maybe_hot_edge_p): Update.
4935 (unlikely_executed_edge_p): Update.
4936 (probably_never_executed_edge_p): Update.
4937 (dump_prediction): Update.
4938 (drop_profile): Update.
4939 (propagate_unlikely_bbs_forward): Update.
4940 (determine_unlikely_bbs): Update.
4941 (force_edge_cold): Update.
4942 * profile.c (compute_branch_probabilities): Update.
4943 * reg-stack.c (better_edge): Update.
4944 * shrink-wrap.c (handle_simple_exit): Update.
4945 * tracer.c (better_p): Update.
4946 * trans-mem.c (expand_transaction): Update.
4947 (split_bb_make_tm_edge): Update.
4948 * tree-call-cdce.c: Update.
4949 * tree-cfg.c (gimple_find_sub_bbs): Update.
4950 (gimple_split_edge): Update.
4951 (gimple_duplicate_sese_region): Update.
4952 (gimple_duplicate_sese_tail): Update.
4953 (gimple_flow_call_edges_add): Update.
4954 (insert_cond_bb): Update.
4955 (execute_fixup_cfg): Update.
4956 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4957 * tree-complex.c (expand_complex_div_wide): Update.
4958 * tree-eh.c (lower_resx): Update.
4959 (unsplit_eh): Update.
4960 (cleanup_empty_eh_move_lp): Update.
4961 * tree-inline.c (copy_edges_for_bb): Update.
4962 (freqs_to_counts): Update.
4963 (copy_cfg_body): Update.
4964 * tree-ssa-dce.c (remove_dead_stmt): Update.
4965 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4966 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4967 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4968 (unloop_loops): Update.
4969 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4970 * tree-ssa-loop-split.c (connect_loops): Update.
4971 (split_loop): Update.
4972 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4973 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4974 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4975 * tree-ssa-reassoc.c (branch_fixup): Update.
4976 * tree-ssa-tail-merge.c (replace_block_by): Update.
4977 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4978 (compute_path_counts): Update.
4979 (update_profile): Update.
4980 (recompute_probabilities): Update.
4981 (update_joiner_offpath_counts): Update.
4982 (estimated_freqs_path): Update.
4983 (freqs_to_counts_path): Update.
4984 (clear_counts_path): Update.
4985 (ssa_fix_duplicate_block_edges): Update.
4986 (duplicate_thread_path): Update.
4987 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4988 (case_bit_test_cmp): Update.
4989 (collect_switch_conv_info): Update.
4990 (gen_inbound_check): Update.
4991 (do_jump_if_equal): Update.
4992 (emit_cmp_and_jump_insns): Update.
4993 * tree-tailcall.c (decrease_profile): Update.
4994 (eliminate_tail_call): Update.
4995 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4996 (vect_do_peeling): Update.
4997 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4998 * ubsan.c (ubsan_expand_null_ifn): Update.
4999 (ubsan_expand_ptr_ifn): Update.
5000 * value-prof.c (gimple_divmod_fixed_value): Update.
5001 (gimple_mod_pow2): Update.
5002 (gimple_mod_subtract): Update.
5003 (gimple_ic): Update.
5004 (gimple_stringop_fixed_value): Update.
5006 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5009 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5011 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5013 PR rtl-optimization/82395
5014 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5015 based on non_spilled_static_chain_regno_p.
5017 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5019 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5020 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5021 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5023 2017-10-19 Martin Sebor <msebor@redhat.com>
5025 PR tree-optimization/82596
5026 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5028 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5030 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5031 (get_mem_refs_of_builtin_call): Likewise.
5032 * builtins.c (expand_builtin_apply): Adjust call to
5033 allocate_dynamic_stack_space.
5034 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5035 the third argument to allocate_dynamic_stack_space, otherwise -1.
5036 (expand_builtin): Deal with all alloca variants.
5037 (is_inexpensive_builtin): Likewise.
5038 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5039 * calls.c (special_function_p): Deal with all alloca variants.
5040 (initialize_argument_information): Adjust call to
5041 allocate_dynamic_stack_space.
5042 (expand_call): Likewise.
5043 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5044 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5045 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5046 use it for the stack usage computation.
5047 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5048 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5049 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5050 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5051 (in_loop_p): Remove first argument and useless check.
5052 (pass_walloca::execute): Remove useless test and adjust call to above.
5053 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5054 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5055 (gimplify_call_expr): Deal with all alloca variants.
5056 * hsa-gen.c (gen_hsa_alloca): Likewise.
5057 (gen_hsa_insns_for_call): Likewise.
5058 * ipa-pure-const.c (special_builtin_state): Likewise.
5059 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5060 * tree-object-size.c (alloc_object_size): Likewise.
5061 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5062 (call_may_clobber_ref_p_1): Likewise.
5063 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5064 (ccp_fold_stmt): Likewise.
5065 (optimize_stack_restore): Likewise.
5066 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5067 (mark_all_reaching_defs_necessary_1): Likewise.
5068 (propagate_necessity): Likewise.
5069 (eliminate_unnecessary_stmts): Likewise.
5070 * tree.c (build_common_builtin_nodes): Build
5071 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5072 (build_alloca_call_expr): New function.
5073 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5074 (CASE_BUILT_IN_ALLOCA): Likewise.
5075 (build_alloca_call_expr): Declare.
5076 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5078 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5081 * dwarf2out.c (new_die_raw): New static inline function.
5082 (new_die): Use it to create the DIE.
5083 (add_AT_external_die_ref): Likewise.
5084 (clone_die): Likewise.
5085 (clone_as_declaration): Likewise.
5086 (dwarf2out_vms_debug_main_pointer): Likewise.
5087 (base_type_die): Likewise. Remove early return for corner cases.
5088 Do not call add_pubtype on the DIE here.
5089 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5090 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5091 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5092 native order exists for base types, attach the DIE manually and call
5093 add_pubtype on it. Do not equate a reverse order DIE to the type.
5095 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5097 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5098 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5100 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5102 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5104 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5105 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5106 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5107 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5108 * opts.c (sanitizer_opts): Add builtin.
5109 * ubsan.c (instrument_builtin): New function.
5110 (pass_ubsan::execute): Call it.
5111 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5112 * doc/invoke.texi: Document -fsanitize=builtin.
5114 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5115 builtins, store max (log2 (align), 0) into uchar field instead of
5116 align into uptr field.
5117 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5118 store uchar 0 field instead of uptr 0 field.
5119 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5120 instead of passing one address of struct with 2 locations pass
5121 two addresses of structs with 1 location each.
5122 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5123 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5124 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5125 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5126 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5127 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5128 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5129 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5131 2017-10-19 Martin Liska <mliska@suse.cz>
5134 * file-find.c (remove_prefix): Remove.
5135 * file-find.h (remove_prefix): Likewise.
5136 * gcc-ar.c: Remove smartness of lookup.
5138 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5140 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5141 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5142 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5144 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5147 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5148 (setcc + and to xor + setcc): New peephole2.
5150 2017-10-19 Tom de Vries <tom@codesourcery.com>
5152 * doc/sourcebuild.texi (Test Directives, Variants of
5153 dg-require-support): Add dg-require-stack-size.
5155 2017-10-19 Martin Liska <mliska@suse.cz>
5158 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5159 that have a large alignment.
5160 (gimplify_target_expr): Likewise.
5162 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5164 PR rtl-optimization/82602
5165 * ira.c (rtx_moveable_p): Return false for volatile asm.
5167 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5170 * config/i386/i386-modes.def (CCGZ): New CC mode.
5171 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5172 * config/i386/predicates.md (ix86_comparison_operator):
5174 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5175 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5176 with double-word subtraction.
5177 (put_condition_code): Handle CCGZmode.
5179 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5181 * wide-int.cc (debug (const wide_int &)): New.
5182 (debug (const wide_int *)): New.
5183 (debug (const widest_int &)): New.
5184 (debug (const widest_int *)): New.
5186 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5189 * lra-constraints.c (curr_insn_transform): Use non-input operand
5190 instead of output one for matched reload.
5192 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5194 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5195 (tree-ssa-loop-ivopts.h): New header file.
5196 (struct builtin_info): New fields.
5197 (classify_builtin_1): Compute and record base and offset parts for
5198 memset builtin partition by calling strip_offset.
5199 (offset_cmp, fuse_memset_builtins): New functions.
5200 (finalize_partitions): Fuse adjacent memset partitions by calling
5202 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5203 Expose the interface.
5204 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5206 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5208 PR tree-optimization/82574
5209 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5210 that data reference must be executed exactly once per iteration
5211 against the outermost loop in nest.
5212 (classify_partition): Update call to above function.
5214 2017-10-18 Richard Biener <rguenther@suse.de>
5216 PR tree-optimization/82591
5217 * graphite.c (graphite_transform_loops): Move code gen message
5219 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5220 Here. Handle scop_to_isl_ast failing.
5221 (scop_to_isl_ast): Limit the number of ISL operations.
5223 2017-10-18 Richard Biener <rguenther@suse.de>
5225 * graphite-isl-ast-to-gimple.c
5226 (translate_isl_ast_to_gimple::set_rename): Simplify.
5227 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5228 (graphite_copy_stmts_from_block): ... here.
5229 (copy_bb_and_scalar_dependences): Simplify.
5230 (add_parameters_to_ivs_params): Canonicalize.
5231 (generate_entry_out_of_ssa_copies): Simplify.
5232 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5233 by passing in ISL dimension.
5234 (parameter_index_in_region_1): Rename to ...
5235 (parameter_index_in_region): ... this.
5236 (extract_affine): Adjust assert, pass down parameter index.
5237 (add_param_constraints): Use range-info when available.
5238 (build_scop_context): Adjust.
5239 * sese.c (new_sese_info): Adjust.
5240 (free_sese_info): Likewise.
5241 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5242 Remove unused typedefs.
5243 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5245 2017-10-18 Martin Liska <mliska@suse.cz>
5247 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5249 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5251 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5252 (s390_sched_init): Do not reset s390_sched_state if we entered the
5253 current basic block via a fallthru edge and all others are unlikely.
5255 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5257 * config/s390/s390.c (NUM_SIDES): New variable.
5258 (LONGRUNNING_THRESHOLD): New variable.
5259 (LATENCY_FACTOR): New variable.
5260 (s390_sched_score): Decrease score for long-running instructions on
5262 (s390_sched_variable_issue): Perform bookkeeping for long-running
5265 2017-10-18 Richard Biener <rguenther@suse.de>
5267 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5268 Simplify with removal of the parameter rename map.
5269 (set_rename): Likewise.
5270 (should_copy_to_new_region): Likewise.
5271 (graphite_copy_stmts_from_block): Likewise.
5272 (copy_bb_and_scalar_dependences): Remove initialization of
5273 unused copied_bb_map.
5275 (copy_internal_parameters): Likewise.
5276 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5277 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5278 Use INTEGRAL_TYPE_P.
5279 (parameter_index_in_region_1): Rename to ...
5280 (assign_parameter_index_in_region): ... this. Assert we have
5281 a parameter we handle.
5282 (scan_tree_for_params): Adjust.
5283 * sese.h (parameter_rename_map_t): Remove.
5284 (struct sese_info_t): Remove unused parameter_rename_map and
5285 copied_bb_map members.
5286 * sese.c (new_sese_info): Adjust.
5287 (free_sese_info): Likewise.
5289 2017-10-18 Martin Liska <mliska@suse.cz>
5292 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5293 on an abnormal edge.
5295 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5297 * doc/invoke.texi (ffunction-sections and fdata-sections):
5300 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5302 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5303 the use statement can throw internally.
5305 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5307 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5308 any RTX present on the RHS of a SET.
5309 * compare-elim.c (try_eliminate_compare): Restore comment.
5311 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5313 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5314 may be also called on tcc_type nodes.
5315 * langhooks.c (lhd_tree_size): Likewise.
5317 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5319 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5320 format_warning_at_substring.
5321 (maybe_warn): Convert source_range * param to a location_t. Pass
5322 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5323 (format_directive): Remove code to extract source_ranges and
5324 source_range * in favor of just a location_t.
5325 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5327 * substring-locations.c (format_warning_va): Convert
5328 source_range * param to a location_t.
5329 (format_warning_at_substring): Likewise.
5330 * substring-locations.h (format_warning_va): Likewise.
5331 (format_warning_at_substring): Likewise.
5333 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5335 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5337 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5338 and scatter/gather ops.
5340 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5341 vec_gather_load and vec_scatter_store.
5342 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5343 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5344 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5345 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5346 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5347 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5349 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5351 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5352 Detect FTST instruction and handle its register pops. Only pop
5353 second operand if can_pop_second_op is true.
5354 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5355 set can_pop_second_op to false in the compare_for_stack_reg call.
5357 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5358 output_fp_compare for stack register operands.
5359 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5360 instructions here. Do not emit stack register pops here. Assert
5361 that FCOMPP pops next to top stack register. Rewrite function.
5363 2017-10-17 Nathan Sidwell <nathan@acm.org>
5366 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5367 use DECL_ASSEMBLER_NAME_RAW.
5370 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5373 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5374 Wilco Dijkstra <wilco.dijkstra@arm.com>
5376 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5377 converstion to Pmode from the buf_addr.
5379 2017-10-17 Richard Biener <rguenther@suse.de>
5381 * graphite-dependences.c (scop_get_reads_and_writes): Change
5382 output parameters to references.
5384 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5386 PR 71026/tree-optimization
5387 * fold-const.c (distribute_real_division): Removed.
5388 (fold_binary_loc): Remove calls to distribute_real_divison.
5390 2017-10-17 Richard Biener <rguenther@suse.de>
5392 * graphite-scop-detection.c
5393 (scop_detection::stmt_has_simple_data_refs_p): Always use
5394 the full nest as region.
5395 (try_generate_gimple_bb): Likewise.
5396 * sese.c (scalar_evolution_in_region): Simplify now that
5397 SCEV can handle instantiation in regions.
5398 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5399 in the non-loop part of a function if requested.
5401 2017-10-17 Richard Biener <rguenther@suse.de>
5403 PR tree-optimization/82563
5404 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5406 (graphite_regenerate_ast_isl): Call it.
5407 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5409 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5411 PR tree-optimization/82549
5412 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5413 Formatting fixes. Instead of calling make_bit_field_ref with negative
5416 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5418 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5419 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5420 _mm_maskz_reduce_ss): New.
5421 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5422 __builtin_ia32_reducess_mask): Ditto..
5423 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5424 * config/i386/sse.md (reduces<mode>): Renamed to ...
5425 (reduces<mode><mask_scalar_name>): ... this.
5426 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5428 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5429 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5431 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5433 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5434 * selftest-run-tests.c (selftest::run_tests): Call
5435 selftest::unique_ptr_tests_cc_tests.
5436 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5437 * unique-ptr-tests.cc: New file.
5439 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5442 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5444 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5445 (make_hard_regno_born, make_hard_regno_dead): Update
5446 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5448 2017-10-16 Jeff Law <law@redhat.com>
5450 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5452 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5454 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5456 2017-10-16 Olivier Hainque <hainque@adacore.com>
5458 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5459 with_cpu if we were configured for an e500v2 target cpu name.
5461 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5463 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5464 * doc/invoke.texi: Document +nodsp as a valid extension for
5467 2017-10-16 Martin Liska <mliska@suse.cz>
5469 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5470 (test_set_range): Likewise.
5471 (test_range_functions): Rename to ...
5472 (test_bit_in_range): ... this.
5473 (sbitmap_c_tests): Add new test.
5475 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5477 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5479 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5480 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5482 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5484 * config/aarch64/aarch64-builtins.c
5485 (aarch64_types_quadopu_lane_qualifiers): New.
5486 (TYPES_QUADOPU_LANE): New.
5487 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5488 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5489 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5490 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5491 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5492 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5493 (Vdottype, DOTPROD): New.
5494 (sur): Add SDOT and UDOT.
5496 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5498 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5499 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5500 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5502 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5503 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5504 Enable TARGET_DOTPROD.
5505 (cortex-a75.cortex-a55): Likewise.
5506 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5508 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5510 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5511 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5513 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5515 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5516 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5517 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5518 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5519 * config/arm/types.md (neon_dot, neon_dot_q): New.
5520 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5522 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5524 * config/arm/arm.h (TARGET_DOTPROD): New.
5525 * config/arm/arm.c (arm_arch_dotprod): New.
5526 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5527 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5528 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5529 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5530 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5531 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5532 * doc/invoke.texi (armv8.2-a): Document dotprod
5534 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5536 * i386.c (ix86_vec_cost): New function.
5537 (ix86_rtx_costs): Handle vector operations better.
5538 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5539 * x86-tune-costs.h: Add new costs to all tables.
5541 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5543 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5545 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5546 divsd, sqrtss and sqrtsd
5547 * x86-tune-costs.h: Add new entries to all costs.
5548 (znver1_cost): Fix to match real instruction latencies.
5550 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5551 Michael Collison <michael.collison@arm.com>
5553 * compare-elim.c: Include emit-rtl.h.
5554 (can_merge_compare_into_arith): New function.
5555 (try_validate_parallel): Likewise.
5556 (try_merge_compare): Likewise.
5557 (try_eliminate_compare): Call the above when no previous clobber
5559 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5562 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5566 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5567 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5571 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5572 to be any operand_equal_p operands. For & (B - 1) require
5573 B to be power of 2. Recognize
5574 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5576 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5579 * optabs.c (expand_memory_blockage): Fix call of
5580 targetm.have_memory_blockage.
5582 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5585 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5586 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5587 objects to extra_objs instead of overwriting it.
5589 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5591 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5592 Use any_fp_register_operand as operand[3] predicate. Simplify
5593 equality test for operands[2] and operands[4] memory location.
5594 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5595 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5596 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5597 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5598 any_fp_register_operand as operand[1] predicate. Simplify
5599 equality test for operands[0] and operands[3] memory location.
5600 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5601 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5602 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5604 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5606 * target-insns.def: Add memory_blockage.
5607 * optabs.c (expand_memory_blockage): New function.
5608 (expand_asm_memory_barrier): Rename ...
5609 (expand_asm_memory_blockage): ... to this.
5610 (expand_mem_thread_fence): Call expand_memory_blockage
5611 instead of expand_asm_memory_barrier.
5612 (expand_mem_singnal_fence): Ditto.
5613 (expand_atomic_load): Ditto.
5614 (expand_atomic_store): Ditto.
5615 * doc/md.texi (Standard Pattern Names For Generation):
5616 Document memory_blockage instruction pattern.
5618 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5620 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5621 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5622 * config/rl78/rl78.md: New define_expand "adddi3".
5624 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5626 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5629 2017-10-13 Jeff Law <law@redhat.com>
5631 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5633 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5636 * internal-fn.c (expand_mul_overflow): If both operands have
5637 the same highpart of -1 or 0 and the topmost bit of lowpart
5638 is different, overflow is if res <= 0 rather than res < 0.
5640 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5642 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5643 TARGET_P9_VECTOR code for unaligned_load case.
5645 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5647 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5649 2017-10-13 Nathan Sidwell <nathan@acm.org>
5651 * tree-core.h (tree_contains_struct): Make bool.
5652 * tree.c (tree_contains_struct): Likewise.
5653 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5654 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5655 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5656 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5657 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5659 2017-10-13 Richard Biener <rguenther@suse.de>
5661 * graphite-isl-ast-to-gimple.c
5662 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5663 parameters and dominance check.
5664 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5665 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5666 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5667 Do not update SSA form here or do intermediate IL verification.
5668 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5669 (graphite_initialize): Remove check on the number of loops in
5670 the function and inline into graphite_transform_loops.
5671 (graphite_finalize): Inline into graphite_transform_loops.
5672 (graphite_transform_loops): Perform SSA update and IL verification
5674 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5676 2017-10-13 Richard Biener <rguenther@suse.de>
5678 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5679 graphite_expression_type_precision): Avoid global constructor
5681 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5682 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5683 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5684 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5685 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5686 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5688 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5691 * config/i386/i386.h (ix86_red_zone_size): New.
5692 * config/i386/i386.md (push peephole2s): Replace
5693 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5695 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5696 Alan Hayward <alan.hayward@arm.com>
5697 David Sherwood <david.sherwood@arm.com>
5699 * combine.c (can_change_dest_mode): Reject changes in
5700 REGMODE_NATURAL_SIZE.
5702 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5703 Alan Hayward <alan.hayward@arm.com>
5704 David Sherwood <david.sherwood@arm.com>
5706 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5707 (expand_debug_source_expr): Likewise.
5708 * combine.c (combine_simplify_rtx): Likewise.
5709 * cse.c (fold_rtx): Likewise.
5710 * fwprop.c (canonicalize_address): Likewise.
5711 * targhooks.c (default_shift_truncation_mask): Likewise.
5713 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5714 Alan Hayward <alan.hayward@arm.com>
5715 David Sherwood <david.sherwood@arm.com>
5717 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5718 (widened_mode): Likewise.
5719 (expand_unop): Likewise.
5720 * ree.c (transform_ifelse): Likewise.
5721 (merge_def_and_ext): Likewise.
5722 (combine_reaching_defs): Likewise.
5723 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5725 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5726 Alan Hayward <alan.hayward@arm.com>
5727 David Sherwood <david.sherwood@arm.com>
5729 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5730 * combine.c (gen_lowpart_for_combine): Likewise.
5731 * dwarf2out.c (rtl_for_decl_location): Likewise.
5732 * final.c (alter_subreg): Likewise.
5733 * rtlhooks.c (gen_lowpart_general): Likewise.
5734 (gen_lowpart_if_possible): Likewise.
5736 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5737 Alan Hayward <alan.hayward@arm.com>
5738 David Sherwood <david.sherwood@arm.com>
5740 * calls.c (expand_call): Use subreg_lowpart_offset.
5741 * cse.c (cse_insn): Likewise.
5742 * regcprop.c (copy_value): Likewise.
5743 (copyprop_hardreg_forward_1): Likewise.
5745 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5748 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5749 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5750 =Q constraints to +Q and into insn condition add check
5751 that operands[0] and operands[1] are equal.
5752 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5753 =Q constraints to +Q and into insn condition add check
5754 that operands[0] is equal to either operands[1] or operands[2].
5757 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5758 instead of handling MINUS_EXPR twice (once for each argument),
5759 canonicalize operand order and handle just once, use rtype where
5760 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5763 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5764 any values of __C while still being pattern recognizable as a simple
5767 2017-10-13 Richard Biener <rguenther@suse.de>
5769 PR tree-optimization/82451
5771 2017-10-02 Richard Biener <rguenther@suse.de>
5773 PR tree-optimization/82355
5774 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5775 a mapping for the enclosing loop but avoid generating one for
5777 (copy_bb_and_scalar_dependences): Remove premature codegen
5778 error on PHIs in blocks duplicated into multiple places.
5779 * graphite-scop-detection.c
5780 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5781 in the region use it as loop and nest to analyze the DR in.
5782 (try_generate_gimple_bb): Likewise.
5783 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5784 (add_loop_constraints): For blocks in a loop not in the region
5785 create a dimension with a single iteration.
5786 * sese.h (gbb_loop_at_index): Remove assert.
5788 * cfgloop.c (loop_preheader_edge): For the loop tree root
5789 return the single successor of the entry block.
5790 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5791 Reset the SCEV hashtable and niters.
5792 * graphite-scop-detection.c
5793 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5794 assert that we only have POLYNOMIAL_CHREC that vary in loops
5795 contained in the region.
5796 (scop_detection::graphite_can_represent_expr): Adjust.
5797 (scop_detection::stmt_has_simple_data_refs_p): For loops
5798 not in the region set loop to NULL. The nest is now the
5799 entry edge to the region.
5800 (try_generate_gimple_bb): Likewise.
5801 * sese.c (scalar_evolution_in_region): Adjust for
5802 instantiate_scev change.
5803 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5804 Make nest parameter the edge into the region.
5805 (create_data_ref): Likewise.
5806 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5807 entry edge into a region and adjust instantiate_scev calls.
5808 (create_data_ref): Likewise.
5809 (graphite_find_data_references_in_stmt): Likewise.
5810 (find_data_references_in_stmt): Pass the loop preheader edge
5811 from the nest argument.
5812 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5813 parameter the edge into the region.
5814 (instantiate_parameters): Use the loop preheader edge as entry.
5815 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5817 (get_instantiated_value_entry): Make instantiate_below parameter
5818 the edge into the region.
5819 (instantiate_scev_name): Likewise. Adjust dominance checks,
5820 when we cannot use loop-based instantiation instantiate by
5821 walking use-def chains.
5822 (instantiate_scev_poly): Adjust.
5823 (instantiate_scev_binary): Likewise.
5824 (instantiate_scev_convert): Likewise.
5825 (instantiate_scev_not): Likewise.
5826 (instantiate_array_ref): Remove.
5827 (instantiate_scev_3): Likewise.
5828 (instantiate_scev_2): Likewise.
5829 (instantiate_scev_1): Likewise.
5830 (instantiate_scev_r): Do not blindly handle N-operand trees.
5831 Do not instantiate array-refs. Handle all constants and invariants.
5832 (instantiate_scev): Make instantiate_below parameter
5833 the edge into the region.
5834 (resolve_mixers): Use the loop preheader edge for the region
5835 parameter to instantiate_scev_r.
5836 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5838 2017-10-13 Richard Biener <rguenther@suse.de>
5840 PR tree-optimization/82525
5841 * graphite-isl-ast-to-gimple.c
5842 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5844 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5845 Fail code generation when we cannot represent the isl integer.
5846 (binary_op_to_tree): Elide modulo operations that are no-ops
5847 in the type we code generate. Remove now superfluous code
5850 2017-10-13 Richard Biener <rguenther@suse.de>
5852 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5853 (scop_detection::harmful_loop_in_region): Remove premature
5854 IV type restriction.
5855 (scop_detection::graphite_can_represent_scev): We can handle
5856 pointer IVs just fine.
5858 2017-10-13 Alan Modra <amodra@gmail.com>
5860 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5861 "Clobbers and Scratch Registers". Add paragraph on
5862 alternative to clobbers for scratch registers and OpenBLAS
5865 2017-10-13 Alan Modra <amodra@gmail.com>
5867 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5868 example of a memory input for a string of known length. Move
5869 commentary out of table. Add a number of new examples
5870 covering array memory inputs.
5872 2017-10-12 Martin Liska <mliska@suse.cz>
5874 PR tree-optimization/82493
5875 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5876 (test_range_functions): New function.
5877 (sbitmap_c_tests): Likewise.
5878 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5879 * selftest.h (sbitmap_c_tests): New function.
5881 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5883 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5885 * config/rs6000/amo.h: Fix spacing issue.
5887 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5890 * config/i386/i386.md (*ashl<mode>3_mask_1,
5891 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5892 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5895 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5897 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5898 (profile_probability): Set max_probability
5899 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5901 * profile-count.c (profile_probability::differs_from_p): Do not
5902 rely on max_probaiblity == 10000
5904 2017-10-12 Jeff Law <law@redhat.com>
5906 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5909 2017-10-12 Martin Sebor <msebor@redhat.com>
5913 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5914 (handle_alias_pairs): Call it.
5915 * common.opt (-Wattribute-alias): New option.
5916 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5917 * doc/invoke.texi (-Wattribute-alias): Document.
5919 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5922 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5924 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5926 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5927 (make_hard_regno_born, make_hard_regno_dead): Update
5928 bb_killed_pseudos and bb_gen_pseudos.
5930 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5932 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5934 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5936 * config/alpha/alpha.c (alpha_split_conditional_move):
5937 Use std::swap instead of manually swapping.
5938 (alpha_stdarg_optimize_hook): Ditto.
5939 (alpha_canonicalize_comparison): Ditto.
5941 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5943 * tree-loop-distribution.c (struct builtin_info): New struct.
5944 (struct partition): Refactor fields into struct builtin_info.
5945 (partition_free): Free struct builtin_info.
5946 (build_size_arg_loc, build_addr_arg_loc): Delete.
5947 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5948 information from struct builtin_info.
5949 (find_single_drs): New function refactored from classify_partition.
5950 Also moved builtin validity checks to this function.
5951 (compute_access_range, alloc_builtin): New functions.
5952 (classify_builtin_st, classify_builtin_ldst): New functions.
5953 (classify_partition): Refactor code into functions find_single_drs,
5954 classify_builtin_st and classify_builtin_ldst.
5955 (distribute_loop): Don't do runtime alias check when distributing
5957 (find_seed_stmts_for_distribution): New function.
5958 (pass_loop_distribution::execute): Refactor code finding seed
5959 stmts into above function. Support distribution for the innermost
5960 two-level loop nest. Adjust dump information.
5962 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5964 * tree-loop-distribution.c: Adjust the general comment.
5965 (NUM_PARTITION_THRESHOLD): New macro.
5966 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5967 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5968 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5969 in call to build_partition_graph.
5970 (finalize_partitions): New parameter. Make loop distribution more
5971 conservative by fusing more partitions.
5972 (distribute_loop): Don't do runtime alias check in case of loop nest
5974 (find_seed_stmts_for_distribution): New function.
5975 (prepare_perfect_loop_nest): New function.
5976 (pass_loop_distribution::execute): Refactor code finding seed stmts
5977 and loop nest into above functions. Support loop nest distribution.
5978 Adjust dump information accordingly.
5980 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5982 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5983 and set PTYPE_SEQUENTIAL for merged partition.
5985 2017-10-12 Richard Biener <rguenther@suse.de>
5987 PR tree-optimization/69728
5989 2017-09-19 Richard Biener <rguenther@suse.de>
5991 PR tree-optimization/69728
5992 * graphite-sese-to-poly.c (schedule_error): New global.
5993 (add_loop_schedule): Handle empty domain by failing the
5995 (build_original_schedule): Handle schedule_error.
5997 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5998 domain by returning an unchanged schedule.
6000 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6002 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6003 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6005 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6007 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6010 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6013 * expr.c (store_field): Don't optimize away bitsize == 0 store
6014 from CALL_EXPR with addressable return type.
6016 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6018 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6019 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6020 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6021 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6022 TARGET_ISEL instead of TARGET_ISEL<sel>.
6024 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6026 * config/rs6000/rs6000.c
6027 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6029 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6031 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6033 (reg_or_cint_operand): Fix comment.
6034 (reg_or_zero_operand): New predicate.
6035 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6036 * config/rs6000/rs6000.c (output_isel): Delete.
6037 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6038 instead of reg_or_cint_operand. Output instruction directly (not via
6040 (isel_unsigned_<mode>): Ditto.
6041 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6042 gpc_reg_operand. Add an instruction alternative for this. Output
6043 instruction directly.
6044 (*isel_reversed_unsigned_<mode>): Ditto.
6046 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6048 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6049 (TARGET_CANONICALIZE_COMPARISON): Define.
6051 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6054 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6055 Check whether the dest is REG before adding REG_EQUIV note.
6057 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6060 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6062 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6063 (make_hard_regno_born, make_hard_regno_dead): Update
6064 bb_killed_pseudos and bb_gen_pseudos.
6066 2017-10-11 Nathan Sidwell <nathan@acm.org>
6068 * incpath.h (enum incpath_kind): Name enum, prefix values.
6069 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6070 * incpath.c (heads, tails): Use INC_MAX.
6071 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6072 (merge_include_chains, split_quote_chain,
6073 register_include_chains): Update incpath_kind names.
6074 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6075 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6077 (add_framework_path, darwin_register_objc_includes): Likewise.
6078 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6080 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6082 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6083 Do not use float_operator operator predicate.
6084 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6085 * config/i386/predicates.md (float_operator): Remove predicate.
6087 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6089 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6090 (*jccxf_i387): Ditto.
6091 (*jcc<mode>_i387): Ditto.
6092 (*jccu<mode>_i387): Ditto.
6093 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6094 (*jcc_*_i387 splitters): Remove.
6095 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6096 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6097 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6100 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6102 * profile-count.h (slow_safe_scale_64bit): New function.
6103 (safe_scale_64bit): New inline.
6104 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6105 * profile-count.c: Include wide-int.h
6106 (slow_safe_scale_64bit): New.
6108 2017-10-11 Nathan Sidwell <nathan@acm.org>
6110 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6111 HAS_DECL_ASSEMBLER_NAME_P.
6112 * gimple-expr.c (gimple_decl_printable_name: Check
6113 HAS_DECL_ASSEMBLER_NAME_P too.
6114 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6115 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6116 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6117 * passes.c (rest_of_decl_compilation): Check
6118 HAS_DECL_ASSEMBLER_NAME_P too.
6119 * recog.c (verify_changes): Likewise.
6120 * tree-pretty-print.c (dump_decl_name): Likewise.
6121 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6123 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6124 (SET_DECL_ASSEMBLER_NAME): Use it.
6125 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6126 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6127 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6129 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6131 * config.gcc (i386, x86_64): Add extra objects.
6132 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6133 (ix86_min_insn_size): Declare.
6134 (ix86_issue_rate): Declare.
6135 (ix86_adjust_cost): Declare.
6136 (ia32_multipass_dfa_lookahead): Declare.
6137 (ix86_macro_fusion_p): Declare.
6138 (ix86_macro_fusion_pair_p): Declare.
6139 (ix86_bd_has_dispatch): Declare.
6140 (ix86_bd_do_dispatch): Declare.
6141 (ix86_core2i7_init_hooks): Declare.
6142 (ix86_atom_sched_reorder): Declare.
6143 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6144 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6145 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6146 (rip_relative_addr_p): Rename to ...
6147 (ix86_rip_relative_addr_p): ... this one; export.
6148 (memory_address_length): Update.
6149 (ix86_issue_rate): Move to x86-tune-sched.c.
6150 (ix86_flags_dependent): Move to x86-tune-sched.c.
6151 (ix86_agi_dependent): Move to x86-tune-sched.c.
6152 (exact_dependency_1): Move to x86-tune-sched.c.
6153 (exact_store_load_dependency): Move to x86-tune-sched.c.
6154 (ix86_adjust_cost): Move to x86-tune-sched.c.
6155 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6156 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6157 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6158 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6159 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6160 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6161 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6162 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6163 (min_insn_size): Rename to ...
6164 (ix86_min_insn_size): ... this one; export.
6165 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6166 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6167 (core2i7_first_cycle_multipass_backtrack): Move to
6168 x86-tune-sched-core.c.
6169 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6170 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6171 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6172 (ix86_avoid_jump_mispredicts): Update.
6173 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6174 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6175 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6176 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6177 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6178 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6179 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6180 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6181 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6182 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6183 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6184 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6185 (BIG): Move to ix86-tune-sched-bd.c.
6186 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6187 (enum insn_path): Move to ix86-tune-sched-bd.c.
6188 (get_mem_group): Move to ix86-tune-sched-bd.c.
6189 (is_cmp): Move to ix86-tune-sched-bd.c.
6190 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6191 (is_branch): Move to ix86-tune-sched-bd.c.
6192 (is_prefetch): Move to ix86-tune-sched-bd.c.
6193 (init_window): Move to ix86-tune-sched-bd.c.
6194 (allocate_window): Move to ix86-tune-sched-bd.c.
6195 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6196 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6197 (process_end_window): Move to ix86-tune-sched-bd.c.
6198 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6199 (find_constant): Move to ix86-tune-sched-bd.c.
6200 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6201 (has_immediate): Move to ix86-tune-sched-bd.c.
6202 (get_insn_path): Move to ix86-tune-sched-bd.c.
6203 (get_insn_group): Move to ix86-tune-sched-bd.c.
6204 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6205 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6206 (add_insn_window): Move to ix86-tune-sched-bd.c.
6207 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6208 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6209 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6210 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6211 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6212 (do_dispatch): Move to ix86-tune-sched-bd.c.
6213 (has_dispatch): Move to ix86-tune-sched-bd.c.
6214 * config/i386/t-i386: Add new object files.
6215 * config/i386/x86-tune-costs.h: New file.
6216 * config/i386/x86-tune-sched-atom.c: New file.
6217 * config/i386/x86-tune-sched-bd.c: New file.
6218 * config/i386/x86-tune-sched-core.c: New file.
6219 * config/i386/x86-tune-sched.c: New file.
6221 2017-10-11 Liu Hao <lh_mouse@126.com>
6223 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6224 the generic version below instead.
6225 (should_colorize): Recognize Windows consoles as terminals
6227 * pretty-print.c [__MINGW32__] (write_all): New function.
6228 [__MINGW32__] (find_esc_head): Likewise.
6229 [__MINGW32__] (find_esc_terminator): Likewise.
6230 [__MINGW32__] (eat_esc_sequence): Likewise.
6231 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6233 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6236 2017-10-11 Richard Biener <rguenther@suse.de>
6238 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6239 Properly call analyze_scalar_evolution with the loop of the stmt.
6241 2017-10-11 Richard Biener <rguenther@suse.de>
6243 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6244 * tree-core.h (tree_base): Add chrec_var union member.
6245 * tree.h (CHREC_VAR): Remove.
6246 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6247 * tree-chrec.h (build_polynomial_chrec): Adjust.
6248 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6249 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6251 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6253 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6254 * match.pd: ... here.
6255 ((T) X == (T) Y): Relax condition.
6257 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6259 PR tree-optimization/82472
6260 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6262 (break_alias_scc_partitions): Update postorder number.
6264 2017-10-11 Martin Liska <mliska@suse.cz>
6267 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6269 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6271 2017-10-11 Martin Liska <mliska@suse.cz>
6273 * print-rtl.c (print_insn): Move declaration of idbuf
6274 to same scope as name.
6276 2017-10-11 Martin Liska <mliska@suse.cz>
6281 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6284 2017-10-11 Martin Liska <mliska@suse.cz>
6287 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6290 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6292 * config.gcc (default_gnu_indirect_function): Default to yes for
6293 arm*-*-linux* with glibc.
6295 2017-10-11 Richard Biener <rguenther@suse.de>
6297 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6298 default-defs and types we do not want to analyze.
6299 (interpret_loop_phi): Replace unreachable code with an assert.
6300 (compute_scalar_evolution_in_loop): Remove and inline ...
6301 (analyze_scalar_evolution_1): ... here, replacing condition with
6302 what makes the intent clearer. Remove handling of cases
6303 get_scalar_evolution now handles.
6305 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6307 PR rtl-optimization/81434
6308 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6309 comment for main loop. In sched_group_found if, also add checks for
6310 pass and min_cost_group.
6312 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6314 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6315 (rs6000_insn_cost): New function.
6316 * config/rs6000/rs6000.md (cost): New attribute.
6318 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6319 H.J. Lu <hongjiu.lu@intel.com>
6323 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6324 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6325 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6326 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6327 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6328 OPTION_MASK_ISA_MMX and some other bit set require both
6329 mmx and the ISAs without the mmx bit.
6330 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6331 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6332 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6333 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6334 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6335 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6336 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6337 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6338 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6339 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6340 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6341 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6342 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6343 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6344 Add OPTION_MASK_ISA_MMX.
6346 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6348 * config.gcc (armv7*-*-freebsd*): New target.
6349 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6351 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6353 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6354 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6357 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6359 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6360 a template parameter.
6361 (WIDE_INT_REF_FOR): Update accordingly.
6362 * tree.h (wi::int_traits <const_tree>): Delete.
6363 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6364 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6365 (wi::tree_to_wide_ref): New typedef.
6366 (wi::to_wide): New function.
6367 * calls.c (get_size_range): Use wi::to_wide when operating on
6369 * cgraph.c (cgraph_node::create_thunk): Likewise.
6370 * config/i386/i386.c (ix86_data_alignment): Likewise.
6371 (ix86_local_alignment): Likewise.
6372 * dbxout.c (stabstr_O): Likewise.
6373 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6374 * expr.c (const_vector_from_tree): Likewise.
6375 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6376 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6377 (fold_negate_expr_1, int_const_binop_1, const_binop)
6378 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6379 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6380 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6381 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6382 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6383 (fold_not_const, round_up_loc): Likewise.
6384 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6385 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6386 (alloca_call_type): Likewise.
6387 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6388 * godump.c (go_output_typedef): Likewise.
6389 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6390 * internal-fn.c (get_min_precision): Likewise.
6391 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6392 * ipa-polymorphic-call.c
6393 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6394 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6395 (ipa_modify_call_arguments): Likewise.
6396 * match.pd: Likewise.
6397 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6398 * print-tree.c (print_node_brief, print_node): Likewise.
6399 * stmt.c (expand_case): Likewise.
6400 * stor-layout.c (layout_type): Likewise.
6401 * tree-affine.c (tree_to_aff_combination): Likewise.
6402 * tree-cfg.c (group_case_labels_stmt): Likewise.
6403 * tree-data-ref.c (dr_analyze_indices): Likewise.
6404 (prune_runtime_alias_test_list): Likewise.
6405 * tree-dump.c (dequeue_and_dump): Likewise.
6406 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6407 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6408 * tree-pretty-print.c (dump_generic_node): Likewise.
6409 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6410 (simple_iv_with_niters): Likewise.
6411 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6412 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6413 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6414 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6415 (refine_value_range_using_guard, number_of_iterations_ne_max)
6416 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6417 (get_cst_init_from_scev, record_nonwrapping_iv)
6418 (scev_var_range_cant_overflow): Likewise.
6419 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6420 * tree-ssa-pre.c (compute_avail): Likewise.
6421 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6422 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6423 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6424 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6425 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6426 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6427 (dump_case_nodes, try_switch_expansion): Likewise.
6428 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6429 (vect_do_peeling): Likewise.
6430 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6431 * tree-vect-stmts.c (vectorizable_load): Likewise.
6432 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6433 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6434 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6435 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6436 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6437 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6438 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6439 (evrp_dom_walker::before_dom_children): Likewise.
6440 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6441 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6442 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6443 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6444 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6445 (get_range_pos_neg): Likewise.
6446 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6447 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6448 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6449 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6450 * config/avr/avr.c (avr_fold_builtin): Likewise.
6451 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6452 * config/msp430/msp430.c (msp430_attr): Likewise.
6453 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6454 * config/powerpcspe/powerpcspe-c.c
6455 (altivec_resolve_overloaded_builtin): Likewise.
6456 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6457 (rs6000_expand_ternop_builtin): Likewise.
6458 * config/rs6000/rs6000-c.c
6459 (altivec_resolve_overloaded_builtin): Likewise.
6460 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6461 (rs6000_expand_ternop_builtin): Likewise.
6462 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6464 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6466 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6467 when copying loop nest with only one inner loop.
6469 2017-10-10 Richard Biener <rguenther@suse.de>
6471 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6472 blocks if SCEV is active.
6473 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6475 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6476 (scev_initialize): Assert we are not yet initialized.
6478 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6480 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6481 inner loop's exit stmt by making it always exit the loop, otherwise
6482 we would generate an infinite empty loop.
6484 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6486 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6487 renaming variables in new preheader if it's deleted.
6489 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6491 * tree-loop-distribution.c (struct partition): Remove unused field
6492 loops of the structure.
6493 (partition_alloc, partition_free): Ditto.
6494 (build_rdg_partition_for_vertex): Ditto.
6496 2017-10-09 Jeff Law <law@redhat.com>
6498 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6499 return type to match prototype and documentation.
6501 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6503 * config/rs6000/rs6000.c (processor_costs): Move to ...
6504 * config/rs6000/rs6000.h: ... here.
6505 (rs6000_cost): Declare.
6507 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6509 * except.c (setjmp_fn): New global variable.
6510 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6511 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6512 if DONT_USE_BUILTIN_SETJMP is defined.
6514 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6516 * target.def (insn_cost): New hook.
6517 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6518 * doc/tm.texi: Regenerate.
6519 * rtlanal.c (insn_cost): Use the new hook.
6521 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6523 * combine.c (combine_validate_cost): Compute the new insn_cost,
6524 not just pattern_cost.
6525 (try_combine): Adjust comment.
6527 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6529 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6531 * combine.c (uid_insn_cost): Adjust comment.
6532 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6534 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6535 * dse.c (find_shift_sequence): Ditto.
6536 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6537 (bb_valid_for_noce_process_p): Use pattern_cost.
6538 * rtl.h (insn_rtx_cost): Delete.
6539 (pattern_cost): New prototype.
6540 (insn_cost): New prototype.
6541 * rtlanal.c (insn_rtx_cost): Rename to...
6542 (pattern_cost): ... this.
6545 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6547 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6548 (*jcc<mode>_0_r_i387): Ditto.
6549 (*jccxf_r_i387): Ditto.
6550 (*jcc<mode>_r_i387): Ditto.
6551 (*jccu<mode>_r_i387): Ditto.
6552 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6553 (*jcc): Rename from *jcc_1.
6555 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6557 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6558 deferred rescans after the lvx/stvx recombination pre-pass.
6560 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6562 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6563 memory operation instruction support.
6564 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6565 (rs6000-ibm-aix[789]*): Likewise.
6566 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6567 Document new functions.
6569 2017-10-09 Richard Biener <rguenther@suse.de>
6571 PR tree-optimization/82397
6572 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6573 equality only for semantically equal trees.
6575 2017-10-09 Richard Biener <rguenther@suse.de>
6577 PR tree-optimization/82449
6578 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6579 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6580 allow constant addresses.
6581 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6584 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6586 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6589 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6592 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6595 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6598 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6600 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6603 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6604 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6606 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6608 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6609 (WI_BINARY_PREDICATE_RESULT): Likewise.
6610 (wi::binary_traits::operator_result): New type.
6611 (wi::binary_traits::predicate_result): Likewise.
6612 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6613 (generic_wide_int::operator==, generic_wide_int::operator!=)
6614 (generic_wide_int::operator&, generic_wide_int::and_not)
6615 (generic_wide_int::operator|, generic_wide_int::or_not)
6616 (generic_wide_int::operator^, generic_wide_int::operator+
6617 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6619 (operator~, unary operator-, operator==, operator!=, operator&)
6620 (operator|, operator^, operator+, binary operator-, operator*): New
6622 * expr.c (get_inner_reference): Use wi::bit_and_not.
6623 * fold-const.c (fold_binary_loc): Likewise.
6624 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6625 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6626 (bit_value_binop): Likewise.
6627 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6628 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6629 (extract_range_from_binary_expr_1): Likewise.
6630 (masked_increment): Likewise.
6631 (simplify_bit_ops_using_ranges): Likewise.
6633 2017-10-09 Martin Jambor <mjambor@suse.cz>
6636 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6637 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6638 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6640 (hsa_fixup_mov_insn_type): New function.
6641 (hsa_op_with_type::get_in_type): Use it.
6642 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6643 immediates in an assert.
6644 (hsa_op_with_type::extend_int_to_32bit): New method.
6645 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6646 types. Convert to dest type if necessary.
6647 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6648 (reg_for_gimple_ssa): Pass false as min32int to
6649 hsa_type_for_scalar_tree_type.
6650 (gen_hsa_addr): Fixup type when creating addresable temporary.
6651 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6652 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6653 necessary. Call hsa_fixup_mov_insn_type.
6654 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6655 extend operands and convert to dest type if necessary.
6656 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6657 to dest type if necessary.
6658 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6659 if conversion nt necessary and size matches.
6660 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6661 to dest type if necessary.
6662 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6663 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6665 (gen_hsa_clrsb): Likewise.
6666 (gen_hsa_ffs): Likewise.
6667 (gen_hsa_divmod): Extend operands and convert to dest type if
6669 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6671 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6673 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6674 Remove empty default arguments. Use a brace block as output
6676 (conditional return): Ditto.
6678 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6679 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6681 (group_ending_nop): Ditto.
6682 (doloop_end): Ditto.
6683 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6684 (splitters for those): Ditto.
6686 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6688 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6689 a conditional jump (and the compare for it) so that pc_rtx is the
6691 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6692 for the deleted and renamed ctr<mode>_internal[234] patterns.
6693 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6694 Delete second conditional return pattern.
6695 (ctr<mode>_internal2): Delete this second bdnz pattern.
6696 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6697 (ctr<mode>_internal4): Delete this second bdz pattern.
6699 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6701 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6702 (always_initialized_rtx_for_ssa_name_p): New predicate.
6703 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6704 (finish_out_of_ssa): Free new field of SA.
6705 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6706 * tree-ssa-coalesce.c: Include tree-ssa.h.
6707 (get_parm_default_def_partitions): Remove extern keyword.
6708 (get_undefined_value_partitions): New function.
6709 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6710 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6712 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6714 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6716 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6718 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6720 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6721 for targets that preffer 128bit.
6723 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6725 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6727 2017-10-08 Olivier Hainque <hainque@adacore.com>
6729 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6730 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6732 (thumb_set_return_address): Likewise.
6734 2017-10-08 Olivier Hainque <hainque@adacore.com>
6736 * common/config/arm/arm-common.c (arm_except_unwind_info):
6737 Handle DWARF2_UNWIND_INFO.
6739 2017-10-07 Michael Collison <michael.collison@arm.com>
6741 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6744 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6746 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6747 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6748 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6749 defined, force the creation of a new block for a dispatch label.
6751 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6753 * invoke.texi (Wsuggest-attribute=cold): Document.
6754 * common.opt (Wsuggest-attribute=cold): New
6755 * ipa-pure-const.c (warn_function_cold): New function.
6756 * predict.c (compute_function_frequency): Use it.
6757 * predict.h (warn_function_cold): Declare.
6759 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6761 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6764 2017-10-06 Martin Liska <mliska@suse.cz>
6766 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6767 keyword for member functions.
6768 (struct sanopt_tree_couple): New struct.
6769 (struct sanopt_tree_couple_hash): New function.
6770 (struct sanopt_ctx): Add new hash_map.
6771 (has_dominating_ubsan_ptr_check): New function.
6772 (record_ubsan_ptr_check_stmt): Likewise.
6773 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6774 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6775 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6777 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6780 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6781 aarch64_simd_valid_immediate on CONST_VECTORs.
6782 (aarch64_reg_or_bic_imm): Likewise.
6784 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6786 PR rtl-optimization/82396
6787 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6789 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6791 * graphite-dependences.c (scop_get_reads): Move code to...
6792 (scop_get_must_writes): Move code to...
6793 (scop_get_may_writes): Move code to...
6794 (scop_get_reads_and_writes): ... here.
6795 (scop_get_dependences): Call scop_get_reads_and_writes.
6797 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6799 PR tree-optimization/82434
6800 * fold-const.h (can_native_encode_type_p,
6801 can_native_encode_string_p): Remove.
6802 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6803 don't encode anything, just return what would be otherwise returned.
6804 (native_encode_fixed, native_encode_complex, native_encode_vector):
6806 (native_encode_string): Likewise. Inline by hand
6807 can_native_encode_string_p.
6808 (can_native_encode_type_p): Remove.
6809 (can_native_encode_string_p): Remove.
6810 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6811 STRING_CSTs using can_native_encode_string_p, test all
6812 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6813 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6814 argument from native_encode_expr.
6815 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6816 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6819 2017-10-06 Richard Biener <rguenther@suse.de>
6821 PR tree-optimization/82397
6822 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6823 operand_equal_p but rely on data_ref_compare_tree for detecting
6825 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6826 to match up with dr_group_sort_cmp.
6828 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6831 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6833 * config/s390/s390-builtin-types.def: Regenerate.
6835 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6838 * config/s390/s390-builtin-types.def: Regenerate.
6839 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6840 Change flag from B_VXE to B_VX.
6841 (s390_vec_min_dbl): Remove B_VXE flag.
6843 2017-10-06 Richard Biener <rguenther@suse.de>
6845 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6846 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6847 translate_isl_ast_to_gimple::is_valid_rename,
6848 translate_isl_ast_to_gimple::get_rename,
6849 translate_isl_ast_to_gimple::get_def_bb_for_const,
6850 translate_isl_ast_to_gimple::get_new_name,
6851 translate_isl_ast_to_gimple::collect_all_ssa_names,
6852 translate_isl_ast_to_gimple::copy_loop_phi_args,
6853 translate_isl_ast_to_gimple::collect_all_ssa_names,
6854 translate_isl_ast_to_gimple::copy_loop_phi_args,
6855 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6856 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6857 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6858 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6859 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6860 translate_isl_ast_to_gimple::copy_cond_phi_args,
6861 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6862 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6863 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6864 translate_isl_ast_to_gimple::rename_uses,
6865 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6866 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6867 (set_rename_for_each_def): Likewise.
6868 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6869 here. Handle rewriting SCEV analyzable uses here.
6870 (copy_bb_and_scalar_dependences): Generate code for PHI
6872 (graphite_regenerate_ast_isl): Adjust.
6873 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6874 (add_write, add_read): New functions.
6875 (build_cross_bb_scalars_def): Use it and simplify.
6876 (build_cross_bb_scalars_use): Likewise.
6877 (graphite_find_cross_bb_scalar_vars): Inline into...
6878 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6879 simulating out-of-SSA. Compute liveout and add dependencies.
6880 (build_scops): Force an empty entry block.
6881 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6883 (sese_build_liveouts): Declare.
6884 (sese_trivially_empty_bb_p): Likewise.
6885 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6886 compute liveout and debug_liveout.
6887 (sese_bad_liveouts_use): Remove.
6888 (sese_reset_debug_liveouts_bb): Likewise.
6889 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6890 (sese_build_liveouts): Build liveout and debug_liveout and store
6892 (new_sese_info): Adjust.
6893 (free_sese_info): Likewise.
6894 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6895 do not build liveout here.
6896 (move_sese_in_condition): Adjust region entry.
6897 (scev_analyzable_p): Match up with chrec_apply requirements.
6898 (sese_trivially_empty_bb_p): New.
6899 * tree-into-ssa.c (get_reaching_def): Properly support generating
6900 default-defs for incremental rewrite of anonymous names.
6902 2017-10-06 Richard Biener <rguenther@suse.de>
6904 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6905 precision do not perform modulo reduction.
6907 2017-10-06 Richard Biener <rguenther@suse.de>
6909 PR tree-optimization/82436
6910 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6911 conservatively choose the vectorization factor when checking
6912 whether we can perform the required load permutation.
6913 (vect_transform_slp_perm_load): Assert when we may not fail.
6915 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6917 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6918 message for incompatible -msdata=* and -mcall-* options.
6920 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6922 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6923 rate for post-reload scheduling.
6925 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6927 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6929 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6931 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6932 to improve monte carlo in scimark.
6934 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6936 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6937 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6938 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6939 pentium4_cost, nocona_cost): Set reassociation width to 1.
6940 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6941 width to 2 for fp operations and 1 otherwise.
6942 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6944 (atom_cost): Set reassociation width to 2.
6945 (slm_cost, generic_cost): Set fp reassociation width
6946 to 2 and 1 otherwise.
6947 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6948 (core_cost): Set fp reassociation width to 4 and vector to 2.
6949 (ix86_reassociation_width): Rewrite using cost table; special case
6950 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6951 and TARGET_AVX128_OPTIMAL.
6952 * config/i386/i386.h (processor_costs): Add
6953 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6954 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6955 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6956 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6957 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6958 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6960 2017-10-05 Nathan Sidwell <nathan@acm.org>
6962 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6964 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6966 * config/arm/arm.c (arm_test_fpu_data): New.
6967 (arm_run_selftests): Call arm_test_fpu_data.
6969 2017-10-04 Nathan Sidwell <nathan@acm.org>
6971 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6973 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6975 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6977 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6978 check type for aarch64_simd_valid_immediate.
6979 (aarch64_output_simd_mov_immediate): Update prototype.
6980 (aarch64_simd_valid_immediate): Update prototype.
6981 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6982 support for ORR-immediate.
6983 (and<mode>3): modified pattern to add support for BIC-immediate.
6984 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6985 now checks for valid immediate for BIC and ORR based on new enum
6987 (aarch64_output_simd_mov_immediate): Function now used to output
6988 BIC/ORR imm as well based on new enum argument.
6989 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6991 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6992 (aarch64_reg_or_bic_imm): Likewise.
6994 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6996 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6997 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6999 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7003 PR rtl-optimization/82396
7004 * haifa-sched.c (autopref_multipass_init): Simplify
7006 (autopref_rank_data): Simplify sort order.
7007 * sched-int.h (autopref_multipass_data_): Remove
7008 multi_mem_insn_p, min_offset and max_offset.
7010 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7012 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7014 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7016 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7019 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7021 * doc/sourcebuild.texi: Document vect_long_mult.
7023 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7025 PR tree-optimization/82413
7026 * fold-const.c (build_range_check): Use widest_int when comparing
7027 the maximum ETYPE value with HIGH.
7029 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7031 PR rtl-optimization/82396
7032 * haifa-sched.c (autopref_multipass_init): Simplify
7034 (autopref_rank_data): Simplify sort order.
7035 * sched-int.h (autopref_multipass_data_): Remove
7036 multi_mem_insn_p, min_offset and max_offset.
7038 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7040 PR tree-optimization/82381
7041 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7042 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7045 PR tree-optimization/82374
7046 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7047 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7048 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7049 current_function_decl to the new decl.
7051 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7053 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7054 helper macro for IEEE float128 hardware built-in functions.
7055 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7057 (TRUNCF128_ODD): Likewise.
7058 (ADDF128_ODD): Likewise.
7059 (SUBF128_ODD): Likewise.
7060 (MULF128_ODD): Likewise.
7061 (DIVF128_ODD): Likewise.
7062 (FMAF128_ODD): Likewise.
7063 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7064 UNSPEC_TRUNC_ROUND_TO_ODD.
7065 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7066 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7067 floating point round to odd instructions.
7068 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7069 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7070 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7071 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7072 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7073 (trunc<mode>sf2_hw): Change the truncate with round to odd
7074 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7075 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7076 to odd hardware instructions.
7077 (sub<mode>3_odd): Likewise.
7078 (mul<mode>3_odd): Likewise.
7079 (div<mode>3_odd): Likewise.
7080 (sqrt<mode>2_odd): Likewise.
7081 (fma<mode>4_odd): Likewise.
7082 (fms<mode>4_odd): Likewise.
7083 (nfma<mode>4_odd): Likewise.
7084 (nfms<mode>4_odd): Likewise.
7085 (trunc<mode>df2_odd): Change the truncate with round to odd
7086 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7088 * doc/extend.texi (PowerPC built-in functions): Update documentation
7089 for existing IEEE float128-bit built-in functions. Add built-in
7090 functions that generate the IEEE 128-bit floating point round to
7093 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7095 PR rtl-optimization/77729
7096 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7097 to (X&(C1&~C2))|C2 transformations.
7099 2017-10-03 Martin Jambor <mjambor@suse.cz>
7101 PR tree-optimization/82363
7102 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7103 mismatch, mark lacc written regardless of racc.
7105 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7107 PR tree-optimization/82381
7108 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7109 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7110 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7111 fallthrough into reassoc_stmt_dominates_stmt_p.
7114 * combine.c (combine_instructions): Don't combine in unreachable
7117 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7120 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7121 function to not use the have_cpu variable. Do not set cpu_index,
7122 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7124 (rs6000_valid_attribute_p): Remove duplicate initializations of
7125 old_optimize and func_optimize.
7126 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7127 (rs6000_activate_target_options): Make global.
7128 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7131 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7133 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7134 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7135 Set *poffset to 0 and *psize and *pmax_size to -1 if
7136 *poffset + *psize overflows in HOST_WIDE_INT.
7138 PR tree-optimization/82387
7139 PR tree-optimization/82388
7140 PR tree-optimization/82389
7141 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7142 instead of live_bytes non-NULL.
7144 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7147 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7150 2017-10-02 Richard Biener <rguenther@suse.de>
7152 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7153 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7154 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7156 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7158 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7159 requested precision matches the type's.
7160 * calls.c (alloc_max_size): Calculate the new candidate size as
7161 a widest_int and use wi::to_widest when comparing it with the
7162 current candidate size.
7163 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7164 zero rather than integer_zero_node.
7165 * match.pd: Check for a no-op conversion before using wi::add
7166 rather than after. Use tree_to_uhwi when summing small shift
7167 counts into an unsigned int.
7169 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7170 Alan Hayward <alan.hayward@arm.com>
7171 David Sherwood <david.sherwood@arm.com>
7174 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7175 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7176 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7177 POINTER_AND_FP_REGS.
7179 2017-10-02 Richard Biener <rguenther@suse.de>
7181 PR tree-optimization/82355
7182 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7183 a mapping for the enclosing loop but avoid generating one for
7185 (copy_bb_and_scalar_dependences): Remove premature codegen
7186 error on PHIs in blocks duplicated into multiple places.
7187 * graphite-scop-detection.c
7188 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7189 in the region use it as loop and nest to analyze the DR in.
7190 (try_generate_gimple_bb): Likewise.
7191 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7192 (add_loop_constraints): For blocks in a loop not in the region
7193 create a dimension with a single iteration.
7194 * sese.h (gbb_loop_at_index): Remove assert.
7196 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7198 * omp-expand.c (adjust_context_scope): New function.
7199 (expand_parallel_call): Call adjust_context_scope.
7201 2017-10-01 Jeff Law <law@redhat.com>
7203 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7204 dom_opt_dom_walker class with direct access to private members.
7205 Add comments. Call test_for_singularity.
7206 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7207 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7208 m_dummy_cond anymore.
7209 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7211 (pass_dominator:execute): Build the dummy_cond here and pass it
7212 to the dom_opt_dom_walker ctor.
7213 (test_for_singularity): New function.
7215 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7216 Maya Rashish <coypu@sdf.org>
7218 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7219 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7220 (alpha*-*-netbsd*) Use nbsd_tm_file.
7221 (arm*-*-netbsdelf*) Likewise.
7222 (i[34567]86-*-netbsdelf*) Likewise.
7223 (x86_64-*-netbsd*) Likewise.
7224 (mips*-*-netbsd*) Likewise.
7225 (powerpc-*-netbsd*) Likewise.
7226 (sh*-*-netbsd*) Likewise.
7227 (sparc-*-netbsdelf*) Likewise.
7228 (sparc64-*-netbsd*) Likewise.
7229 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7231 (vax-*-netbsdelf*) Likewise.
7232 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7233 (UINT_FAST8_TYPE) Likewise.
7234 (INT_FAST16_TYPE) Check CHAR_FAST16.
7235 (UINT_FAST16_TYPE) Likewise.
7237 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7240 * config/i386/i386.md
7241 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7242 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7243 *divmodsi4_zext_2): New define_insn_and_split.
7244 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7245 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7246 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7247 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7248 New define_insn_and_split.
7249 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7250 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7251 operands[1] having DImode when mode is SImode.
7253 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7254 always SImode for DIV and MOD in REG_EQUAL notes.
7256 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7259 * match.pd: Fix handling of NaNs in pattern.
7261 2017-09-29 Jeff Law <law@redhat.com>
7263 * sbitmap.c (bitmap_bit_in_range_p): New function.
7264 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7265 * tree-ssa-dse.c (live_bytes_read): New function.
7266 (dse_classify_store): Ignore reads of dead bytes.
7268 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7269 typos and whitespace errors.
7270 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7271 * config/s390/s390.c (s390_emit_prologue): Likewise.
7273 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7276 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7277 with a symbol for LRA.
7279 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7281 PR rtl-optimization/82338
7282 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7284 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7286 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7287 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7288 (qsort_chk): Declare.
7289 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7290 (qsort_chk): New function.
7292 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7294 PR tree-optimization/82337
7295 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7296 phi definition if the PHI result appears in an abnormal PHI.
7297 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7298 basis appears in an abnormal PHI.
7300 2017-09-29 Richard Biener <rguenther@suse.de>
7302 * graphite-isl-ast-to-gimple.c
7303 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7304 (binary_op_to_tree): Use it.
7305 (get_rename_from_scev): Likewise.
7306 (copy_loop_phi_nodes): Likewise.
7307 (copy_bb_and_scalar_dependences): Likewise.
7308 (translate_pending_phi_nodes): Likewise.
7310 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7313 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7314 for movabsq $(i32 << shift), r64.
7316 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7318 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7319 index when encoding %esp as %rsp to avoid 0x67 prefix.
7321 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7323 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7324 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7326 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7328 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7329 Extensions with more than 16 double VFP registers.
7330 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7331 to_clear_mask and all code related to it. Replace the remaining
7332 entry by a sbitmap and adapt code accordingly.
7334 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7336 * brig-builtins.def: Change pure attributes to const.
7338 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7340 * config.gcc (default_gnu_indirect_function): Default to yes for
7341 sparc*-*-linux* with glibc.
7343 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7345 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7346 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7347 when creating .init_array and .fini_array sections with priority
7350 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7353 * config/aarch64/aarch64.c
7354 (aarch64_builtin_support_vector_misalignment): Always return false
7355 when misalignment is unknown.
7357 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7359 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7360 this function to return false if the definition used by the swap
7361 instruction is artificial, or if the memory address from which the
7362 constant value is loaded is not represented by a base address held
7363 in a register or if the base address register is a frame or stack
7364 pointer. Additionally, return false if the base address of the
7365 loaded constant is a SYMBOL_REF but is not considered to be a
7367 (replace_swapped_load_constant): New function.
7368 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7369 loaded constant vector with a load of a swapped constant vector.
7371 2017-09-27 Carl Love <cel@us.ibm.com>
7373 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7374 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7375 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7378 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7380 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7381 first, always call autopref_rank_data otherwise.
7383 2017-09-27 Richard Biener <rguenther@suse.de>
7385 * graphite-scop-detection.c (find_scop_parameters): Move
7386 loop bound handling ...
7387 (gather_bbs::before_dom_children): ... here, avoiding the need
7388 to build scop_info->loop_nest.
7389 (record_loop_in_sese): Remove.
7390 * sese.h (sese_info_t::loop_nest): Remove.
7391 * sese.c (new_sese_info): Do not allocate loop_nest.
7392 (free_sese_info): Do not free loop_nest.
7394 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7397 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7398 lhs from calls if the lhs has addressable type.
7400 2017-09-27 Richard Biener <rguenther@suse.de>
7402 * graphite.h (scop::max_alias_set): New member.
7403 * graphite-scop-detection.c: Remove references to non-existing
7404 --param in comments.
7405 (build_alias_sets): Record the maximum alias set used for drs.
7406 (build_scops): Support zero as unlimited for
7407 --param graphite-max-arrays-per-scop.
7408 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7410 (build_poly_sr_1): ... here. Compute alias set based on the
7411 maximum alias set used for drs rather than
7412 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7414 2017-09-27 Richard Biener <rguenther@suse.de>
7416 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7417 --param loop-block-tile-size=0 to disable tiling.
7419 2017-09-27 Richard Biener <rguenther@suse.de>
7421 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7422 (graphite-max-nb-scop-params): Document special value zero.
7423 * domwalk.h (dom_walker::STOP): New symbolical constant.
7424 (dom_walker::dom_walker): Add optional parameter for bb to
7426 (dom_walker::~dom_walker): Declare.
7427 (dom_walker::before_dom_children): Document STOP return value.
7428 (dom_walker::m_user_bb_to_rpo): New member.
7429 (dom_walker::m_bb_to_rpo): Likewise.
7430 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7431 mapping here if not provided by the user.
7432 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7433 provided by the user.
7434 (dom_walker::STOP): Define.
7435 (dom_walker::walk): Do not compute bb to RPO mapping here.
7436 Support STOP return value from before_dom_children to stop
7438 * graphite-optimize-isl.c (optimize_isl): If the schedule
7439 is the same still generate code if -fgraphite-identity
7440 or -floop-parallelize-all are given.
7441 * graphite-scop-detection.c: Include cfganal.h.
7442 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7444 (gather_bbs::before_dom_children): Return STOP for BBs
7446 (build_scops): Compute bb to RPO mapping and pass it to
7447 the domwalk. Treat --param graphite-max-nb-scop-params=0
7448 as not limiting the number of params.
7449 * graphite.c (graphite_initialize): Remove limit on the
7450 number of basic-blocks in a function.
7451 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7452 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7453 default value of 10.
7455 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7457 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7458 Adjust code to eliminate needing to do the shift right 32-bits
7459 operation after XSCVDPSPN.
7461 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7463 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7464 ((X / Y) != 0 -> X >= Y): Likewise.
7466 2017-09-26 Carl Love <cel@us.ibm.com>
7468 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7469 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7470 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7471 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7472 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7473 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7474 definitions and overloading.
7475 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7476 statement for P9V_BUILTIN_XST_LEN_R.
7477 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7478 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7479 define_expand and define_insn for the instructions and builtins.
7480 * doc/extend.texi: Update the built-in documentation file for the new
7482 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7483 define_insn for the instructions
7485 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7488 * gcc/config/netbsd-protos.h: New file.
7489 * gcc/config/netbsd.c: New file.
7490 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7491 * gcc/config/t-netbsd: New file.
7492 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7493 (tmake_file) Add t-netbsd.
7494 (extra_objs) Add netbsd.o.
7496 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7500 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7502 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7504 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7505 sign extension from a vector register to a GPR by doing a 32-bit
7506 direct move and then an EXTSW.
7507 (extendsi<mode>2 splitter): Likewise.
7508 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7509 right or vector extract after doing XSCVDPSPN. Use
7510 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7512 (movdi_from_sf_zero_ext): Likewise.
7513 (reload_gpr_from_vsxsf): Likewise.
7514 (p8_mfvsrd_4_disf): Delete, no longer used.
7515 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7516 then needing to move the SFmode to a GPR to use the XSCVDPSP
7517 instruction instead of FRSP and XSCVDPSPN.
7518 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7519 it is adjacent to the other XSCVSPDP insns.
7520 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7521 SFmode to be in traditional Altivec registers.
7522 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7523 (vsx_xscvspdpn): Likewise.
7524 (vsx_xscvspdpn_scalar): Likewise.
7526 2017-09-26 Martin Jambor <mjambor@suse.cz>
7528 * tree-sra.c (compare_access_positions): Put integral types first,
7529 stabilize sorting of integral types, remove conditions putting
7530 non-full-precision integers last.
7531 (sort_and_splice_var_accesses): Disable scalarization if a
7532 non-integert would be represented by a non-full-precision integer.
7534 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7536 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7537 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7538 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7539 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7540 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7541 conditionals inside the function instead of around it. Call
7542 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7543 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7545 2017-09-26 Richard Biener <rguenther@suse.de>
7547 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7549 (scop_detection::build_scop_breadth): ... this. Removed.
7550 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7551 (scop_detection::harmful_stmt_in_bb): Likewise.
7552 (scop_detection::graphite_can_represent_stmt): Likewise.
7553 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7554 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7555 (scop_detection::can_represent_loop_1): ... this. Removed.
7556 (scop_detection::harmful_loop_in_region): Simplify after inlining
7557 the above and remove more quadraticness.
7558 (build_scops): Adjust.
7559 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7562 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7565 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7566 REGNO (base) == SP_REG if base is a REG.
7569 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7570 if it is different SSA_NAME.
7571 (optimize_range_tests_cmp_bitwise): New function.
7572 (optimize_range_tests): Call it.
7574 2017-09-26 Richard Biener <rguenther@suse.de>
7576 PR tree-optimization/82321
7577 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7578 for the def being inside the loop.
7580 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7582 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7584 * config/s390/s390-builtins.def: Fix constraint on op4.
7586 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7588 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7589 independent expanders.
7590 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7591 ("vec_ordered", "vec_unordered"): New expanders.
7593 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7595 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7598 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7600 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7601 vec_unpacks_lo_v16qi.
7602 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7604 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7606 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7607 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7608 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7610 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7612 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7614 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7616 ("*vec_slb<mode>"): New insn pattern.
7617 ("vec_shr_<mode>"): New expander.
7618 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7619 and force the shift count operand to V16QImode.
7620 ("vec_srb<mode>"): Set shift count mode to V16QI.
7622 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7624 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7625 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7626 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7628 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7631 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7633 2017-09-26 Richard Biener <rguenther@suse.de>
7635 PR tree-optimization/82320
7636 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7639 2017-09-25 Jeff Law <law@redhat.com>
7641 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7642 prototype for new argument.
7643 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7644 mostly extracted from rs6000_emit_allocate_stack.
7645 (rs6000_emit_probe_stack_range_stack_clash): New function.
7646 (rs6000_emit_allocate_stack): Call
7647 rs6000_emit_probe_stack_range_stack_clash as needed.
7648 (rs6000_emit_probe_stack_range): Add additional argument
7649 to call to gen_probe_stack_range{si,di}.
7650 (output_probe_stack_range): New.
7651 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7652 (output_probe_stack_range_stack_clash): New.
7653 (rs6000_emit_prologue): Emit notes into dump file as requested.
7654 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7655 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7656 Add additional operand and pass it to output_probe_stack_range.
7658 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7660 PR tree-optimization/82163
7661 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7662 (checking_verify_loop_closed_ssa): New parameter.
7663 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7664 (check_loop_closed_ssa_stmt): Delete.
7665 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7666 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7667 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7670 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7672 * brig-builtins.def: Treat HSAIL barrier builtins as
7673 setjmp/longjump style functions.
7675 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7677 * target.def (constant_alignment): New hook.
7678 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7679 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7680 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7681 * doc/tm.texi: Regenerate.
7682 * targhooks.h (default_constant_alignment): Declare.
7683 (constant_alignment_word_strings): Likewise.
7684 * targhooks.c (default_constant_alignment): New function.
7685 (constant_alignment_word_strings): Likewise.
7686 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7687 instead of CONSTANT_ALIGNMENT.
7688 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7689 (force_const_mem): Likewise.
7690 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7691 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7692 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7693 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7694 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7696 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7697 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7698 constant_alignment_word_strings.
7699 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7700 (CONSTANT_ALIGNMENT): Likewise.
7701 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7702 (arm_constant_alignment): New function.
7703 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7704 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7705 constant_alignment_word_strings.
7706 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7707 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7708 constant_alignment_word_strings.
7709 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7710 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7711 (cris_constant_alignment): New function.
7712 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7713 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7714 (epiphany_constant_alignment): New function.
7715 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7716 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7717 constant_alignment_word_strings.
7718 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7719 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7720 constant_alignment_word_strings.
7721 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7722 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7723 constant_alignment_word_strings.
7724 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7725 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7726 * config/i386/i386.c (ix86_constant_alignment): Make static.
7727 Use the same interface as the target hook.
7728 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7729 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7730 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7731 constant_alignment_word_strings.
7732 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7733 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7734 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7735 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7736 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7737 constant_alignment_word_strings.
7738 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7739 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7740 constant_alignment_word_strings.
7741 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7742 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7743 constant_alignment_word_strings.
7744 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7745 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7747 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7748 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7749 * config/mips/mips.c (mips_constant_alignment): New function.
7750 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7751 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7752 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7753 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7754 (mmix_constant_alignment): Make static. Use the same interface
7756 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7757 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7758 constant_alignment_word_strings.
7759 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7760 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7761 constant_alignment_word_strings.
7762 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7763 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7764 constant_alignment_word_strings.
7765 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7766 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7767 (rs6000_constant_alignment): New function.
7768 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7769 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7770 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7771 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7772 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7773 (rs6000_constant_alignment): New function.
7774 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7775 * config/s390/s390.c (s390_constant_alignment): New function.
7776 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7777 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7778 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7779 constant_alignment_word_strings.
7780 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7781 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7782 (sparc_constant_alignment): New function.
7783 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7784 * config/spu/spu.c (spu_constant_alignment): New function.
7785 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7786 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7787 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7788 constant_alignment_word_strings.
7789 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7790 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7791 constant_alignment_word_strings.
7792 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7793 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7794 constant_alignment_word_strings.
7795 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7796 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7797 (visium_constant_alignment): New function.
7798 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7799 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7800 (xtensa_constant_alignment): New function.
7801 * system.h (CONSTANT_ALIGNMENT): Poison.
7803 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7805 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7806 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7807 (rs6000_builtin_valid_without_lhs): New helper function.
7808 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7809 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7811 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7813 * target.h (vec_perm_indices): Use unsigned short rather than
7815 (auto_vec_perm_indices): Likewise.
7816 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7817 Use unsigned int rather than unsigned char.
7818 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7820 2017-09-25 Richard Biener <rguenther@suse.de>
7822 * cfgloop.h (sort_sibling_loops): Declare.
7823 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7824 (sort_sibling_loops): New function sorting the sibling loop list
7826 * graphite.c (graphite_transform_loops): Sort sibling loops.
7828 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7830 * target.def (vec_perm_const_ok): Change sel parameter to
7832 * optabs-query.c (can_vec_perm_p): Update accordingly.
7833 * doc/tm.texi: Regenerate.
7834 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7835 auto_vec_perm_indices and remove separate nelt field.
7836 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7837 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7838 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7839 (aarch64_expand_vec_perm_const): Update accordingly.
7840 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7841 to vec_perm_indices.
7842 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7843 auto_vec_perm_indices and remove separate nelt field.
7844 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7845 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7846 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7848 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7849 to vec_perm_indices.
7850 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7851 sel to vec_perm_indices.
7852 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7853 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7854 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7856 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7859 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7862 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7863 on the FUNCTION_DECL function context if it has a DIE that is a
7866 2017-09-25 Richard Biener <rguenther@suse.de>
7868 PR tree-optimization/82285
7869 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7872 2017-09-25 Tom de Vries <tom@codesourcery.com>
7876 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7879 2017-09-25 Richard Biener <rguenther@suse.de>
7881 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7882 ISL errors other than isl_error_quota happen. Dump if the
7883 schedule is the same.
7884 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7885 errors instead of aborting inside ISL.
7887 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7890 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7891 of libgcc_eh for m64.
7892 * config/i386/darwin64.h: Likewise.
7894 2017-09-25 Richard Biener <rguenther@suse.de>
7897 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7898 attribute for incomplete types nor twice for complete ones.
7900 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7903 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7904 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7907 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7910 * config/i386/i386.opt (mprefer-avx256): Use
7911 ix86_target_flags variable.
7912 * config/i386/i386.c (ix86_target_string): Move
7913 -mprefer-avx256 to flag2_opts.
7915 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7918 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7919 and x != -1 | y != -1 into (x & y) != -1.
7921 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7923 * config.gcc: Add new case statement to set
7924 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7925 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7926 s390x-*-linux* case statements. Added aarch64 to the list of
7927 supported architectures.
7929 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7931 PR tree-optimization/82289
7932 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7933 STMT_VINFO_RELEVANT_P.
7935 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7936 Alan Hayward <alan.hayward@arm.com>
7937 David Sherwood <david.sherwood@arm.com>
7939 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7940 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7941 (extract_range_from_binary_expr_1): Don't call
7942 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7944 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7945 Alan Hayward <alan.hayward@arm.com>
7946 David Sherwood <david.sherwood@arm.com>
7948 * target.def (preferred_vector_alignment): New hook.
7949 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7951 * doc/tm.texi: Regenerate.
7952 * targhooks.h (default_preferred_vector_alignment): Declare.
7953 * targhooks.c (default_preferred_vector_alignment): New function.
7954 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7956 (DR_TARGET_ALIGNMENT): New macro.
7957 (aligned_access_p): Update commentary.
7958 (vect_known_alignment_in_bytes): New function.
7959 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7961 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7962 Calculate the misalignment based on the target alignment rather than
7964 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7965 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7966 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7967 the target alignment, rather than masking the element misalignment
7968 with the number of elements in a vector. Also use the target
7969 alignment when calculating the maximum number of peels.
7970 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7971 instead of TYPE_ALIGN_UNIT.
7972 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7973 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7974 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7975 (vect_create_data_ref_ptr): Likewise.
7976 (vect_setup_realignment): Realign by ANDing with
7977 -DR_TARGET_MISALIGNMENT.
7978 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7979 the number of peels based on DR_TARGET_ALIGNMENT.
7980 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7981 with the guaranteed alignment boundary when deciding whether
7983 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7984 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7985 (ensure_base_align): Remove stmt_info parameter. Get the
7986 target base alignment from DR_TARGET_ALIGNMENT.
7987 (vectorizable_store): Update call accordingly. Interpret
7988 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7990 (vectorizable_load): Likewise.
7992 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7993 Alan Hayward <alan.hayward@arm.com>
7994 David Sherwood <david.sherwood@arm.com>
7996 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7997 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7998 (vect_enhance_data_refs_alignment): Likewise.
8000 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8002 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8003 error. Only quit immediately if parsing is complete.
8004 (BEGIN): Initialize fatal_err and parse_done.
8005 (begin fpu, end fpu): Check number of arguments.
8006 (begin arch, end arch): Likewise.
8007 (begin cpu, end cpu): Likewise.
8008 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8009 (optalias): Likewise.
8011 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8013 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8014 * config/arm/arm-isa.h: Delete. Move definitions to ...
8015 * arm-cpus.in: ... here. Use new feature and fgroup values.
8016 * config/arm/arm.c (arm_option_override): Use lower case for feature
8018 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8019 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8020 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8022 (print_isa_bits_for): New function.
8023 (gen_isa): New function.
8024 (gen_comm_data): Use print_isa_bits_for.
8025 (define feature): New keyword.
8026 (define fgroup): New keyword.
8027 * config/arm/t-arm (TM_H): Remove.
8028 (GTM_H): Add arm-isa.h.
8029 (arm-isa.h): Add rule to generate file.
8030 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8031 case for feature bit names.
8033 2017-09-22 Richard Biener <rguenther@suse.de>
8035 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8037 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8038 print of no dependency loops ...
8039 * graphite.c (graphite_transform_loops): ... here.
8040 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8042 (same_close_phi_node, remove_duplicate_close_phi,
8043 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8044 (canonicalize_loop_closed_ssa): ... here and simplify.
8045 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8046 (optimize_isl): Use dump_printf_loc to tell when we stopped
8047 optimizing because of an ISL timeout.
8049 2017-09-22 Richard Biener <rguenther@suse.de>
8051 PR tree-optimization/82291
8052 * tree-if-conv.c (predicate_mem_writes): Make sure to
8053 remove writes in blocks predicated with false.
8055 2017-09-22 Richard Biener <rguenther@suse.de>
8057 * sese.c: Include cfganal.h.
8058 (if_region_set_false_region): Remove.
8059 (create_if_region_on_edge): Likewise.
8060 (move_sese_in_condition): Re-implement without destroying
8063 2017-09-22 Richard Biener <rguenther@suse.de>
8065 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8066 Verify both BBs contain loop PHI nodes before dispatching to
8068 (graphite_regenerate_ast_isl): Do not recompute dominators,
8069 do not verify three times. Restructure for clarity.
8070 * graphite-scop-detection.c (same_close_phi_node,
8071 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8072 defined_in_loop_p, canonicalize_loop_closed_ssa,
8073 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8074 checking and SSA rewrite, move to ...
8075 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8076 (graphite_initialize): Do not pass in ctx, do not reset the
8077 SCEV cache, compute only dominators.
8078 (graphite_transform_loops): Allocate ISL ctx after
8079 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8080 Maintain post-dominators only around build_scops.
8081 * sese.c (if_region_set_false_region): Make static. Free
8082 and recompute dominators.
8083 (move_sese_in_condition): Assert we don't get called with
8084 post-dominators computed.
8085 * sese.h (if_region_set_false_region): Remove.
8087 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8089 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8090 mode attribute for TARGET_AVX512VL.
8092 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8094 * config/i386/i386.opt (mprefer-avx256): New option.
8095 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8097 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8098 for TARGET_PREFER_AVX256.
8099 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8101 2017-09-21 Jeff Law <law@redhat.com>
8103 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8104 Fix dump output if the only stack space is for pushed registers.
8106 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8108 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8111 2017-09-21 Martin Sebor <msebor@redhat.com>
8114 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8115 code (in C++) or code that triggers warnings.
8117 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8119 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8121 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8123 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8124 * sched-rgn.c: Ditto.
8125 * sel-sched-ir.c: Ditto.
8127 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8129 * toplev.h (set_random_seed): Adjust return type.
8130 * toplev.c (init_local_tick): Move eager initialization of random_seed
8131 to get_random_seed. Adjust comment.
8132 (init_random_seed): Inline to get_random_seed, delete.
8133 (get_random_seed): Initialize random_seed lazily.
8134 (set_random_seed): Do not return previous value.
8135 (print_switch_value): Do not call get_random_seed.
8137 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8139 * cgraph.c (delete_function_version): New, broken out from...
8140 (cgraph_node::delete_function_version): ...here. Rename to
8141 cgraph_node::delete_function_version_by_decl. Update all uses.
8142 (cgraph_node::remove): Call delete_function_version.
8144 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8147 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8148 VAR_DECLs to which addressable non-volatile parameters are mapped
8149 and for id->retvar after the return value assignment. Clear
8150 id->retval and id->retbnd after inlining.
8152 2017-09-21 Richard Biener <rguenther@suse.de>
8154 PR tree-optimization/82276
8155 PR tree-optimization/82244
8156 * tree-vrp.c (build_assert_expr_for): Set
8157 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8159 (remove_range_assertions): Revert earlier change.
8161 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8164 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8166 2017-09-21 Richard Biener <rguenther@suse.de>
8168 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8169 Restore valid IL after code generation errors.
8170 * graphite.c (graphite_transform_loops): Diagnose code
8171 generation issues as MSG_MISSED_OPTIMIZATION and continue
8172 with processing SCOPs.
8174 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8175 Alan Hayward <alan.hayward@arm.com>
8176 David Sherwood <david.sherwood@arm.com>
8178 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8179 rather than choosing between plus_constant and gen_rtx_<CODE>.
8180 * expr.c (emit_push_insn): Likewise.
8181 (expand_expr_real_2): Likewise.
8183 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8184 Alan Hayward <alan.hayward@arm.com>
8185 David Sherwood <david.sherwood@arm.com>
8187 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8189 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8190 Alan Hayward <alan.hayward@arm.com>
8191 David Sherwood <david.sherwood@arm.com>
8193 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8194 calling tree_to_uhwi.
8196 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8197 Alan Hayward <alan.hayward@arm.com>
8198 David Sherwood <david.sherwood@arm.com>
8200 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8201 INTEGER_CST rather than a negative test for ADDR_EXPR.
8203 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8204 Alan Hayward <alan.hayward@arm.com>
8205 David Sherwood <david.sherwood@arm.com>
8207 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8208 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8210 2017-09-21 Richard Biener <rguenther@suse.de>
8212 PR tree-optimization/71351
8213 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8214 graphite_create_new_loop_guard): Remove, fold remaining parts
8216 (translate_isl_ast_node_for): ... here and simplify.
8218 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8221 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8222 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8223 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8224 alternative always use QI mode, for -Os imov (=R,R) alternative
8225 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8228 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8229 Jeff Law <law@redhat.com>
8231 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8232 (allocate_stack_space): New function, partially extracted from
8234 (s390_emit_prologue): Track offset to most recent stack probe.
8235 Code to allocate space moved into allocate_stack_space.
8236 Dump actions when no stack is allocated.
8237 (s390_prologue_plus_offset): New function.
8238 (s390_emit_stack_probe): Likewise.
8240 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8242 * common.opt (Wa, Wl, Wp, g, gz=): Add
8244 (gno-column-info): Remove.
8245 (gcolumn-info): Drop RejectNegative.
8247 (gno-record-gcc-switches): Remove.
8248 (grecord-gcc-switches): Drop RejectNegative.
8249 (gno-split-dwarf): Remove.
8250 (gsplit-dwarf): Drop RejectNegative.
8251 (gno-strict-dwarf): Remove.
8252 (gstrict-dwarf): Drop RejectNegative.
8253 * config/darwin.opt (gfull, gused): Add RejectNegative.
8254 * dwarf2out.c (gen_producer_string): Drop
8255 gno-record-gcc-switches handler.
8256 * optc-gen.awk: Add g to prefixes with negative forms.
8257 * opts-common.c (remapping_prefix_p): New.
8258 (find_opt): Check it.
8259 (generate_canonical_option): Test g prefix.
8260 (option_map): Add -gno- mapping.
8261 (add_misspelling_candidates): Check remapping_prefix_p.
8263 2017-09-20 Jeff Law <law@redhat.com>
8265 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8266 thinko in stack clash protection support.
8268 * explow.c (compute_stack_clash_protection_loop_data): Use
8269 CONST_INT_P instead of explicit test. Verify object is a
8270 CONST_INT_P before looking at INTVAL.
8271 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8272 instead of explicit test.
8274 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8277 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8278 address instead of to r1 and r11.
8280 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8282 * config.gcc: Support "knm".
8283 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8284 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8286 * config/i386/i386.c (m_KNM): Define.
8287 (processor_target_table): Add "knm".
8289 (ix86_option_override_internal): Add "knm".
8290 (ix86_issue_rate): Add PROCESSOR_KNM.
8291 (ix86_adjust_cost): Ditto.
8292 (ia32_multipass_dfa_lookahead): Ditto.
8293 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8294 (fold_builtin_cpu): Add M_INTEL_KNM.
8295 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8296 (processor_type): Add PROCESSOR_KNM.
8297 * config/i386/x86-tune.def: Add m_KNM.
8298 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8300 2017-09-20 Richard Biener <rguenther@suse.de>
8302 PR tree-optimization/80213
8303 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8304 are allowed in empty BBs as well.
8305 (canonicalize_loop_closed_ssa): Also look for other complex
8307 (scop_detection::get_sese): Include the loop-closed PHI block
8309 (scop_detection::merge_sese): Remove code adding extra blocks.
8310 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8311 (build_scops): Assert the final returned scop is invalid.
8313 2017-09-20 Richard Biener <rguenther@suse.de>
8315 PR tree-optimization/82264
8316 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8318 (vn_phi_lookup): Likewise.
8319 (vn_phi_insert): Likewise.
8321 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8323 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8324 that fits into uhwi or shwi, add DW_AT_const_value regardless
8325 of early_dwarf without going through RTL, using add_AT_unsigned
8328 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8329 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8330 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8331 (DEBUG_MACRO_SECTION): Likewise.
8332 (DEBUG_LTO_MACRO_SECTION): Likewise.
8333 (DEBUG_STR_DWO_SECTION): Likewise.
8334 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8335 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8336 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8337 (DEBUG_LTO_LINE_STR_SECTION): Define.
8338 (init_sections_and_labels): Initialize debug_line_str_section
8339 variable. Initialize debug_loc_section for -gdwarf-5 to
8340 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8342 2017-09-20 Richard Biener <rguenther@suse.de>
8344 * graphite-sese-to-poly.c (extract_affine): Properly handle
8345 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8347 2017-09-20 Richard Biener <rguenther@suse.de>
8349 PR tree-optimization/81373
8350 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8351 Force SESE live-out defs to be handled even if they are
8354 2017-09-19 Jeff Law <law@redhat.com>
8356 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8357 nothing for stack adjustments with REG_STACK_CHECK.
8358 * sched-deps.c (parse_add_or_inc): Reject insns with
8359 REG_STACK_CHECK from dependency breaking.
8360 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8361 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8362 * reg-notes.def (STACK_CHECK): New note.
8364 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8365 (ix86_expand_prologue): Dump stack clash info as needed.
8366 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8368 * function.c (dump_stack_clash_frame_info): New function.
8369 * function.h (dump_stack_clash_frame_info): Prototype.
8370 (enum stack_clash_probes): New enum.
8372 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8373 flag_stack_clash_protection.
8374 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8375 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8376 (arm_frame_pointer_required): Likewise.
8377 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8378 (ia64_expand_prologue): Likewise.
8379 * config/mips/mips.c (mips_expand_prologue): Likewise.
8380 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8381 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8382 (sparc_flat_expand_prologue): Likewise.
8383 * config/spu/spu.c (spu_expand_prologue): Likewise.
8385 * explow.c: Include "params.h".
8386 (anti_adjust_stack_and_probe_stack_clash): New function.
8387 (get_stack_check_protect): Likewise.
8388 (compute_stack_clash_protection_loop_data): Likewise.
8389 (emit_stack_clash_protection_loop_start): Likewise.
8390 (emit_stack_clash_protection_loop_end): Likewise.
8391 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8392 Use anti_adjust_stack_and_probe_stack_clash.
8393 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8394 (emit_stack_clash_protection_loop_start): Likewise.
8395 (emit_stack_clash_protection_loop_end): Likewise.
8396 * rtl.h (get_stack_check_protect): Prototype.
8397 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8398 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8399 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8401 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8403 * doc/tm.texi: Rebuilt.
8404 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8405 get_stack_check_protect.
8406 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8407 * config/arm/arm.c (arm_expand_prologue): Likewise.
8408 (arm_frame_pointer_required): Likewise.
8409 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8410 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8411 * config/mips/mips.c (mips_expand_prologue): Likewise.
8412 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8413 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8414 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8415 (sparc_flat_expand_prologue): Likewise.
8417 * common.opt (-fstack-clash-protection): New option.
8418 * flag-types.h (enum stack_check_type): Note difference between
8419 -fstack-check= and -fstack-clash-protection.
8420 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8421 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8422 * toplev.c (process_options): Issue warnings/errors for cases
8423 not handled with -fstack-clash-protection.
8424 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8425 (-fstack-check): Note additional problem with -fstack-check=generic.
8426 Note that -fstack-check is primarily for Ada and refer users
8427 to -fstack-clash-protection for stack-clash-protection.
8428 Document new params for stack clash protection.
8430 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8432 * config/i386/i386.c (ix86_split_long_move): Do not handle
8433 address used for LEA in a special way.
8435 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8437 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8439 2017-09-19 Martin Sebor <msebor@redhat.com>
8442 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8443 of incompatible types.
8445 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8447 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8448 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8449 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8450 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8452 2017-09-19 Richard Biener <rguenther@suse.de>
8454 PR tree-optimization/82244
8455 * tree-vrp.c (remove_range_assertions): Do not propagate
8456 a constant to abnormals but replace the assert with a copy.
8458 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8460 PR rtl-optimization/57878
8461 PR rtl-optimization/68988
8462 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8463 avoidance test involving non_reload_pseudos. Move frequency test
8464 below the general fragmentation avoidance test.
8466 2017-09-19 Richard Biener <rguenther@suse.de>
8468 PR tree-optimization/69728
8469 * graphite-sese-to-poly.c (schedule_error): New global.
8470 (add_loop_schedule): Handle empty domain by failing the
8472 (build_original_schedule): Handle schedule_error.
8474 2017-09-19 Richard Biener <rguenther@suse.de>
8476 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8477 Do not iterate to sibling loops but only to siblings of inner
8480 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8483 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8484 that the registers are different.
8486 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8488 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8489 to processor_model and "amdfam17h" to arch_names_table.
8490 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8492 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8495 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8497 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8498 Alan Hayward <alan.hayward@arm.com>
8499 David Sherwood <david.sherwood@arm.com>
8501 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8503 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8505 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8506 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8507 vect_schedule_slp_instance.
8508 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8509 Update call to vect_slp_analyze_node_operations. Simplify return
8511 (vect_slp_analyze_bb_1): Update call accordingly.
8512 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8513 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8514 (vect_schedule_slp): Update call accordingly.
8516 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8517 Alan Hayward <alan.hayward@arm.com>
8518 David Sherwood <david.sherwood@arm.com>
8520 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8521 with types that aren't in fact scalar.
8523 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8525 * tree-vect-slp.c (vect_record_max_nunits): New function,
8527 (vect_build_slp_tree_1): ...here.
8528 (vect_build_slp_tree_2): Call it for phis too.
8530 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8532 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8533 to vect_get_vec_def_for_operand when getting the mask operand.
8535 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8536 Alan Hayward <alan.hayward@arm.com>
8537 David Sherwood <david.sherwood@arm.com>
8539 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8542 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8543 Alan Hayward <alan.hayward@arm.com>
8544 David Sherwood <david.sherwood@arm.com>
8546 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8547 calculation for vector booleans.
8549 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8550 Alan Hayward <alan.hayward@arm.com>
8551 David Sherwood <david.sherwood@arm.com>
8553 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8555 (vect_transform_stmt): ...here.
8556 (vect_analyze_stmt): Use it instead of calling
8557 vectorizable_live_operation directly.
8559 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8561 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8562 non-SIMT targets in acc vector loops.
8564 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8566 * configure.ac: Add arc and check if assembler supports gdwarf2.
8567 * configure: Regenerate.
8569 2017-09-18 Richard Biener <rguenther@suse.de>
8571 PR tree-optimization/82220
8572 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8573 epilogue niters from the min_profitable_iters compute.
8575 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8578 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8579 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8580 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8582 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8585 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8586 switching to a new text section.
8588 2017-09-18 Richard Biener <rguenther@suse.de>
8590 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8592 (build_alias_set): Reject aliases with no access function.
8594 2017-09-18 Richard Biener <rguenther@suse.de>
8596 PR tree-optimization/79622
8597 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8599 (build_cross_bb_scalars_use): Likewise.
8601 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8603 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8605 2017-09-18 Alan Modra <amodra@gmail.com>
8608 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8609 stack_pointer_rtx for count 0. Update comments. Break up
8610 large rtl expression.
8612 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8614 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8615 Increase to 20 bytes.
8616 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8617 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8618 or avx version of the stub.
8620 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8623 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8624 compute the minimum stack alignment. Also update preferred stack
8625 boundary for leaf functions.
8627 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8629 PR tree-optimization/82228
8630 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8633 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8635 * common/config/nds32/nds32-common.c
8636 (nds32_option_optimization_table): Refine formatting.
8637 (nds32_option_optimization_table): Use -fsched-pressure and
8638 -fomit-frame-pointer for specific optimization level.
8640 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8642 * config/nds32/nds32.c: Refine formatting and comments.
8643 * config/nds32/nds32.h: Likewise.
8644 * config/nds32/nds32.md: Likewise.
8645 * config/nds32/nds32-cost.c: Likewise.
8646 * config/nds32/nds32-isr.c: Likewise.
8647 * config/nds32/nds32-md-auxiliary.c: Likewise.
8648 * config/nds32/nds32-multiple.md: Likewise.
8649 * config/nds32/nds32-predicates.c: Likewise.
8651 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8652 Jakub Jelinek <jakub@redhat.com>
8654 Add support for -std=c++2a.
8655 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8657 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8659 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8662 * doc/extend.texi (Common Function Attributes): Add
8663 references to ARM, AArch64, and S/390 specific attributes.
8664 (Function Specific Option Pragmas): Add AArch64 and S/390
8665 to list of back ends that support the target pragma.
8667 2017-09-15 Nathan Sidwell <nathan@acm.org>
8669 * doc/standards.texi: Fix C++17 description. Update URLs for
8672 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8674 * common.opt (Wcast-align=strict): New warning option.
8675 * doc/invoke.texi: Document -Wcast-align=strict.
8677 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8679 * cgraph.h (cgraph_thunk_info): Add comments.
8680 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8681 assert for VIRTUAL_* arguments stricter.
8683 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8685 PR tree-optimization/71026
8686 * match.pd: Move RDIV patterns from fold-const.c
8687 * fold-const.c (distribute_real_division): Removed.
8688 (fold_binary_loc): Remove calls to distribute_real_divison.
8690 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8692 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8693 c++1z and gnu++1z as deprecated. Change other references to
8694 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8695 Change -Wc++1z-compat to -Wc++17-compat.
8696 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8697 * dwarf2out.c (highest_c_language): Handle C++17.
8698 (gen_compile_unit_die): Likewise.
8700 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8702 PR rtl-optimization/82192
8703 * combine.c (make_extraction): Don't look through non-paradoxical
8704 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8707 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8708 Alan Hayward <alan.hayward@arm.com>
8709 David Sherwood <david.sherwood@arm.com>
8711 * target.def (function_arg_offset): New hook.
8712 * targhooks.h (default_function_arg_offset): Declare.
8713 * targhooks.c (default_function_arg_offset): New function.
8714 * function.c (locate_and_pad_parm): Use
8715 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8716 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8717 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8718 * doc/tm.texi: Regenerate.
8719 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8720 * config/spu/spu.c (spu_function_arg_offset): New function.
8721 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8722 * system.h (FUNCTION_ARG_OFFSET): Poison.
8724 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8725 Alan Hayard <alan.hayward@arm.com>
8726 David Sherwood <david.sherwood@arm.com>
8728 * target.def (truly_noop_truncation): New hook.
8729 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8730 than TRULY_NOOP_TRUNCATION.
8731 * hooks.h (hook_bool_uint_uint_true): Declare.
8732 * hooks.c (hook_bool_uint_uint_true): New function.
8733 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8734 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8735 * doc/tm.texi: Regenerate.
8736 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8737 rather than TRULY_NOOP_TRUNCATION in comments.
8738 (simplify_comparison): Likewise.
8739 (record_truncated_value): Likewise.
8740 * expmed.c (extract_bit_field_1): Likewise.
8741 (extract_split_bit_field): Likewise.
8742 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8743 instead of TRULY_NOOP_TRUNCATION.
8744 * function.c (assign_parm_setup_block): Likewise.
8745 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8746 * rtlhooks.c: Include target.h.
8747 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8748 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8749 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8750 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8751 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8752 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8753 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8754 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8755 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8756 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8757 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8758 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8759 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8760 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8761 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8762 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8763 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8764 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8765 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8766 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8767 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8768 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8769 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8770 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8771 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8772 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8773 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8774 rather than TRULY_NOOP_TRUNCATION in comments.
8775 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8776 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8777 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8778 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8779 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8780 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8781 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8782 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8783 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8784 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8785 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8786 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8787 rather than TRULY_NOOP_TRUNCATION in comments.
8788 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8789 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8790 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8791 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8792 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8793 TRULY_NOOP_TRUNCATION condition.
8794 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8795 (TRULY_NOOP_TRUNCATION): Delete.
8796 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8797 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8798 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8799 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8800 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8801 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8802 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8803 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8804 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8805 rather than TRULY_NOOP_TRUNCATION in comments.
8806 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8807 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8808 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8809 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8810 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8811 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8813 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8816 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8817 (*cmp_ior): Likewise.
8818 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8819 (*ior_scc_scc_cmp): Likewise.
8820 (*and_scc_scc): Likewise.
8821 (*and_scc_scc_cmp): Likewise.
8823 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8824 Alan Hayard <alan.hayward@arm.com>
8825 David Sherwood <david.sherwood@arm.com>
8827 * target.def (can_change_mode_class): New hook.
8828 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8829 (hard_regno_nregs): Likewise.
8830 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8831 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8832 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8833 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8834 (LOAD_EXTEND_OP): Update accordingly.
8835 * doc/tm.texi: Regenerate.
8836 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8837 CANNOT_CHANGE_MODE_CLASS.
8838 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8839 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8840 * combine.c (simplify_set): Update accordingly.
8841 * emit-rtl.c (validate_subreg): Likewise.
8842 * recog.c (general_operand): Likewise.
8843 * regcprop.c (mode_change_ok): Likewise.
8844 * reload1.c (choose_reload_regs): Likewise.
8845 (inherit_piecemeal_p): Likewise.
8846 * rtlanal.c (simplify_subreg_regno): Likewise.
8847 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8848 instead of CANNOT_CHANGE_MODE_CLASS.
8849 (reload_cse_simplify_operands): Likewise.
8850 * reload.c (push_reload): Use targetm.can_change_mode_class
8851 instead of CANNOT_CHANGE_MODE_CLASS.
8852 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8853 REG_CANNOT_CHANGE_MODE_P.
8854 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8855 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8856 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8857 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8858 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8859 (arm_can_change_mode_class): New function.
8860 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8861 than CANNOT_CHANGE_MODE_CLASS in comments.
8862 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8863 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8864 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8865 (ix86_can_change_mode_class): ...this new function, inverting the
8866 sense of the return value.
8867 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8868 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8869 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8870 (ia64_can_change_mode_class): New function.
8871 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8872 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8873 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8874 (m32c_can_change_mode_class): ...this new function, inverting the
8875 sense of the return value.
8876 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8877 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8878 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8879 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8880 (mips_can_change_mode_class): ...this new function, inverting the
8881 sense of the return value.
8882 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8883 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8884 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8885 (msp430_can_change_mode_class): New function.
8886 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8887 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8888 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8889 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8890 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8891 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8892 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8893 (pa_cannot_change_mode_class): Replace with...
8894 (pa_can_change_mode_class): ...this new function, inverting the
8895 sense of the return value.
8896 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8897 than CANNOT_CHANGE_MODE_CLASS in comments.
8898 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8899 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8900 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8901 (pdp11_cannot_change_mode_class): Replace with...
8902 (pdp11_can_change_mode_class): ...this new function, inverting the
8903 sense of the return value.
8904 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8905 * config/powerpcspe/powerpcspe-protos.h
8906 (rs6000_cannot_change_mode_class_ptr): Delete.
8907 * config/powerpcspe/powerpcspe.c
8908 (rs6000_cannot_change_mode_class_ptr): Delete.
8909 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8910 (rs6000_option_override_internal): Assign to
8911 targetm.can_change_mode_class instead of
8912 rs6000_cannot_change_mode_class_ptr.
8913 (rs6000_cannot_change_mode_class): Replace with...
8914 (rs6000_can_change_mode_class): ...this new function, inverting the
8915 sense of the return value.
8916 (rs6000_debug_cannot_change_mode_class): Replace with...
8917 (rs6000_debug_can_change_mode_class): ...this new function.
8918 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8919 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8920 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8921 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8922 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8924 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8925 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8926 (rs6000_option_override_internal): Assign to
8927 targetm.can_change_mode_class instead of
8928 rs6000_cannot_change_mode_class_ptr.
8929 (rs6000_cannot_change_mode_class): Replace with...
8930 (rs6000_can_change_mode_class): ...this new function, inverting the
8931 sense of the return value.
8932 (rs6000_debug_cannot_change_mode_class): Replace with...
8933 (rs6000_debug_can_change_mode_class): ...this new function.
8934 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8935 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8936 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8937 (s390_can_change_mode_class): ...this new function, inverting the
8938 sense of the return value.
8939 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8940 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8941 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8942 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8943 (sh_cannot_change_mode_class): Replace with...
8944 (sh_can_change_mode_class): ...this new function, inverting the
8945 sense of the return value.
8946 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8947 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8948 (sparc_can_change_mode_class): New function.
8949 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8950 * config/spu/spu.c (spu_can_change_mode_class): New function.
8951 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8952 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8953 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8954 (visium_can_change_mode_class): New function.
8955 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8957 2017-09-15 Richard Biener <rguenther@suse.de>
8959 PR tree-optimization/82217
8960 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8961 but not undefined case.
8963 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8966 * postreload.c (reload_cse_simplify_operands): Skip
8967 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8969 2017-09-15 Richard Biener <rguenther@suse.de>
8971 PR tree-optimization/68823
8972 * graphite-scop-detection.c (build_alias_set): If we have a
8973 possible dependence check whether we can handle them by just
8974 looking at the DRs DR_ACCESS_FNs.
8975 (build_scops): If build_alias_set fails, fail the SCOP.
8977 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8979 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8980 to support float128 built-in functions that require the ISA 3.0
8982 (BU_FLOAT128_3_HW): Likewise.
8983 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8985 (FMAF128): Likewise.
8987 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8988 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8989 floating point instructions.
8990 (rs6000_invalid_builtin): Likewise.
8991 (rs6000_builtin_mask_names): Likewise.
8992 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8993 (RS6000_BTM_FLOAT128_HW): Likewise.
8994 (RS6000_BTM_COMMON): Likewise.
8995 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8997 * doc/extend.texi (RS/6000 built-in functions): Document the
8998 IEEE 128-bit floating point square root and fused multiply-add
9001 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9003 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9004 reg (r2) isn't in the set of registers defined in the prologue.
9006 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9007 Alan Hayward <alan.hayward@arm.com>
9008 David Sherwood <david.sherwood@arm.com>
9010 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9011 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9012 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9013 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9014 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9016 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9017 max_vectorization_factor.
9018 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9020 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9021 Alan Hayward <alan.hayward@arm.com>
9022 David Sherwood <david.sherwood@arm.com>
9024 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9025 (vect_worthwhile_without_simd_p): Declare.
9026 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9027 (vectorizable_reduction): Use it.
9028 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9029 (vectorizable_operation): Likewise.
9031 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9032 Alan Hayward <alan.hayward@arm.com>
9033 David Sherwood <david.sherwood@arm.com>
9035 * tree-vectorizer.h (vect_get_num_copies): New function.
9036 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9037 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9038 (vectorizable_induction): Likewise.
9039 (vectorizable_live_operation): Likewise.
9040 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9041 (vectorizable_bswap): Likewise.
9042 (vectorizable_call): Likewise.
9043 (vectorizable_conversion): Likewise.
9044 (vectorizable_assignment): Likewise.
9045 (vectorizable_shift): Likewise.
9046 (vectorizable_operation): Likewise.
9047 (vectorizable_store): Likewise.
9048 (vectorizable_load): Likewise.
9049 (vectorizable_condition): Likewise.
9050 (vectorizable_comparison): Likewise.
9051 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9053 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9054 Alan Hayward <alan.hayward@arm.com>
9055 David Sherwood <david.sherwood@arm.com>
9057 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9058 of vect_init_vector.
9060 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9061 Alan Hayward <alan.hayward@arm.com>
9062 David Sherwood <david.sherwood@arm.com>
9064 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9065 an inline wrapper that provides a location.
9066 (gimple_build_vector): Likewise.
9067 * gimple-fold.c (gimple_build_vector_from_val): New function.
9068 (gimple_build_vector): Likewise.
9069 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9070 functions to build the initial value. Always return a gimple value.
9071 (get_initial_defs_for_reduction): Likewise. Only compute
9073 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9074 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9075 (vectorizable_induction): Use gimple_build_vector rather than
9078 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9079 Alan Hayward <alan.hayward@arm.com>
9080 David Sherwood <david.sherwood@arm.com>
9082 * target.h (vec_perm_indices): New typedef.
9083 (auto_vec_perm_indices): Likewise.
9084 * optabs-query.h: Include target.h
9085 (can_vec_perm_p): Take a vec_perm_indices *.
9086 * optabs-query.c (can_vec_perm_p): Likewise.
9087 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9088 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9089 * tree-vect-generic.c (lower_vec_perm): Likewise.
9090 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9091 (vect_grouped_load_supported): Likewise.
9092 (vect_shift_permute_load_chain): Likewise.
9093 (vect_permute_store_chain): Use auto_vec_perm_indices.
9094 (vect_permute_load_chain): Likewise.
9095 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9096 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9097 Update uses of can_vec_perm_p.
9098 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9099 mode with a number of elements. Take a vec_perm_indices *.
9100 (vect_create_epilog_for_reduction): Update accordingly.
9101 Use auto_vec_perm_indices.
9102 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9103 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9104 (vect_transform_slp_perm_load): Likewise.
9105 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9106 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9107 (vect_gen_perm_mask_checked): Likewise.
9108 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9109 (vect_gen_perm_mask_checked): Likewise.
9110 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9111 (vectorizable_store): Likewise.
9112 (vectorizable_load): Likewise.
9113 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9114 (vectorizable_bswap): Likewise.
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 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9121 * tree.c (build_vector): Likewise.
9122 (build_vector_from_ctor): Update accordingly.
9123 (build_vector_from_val): Likewise.
9124 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9125 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9126 * tree-vect-generic.c (add_rshift): Likewise.
9127 (expand_vector_divmod): Likewise.
9128 (optimize_vector_constructor): Likewise.
9129 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9130 (vect_transform_slp_perm_load): Likewise.
9131 (vect_schedule_slp_instance): Likewise.
9132 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9133 (vectorizable_call): Likewise.
9134 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9135 * expmed.c (make_tree): Likewise.
9136 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9137 a vector passed to build_vector.
9138 (fold_convert_const): Likewise.
9139 (exact_inverse): Likewise.
9140 (fold_ternary_loc): Likewise.
9141 (fold_relational_const): Likewise.
9142 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9143 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9144 (const_unop): Likewise. Store the reduction accumulator in a
9145 variable rather than an array.
9146 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9147 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9148 the new vector, rather than constructing it after the input arrays.
9149 (native_interpret_vector): Use auto_vec<tree> when building
9150 a vector passed to build_vector. Add elements in order.
9151 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9152 auto_vec<tree> when building a vector passed to build_vector.
9153 (vect_create_epilog_for_reduction): Likewise.
9154 (vectorizable_induction): Likewise.
9155 (get_initial_def_for_reduction): Likewise. Fix indentation of
9157 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9159 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9160 passed to build_vector.
9162 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9163 Alan Hayward <alan.hayward@arm.com>
9164 David Sherwood <david.sherwood@arm.com>
9166 * tree-core.h (tree_base::u): Add an "nelts" field.
9167 (tree_vector): Use VECTOR_CST_NELTS as the length.
9168 * tree.c (tree_size): Likewise.
9169 (make_vector): Initialize VECTOR_CST_NELTS.
9170 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9171 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9172 TYPE_VECTOR_SUBPARTS.
9173 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9174 as the number of units, setting it from VECTOR_CST_NELTS.
9175 (const_vector_from_tree): Likewise.
9176 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9177 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9178 (fold_negate_expr_1): Likewise.
9179 (fold_convert_const): Likewise.
9180 (const_binop): Likewise. Differentiate the number of output and
9182 (const_unop): Likewise.
9183 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9184 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9185 in cases that did the opposite.
9187 2017-09-14 Richard Biener <rguenther@suse.de>
9189 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9192 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9194 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9196 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9199 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9200 if and where to split a bb, except for splitting before debug insn
9201 sequences followed by non-label real insn. Delete debug insns
9202 in between basic blocks.
9204 * combine.c (make_compound_operation_int): Formatting fixes.
9206 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9207 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9208 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9209 * config/sol2.h (LINK_EH_SPEC): Likewise.
9210 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9211 * config/s390/linux.h (LINK_SPEC): Likewise.
9212 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9213 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9214 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9215 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9216 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9217 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9218 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9219 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9221 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9223 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9225 (ENDFILE_LINUX_SPEC): Likewise.
9226 (LINK_EH_SPEC): Likewise.
9227 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9228 (LINK_OS_LINUX_SPEC32): Likewise.
9229 (LINK_OS_LINUX_SPEC64): Likewise.
9230 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9231 (LINK_OS_LINUX_SPEC): Likewise.
9233 2017-09-13 Martin Liska <mliska@suse.cz>
9236 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9237 CASE_HIGH is NULL_TREE.
9239 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9240 Alan Hayward <alan.hayward@arm.com>
9241 David Sherwood <david.sherwood@arm.com>
9243 * target.def (secondary_memory_needed): New hook.
9244 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9245 instead of SECONDARY_MEMORY_NEEDED.
9246 (secondary_memory_needed_mode): Likewise.
9247 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9248 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9249 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9250 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9251 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9252 * doc/tm.texi: Regenerate.
9253 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9254 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9255 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9256 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9257 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9258 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9259 mode argument first and change the reg_class arguments to reg_class_t.
9260 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9261 Make static. Update the call to inline_secondary_memory_needed.
9262 (ix86_register_move_cost): Update the call to
9263 inline_secondary_memory_needed.
9264 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9265 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9267 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9268 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9270 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9271 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9272 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9273 and match hook interface. Add comment from mips.h.
9274 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9275 * config/mmix/mmix.md (truncdfsf2): Refer to
9276 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9278 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9279 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9280 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9281 (pa_secondary_memory_needed): New function.
9282 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9283 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9284 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9285 (pdp11_secondary_memory_needed): Make static and match hook interface.
9286 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9287 * config/powerpcspe/powerpcspe-protos.h
9288 (rs6000_secondary_memory_needed_ptr): Delete.
9289 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9291 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9292 (rs6000_option_override_internal): Assign to
9293 targetm.secondary_memory_needed rather than
9294 rs6000_secondary_memory_needed_ptr.
9295 (rs6000_secondary_memory_needed): Match hook interface.
9296 (rs6000_debug_secondary_memory_needed): Likewise.
9297 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9298 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9299 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9300 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9301 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9302 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9304 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9305 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9306 (rs6000_option_override_internal): Assign to
9307 targetm.secondary_memory_needed rather than
9308 rs6000_secondary_memory_needed_ptr.
9309 (rs6000_secondary_memory_needed): Match hook interface.
9310 (rs6000_debug_secondary_memory_needed): Likewise.
9311 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9312 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9313 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9314 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9315 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9316 (sparc_secondary_memory_needed): New function.
9317 * lra-constraints.c (check_and_process_move): Refer to
9318 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9320 (curr_insn_transform): Likewise.
9321 (process_alt_operands): Use targetm.secondary_memory_needed
9322 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9323 (check_secondary_memory_needed_p): Likewise.
9324 (choose_split_class): Likewise.
9325 * reload.c: Unconditionally include code that was previously
9326 conditional on SECONDARY_MEMORY_NEEDED.
9327 (push_secondary_reload): Use targetm.secondary_memory_needed
9328 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9329 (push_reload): Likewise.
9330 * reload1.c: Unconditionally include code that was previously
9331 conditional on SECONDARY_MEMORY_NEEDED.
9332 (choose_reload_regs): Use targetm.secondary_memory_needed
9333 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9334 (gen_reload): Likewise.
9335 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9337 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9338 Alan Hayward <alan.hayward@arm.com>
9339 David Sherwood <david.sherwood@arm.com>
9341 * target.def (secondary_memory_needed_mode): New hook:
9342 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9343 * targhooks.h (default_secondary_memory_needed_mode): New function.
9344 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9345 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9346 * doc/tm.texi: Regenerate.
9347 * lra-constraints.c (check_and_process_move): Use
9348 targetm.secondary_memory_needed_mode instead of
9349 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9350 (curr_insn_transform): Likewise.
9351 * reload.c (get_secondary_mem): Likewise.
9352 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9353 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9355 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9356 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9357 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9358 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9359 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9361 * config/powerpcspe/powerpcspe-protos.h
9362 (rs6000_secondary_memory_needed_mode): Delete.
9363 * config/powerpcspe/powerpcspe.c
9364 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9365 (rs6000_secondary_memory_needed_mode): Make static.
9366 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9367 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9369 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9371 (rs6000_secondary_memory_needed_mode): Make static.
9372 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9373 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9374 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9375 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9376 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9378 (sparc_secondary_memory_needed_mode): New function.
9379 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9381 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9383 * config/aarch64/constraints.md (Umq): New constraint.
9384 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9386 (mov<mode>): Update condition.
9388 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9390 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9391 when bitposition is the same.
9393 2017-09-13 Richard Biener <rguenther@suse.de>
9395 * dwarf2out.c (output_die_symbol): Remove.
9396 (output_die): Do not output a DIEs symbol.
9398 2017-09-13 Richard Biener <rguenther@suse.de>
9401 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9402 default-def to avoid breaking iterator update with the weird
9403 interaction with cgraph_update_edges_for_call_stmt_node.
9405 2017-09-13 Richard Biener <rguenther@suse.de>
9407 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9408 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9409 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9410 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9412 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9414 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9415 Disable pc relative literal load irrespective of
9416 TARGET_FIX_ERR_A53_84341 for default.
9418 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9420 * config/sparc/sparc.c (output_return): Output the source location of
9421 the insn in the delay slot, if any.
9422 (output_sibcall): Likewise.
9424 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9427 * common.opt (-static-pie): New alias.
9428 (shared): Negate static-pie.
9429 (-no-pie): Update help text.
9431 (static-pie): New option.
9432 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9433 -static-pie support.
9434 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9435 (LINK_EH_SPEC): Likewise.
9436 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9437 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9438 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9439 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9440 (init_gcc_specs): Likewise.
9441 (init_spec): Likewise.
9442 (display_help): Update help message for -pie.
9443 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9446 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9448 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9449 (movdi_aarch64): Likewise.
9450 (movti_aarch64): Likewise.
9452 2017-09-12 Simon Wright <simon@pushface.org>
9455 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9456 calculation of the minor version, always output as 0.
9458 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9461 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9462 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9463 on it early, rather than manual conversion late. For
9464 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9465 instead of performing manual conversion.
9467 2017-09-12 Carl Love <cel@us.ibm.com>
9469 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9470 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9472 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9473 VMULOSW): Add definitions.
9474 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9475 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9476 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9477 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9478 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9480 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9482 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9484 (movti_aarch64): Likewise.
9485 (movdf_aarch64): Likewise.
9486 (movtf_aarch64): Likewise.
9487 (load_pairdi): Likewise.
9488 (store_pairdi): Likewise.
9489 (load_pairdf): Likewise.
9490 (store_pairdf): Likewise.
9491 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9492 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9493 (ldr_got_small_<mode>): Likewise.
9494 (ldr_got_small_28k_<mode>): Likewise.
9495 (ldr_got_tiny): Likewise.
9496 * config/aarch64/iterators.md (ldst_sz): New.
9497 (ldpstp_sz): Likewise.
9498 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9500 (thunderx_load): Split load_8 to load_16.
9501 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9503 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9504 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9505 (xgene1_store_pair): Split store_8 to store_16.
9506 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9507 (falkor_st_0_st_sd): Split store_8 to store_16.
9509 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9511 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9512 and store1/2/3/4 to store_4/8/12/16.
9513 * config/aarch64/aarch64.md: Update for rename.
9514 * config/arm/arm.md: Likewise.: Likewise.
9515 * config/arm/arm.c: Likewise.
9516 * config/arm/thumb1.md: Likewise.
9517 * config/arm/thumb2.md: Likewise.
9518 * config/arm/vfp.md: Likewise.
9519 * config/arm/arm-generic.md: Likewise.
9520 * config/arm/arm1020e.md: Likewise.
9521 * config/arm/arm1026ejs.md: Likewise.
9522 * config/arm/arm1136jfs.md: Likewise.
9523 * config/arm/arm926ejs.md: Likewise.
9524 * config/arm/cortex-a15.md: Likewise.
9525 * config/arm/cortex-a17.md: Likewise.
9526 * config/arm/cortex-a5.md: Likewise.
9527 * config/arm/cortex-a53.md: Likewise.
9528 * config/arm/cortex-a57.md: Likewise.
9529 * config/arm/cortex-a7.md: Likewise.
9530 * config/arm/cortex-a8.md: Likewise.
9531 * config/arm/cortex-a9.md: Likewise.
9532 * config/arm/cortex-m4.md: Likewise.
9533 * config/arm/cortex-m7.md: Likewise.
9534 * config/arm/cortex-r4.md: Likewise.
9535 * config/arm/exynos-m1.md: Likewise.
9536 * config/arm/fa526.md: Likewise.
9537 * config/arm/fa606te.md: Likewise.
9538 * config/arm/fa626te.md: Likewise.
9539 * config/arm/fa726te.md: Likewise.
9540 * config/arm/fmp626.md: Likewise.
9541 * config/arm/iwmmxt.md: Likewise.
9542 * config/arm/ldmstm.md: Likewise.
9543 * config/arm/marvell-pj4.md: Likewise.
9544 * config/arm/xgene1.md: Likewise.
9545 * config/aarch64/thunderx.md: Likewise.
9546 * config/aarch64/thunderx2t99.md: Likewise.
9547 * config/aarch64/falkor.md: Likewise.
9549 2017-09-12 Martin Liska <mliska@suse.cz>
9551 * attribs.c (private_lookup_attribute): New function.
9552 * attribs.h (private_lookup_attribute): Declared here.
9553 (lookup_attribute): Called from this place.
9555 2017-09-12 Richard Biener <rguenther@suse.de>
9557 PR tree-optimization/82157
9558 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9559 stmts with side-effects.
9561 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9562 Alan Hayward <alan.hayward@arm.com>
9563 David Sherwood <david.sherwood@arm.com>
9565 * target.def (hard_regno_nregs): New hook.
9566 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9567 * targhooks.h (default_hard_regno_nregs): Declare.
9568 * targhooks.c (default_hard_regno_nregs): New function.
9569 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9570 (TARGET_HARD_REGNO_NREGS): ...this hook.
9571 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9572 (CLASS_MAX_NREGS): Likewise.
9573 * doc/tm.texi: Regenerate.
9574 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9575 instead of HARD_REGNO_NREGS.
9576 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9577 HARD_REGNO_NREGS in the comment.
9578 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9579 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9580 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9581 Return an unsigned int.
9582 (TARGET_HARD_REGNO_NREGS): Redefine.
9583 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9584 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9585 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9586 (arc_hard_regno_nregs): New function.
9587 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9588 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9589 (arm_hard_regno_nregs): New function.
9590 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9591 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9592 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9593 (TARGET_HARD_REGNO_NREGS): Redefine.
9594 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9595 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9596 (HARD_REGNO_NREGS): Delete.
9597 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9598 (cr16_hard_regno_nregs): New function.
9599 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9600 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9601 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9602 (cris_hard_regno_nregs): New function.
9603 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9604 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9605 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9606 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9607 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9608 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9609 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9610 (frv_hard_regno_nregs): Make static. Take and return an
9612 (frv_class_max_nregs): Remove outdated copy of documentation.
9613 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9614 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9615 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9616 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9617 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9618 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9619 (TARGET_HARD_REGNO_NREGS): Redefine.
9620 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9621 (CLASS_MAX_NREGS): Update comment.
9622 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9623 (ia64_hard_regno_nregs): New function.
9624 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9625 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9626 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9627 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9628 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9630 (m32c_hard_regno_nregs): Likewise. Make static.
9631 (TARGET_HARD_REGNO_NREGS): Redefine.
9632 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9633 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9634 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9635 (m68k_hard_regno_nregs): New function.
9636 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9637 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9638 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9639 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9640 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9641 Take and return an unsigned int.
9642 (TARGET_HARD_REGNO_NREGS): Redefine.
9643 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9644 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9645 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9646 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9647 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9648 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9649 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9650 (msp430_hard_regno_nregs): Make static. Take and return an
9652 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9653 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9654 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9655 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9656 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9657 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9658 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9659 (TARGET_HARD_REGNO_NREGS): Redefine.
9660 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9661 (PA_HARD_REGNO_NREGS): ...this.
9662 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9663 (PA_HARD_REGNO_NREGS): ...this.
9664 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9665 (pa_hard_regno_nregs): New function.
9666 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9667 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9668 (pdp11_hard_regno_nregs): New function.
9669 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9670 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9671 (rs6000_hard_regno_nregs_hook): New function.
9672 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9673 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9674 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9675 Take and return an unsigned int. Move earlier in file.
9676 (TARGET_HARD_REGNO_NREGS): Redefine.
9677 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9678 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9679 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9680 (rl78_hard_regno_nregs): Make static. Take and return an
9682 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9683 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9684 (rs6000_hard_regno_nregs_hook): New function.
9685 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9686 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9687 (TARGET_HARD_REGNO_NREGS): Redefine.
9688 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9689 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9690 instead of HARD_REGNO_NREGS.
9691 (s390_hard_regno_nregs): New function.
9692 (s390_hard_regno_mode_ok): Add comment from s390.h.
9693 (TARGET_HARD_REGNO_NREGS): Redefine.
9694 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9695 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9696 (sh_hard_regno_nregs): New function.
9697 (sh_pass_in_reg_p): Use it.
9698 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9699 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9700 (sparc_hard_regno_nregs): New function.
9701 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9702 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9703 (spu_function_arg_advance): Use it, supplying a valid register number.
9704 (TARGET_HARD_REGNO_NREGS): Redefine.
9705 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9706 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9707 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9708 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9709 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9710 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9711 (CLASS_MAX_NREGS): Remove copy of old documentation.
9712 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9713 (visium_hard_regno_nregs): New function.
9714 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9715 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9716 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9717 xtensa_hard_regno_nregs): New function.
9718 * system.h (HARD_REGNO_NREGS): Poison.
9720 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9722 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9723 hard_regno_nregs instead of HARD_REGNO_NREGS.
9724 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9725 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9726 (c6x_expand_epilogue): Likewise.
9727 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9728 (frv_read_iacc_argument): Likewise.
9729 * config/sh/sh.c: Include regs.h.
9730 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9731 (regs_used): Likewise.
9732 (output_stack_adjust): Likewise.
9733 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9734 * expmed.c: Include regs.h.
9735 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9736 * ree.c: Include regs.h.
9737 (combine_reaching_defs): Use hard_regno_nregs instead of
9739 (add_removable_extension): Likewise.
9741 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9743 * regs.h (hard_regno_nregs): Turn into a function.
9744 (end_hard_regno): Update accordingly.
9745 * caller-save.c (setup_save_areas): Likewise.
9746 (save_call_clobbered_regs): Likewise.
9747 (replace_reg_with_saved_mem): Likewise.
9748 (insert_restore): Likewise.
9749 (insert_save): Likewise.
9750 * combine.c (can_change_dest_mode): Likewise.
9751 (move_deaths): Likewise.
9752 (distribute_notes): Likewise.
9753 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9754 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9755 (rs6000_split_multireg_move): Likewise.
9756 (rs6000_register_move_cost): Likewise.
9757 (rs6000_memory_move_cost): Likewise.
9758 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9759 (rs6000_split_multireg_move): Likewise.
9760 (rs6000_register_move_cost): Likewise.
9761 (rs6000_memory_move_cost): Likewise.
9762 * cselib.c (cselib_reset_table): Likewise.
9763 (cselib_lookup_1): Likewise.
9764 * emit-rtl.c (set_mode_and_regno): Likewise.
9765 * function.c (aggregate_value_p): Likewise.
9766 * ira-color.c (setup_profitable_hard_regs): Likewise.
9767 (check_hard_reg_p): Likewise.
9768 (calculate_saved_nregs): Likewise.
9769 (assign_hard_reg): Likewise.
9770 (improve_allocation): Likewise.
9771 (calculate_spill_cost): Likewise.
9772 * ira-emit.c (modify_move_list): Likewise.
9773 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9774 (ira_hard_reg_in_set_p): Likewise.
9775 * ira.c (setup_reg_mode_hard_regset): Likewise.
9776 (clarify_prohibited_class_mode_regs): Likewise.
9777 (check_allocation): Likewise.
9778 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9779 (lra_setup_reg_renumber): Likewise.
9780 (setup_try_hard_regno_pseudos): Likewise.
9781 (spill_for): Likewise.
9782 (assign_hard_regno): Likewise.
9783 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9784 * lra-constraints.c (in_class_p): Likewise.
9785 (lra_constraint_offset): Likewise.
9786 (simplify_operand_subreg): Likewise.
9787 (lra_constraints): Likewise.
9788 (split_reg): Likewise.
9789 (split_if_necessary): Likewise.
9790 (invariant_p): Likewise.
9791 (inherit_in_ebb): Likewise.
9792 * lra-lives.c (process_bb_lives): Likewise.
9793 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9794 (get_hard_regs): Likewise.
9795 (do_remat): Likewise.
9796 * lra-spills.c (assign_spill_hard_regs): Likewise.
9797 * mode-switching.c (create_pre_exit): Likewise.
9798 * postreload.c (reload_combine_recognize_pattern): Likewise.
9799 * recog.c (peep2_find_free_register): Likewise.
9800 * regcprop.c (kill_value_regno): Likewise.
9801 (set_value_regno): Likewise.
9802 (copy_value): Likewise.
9803 (maybe_mode_change): Likewise.
9804 (find_oldest_value_reg): Likewise.
9805 (copyprop_hardreg_forward_1): Likewise.
9806 * regrename.c (check_new_reg_p): Likewise.
9807 (regrename_do_replace): Likewise.
9808 * reload.c (push_reload): Likewise.
9809 (combine_reloads): Likewise.
9810 (find_dummy_reload): Likewise.
9811 (operands_match_p): Likewise.
9812 (find_reloads): Likewise.
9813 (find_equiv_reg): Likewise.
9814 (reload_adjust_reg_for_mode): Likewise.
9815 * reload1.c (count_pseudo): Likewise.
9816 (count_spilled_pseudo): Likewise.
9817 (find_reg): Likewise.
9818 (clear_reload_reg_in_use): Likewise.
9819 (free_for_value_p): Likewise.
9820 (allocate_reload_reg): Likewise.
9821 (choose_reload_regs): Likewise.
9822 (reload_adjust_reg_for_temp): Likewise.
9823 (emit_reload_insns): Likewise.
9824 (delete_output_reload): Likewise.
9825 * rtlanal.c (subreg_get_info): Likewise.
9826 * sched-deps.c (sched_analyze_reg): Likewise.
9827 * sel-sched.c (init_regs_for_mode): Likewise.
9828 (mark_unavailable_hard_regs): Likewise.
9829 (choose_best_reg_1): Likewise.
9830 (verify_target_availability): Likewise.
9831 * valtrack.c (dead_debug_insert_temp): Likewise.
9832 * var-tracking.c (track_loc_p): Likewise.
9833 (emit_note_insn_var_location): Likewise.
9834 * varasm.c (make_decl_rtl): Likewise.
9835 * reginfo.c (choose_hard_reg_mode): Likewise.
9836 (init_reg_modes_target): Refer directly to
9837 this_target_regs->x_hard_regno_nregs.
9839 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9841 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9842 instead of hard_regno_nregs.
9844 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9846 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9847 end_hard_regno instead of hard_regno_nregs.
9848 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9849 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9850 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9851 * ira-color.c (improve_allocation): Likewise.
9852 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9853 * lra-lives.c (mark_regno_live): Likewise.
9854 (mark_regno_dead): Likewise.
9855 * lra-remat.c (operand_to_remat): Likewise.
9856 * lra.c (collect_non_operand_hard_regs): Likewise.
9857 * postreload.c (reload_combine_note_store): Likewise.
9858 (move2add_valid_value_p): Likewise.
9859 * reload.c (regno_clobbered_p): Likewise.
9861 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9863 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9865 * config/v850/v850.c (v850_reorg): Likewise.
9866 * reload.c (refers_to_regno_for_reload_p): Likewise.
9867 (find_equiv_reg): Likewise.
9868 * reload1.c (reload_reg_reaches_end_p): Likewise.
9870 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9872 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9874 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9875 * config/arm/arm.c (output_move_neon): Likewise.
9876 (arm_attr_length_move_neon): Likewise.
9877 (neon_split_vcombine): Likewise.
9878 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9879 (c6x_mark_reg_written): Likewise.
9880 (c6x_dwarf_register_span): Likewise.
9881 * config/i386/i386.c (ix86_save_reg): Likewise.
9882 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9883 (rws_access_reg): Likewise.
9884 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9885 * mode-switching.c (create_pre_exit): Likewise.
9886 * ree.c (combine_reaching_defs): Likewise.
9887 (add_removable_extension): Likewise.
9888 * regcprop.c (find_oldest_value_reg): Likewise.
9889 (copyprop_hardreg_forward_1): Likewise.
9890 * reload.c (reload_inner_reg_of_subreg): Likewise.
9891 (push_reload): Likewise.
9892 (combine_reloads): Likewise.
9893 (find_dummy_reload): Likewise.
9894 (reload_adjust_reg_for_mode): Likewise.
9895 * reload1.c (find_reload_regs): Likewise.
9896 (forget_old_reloads_1): Likewise.
9897 (reload_reg_free_for_value_p): Likewise.
9898 (reload_adjust_reg_for_temp): Likewise.
9899 (emit_reload_insns): Likewise.
9900 (delete_output_reload): Likewise.
9901 * sel-sched.c (choose_best_reg_1): Likewise.
9902 (choose_best_pseudo_reg): Likewise.
9904 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9905 Alan Hayward <alan.hayward@arm.com>
9906 David Sherwood <david.sherwood@arm.com>
9908 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9909 * target.def (slow_unaligned_access): New hook.
9910 * targhooks.h (default_slow_unaligned_access): Declare.
9911 * targhooks.c (default_slow_unaligned_access): New function.
9912 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9913 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9914 * doc/tm.texi: Regenerate.
9915 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9916 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9917 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9919 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9920 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9922 (rs6000_slow_unaligned_access): New function.
9923 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9924 (expand_block_compare): Likewise.
9925 (expand_strn_compare): Likewise.
9926 (rs6000_rtx_costs): Likewise.
9927 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9928 (riscv_slow_unaligned_access): Likewise.
9929 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9930 (riscv_slow_unaligned_access_p): ...this and make static.
9931 (riscv_option_override): Update accordingly.
9932 (riscv_slow_unaligned_access): New function.
9933 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9934 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9935 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9936 (rs6000_slow_unaligned_access): New function.
9937 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9938 (rs6000_rtx_costs): Likewise.
9939 * config/rs6000/rs6000-string.c (expand_block_compare)
9940 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9941 of SLOW_UNALIGNED_ACCESS.
9942 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9943 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9944 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9945 of SLOW_UNALIGNED_ACCESS.
9946 * expmed.c (simple_mem_bitfield_p): Likewise.
9947 * expr.c (alignment_for_piecewise_move): Likewise.
9948 (emit_group_load_1): Likewise.
9949 (emit_group_store): Likewise.
9950 (copy_blkmode_from_reg): Likewise.
9951 (emit_push_insn): Likewise.
9952 (expand_assignment): Likewise.
9953 (store_field): Likewise.
9954 (expand_expr_real_1): Likewise.
9955 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9956 * lra-constraints.c (simplify_operand_subreg): Likewise.
9957 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9958 * gimple-ssa-store-merging.c: Likewise in block comment at start
9960 * tree-ssa-strlen.c: Include target.h.
9961 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9962 of SLOW_UNALIGNED_ACCESS.
9963 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9965 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9967 PR rtl-optimization/82185
9968 * expmed.c (emit_store_flag_int): Only test tem if it has been
9971 2017-09-12 Richard Biener <rguenther@suse.de>
9974 * match.pd ((FTYPE) N CMP CST): Fix typo.
9976 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9978 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9980 (mips_near_type_p): Add 'short_call' attribute as a synonym
9982 * doc/extend.texi (short_call): Document new function attribute.
9984 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9987 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9988 assertion check that in the condition.
9989 (get_atomic_generic_size): Likewise. Before testing if parameter
9990 has pointer type, if it has array type, call for C++
9991 default_conversion to perform array-to-pointer conversion.
9993 2017-09-12 Richard Biener <rguenther@suse.de>
9995 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9996 for operations we cannot scalarize.
9998 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10000 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10001 vectors heap vectors. Clean up comments.
10002 Make visited_bbs a reference.
10003 (profitable_jump_thread_path): Make GC
10004 vectors heap vectors. Clean up comments.
10006 (convert_and_register_jump_thread_path): Make GC vectors heap
10008 (check_subpath_and_update_thread_path): Same. Clean up comments.
10009 Make visited_bbs a reference.
10010 (handle_phi): Abstract common code to to
10011 register_jump_thread_path_if_profitable.
10012 Rename VAR_BB to DEF_BB.
10014 Make GC vectors heap vectors.
10015 Make visited_bbs a reference.
10016 (handle_assignment): Same.
10017 (register_jump_thread_path_if_profitable): New.
10018 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10020 Make GC vectors heap vectors. Clean up comments.
10021 Make visited_bbs a reference.
10022 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10023 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10026 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10029 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10030 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10032 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10034 Revert r251800 and r251799.
10036 2017-09-11 Martin Jambor <mjambor@suse.cz>
10039 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10040 arguments in advance.
10041 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10042 use it to find predecessor edges.
10043 (naive_outof_ssa): Collect vector of predecessors.
10045 2017-09-08 Jason Merrill <jason@redhat.com>
10047 PR c++/70029 - ICE with ref-qualifier and -flto
10048 * langhooks.h (struct lang_hooks_for_types): Add
10049 copy_lang_qualifiers.
10050 * attribs.c (build_type_attribute_qual_variant): Use it.
10051 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10053 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10054 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10056 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10059 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10060 (*mulsi3_sp64): New instruction.
10061 (mulsi3): New expander.
10063 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10065 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10067 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10069 * sancov.c: Include memmodel.h.
10071 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10074 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10077 2017-09-07 Carl Love <cel@us.ibm.com>
10079 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10080 the sldi instruction.
10082 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10084 * sancov.c: Include tm_p.h.
10086 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10089 * output.h (switch_to_other_text_partition): New declaration.
10090 * varasm.c (switch_to_other_text_partition): New function.
10091 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10092 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10093 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10094 to the other text partition before emitting LCL label and switch back
10095 after emitting the word after it.
10097 2017-09-07 Richard Biener <rguenther@suse.de>
10099 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10100 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10101 critical edges here, after loop init.
10102 (pass_data_pre): Remove PROP_no_crit_edges flags.
10103 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10104 for valueization of call args to avoid leaking VN_TOP.
10105 (visit_use): Assert we do not visit default defs.
10106 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10107 Use error_mark_node to more easily detect leaking VN_TOP.
10108 All default-defs are varying, not VN_TOP. Mark them visited.
10109 (run_scc_vn): Make code match comment.
10111 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10113 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10114 OPTION_MASK_FLOAT128_KEYWORD.
10115 (POWERPC_MASKS): Likewise.
10116 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10117 support for the -mfloat128-type option, and make -mfloat128
10118 default on PowerPC Linux systems. Define or undefine
10119 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10120 Define __float128 to be __ieee128 if IEEE 128-bit support is
10121 enabled, or undefine it.
10122 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10123 Delete defining __FLOAT128_TYPE__.
10124 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10125 -mfloat128-type option and make -mfloat128 default on PowerPC
10127 (TARGET_FLOAT128_TYPE): Likewise.
10128 (-mfloat128-type): Likewise.
10129 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10130 Delete the -mfloat128-type option and make -mfloat128 default on
10131 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10132 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10133 128-bit floating point is enabled. Change tests from using
10134 -mfloat128-type to -mfloat128.
10135 (rs6000_mangle_type): Use the correct mangling for the __float128
10136 type even if normal long double is restricted to 64-bits.
10137 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10139 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10140 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10141 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10143 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10144 documentation for -mfloat128.
10146 2017-09-06 Olivier Hainque <hainque@adacore.com>
10148 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10150 2017-09-06 Wish Wu <wishwu007@gmail.com>
10151 Jakub Jelinek <jakub@redhat.com>
10153 * asan.c (initialize_sanitizer_builtins): Add
10154 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10155 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10156 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10157 BT_FN_VOID_UINT64_PTR variables.
10158 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10159 (BT_FN_VOID_UINT16_UINT16): Likewise.
10160 (BT_FN_VOID_UINT32_UINT32): Likewise.
10161 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10162 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10163 (BT_FN_VOID_UINT64_PTR): Likewise.
10164 * common.opt (flag_sanitize_coverage): New variable.
10165 (fsanitize-coverage=trace-pc): Remove.
10166 (fsanitize-coverage=): Add.
10167 * flag-types.h (enum sanitize_coverage_code): New enum.
10168 * fold-const.c (fold_range_test): Disable non-short-circuit
10169 optimization if flag_sanitize_coverage.
10170 (fold_truth_andor): Likewise.
10171 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10172 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10173 (coverage_sanitizer_opts): New array.
10174 (get_closest_sanitizer_option): Add OPTS argument, handle also
10175 OPT_fsanitize_coverage_.
10176 (parse_sanitizer_options): Adjusted to also handle
10177 OPT_fsanitize_coverage_.
10178 (common_handle_option): Add OPT_fsanitize_coverage_.
10179 * sancov.c (instrument_comparison, instrument_switch): New function.
10180 (sancov_pass): Add trace-cmp support.
10181 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10182 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10183 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10184 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10185 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10186 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10187 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10188 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10189 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10190 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10192 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10194 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10195 error. Only quit immediately if parsing is complete.
10196 (BEGIN): Initialize fatal_err and parse_done.
10197 (begin fpu, end fpu): Check number of arguments.
10198 (begin arch, end arch): Likewise.
10199 (begin cpu, end cpu): Likewise.
10200 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10201 (optalias): Likewise.
10203 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10205 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10206 * config/arm/arm-isa.h: Delete. Move definitions to ...
10207 * arm-cpus.in: ... here. Use new feature and fgroup values.
10208 * config/arm/arm.c (arm_option_override): Use lower case for feature
10210 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10211 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10212 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10214 (print_isa_bits_for): New function.
10215 (gen_isa): New function.
10216 (gen_comm_data): Use print_isa_bits_for.
10217 (define feature): New keyword.
10218 (define fgroup): New keyword.
10219 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10220 (arm-isa.h): Add rule to generate file.
10221 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10222 case for feature bit names.
10224 2017-09-06 Richard Biener <rguenther@suse.de>
10226 * tree-ssa-pre.c (NECESSARY): Remove.
10227 (create_expression_by_pieces): Do not touch pass-local flags.
10228 (insert_into_preds_of_block): Likewise.
10229 (do_pre_regular_insertion): Likewise.
10230 (eliminate_insert): Likewise.
10231 (eliminate_dom_walker::before_dom_children): Likewise.
10232 (fini_eliminate): Do not look at inserted_exprs.
10233 (mark_operand_necessary): Remove.
10234 (remove_dead_inserted_code): Replace with simple work-list
10235 algorithm based on inserted_exprs and SSA uses.
10236 (pass_pre::execute): Re-order fini_eliminate and
10237 remove_dead_inserted_code.
10239 2017-09-06 Olivier Hainque <hainque@adacore.com>
10241 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10242 for VxWorks 7. Adjust surrounding comments.
10244 2017-09-06 Richard Biener <rguenther@suse.de>
10246 * gimple-ssa-strength-reduction.c
10247 (find_candidates_dom_walker::before_dom_children): Also allow
10250 2017-09-06 Richard Biener <rguenther@suse.de>
10252 PR tree-optimization/82108
10253 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10254 for gap in the non-permutation SLP case.
10256 2017-09-06 Martin Jambor <mjambor@suse.cz>
10258 PR tree-optimization/82078
10259 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10260 add_access_to_work_queue...
10261 (build_accesses_from_assign): ...here.
10262 (propagate_all_subaccesses): Make sure racc is the group
10263 representative, if there is one.
10265 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10267 PR middle-end/82095
10268 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10271 2017-09-06 Richard Biener <rguenther@suse.de>
10273 * gimple-ssa-strength-reduction.c
10274 (find_candidates_dom_walker::before_doom_children): Use a
10275 type and not a mode check.
10277 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10280 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10282 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10284 2017-09-05 Jeff Law <law@redhat.com>
10286 PR tree-optimization/64910
10287 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10288 cases where we have 3 or more operands.
10290 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10292 PR middle-end/81768
10293 * omp-low.c (lower_omp_for): Recompute tree invariant if
10294 gimple_omp_for_initial/final is ADDR_EXPR.
10296 PR middle-end/81768
10297 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10298 into gimple val before gimplification fo the COND_EXPR.
10300 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10302 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10303 REGION_COPY argument.
10304 (thread_through_all_blocks): Remove unused argument to
10305 duplicate_thread_path.
10307 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10308 Alan Hayward <alan.hayward@arm.com>
10309 David Sherwood <david.sherwood@arm.com>
10311 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10312 Take a scalar_mode rather than a machine_mode.
10313 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10314 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10315 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10316 (aarch64_gen_adjusted_ldpstp): Likewise.
10317 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10319 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10320 Alan Hayward <alan.hayward@arm.com>
10321 David Sherwood <david.sherwood@arm.com>
10323 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10324 Take a scalar_int_mode instead of a machine_mode.
10325 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10326 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10327 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10328 (aarch64_simd_attr_length_rglist): Delete.
10329 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10330 a scalar_int_mode instead of a machine_mode.
10331 (aarch64_add_offset): Likewise.
10332 (aarch64_internal_mov_immediate): Likewise
10333 (aarch64_add_constant_internal): Likewise.
10334 (aarch64_add_constant): Likewise.
10335 (aarch64_movw_imm): Likewise.
10336 (aarch64_rtx_arith_op_extract_p): Likewise.
10337 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10338 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10339 Remove assert that the mode isn't a vector.
10340 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10341 (aarch64_expand_mov_immediate): Update calls after above changes.
10342 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10343 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10344 (aarch64_move_imm): Likewise.
10345 (aarch64_can_const_movi_rtx_p): Likewise.
10346 (aarch64_strip_extend): Likewise.
10347 (aarch64_extr_rtx_p): Likewise.
10348 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10349 a CONST_INT when the mode parameter is VOIDmode.
10350 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10352 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10354 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10355 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10356 (bitwise_type_for_mode): Update accordingly.
10358 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10360 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10361 * stor-layout.c (mode_for_size_tree): Likewise.
10362 (mode_for_array): Update accordingly.
10363 (layout_decl): Likewise.
10364 (compute_record_mode): Likewise. Only set the mode once.
10366 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10368 * target.def (get_mask_mode): Change return type to opt_mode.
10370 * doc/tm.texi: Regenerate.
10371 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10372 * targhooks.c (default_get_mask_mode): Likewise.
10373 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10374 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10375 targetm.get_mask_mode.
10376 * tree.c (build_truth_vector_type): Likewise.
10378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10380 * machmode.h (mode_for_vector): Return an opt_mode.
10381 * stor-layout.c (mode_for_vector): Likewise.
10382 (mode_for_int_vector): Update accordingly.
10383 (layout_type): Likewise.
10384 * config/i386/i386.c (emit_memmov): Likewise.
10385 (ix86_expand_set_or_movmem): Likewise.
10386 (ix86_expand_vector_init): Likewise.
10387 (ix86_get_mask_mode): Likewise.
10388 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10390 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10391 * expmed.c (extract_bit_field_1): Likewise.
10392 * expr.c (expand_expr_real_2): Likewise.
10393 * optabs-query.c (can_vec_perm_p): Likewise.
10394 (can_vec_mask_load_store_p): Likewise.
10395 * optabs.c (expand_vec_perm): Likewise.
10396 * targhooks.c (default_get_mask_mode): Likewise.
10397 * tree-vect-stmts.c (vectorizable_store): Likewise.
10398 (vectorizable_load): Likewise.
10399 (get_vectype_for_scalar_type_and_size): Likewise.
10401 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10403 * machmode.h (mode_for_int_vector): New function.
10404 * stor-layout.c (mode_for_int_vector): Likewise.
10405 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10406 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10407 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10408 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10409 (s390_expand_vcond): Likewise.
10411 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10413 * machmode.h (opt_machine_mode): New type.
10414 (opt_mode<T>): Allow construction from anything that can be
10416 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10417 (mode_for_size): Return an opt_machine_mode.
10418 * stor-layout.c (mode_for_size): Likewise.
10419 (mode_for_size_tree): Update call accordingly.
10420 (bitwise_mode_for_mode): Likewise.
10421 (make_fract_type): Likewise.
10422 (make_accum_type): Likewise.
10423 * caller-save.c (replace_reg_with_saved_mem): Update call
10425 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10426 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10427 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10428 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10429 * expmed.c (extract_bit_field_1): Likewise.
10430 * reload.c (get_secondary_mem): Likewise.
10431 * varasm.c (assemble_integer): Likewise.
10432 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10435 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10437 * machmode.h (decimal_float_mode_for_size): New function.
10438 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10439 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10440 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10441 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10442 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10443 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10445 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10447 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10448 (get_builtin_sync_mode): Likewise.
10449 (expand_ifn_atomic_compare_exchange): Likewise.
10450 (expand_builtin_atomic_clear): Likewise.
10451 (expand_builtin_atomic_test_and_set): Likewise.
10452 (fold_builtin_atomic_always_lock_free): Likewise.
10453 * calls.c (compute_argument_addresses): Likewise.
10454 (emit_library_call_value_1): Likewise.
10455 (store_one_arg): Likewise.
10456 * combine.c (combine_instructions): Likewise.
10457 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10458 * config/arm/arm.c (arm_function_value): Likewise.
10459 (aapcs_allocate_return_reg): Likewise.
10460 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10461 * config/i386/i386.c (construct_container): Likewise.
10462 (ix86_gimplify_va_arg): Likewise.
10463 (ix86_expand_sse_cmp): Likewise.
10464 (emit_memmov): Likewise.
10465 (emit_memset): Likewise.
10466 (expand_small_movmem_or_setmem): Likewise.
10467 (ix86_expand_pextr): Likewise.
10468 (ix86_expand_pinsr): Likewise.
10469 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10470 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10472 * config/mips/mips.c (mips_function_value_1) Likewise.
10473 (mips_block_move_straight): Likewise.
10474 (mips_expand_ins_as_unaligned_store): Likewise.
10475 * config/powerpcspe/powerpcspe.c
10476 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10477 (rs6000_darwin64_record_arg_flush): Likewise.
10478 * config/rs6000/rs6000.c
10479 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10480 (rs6000_darwin64_record_arg_flush): Likewise.
10481 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10482 (sparc_function_value_1): Likewise.
10483 * config/spu/spu.c (adjust_operand): Likewise.
10484 (spu_emit_branch_or_set): Likewise.
10485 (arith_immediate_p): Likewise.
10486 * emit-rtl.c (gen_lowpart_common): Likewise.
10487 * expr.c (expand_expr_real_1): Likewise.
10488 * function.c (assign_parm_setup_block): Likewise.
10489 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10490 * reload1.c (alter_reg): Likewise.
10491 * stor-layout.c (mode_for_vector): Likewise.
10492 (layout_type): Likewise.
10494 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10496 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10497 (spu_convert_move): Likewise.
10498 * lower-subreg.c (resolve_simple_move): Likewise.
10500 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10503 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10504 define_insn to a define_expand.
10505 (altivec_vsum2sws_direct): New define_insn.
10506 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10508 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10510 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10513 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10518 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10519 realign stack if stack alignment needed is less than incoming
10522 2017-09-05 Marek Polacek <polacek@redhat.com>
10525 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10528 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10530 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10532 2017-09-05 Richard Biener <rguenther@suse.de>
10534 PR tree-optimization/82084
10535 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10537 2017-09-05 Richard Biener <rguenther@suse.de>
10539 PR tree-optimization/82102
10540 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10542 2017-09-05 Martin Liska <mliska@suse.cz>
10544 PR tree-optimization/82032
10545 * tree-cfg.c (generate_range_test): New function.
10546 * tree-cfg.h (generate_range_test): Declared here.
10547 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10548 (cleanup_control_expr_graph): Use it.
10549 * tree-switch-conversion.c (try_switch_expansion): Remove
10551 (emit_case_nodes): Use generate_range_test.
10553 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10556 * config/i386/i386.md (*<btsc><mode>_mask): Add
10557 TARGET_USE_BT to insn constraint.
10558 (*btr<mode>_mask): Ditto.
10560 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10562 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10563 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10565 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10568 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10569 TARGET_NEON and TARGET_IWMMXT.
10570 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10571 TARGET_NEON and TARGET_IWMMXT.
10572 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10574 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10576 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10577 (ix86_rewrite_tls_address): Ditto.
10578 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10579 (ix86_rewrite_tls_address_1): Ditto.
10580 (ix86_rewrite_tls_address): Ditto.
10581 * config/i386/predicates.md (tls_address_pattern): New predicate.
10582 * config/i386/i386.md (TLS address splitter): New splitter.
10584 2017-09-04 Richard Biener <rguenther@suse.de>
10586 PR tree-optimization/82084
10587 * fold-const.h (can_native_encode_string_p): Declare.
10588 * fold-const.c (can_native_encode_string_p): Factor out from ...
10589 (native_encode_string): ... here.
10590 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10591 vectorizing stores from constants we later cannot handle.
10593 2017-09-04 Marek Polacek <polacek@redhat.com>
10596 * doc/invoke.texi: Update -Wtautological-compare documentation.
10598 2017-09-04 Jeff Law <law@redhat.com>
10600 PR tree-optimization/64910
10601 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10602 swap the first and last operand if the last is a constant.
10604 2017-09-04 Marek Polacek <polacek@redhat.com>
10607 * convert.c (do_narrow): When sanitizing signed integer overflows,
10608 bail out for signed types.
10609 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10611 2017-09-04 Richard Biener <rguenther@suse.de>
10613 PR tree-optimization/82060
10614 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10615 Move devirtualization after stmt folding and before EH/AB/noreturn
10616 cleanup to get the stmt refs canonicalized. Use a bool instead
10617 of gimple_modified_p since that doesn't work for NOPs. Schedule
10618 NOPs generated by folding for removal.
10620 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * coretypes.h (pad_direction): New enum.
10625 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10626 (FUNCTION_ARG_PADDING): Likewise.
10627 * target.def (function_arg_padding): New hook.
10628 * targhooks.h (default_function_arg_padding): Declare.
10629 * targhooks.c (default_function_arg_padding): New function.
10630 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10631 (TARGET_FUNCTION_ARG_PADDING): ...this.
10632 * doc/tm.texi: Regenerate.
10633 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10634 instead of direction.
10635 (compute_argument_addresses): Likewise.
10636 (load_register_parameters): Likewise.
10637 (emit_library_call_value_1): Likewise.
10638 (store_one_arg): Use targetm.calls.function_arg_padding instead
10639 of FUNCTION_ARG_PADDING.
10640 (must_pass_in_stack_var_size_or_pad): Likewise.
10641 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10642 (emit_group_store): Likewise.
10643 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10644 instead of FUNCTION_ARG_PADDING.
10645 (emit_push_insn): Likewise, and propagate enum change throughout
10647 * function.h (direction): Delete.
10648 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10650 * function.c (assign_parm_find_stack_rtl): Likewise.
10651 (assign_parm_setup_block_p): Likewise.
10652 (assign_parm_setup_block): Likewise.
10653 (gimplify_parameters): Likewise.
10654 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10655 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10657 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10658 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10659 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10660 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10661 (aarch64_function_arg_padding): ...this new function.
10662 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10663 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10664 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10665 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10666 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10667 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10668 (arm_pad_arg_upward): Replace with...
10669 (arm_function_arg_padding): ...this new function.
10670 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10672 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10673 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10674 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10675 (ia64_hpux_function_arg_padding): Replace with...
10676 (ia64_function_arg_padding): ...this new function. Use pad_direction
10677 instead of direction. Check for TARGET_HPUX.
10678 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10679 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10680 (iq2000_function_arg_padding): New function.
10681 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10682 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10683 (mips_function_arg_padding): ...this new function.
10684 (mips_pad_reg_upward): Update accordingly.
10685 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10686 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10687 targetm.calls.function_arg_padding.
10688 (FUNCTION_ARG_PADDING): Delete.
10689 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10690 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10691 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10692 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10693 (nios2_block_reg_padding): Return pad_direction instead of direction.
10694 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10695 instead of direction.
10696 (nios2_function_arg_padding): Likewise. Make static.
10697 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10698 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10699 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10700 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10701 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10702 (pa_function_arg_padding): Make static. Return pad_direction instead
10704 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10705 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10706 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10707 instead of direction. Use targetm.calls.function_arg_padding.
10708 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10709 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10710 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10711 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10712 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10714 (function_arg_padding): Rename to...
10715 (rs6000_function_arg_padding): ...this. Make static. Return
10716 pad_direction instead of direction.
10717 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10718 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10719 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10720 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10721 instead of direction. Use targetm.calls.function_arg_padding.
10722 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10723 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10724 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10725 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10726 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10727 (function_arg_padding): Rename to...
10728 (rs6000_function_arg_padding): ...this. Make static. Return
10729 pad_direction instead of direction.
10730 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10731 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10732 * config/s390/s390.c (s390_function_arg_padding): New function.
10733 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10734 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10735 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10736 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10737 (function_arg_padding): Rename to...
10738 (sparc_function_arg_padding): ...this. Make static. Return
10739 pad_direction instead of direction.
10740 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10741 * config/spu/spu.c (spu_function_arg_padding): New function.
10742 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10743 * system.h (FUNCTION_ARG_PADDING): Poison.
10745 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10746 Alan Hayward <alan.hayward@arm.com>
10747 David Sherwood <david.sherwood@arm.com>
10749 * target.def (modes_tieable_p): New hook.
10750 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10751 (TARGET_MODES_TIEABLE_P): ...this.
10752 * doc/tm.texi.in: Regenerate.
10753 * hooks.h (hook_bool_mode_mode_true): Declare.
10754 * hooks.c (hook_bool_mode_mode_true): New function.
10755 * combine.c (subst): Use targetm.modes_tieable_p instead of
10757 * dse.c (find_shift_sequence): Likewise.
10758 * expmed.c (extract_low_bits): Likewise.
10759 * lower-subreg.c: Include target.h.
10760 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10762 * rtlanal.c (rtx_cost): Likewise.
10763 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10764 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10765 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10766 (TARGET_MODES_TIEABLE_P): Redefine.
10767 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10768 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10769 (TARGET_MODES_TIEABLE_P): Redefine.
10770 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10771 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10772 (arc_modes_tieable_p): New function.
10773 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10774 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10775 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10776 (arm_modes_tieable_p): Make static.
10777 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10778 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10779 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10780 (TARGET_MODES_TIEABLE_P): Redefine.
10781 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10782 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10783 (TARGET_MODES_TIEABLE_P): Redefine.
10784 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10785 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10786 (cr16_modes_tieable_p): New function.
10787 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10788 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10789 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10790 (TRULY_NOOP_TRUNCATION): Update comment.
10791 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10792 (TRULY_NOOP_TRUNCATION): Update comment.
10793 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10794 (frv_modes_tieable_p): New function.
10795 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10796 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10797 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10798 (TARGET_MODES_TIEABLE_P): Redefine.
10799 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10800 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10801 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10802 (TARGET_MODES_TIEABLE_P): Redefine.
10803 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10804 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10805 (ia64_modes_tieable_p): New function.
10806 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10807 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10808 (iq2000_modes_tieable_p): New function.
10809 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10810 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10811 (lm32_modes_tieable_p): New function.
10812 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10813 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10814 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10815 (TARGET_MODES_TIEABLE_P): Redefine.
10816 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10817 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10818 (m32r_modes_tieable_p): New function.
10819 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10820 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10821 (m68k_modes_tieable_p): New function.
10822 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10823 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10824 (mcore_modes_tieable_p): New function.
10825 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10826 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10828 (TARGET_MODES_TIEABLE_P): Redefine.
10829 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10830 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10831 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10832 (TARGET_MODES_TIEABLE_P): Redefine.
10833 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10834 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10835 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10836 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10837 (mn10300_modes_tieable_p): ...this and make static.
10838 (TARGET_MODES_TIEABLE_P): Redefine.
10839 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10840 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10841 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10842 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10843 (msp430_modes_tieable_p): Make static.
10844 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10845 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10846 (TARGET_MODES_TIEABLE_P): Redefine.
10847 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10848 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10849 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10850 (TARGET_MODES_TIEABLE_P): Redefine.
10851 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10852 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10853 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10854 (TARGET_MODES_TIEABLE_P): Redefine.
10855 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10856 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10857 (pdp11_modes_tieable_p): New function.
10858 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10859 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10860 (rs6000_modes_tieable_p): New function.
10861 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10862 * config/powerpcspe/powerpcspe.md: Update comment.
10863 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10864 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10865 (TARGET_MODES_TIEABLE_P): Redefine.
10866 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10867 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10868 (rl78_modes_tieable_p): New function.
10869 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10870 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10871 (rs6000_modes_tieable_p): New function.
10872 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10873 * config/rs6000/rs6000.md: Update comment.
10874 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10875 * config/rx/rx.c (rx_modes_tieable_p): New function.
10876 (TARGET_MODES_TIEABLE_P): Redefine.
10877 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10878 * config/s390/s390.c (s390_modes_tieable_p): New function.
10879 (TARGET_MODES_TIEABLE_P): Redefine.
10880 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10881 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10882 (sh_modes_tieable_p): New function.
10883 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10884 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10885 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10886 (sparc_modes_tieable_p): Make static.
10887 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10888 * config/spu/spu.c (spu_modes_tieable_p): New function.
10889 (TARGET_MODES_TIEABLE_P): Redefine.
10890 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10891 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10892 (TARGET_MODES_TIEABLE_P): Redefine.
10893 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10894 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10895 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10896 * config/v850/v850.c (v850_modes_tieable_p): New function.
10897 (TARGET_MODES_TIEABLE_P): Redefine.
10898 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10899 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10900 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10901 (visium_modes_tieable_p): New function.
10902 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10903 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10904 (xtensa_modes_tieable_p): New function.
10905 * system.h (MODES_TIEABLE_P): Poison.
10907 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10908 Alan Hayward <alan.hayward@arm.com>
10909 David Sherwood <david.sherwood@arm.com>
10911 * target.def (hard_regno_mode_ok): New hook.
10912 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10913 (TARGET_HARD_REGNO_MODE_OK): ...this.
10914 * doc/tm.texi.in: Regenerate.
10915 * hooks.h (hook_bool_uint_mode_true): Declare.
10916 * hooks.c (hook_bool_uint_mode_true): New function.
10917 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10918 HARD_REGNO_MODE_OK.
10919 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10920 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10921 instead of HARD_REGNO_MODE_OK.
10922 * caller-save.c: Include target.h.
10923 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10924 HARD_REGNO_MODE_OK.
10925 * combine.c (can_combine_p): Likewise.
10926 (combinable_i3pat): Likewise.
10927 (can_change_dest_mode): Likewise.
10928 * expr.c (init_expr_target): Likewise.
10929 (convert_move): Likewise.
10930 (convert_modes): Likewise.
10931 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10932 (setup_prohibited_mode_move_regs): Likewise.
10933 * ira.h (target_ira): Likewise.
10934 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10935 * lra-constraints.c (process_alt_operands): Likewise.
10936 (split_reg): Likewise.
10937 * recog.c (peep2_find_free_register): Likewise.
10938 * ree.c (combine_reaching_defs): Likewise.
10939 * regcprop.c (maybe_mode_change): Likewise.
10940 * reginfo.c (init_reg_sets_1): Likewise.
10941 (choose_hard_reg_mode): Likewise.
10942 (simplifiable_subregs): Likewise.
10943 * regrename.c (check_new_reg_p): Likewise.
10944 * reload.c (find_valid_class): Likewise.
10945 (find_valid_class_1): Likewise.
10946 (reload_inner_reg_of_subreg): Likewise.
10947 (push_reload): Likewise.
10948 (combine_reloads): Likewise.
10949 (find_dummy_reload): Likewise.
10950 (find_reloads): Likewise.
10951 * reload1.c (find_reg): Likewise.
10952 (set_reload_reg): Likewise.
10953 (allocate_reload_reg): Likewise.
10954 (choose_reload_regs): Likewise.
10955 (reload_adjust_reg_for_temp): Likewise.
10956 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10957 (simplify_subreg_regno): Likewise.
10958 * sel-sched.c (init_regs_for_mode): Likewise.
10959 * varasm.c (make_decl_rtl): Likewise.
10960 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10961 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10962 HARD_REGNO_MODE_OK.
10963 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10964 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10966 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10967 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10968 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10969 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10970 (arc_mode_class): Delete.
10971 (HARD_REGNO_MODE_OK): Delete.
10972 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10973 (arc_hard_regno_mode_ok): Rename old array to...
10974 (arc_hard_regno_mode_ok_modes): ...this.
10975 (arc_conditional_register_usage): Update accordingly.
10976 (arc_mode_class): Make static.
10977 (arc_hard_regno_mode_ok): New function.
10978 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10979 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10980 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10981 (arm_hard_regno_mode_ok): Make static.
10982 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10983 HARD_REGNO_MODE_OK.
10984 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10985 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10986 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10989 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10990 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10991 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10992 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10993 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10994 * config/bfin/predicates.md (valid_reg_operand): Use
10995 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10996 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10997 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10998 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10999 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11000 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11001 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11002 (cr16_hard_regno_mode_ok): Make static and return a bool.
11003 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11004 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11005 (cris_hard_regno_mode_ok): New function.
11006 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11007 (epiphany_mode_class): Delete.
11008 (HARD_REGNO_MODE_OK): Delete.
11009 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11010 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11011 (hard_regno_mode_ok): Rename to...
11012 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11013 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11014 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11015 HARD_REGNO_MODE_OK.
11016 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11017 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11018 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11019 (frv_hard_regno_mode_ok): Make static and return a bool.
11020 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11021 HARD_REGNO_MODE_OK.
11022 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11023 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11024 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11025 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11027 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11028 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11029 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11030 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11032 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11033 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11034 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11035 (ia64_hard_regno_mode_ok): New function.
11036 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11037 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11038 (iq2000_hard_regno_mode_ok): New function.
11039 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11040 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11041 (lm32_hard_regno_mode_ok): New function.
11042 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11043 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11044 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11045 instead of HARD_REGNO_MODE_OK.
11046 (m32c_hard_regno_ok): Rename to...
11047 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11048 (m32c_cannot_change_mode_class): Update accordingly.
11049 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11050 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11051 (m32r_mode_class): Delete.
11052 (HARD_REGNO_MODE_OK): Delete.
11053 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11054 (m32r_hard_regno_mode_ok): Rename to...
11055 (m32r_hard_regno_modes): ...this.
11056 (m32r_mode_class): Make static.
11057 (m32r_hard_regno_mode_ok): New function.
11058 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11059 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11060 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11061 (m68k_hard_regno_mode_ok): Make static.
11062 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11063 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11064 (mcore_hard_regno_mode_ok): New function.
11065 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11066 (HARD_REGNO_MODE_OK): Delete.
11067 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11069 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11070 (microblaze_hard_regno_mode_ok): New function.
11071 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11072 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11073 (mips_hard_regno_mode_ok): Delete.
11074 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11075 (mips_hard_regno_mode_ok_p): ...this and make static.
11076 (mips_hard_regno_mode_ok_p): Rename to...
11077 (mips_hard_regno_mode_ok_uncached): ...this.
11078 (mips_hard_regno_mode_ok): New function.
11079 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11080 of HARD_REGNO_MODE_OK.
11081 (mips_option_override): Update after above name changes.
11082 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11083 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11084 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11085 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11086 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11087 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11088 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11089 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11090 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11091 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11092 (msp430_hard_regno_mode_ok): Make static and return a bool.
11093 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11094 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11095 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11097 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11098 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11099 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11100 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11101 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11102 (PA_HARD_REGNO_MODE_OK): ...this
11103 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11104 (PA_HARD_REGNO_MODE_OK): ...this.
11105 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11106 (pa_hard_regno_mode_ok): New function.
11107 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11108 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11109 (pdp11_hard_regno_mode_ok): New function.
11110 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11111 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11113 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11115 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11116 (rs6000_hard_regno_mode_ok): Rename to...
11117 (rs6000_hard_regno_mode_ok_uncached): ...this.
11118 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11119 (rs6000_hard_regno_mode_ok): New function.
11120 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11121 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11122 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11123 (riscv_hard_regno_mode_ok): ...this and make static.
11124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11125 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11126 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11127 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11128 (rl78_hard_regno_mode_ok): Make static and return bool.
11129 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11130 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11132 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11133 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11134 (rs6000_hard_regno_mode_ok): Rename to...
11135 (rs6000_hard_regno_mode_ok_uncached): ...this.
11136 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11137 (rs6000_hard_regno_mode_ok): New function.
11138 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11139 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11141 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11142 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11143 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11144 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11145 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11146 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11147 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11148 (sh_hard_regno_mode_ok): Make static.
11149 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11150 instead of HARD_REGNO_MODE_OK.
11151 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11152 (sparc_mode_class): Delete.
11153 (HARD_REGNO_MODE_OK): Delete.
11154 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11155 (hard_regno_mode_classes): Make static.
11156 (sparc_mode_class): Likewise.
11157 (sparc_hard_regno_mode_ok): New function.
11158 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11159 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11160 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11162 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11163 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11164 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11165 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11166 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11167 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11168 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11169 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11170 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11171 (visium_hard_regno_mode_ok): New function.
11172 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11173 instead of HARD_REGNO_MODE_OK.
11174 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11175 (HARD_REGNO_MODE_OK): Delete.
11176 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11177 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11178 (xtensa_option_override): Update accordingly.
11179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11180 (xtensa_hard_regno_mode_ok): New function.
11181 * system.h (HARD_REGNO_MODE_OK): Poison.
11183 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11184 Alan Hayward <alan.hayward@arm.com>
11185 David Sherwood <david.sherwood@arm.com>
11187 * target.def (hard_regno_call_part_clobbered): New hook.
11188 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11189 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11190 * doc/tm.texi: Regenerate.
11191 * hooks.h (hook_bool_uint_mode_false): Declare.
11192 * hooks.c (hook_bool_uint_mode_false): New function.
11193 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11194 * cselib.c (cselib_process_insn): Use
11195 targetm.hard_regno_call_part_clobbered instead of
11196 HARD_REGNO_CALL_PART_CLOBBERED.
11197 * ira-conflicts.c (ira_build_conflicts): Likewise.
11198 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11199 * lra-constraints.c (need_for_call_save_p): Likewise.
11200 * lra-lives.c: Include target.h.
11201 (check_pseudos_live_through_calls): Use
11202 targetm.hard_regno_call_part_clobbered instead of
11203 HARD_REGNO_CALL_PART_CLOBBERED.
11204 * regcprop.c: Include target.h.
11205 (copyprop_hardreg_forward_1): Use
11206 targetm.hard_regno_call_part_clobbered instead of
11207 HARD_REGNO_CALL_PART_CLOBBERED.
11208 * reginfo.c (choose_hard_reg_mode): Likewise.
11209 * regrename.c (check_new_reg_p): Likewise.
11210 * reload.c (find_equiv_reg): Likewise.
11211 * reload1.c (emit_reload_insns): Likewise.
11212 * sched-deps.c (deps_analyze_insn): Likewise.
11213 * sel-sched.c (init_regs_for_mode): Likewise.
11214 (mark_unavailable_hard_regs): Likewise.
11215 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11216 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11217 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11219 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11220 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11221 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11223 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11225 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11226 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11227 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11229 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11230 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11231 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11233 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11234 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11236 * config/powerpcspe/powerpcspe.c
11237 (rs6000_hard_regno_call_part_clobbered): New function.
11238 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11239 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11240 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11242 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11243 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11244 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11246 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11247 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11248 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11250 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11251 Alan Hayward <alan.hayward@arm.com>
11252 David Sherwood <david.sherwood@arm.com>
11254 * rtl.h (subreg_memory_offset): Declare.
11255 * emit-rtl.c (subreg_memory_offset): New function.
11256 * expmed.c (store_bit_field_1): Use it.
11257 * expr.c (undefined_operand_subword_p): Likewise.
11258 * simplify-rtx.c (simplify_subreg): Likewise.
11260 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11262 PR rtl-optimization/57448
11265 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11266 using atomic_load pattern.
11267 (expand_atomic_store): Likewise.
11269 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11272 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11273 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11274 handling. Use replace_call_with_value with NULL instead of
11275 gsi_replace, unlink_stmt_vdef and release_defs.
11277 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11280 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11282 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11285 * rtl.h (emit_library_call_value_1): Declare.
11286 (emit_library_call): Replace declaration with a series of overloads.
11287 Remove the parameter count argument.
11288 (emit_library_call_value): Likewise.
11289 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11290 with an "rtx_mode_t *".
11291 (emit_library_call_value): Delete.
11292 (emit_library_call): Likewise.
11293 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11294 (asan_emit_allocas_unpoison): Likewise.
11295 * builtins.c (expand_builtin_powi): Likewise.
11296 (expand_asan_emit_allocas_unpoison): Likewise.
11297 * cfgexpand.c (expand_main_function): Likewise.
11298 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11299 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11300 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11301 * config/arm/arm.c (arm_trampoline_init): Likewise.
11302 (arm_call_tls_get_addr): Likewise.
11303 (arm_expand_divmod_libfunc): Likewise.
11304 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11305 (smulsi3_highpart): Likewise.
11306 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11307 (c6x_expand_compare): Likewise.
11308 (c6x_expand_movmem): Likewise.
11309 * config/frv/frv.c (frv_trampoline_init): Likewise.
11310 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11311 (ix86_expand_divmod_libfunc): Likewise.
11312 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11313 (ia64_expand_compare): Likewise.
11314 (ia64_profile_hook): Likewise.
11315 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11316 (nonlocal_goto): Likewise.
11317 (restore_stack_nonlocal): Likewise.
11318 * config/m32r/m32r.c (block_move_call): Likewise.
11319 (m32r_trampoline_init): Likewise.
11320 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11321 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11322 (m68k_call_m68k_read_tp): Likewise.
11323 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11324 (microblaze_expand_divide): Likewise.
11325 * config/mips/mips.h (mips_args): Likewise.
11326 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11327 (MIPS_ICACHE_SYNC): Likewise.
11328 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11329 (nios2_trampoline_init): Likewise.
11330 * config/pa/pa.c (hppa_tls_call): Likewise.
11331 (pa_trampoline_init): Likewise.
11332 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11333 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11334 (expand_strn_compare): Likewise.
11335 (rs6000_generate_compare): Likewise.
11336 (rs6000_expand_float128_convert): Likewise.
11337 (output_profile_hook): Likewise.
11338 (rs6000_trampoline_init): Likewise.
11339 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11340 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11341 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11342 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11343 (rs6000_generate_compare): Likewise.
11344 (rs6000_expand_float128_convert): Likewise.
11345 (output_profile_hook): Likewise.
11346 (rs6000_trampoline_init): Likewise.
11347 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11348 * config/sh/sh.c (sh_trampoline_init): Likewise.
11349 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11350 (sparc_emit_float_lib_cmp): Likewise.
11351 (sparc32_initialize_trampoline): Likewise.
11352 (sparc64_initialize_trampoline): Likewise.
11353 (sparc_profile_hook): Likewise.
11354 * config/spu/spu.c (ea_load_store): Likewise.
11355 * config/spu/spu.md (floatunssidf2): Likewise.
11356 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11357 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11358 * config/visium/visium.c (expand_block_move_4): Likewise.
11359 (expand_block_move_2): Likewise.
11360 (expand_block_move_1): Likewise.
11361 (expand_block_set_4): Likewise.
11362 (expand_block_set_2): Likewise.
11363 (expand_block_set_1): Likewise.
11364 (visium_trampoline_init): Likewise.
11365 (visium_profile_hook): Likewise.
11366 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11367 (xtensa_setup_frame_addresses): Likewise.
11368 (xtensa_trampoline_init): Likewise.
11369 * except.c (sjlj_emit_function_enter): Likewise.
11370 (sjlj_emit_function_exit): Likewise.
11371 * explow.c (allocate_dynamic_stack_space): Likewise.
11372 (probe_stack_range): Likewise.
11373 * expr.c (convert_mode_scalar): Likewise.
11374 * optabs.c (expand_binop): Likewise.
11375 (expand_twoval_binop_libfunc): Likewise.
11376 (expand_unop): Likewise.
11377 (prepare_cmp_insn): Likewise.
11378 (prepare_float_lib_cmp): Likewise.
11379 (expand_float): Likewise.
11380 (expand_fix): Likewise.
11381 (expand_fixed_convert): Likewise.
11382 (maybe_emit_sync_lock_test_and_set): Likewise.
11383 (expand_atomic_compare_and_swap): Likewise.
11384 (expand_mem_thread_fence): Likewise.
11385 (expand_atomic_fetch_op): Likewise.
11387 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11389 * doc/generic.texi (OpenACC): Adjust URL.
11390 * doc/invoke.texi (C Dialect Options): Ditto.
11392 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11394 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11395 predicate for operand 1. Add (m,<S>) constraint.
11396 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11397 Prevent memory operand 1 with register operand 2.
11399 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11401 PR rtl-optimization/82024
11402 * combine.c (try_combine): If the combination result is a PARALLEL,
11403 and we only need to retain the SET in there that would be placed
11404 at I2, check that we can place that at I3 instead, before doing so.
11406 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11409 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11411 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11412 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11415 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11416 Jeff Law <law@redhat.com>
11418 * varasm.c (bss_initializer_p): Do not put constants into .bss
11419 (categorize_decl_for_section): Handle bss_initializer_p returning
11420 false when DECL_INITIAL is NULL.
11422 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11425 * config/s390/s390.c (s390_can_inline_p): New function.
11427 2017-09-01 Jeff Law <law@redhat.com>
11429 PR tree-optimization/82052
11430 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11431 Always initialize the returned slot after a hash table miss
11432 when INSERT is true.
11434 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11436 * config/s390/s390.md (mem_signal_fence): Remove.
11437 * doc/md.texi (mem_signal_fence): Remove.
11438 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11440 * target-insns.def (mem_signal_fence): Remove.
11442 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11445 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11448 * asan.c (create_odr_indicator): Strip name encoding from assembler
11449 name before appending it after __odr_asan_.
11451 2017-09-01 Martin Liska <mliska@suse.cz>
11453 PR tree-optimization/82059
11454 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11455 frequency only when an edge is redirected.
11457 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11459 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11460 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11461 (arc_conditional_register_usage): Remove ARC600 lp_count
11463 (arc_file_start): Emit Tag_ARC_CPU_variation.
11464 (arc_can_use_doloop_p): New conditions to use ZOLs.
11465 (hwloop_fail): New function.
11466 (hwloop_optimize): Likewise.
11467 (hwloop_pattern_reg): Likewise.
11468 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11469 (arc_reorg_loops): New function, calls reorg_loops.
11470 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11471 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11473 (arc_loop_hazard): Remove function, functionality moved into
11475 (arc_hazard): Remove arc_loop_hazard call.
11476 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11477 into hwloop_optimize.
11478 (arc_label_align): Remove ZOL handling.
11479 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11480 * config/arc/arc.md (doloop_begin): Remove pattern.
11481 (doloop_begin_i): Likewise.
11482 (doloop_end_i): Likewise.
11483 (doloop_fallback): Likewise.
11484 (doloop_fallback_m): Likewise.
11485 (doloop_end): Reimplement expand.
11486 (arc_lp): New pattern for LP instruction.
11487 (loop_end): New pattern.
11488 (loop_fail): Likewise.
11489 (decrement_and_branch_until_zero): Likewise.
11490 * config/arc/arc.opt (mlpc-width): New option.
11491 * doc/invoke.texi (mlpc-width): Document option.
11493 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11495 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11496 (arc_ccfsm_advance): Fix checking for delay slots.
11497 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11499 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11501 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11503 (movhi_insn): Update store instruction constraint which are saving
11504 6-bit short immediates.
11505 (movsi_insn): Consider also short scaled load operations.
11506 (zero_extendhisi2_i): Use Usd constraint instead of T.
11507 (extendhisi2_i): Add q constraint.
11508 (arc_clzsi2): Add type and length attributes.
11509 (arc_ctzsi2): Likewise.
11510 * config/arc/constraints.md (Usc): Update constraint, the
11511 assembler can parse two relocations for a single instruction.
11513 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11515 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11516 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11518 2017-08-31 Olivier Hainque <hainque@adacore.com>
11520 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11521 match as powerpc-wrs-vxworks*.
11523 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11525 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11526 register constraint for by-element operand.
11527 (aarch64_mls_elt_merge<mode>): Likewise.
11529 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11531 * config/arc/arc.c (arc_can_follow_jump): Check for short
11534 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11536 * config.gcc: Use g.opt for arc.
11537 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11538 functionality moved to ...
11539 (legitimate_scaled_address_p): New function, ...here.
11540 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11541 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11542 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11544 (arc_override_options): Handle G option.
11545 (arc_output_pic_addr_const): Correct function definition.
11546 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11547 (arc_decl_anon_ns_mem_p): Delete.
11548 (arc_in_small_data_p): Overhaul this function to take into
11549 consideration the value given via G option.
11550 (arc_rewrite_small_data_1): Renamed and corrected old
11551 arc_rewrite_small_data function.
11552 (arc_rewrite_small_data): New function.
11553 (small_data_pattern): Don't use pic_offset_table_rtx.
11554 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11555 * config/arc/simdext.md (movmisalignv2hi): Use
11556 prepare_move_operands function.
11558 (movmisalign*): Likewise.
11559 * doc/invoke.texi (ARC options): Document -G option.
11561 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11563 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11565 * config/arc/arc.c (arc_print_operand): Output scalled address for
11566 sdata whenever is possible.
11567 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11568 load/stores are available.
11569 (compact_sda_memory_operand): Check for the alignment required by
11570 code density instructions.
11571 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11573 * config/arc/constraints.md (Usd): Update constraint.
11574 (Us0): New constraint.
11575 (Usc): Update constraint.
11577 2017-08-31 Richard Biener <rguenther@suse.de>
11579 PR middle-end/82054
11580 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11581 function only once.
11583 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11585 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11586 Resize type_signature.
11588 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11589 Alan Hayward <alan.hayward@arm.com>
11590 David Sherwood <david.sherwood@arm.com>
11592 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11593 subregs whose inner modes can be stored in GPRs.
11594 (aarch64_classify_index): Likewise.
11596 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11597 Alan Hayward <alan.hayward@arm.com>
11598 David Sherwood <david.sherwood@arm.com>
11600 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11601 (V_INT_EQUIV): ...this.
11602 (v_cmp_result): Rename to...
11603 (v_int_equiv): ...this.
11604 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11605 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11606 (copysign<mode>3): Likewise.
11607 (aarch64_simd_bsl<mode>_internal): Likewise.
11608 (aarch64_simd_bsl<mode>): Likewise.
11609 (vec_cmp<mode><mode>): Likewise.
11610 (vcond<mode><mode>): Likewise.
11611 (vcond<v_cmp_mixed><mode>): Likewise.
11612 (vcondu<mode><v_cmp_mixed>): Likewise.
11613 (aarch64_cm<optab><mode>): Likewise.
11614 (aarch64_cmtst<mode>): Likewise.
11615 (aarch64_fac<optab><mode>): Likewise.
11616 (vec_perm_const<mode>): Likewise.
11617 (vcond_mask_<mode><v_cmp_result>): Rename to...
11618 (vcond_mask_<mode><v_int_equiv>): ...this.
11619 (vec_cmp<mode><v_cmp_result>): Rename to...
11620 (vec_cmp<mode><v_int_equiv>): ...this.
11622 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11623 Alan Hayward <alan.hayward@arm.com>
11624 David Sherwood <david.sherwood@arm.com>
11626 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11628 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11629 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11630 (UNSPEC_LD4_DREG): New unspecs.
11631 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11632 (aarch64_ld2<mode>_dreg_be): Replace with...
11633 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11635 (aarch64_ld3<mode>_dreg_le)
11636 (aarch64_ld3<mode>_dreg_be): Replace with...
11637 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11639 (aarch64_ld4<mode>_dreg_le)
11640 (aarch64_ld4<mode>_dreg_be): Replace with...
11641 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11644 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11646 PR tree-optimization/81987
11647 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11648 insert an initializer in a location not dominated by the stride
11651 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11653 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11654 on the entire header of the finally block in the fallthru case.
11656 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11658 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11660 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11662 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11663 rs6000_emit_move_from_cr and call renamed function.
11664 (rs6000_emit_prologue): Call renamed functions.
11665 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11666 movesi_from_cr, remove volatile CRs.
11668 2017-08-30 Jon Beniston <jon@beniston.com>
11669 Richard Biener <rguenther@suse.de>
11671 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11672 of VECTOR_MODE_P check.
11673 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11674 element vector types.
11676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11678 * df.h (df_read_modify_subreg_p): Remove in favor of...
11679 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11680 const_rtx instead of an rtx.
11681 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11682 * df-problems.c (df_word_lr_mark_ref): Likewise.
11683 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11684 (mark_pseudo_reg_dead): Likewise.
11685 (mark_ref_dead): Likewise.
11686 * reginfo.c (init_subregs_of_mode): Likewise.
11687 * sched-deps.c (sched_analyze_1): Likewise.
11688 * df-scan.c (df_def_record_1): Likewise.
11689 (df_uses_record): Likewise.
11690 (df_read_modify_subreg_p): Remove in favor of...
11691 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11692 const_rtx instead of an rtx.
11694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11695 Alan Hayward <alan.hayward@arm.com>
11696 David Sherwood <david.sherwood@arm.com>
11698 * rtl.h (partial_subreg_p): New function.
11699 * caller-save.c (save_call_clobbered_regs): Use it.
11700 * calls.c (expand_call): Likewise.
11701 * combine.c (combinable_i3pat): Likewise.
11702 (simplify_set): Likewise.
11703 (make_extraction): Likewise.
11704 (make_compound_operation_int): Likewise.
11705 (gen_lowpart_or_truncate): Likewise.
11706 (force_to_mode): Likewise.
11707 (make_field_assignment): Likewise.
11708 (reg_truncated_to_mode): Likewise.
11709 (record_truncated_value): Likewise.
11710 (move_deaths): Likewise.
11711 * cse.c (record_jump_cond): Likewise.
11712 (cse_insn): Likewise.
11713 * cselib.c (cselib_lookup_1): Likewise.
11714 * expmed.c (extract_bit_field_using_extv): Likewise.
11715 * function.c (assign_parm_setup_reg): Likewise.
11716 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11717 * ira-build.c (create_insn_allocnos): Likewise.
11718 * lra-coalesce.c (merge_pseudos): Likewise.
11719 * lra-constraints.c (match_reload): Likewise.
11720 (simplify_operand_subreg): Likewise.
11721 (curr_insn_transform): Likewise.
11722 * lra-lives.c (process_bb_lives): Likewise.
11723 * lra.c (new_insn_reg): Likewise.
11724 (lra_substitute_pseudo): Likewise.
11725 * regcprop.c (mode_change_ok): Likewise.
11726 (maybe_mode_change): Likewise.
11727 (copyprop_hardreg_forward_1): Likewise.
11728 * reload.c (push_reload): Likewise.
11729 (find_reloads): Likewise.
11730 (find_reloads_subreg_address): Likewise.
11731 * reload1.c (alter_reg): Likewise.
11732 (eliminate_regs_1): Likewise.
11733 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11735 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11737 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11738 back to if statements, including unpack.
11740 2017-08-30 Martin Liska <mliska@suse.cz>
11742 PR inline-asm/82001
11743 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11745 (func_checker::compare_asm_inputs_outputs): ... this function.
11746 (func_checker::compare_gimple_asm): Use the function to compare
11747 also ASM constrains.
11748 * ipa-icf-gimple.h: Rename the function.
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 * coretypes.h (complex_mode): New type.
11755 * gdbhooks.py (build_pretty_printer): Handle it.
11756 * machmode.h (complex_mode): New class.
11757 (complex_mode::includes_p): New function.
11758 (is_complex_int_mode): Likewise.
11759 (is_complex_float_mode): Likewise.
11760 * genmodes.c (get_mode_class): Handle complex mode classes.
11761 * function.c (expand_function_end): Use is_complex_int_mode.
11763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11764 Alan Hayward <alan.hayward@arm.com>
11765 David Sherwood <david.sherwood@arm.com>
11767 * coretypes.h (scalar_mode_pod): New typedef.
11768 * gdbhooks.py (build_pretty_printer): Handle it.
11769 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11770 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11771 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11772 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11773 as_a <scalar_mode>.
11775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11776 Alan Hayward <alan.hayward@arm.com>
11777 David Sherwood <david.sherwood@arm.com>
11779 * machmode.h (mode_for_vector): Take a scalar_mode instead
11781 * stor-layout.c (mode_for_vector): Likewise.
11782 * explow.c (promote_mode): Use as_a <scalar_mode>.
11783 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11786 Alan Hayward <alan.hayward@arm.com>
11787 David Sherwood <david.sherwood@arm.com>
11789 * target.def (preferred_simd_mode): Take a scalar_mode
11790 instead of a machine_mode.
11791 * targhooks.h (default_preferred_simd_mode): Likewise.
11792 * targhooks.c (default_preferred_simd_mode): Likewise.
11793 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11794 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11795 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11796 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11797 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11798 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11799 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11800 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11802 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11803 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11804 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11805 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11806 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11807 * doc/tm.texi: Regenerate.
11808 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11812 Alan Hayward <alan.hayward@arm.com>
11813 David Sherwood <david.sherwood@arm.com>
11815 * target.def (scalar_mode_supported_p): Take a scalar_mode
11816 instead of a machine_mode.
11817 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11818 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11819 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11820 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11821 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11822 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11823 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11824 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11825 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11826 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11827 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11828 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11829 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11830 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11832 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11833 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11834 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11835 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11836 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11838 * doc/tm.texi: Regenerate.
11840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11841 Alan Hayward <alan.hayward@arm.com>
11842 David Sherwood <david.sherwood@arm.com>
11844 * coretypes.h (opt_scalar_mode): New typedef.
11845 * gdbhooks.py (build_pretty_printers): Handle it.
11846 * machmode.h (mode_iterator::get_2xwider): Add overload for
11848 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11850 * expr.c (convert_mode_scalar): Likewise.
11851 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11852 * optabs.c (expand_float): Likewise.
11853 (expand_fix): Likewise.
11854 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11857 Alan Hayward <alan.hayward@arm.com>
11858 David Sherwood <david.sherwood@arm.com>
11860 * optabs.c (expand_float): Explicitly check for scalars before
11861 using a branching expansion.
11862 (expand_fix): Likewise.
11864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11865 Alan Hayward <alan.hayward@arm.com>
11866 David Sherwood <david.sherwood@arm.com>
11868 * expr.c (convert_mode): Split scalar handling out into...
11869 (convert_mode_scalar): ...this new function. Treat the modes
11872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11873 Alan Hayward <alan.hayward@arm.com>
11874 David Sherwood <david.sherwood@arm.com>
11876 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11878 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11881 Alan Hayward <alan.hayward@arm.com>
11882 David Sherwood <david.sherwood@arm.com>
11884 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11885 rather than a machine_mode.
11886 (fixed_from_string): Likewise.
11887 (fixed_convert): Likewise.
11888 (fixed_convert_from_int): Likewise.
11889 (fixed_convert_from_real): Likewise.
11890 (real_convert_from_fixed): Likewise.
11891 * fixed-value.c (fixed_from_double_int): Likewise.
11892 (fixed_from_string): Likewise.
11893 (fixed_convert): Likewise.
11894 (fixed_convert_from_int): Likewise.
11895 (fixed_convert_from_real): Likewise.
11896 (real_convert_from_fixed): Likewise.
11897 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11900 Alan Hayward <alan.hayward@arm.com>
11901 David Sherwood <david.sherwood@arm.com>
11903 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11904 of separate mode class checks. Do not allow vector modes here.
11905 (immed_wide_int_const): Use as_a <scalar_mode>.
11906 * explow.c (trunc_int_for_mode): Likewise.
11907 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11908 (wi::shwi): Likewise.
11909 (wi::min_value): Likewise.
11910 (wi::max_value): Likewise.
11911 * dwarf2out.c (loc_descriptor): Likewise.
11912 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11913 for CONST_WIDE_INT.
11915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11916 Alan Hayward <alan.hayward@arm.com>
11917 David Sherwood <david.sherwood@arm.com>
11919 * tree.h (SCALAR_TYPE_MODE): New macro.
11920 * expr.c (expand_expr_addr_expr_1): Use it.
11921 (expand_expr_real_2): Likewise.
11922 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11923 (fold_convert_const_fixed_from_int): Likewise.
11924 (fold_convert_const_fixed_from_real): Likewise.
11925 (native_encode_fixed): Likewise
11926 (native_encode_complex): Likewise
11927 (native_encode_vector): Likewise.
11928 (native_interpret_fixed): Likewise.
11929 (native_interpret_real): Likewise.
11930 (native_interpret_complex): Likewise.
11931 (native_interpret_vector): Likewise.
11932 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11933 (simd_clone_adjust_argument_types): Likewise.
11934 (simd_clone_init_simd_arrays): Likewise.
11935 (simd_clone_adjust): Likewise.
11936 * stor-layout.c (layout_type): Likewise.
11937 * tree.c (build_minus_one_cst): Likewise.
11938 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11939 * tree-inline.c (estimate_move_cost): Likewise.
11940 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11941 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11942 (vectorizable_reduction): Likewise.
11943 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11944 (vect_recog_mixed_size_cond_pattern): Likewise.
11945 (check_bool_pattern): Likewise.
11946 (adjust_bool_pattern): Likewise.
11947 (search_type_for_mask_1): Likewise.
11948 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11949 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11950 (vectorizable_load): Likewise.
11951 (vectorizable_store): Likewise.
11952 * ubsan.c (ubsan_encode_value): Likewise.
11953 * varasm.c (output_constant): Likewise.
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 * coretypes.h (scalar_mode): New class.
11960 * machmode.h (scalar_mode): Likewise.
11961 (scalar_mode::includes_p): New function.
11962 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11963 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11964 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11965 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11966 * expmed.c (store_bit_field_1): Likewise.
11967 (extract_bit_field_1): Likewise.
11968 * expr.c (write_complex_part): Likewise.
11969 (read_complex_part): Likewise.
11970 (emit_move_complex_push): Likewise.
11971 (expand_expr_real_2): Likewise.
11972 * function.c (assign_parm_setup_reg): Likewise.
11973 (assign_parms_unsplit_complex): Likewise.
11974 * optabs.c (expand_binop): Likewise.
11975 * rtlanal.c (subreg_get_info): Likewise.
11976 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11977 * varasm.c (output_constant_pool_2): Likewise.
11979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11980 Alan Hayward <alan.hayward@arm.com>
11981 David Sherwood <david.sherwood@arm.com>
11983 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11985 (expmed_mult_highpart_optab): Likewise.
11986 (expmed_mult_highpart): Likewise.
11988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11989 Alan Hayward <alan.hayward@arm.com>
11990 David Sherwood <david.sherwood@arm.com>
11992 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11993 instead of a machine_mode.
11994 (builtin_memset_read_str): Likewise.
11995 * builtins.c (c_readstr): Likewise.
11996 (builtin_memcpy_read_str): Likewise.
11997 (builtin_strncpy_read_str): Likewise.
11998 (builtin_memset_read_str): Likewise.
11999 (builtin_memset_gen_str): Likewise.
12000 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12001 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12002 instead of a machine_mode.
12003 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12005 (make_extraction): Likewise.
12006 (try_widen_shift_mode): Take and return scalar_int_modes instead
12008 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12009 a scalar_int_mode instead of a machine_mode.
12010 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12011 (avr_addr_space_pointer_mode): Likewise.
12012 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12013 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12014 (msp430_unwind_word_mode): Likewise.
12015 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12016 (spu_addr_space_pointer_mode): Likewise.
12017 (spu_addr_space_address_mode): Likewise.
12018 (spu_libgcc_cmp_return_mode): Likewise.
12019 (spu_libgcc_shift_count_mode): Likewise.
12020 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12021 (rl78_addr_space_pointer_mode): Likewise.
12022 (fl78_unwind_word_mode): Likewise.
12023 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12025 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12026 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12027 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12028 (mips_valid_pointer_mode): Likewise.
12029 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12030 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12031 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12033 (ft32_addr_space_address_mode): Likewise.
12034 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12035 scalar_int_mode instead of a machine_mode.
12036 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12038 (m32c_addr_space_address_mode): Likewise.
12039 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12040 (rs6000_eh_return_filter_mode): Likewise.
12041 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12042 (rs6000_eh_return_filter_mode): Likewise.
12043 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12044 (s390_libgcc_shift_count_mode): Likewise.
12045 (s390_unwind_word_mode): Likewise.
12046 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12048 * target.def (mode_rep_extended): Likewise.
12049 (valid_pointer_mode): Likewise.
12050 (addr_space.valid_pointer_mode): Likewise.
12051 (eh_return_filter_mode): Return a scalar_int_mode rather than
12053 (libgcc_cmp_return_mode): Likewise.
12054 (libgcc_shift_count_mode): Likewise.
12055 (unwind_word_mode): Likewise.
12056 (addr_space.pointer_mode): Likewise.
12057 (addr_space.address_mode): Likewise.
12058 * doc/tm.texi: Regenerate.
12059 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12061 (do_jump): Use scalar_int_mode for local variables.
12062 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12063 rather than a machine_mode.
12064 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12065 (scompare_loc_descriptor_wide): Likewise.
12066 (scompare_loc_descriptor_narrow): Likewise.
12067 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12069 * except.c (sjlj_emit_dispatch_table): Likewise.
12070 (expand_builtin_eh_copy_values): Likewise.
12071 * explow.c (convert_memory_address_addr_space_1): Likewise.
12072 Take a scalar_int_mode rather than a machine_mode.
12073 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12074 than a machine_mode.
12075 (memory_address_addr_space): Use scalar_int_mode for local variables.
12076 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12077 rather than a machine_mode.
12078 * expmed.c (mask_rtx): Likewise.
12079 (init_expmed_one_conv): Likewise.
12080 (expand_mult_highpart_adjust): Likewise.
12081 (extract_high_half): Likewise.
12082 (expmed_mult_highpart_optab): Likewise.
12083 (expmed_mult_highpart): Likewise.
12084 (expand_smod_pow2): Likewise.
12085 (expand_sdiv_pow2): Likewise.
12086 (emit_store_flag_int): Likewise.
12087 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12089 (extract_low_bits): Likewise.
12090 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12092 * expr.c (pieces_addr::adjust): Likewise.
12093 (can_store_by_pieces): Likewise.
12094 (store_by_pieces): Likewise.
12095 (clear_by_pieces_1): Likewise.
12096 (expand_expr_addr_expr_1): Likewise.
12097 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12098 (expand_expr_real_1): Likewise.
12099 (try_casesi): Likewise.
12100 * final.c (shorten_branches): Likewise.
12101 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12102 type of "mode" to machine_mode.
12103 * internal-fn.c (expand_arith_overflow_result_store): Take a
12104 scalar_int_mode rather than a machine_mode.
12105 (expand_mul_overflow): Use scalar_int_mode for local variables.
12106 * loop-doloop.c (doloop_modify): Likewise.
12107 (doloop_optimize): Likewise.
12108 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12109 than a machine_mode.
12110 (expand_doubleword_shift_condmove): Likewise.
12111 (expand_doubleword_shift): Likewise.
12112 (expand_doubleword_clz): Likewise.
12113 (expand_doubleword_popcount): Likewise.
12114 (expand_doubleword_parity): Likewise.
12115 (expand_absneg_bit): Use scalar_int_mode for local variables.
12116 (prepare_float_lib_cmp): Likewise.
12117 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12118 rather than a machine_mode.
12119 (convert_memory_address_addr_space): Likewise.
12120 (get_mode_bounds): Likewise.
12121 (get_address_mode): Return a scalar_int_mode rather than a
12123 * rtlanal.c (get_address_mode): Likewise.
12124 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12125 than a machine_mode.
12126 * targhooks.c (default_mode_rep_extended): Likewise.
12127 (default_valid_pointer_mode): Likewise.
12128 (default_addr_space_valid_pointer_mode): Likewise.
12129 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12130 than a machine_mode.
12131 (default_libgcc_cmp_return_mode): Likewise.
12132 (default_libgcc_shift_count_mode): Likewise.
12133 (default_unwind_word_mode): Likewise.
12134 (default_addr_space_pointer_mode): Likewise.
12135 (default_addr_space_address_mode): Likewise.
12136 * targhooks.h (default_eh_return_filter_mode): Likewise.
12137 (default_libgcc_cmp_return_mode): Likewise.
12138 (default_libgcc_shift_count_mode): Likewise.
12139 (default_unwind_word_mode): Likewise.
12140 (default_addr_space_pointer_mode): Likewise.
12141 (default_addr_space_address_mode): Likewise.
12142 (default_mode_rep_extended): Take a scalar_int_mode rather than
12144 (default_valid_pointer_mode): Likewise.
12145 (default_addr_space_valid_pointer_mode): Likewise.
12146 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12148 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12149 rather than a machine_mode.
12150 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12151 for local variables.
12152 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12153 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12154 than a machine_mode.
12156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12157 Alan Hayward <alan.hayward@arm.com>
12158 David Sherwood <david.sherwood@arm.com>
12160 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12161 the mode argument to scalar_int_mode.
12162 (do_jump_by_parts_zero_rtx): Likewise.
12163 (do_jump_by_parts_equality_rtx): Likewise.
12164 (do_jump_by_parts_greater): Take a mode argument.
12165 (do_jump_by_parts_equality): Likewise.
12166 (do_jump_1): Update calls accordingly.
12168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12169 Alan Hayward <alan.hayward@arm.com>
12170 David Sherwood <david.sherwood@arm.com>
12172 * is-a.h (safe_dyn_cast): New function.
12173 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12174 (jump_table_for_label): Likewise.
12175 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12176 instead of an rtx_insn *.
12177 (shorten_branches): Use dyn_cast instead of LABEL_P and
12178 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12179 rtx_jump_table_data::get_data_mode.
12180 (final_scan_insn): Likewise.
12182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12183 Alan Hayward <alan.hayward@arm.com>
12184 David Sherwood <david.sherwood@arm.com>
12186 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12187 trying to combine a full-register integer set with a subreg
12190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12191 Alan Hayward <alan.hayward@arm.com>
12192 David Sherwood <david.sherwood@arm.com>
12194 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12195 that is always either address_mode or pointer_mode.
12197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12198 Alan Hayward <alan.hayward@arm.com>
12199 David Sherwood <david.sherwood@arm.com>
12201 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12202 when the two are known to be equal.
12204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12205 Alan Hayward <alan.hayward@arm.com>
12206 David Sherwood <david.sherwood@arm.com>
12208 * simplify-rtx.c (simplify_const_unary_operation): Use
12209 is_a <scalar_int_mode> instead of checking for a nonzero
12210 precision. Forcibly convert op_mode to a scalar_int_mode
12211 in that case. More clearly differentiate the operand and
12212 result modes and use the former when deciding what the value
12213 of a count-bits operation should be. Use is_int_mode instead
12214 of checking for a MODE_INT. Remove redundant check for whether
12215 this mode has a zero precision.
12217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12218 Alan Hayward <alan.hayward@arm.com>
12219 David Sherwood <david.sherwood@arm.com>
12221 * optabs.c (widen_leading): Change the type of the mode argument
12222 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12223 (widen_bswap): Likewise.
12224 (expand_parity): Likewise.
12225 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12226 (expand_ffs): Likewise.
12227 (epand_unop): Check for scalar integer modes before calling the
12230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12231 Alan Hayward <alan.hayward@arm.com>
12232 David Sherwood <david.sherwood@arm.com>
12234 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12236 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12243 value_mode arguments. Use scalar_int_mode internally.
12244 (store_bit_field_1): Rename the new integer mode from imode
12245 to op0_mode and use it instead of GET_MODE (op0). Update calls
12246 to store_split_bit_field, store_bit_field_using_insv and
12247 store_fixed_bit_field.
12248 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12249 Use scalar_int_mode internally. Use a bit count rather than a mode
12250 when calculating the largest bit size for get_best_mode.
12251 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12252 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12253 Remove assertion that OP0 has a scalar integer mode.
12254 (store_split_bit_field): Add op0_mode and value_mode arguments.
12255 Update calls to extract_fixed_bit_field.
12256 (extract_bit_field_using_extv): Add an op0_mode argument.
12257 Use scalar_int_mode internally.
12258 (extract_bit_field_1): Rename the new integer mode from imode to
12259 op0_mode and use it instead of GET_MODE (op0). Update calls to
12260 extract_split_bit_field, extract_bit_field_using_extv and
12261 extract_fixed_bit_field.
12262 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12263 to extract_split_bit_field and extract_fixed_bit_field_1.
12264 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12265 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12266 on the target mode.
12267 (extract_split_bit_field): Add an op0_mode argument. Update call
12268 to extract_fixed_bit_field.
12270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12271 Alan Hayward <alan.hayward@arm.com>
12272 David Sherwood <david.sherwood@arm.com>
12274 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12275 * explow.c (hard_function_value): Likewise.
12276 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12277 convert_to_mode call outside the loop.
12278 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12279 for the mode iterator. Require the mode specified by max_pieces
12281 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12283 (copy_blkmode_to_reg): Likewise.
12284 (set_storage_via_setmem): Likewise.
12285 * optabs.c (prepare_cmp_insn): Likewise.
12286 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12287 * stor-layout.c (finish_bitfield_representative): Likewise.
12289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12290 Alan Hayward <alan.hayward@arm.com>
12291 David Sherwood <david.sherwood@arm.com>
12293 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12294 * expr.c (convert_move): Use them.
12295 (convert_modes): Likewise.
12296 (store_expr_with_bounds): Likewise.
12298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12299 Alan Hayward <alan.hayward@arm.com>
12300 David Sherwood <david.sherwood@arm.com>
12302 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12303 parameter for the mode of "x". Remove the "known_x", "known_mode"
12304 and "known_ret" arguments. Change the type of the mode argument
12305 to scalar_int_mode.
12306 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12307 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12308 (reg_num_sign_bit_copies_for_combine): Likewise.
12309 * rtlanal.c (nonzero_bits1): Likewise.
12310 (num_sign_bit_copies1): Likewise.
12311 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12312 (reg_num_sign_bit_copies_general): Likewise.
12313 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12314 (reg_nonzero_bits_general): Likewise.
12316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12317 Alan Hayward <alan.hayward@arm.com>
12318 David Sherwood <david.sherwood@arm.com>
12320 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12321 than in subroutines. Return 1 for non-integer modes.
12322 (cached_num_sign_bit_copies): Change the type of the mode parameter
12323 to scalar_int_mode.
12324 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12325 classes. Handle CONST_INT_P first and then check whether X also
12326 has a scalar integer mode. Check the same thing for inner registers
12327 of a SUBREG and for values that are being extended or truncated.
12329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12330 Alan Hayward <alan.hayward@arm.com>
12331 David Sherwood <david.sherwood@arm.com>
12333 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12334 in subroutines. Return the mode mask for non-integer modes.
12335 (cached_nonzero_bits): Change the type of the mode parameter
12336 to scalar_int_mode.
12337 (nonzero_bits1): Likewise. Remove early exit for other mode
12338 classes. Handle CONST_INT_P first and then check whether X
12339 also has a scalar integer mode.
12341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12342 Alan Hayward <alan.hayward@arm.com>
12343 David Sherwood <david.sherwood@arm.com>
12345 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12346 Return a scalar_int_mode and assert that the size is greater than
12348 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12349 (op_by_pieces_d::op_by_pieces_d): Likewise.
12350 (op_by_pieces_d::run): Likewise.
12351 (can_store_by_pieces): Likewise.
12353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12354 Alan Hayward <alan.hayward@arm.com>
12355 David Sherwood <david.sherwood@arm.com>
12357 * combine.c (extract_left_shift): Add a mode argument and update
12359 (make_compound_operation_int): Change the type of the mode parameter
12360 to scalar_int_mode and update the call to extract_left_shift.
12362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12363 Alan Hayward <alan.hayward@arm.com>
12364 David Sherwood <david.sherwood@arm.com>
12366 * combine.c (simplify_and_const_int): Change the type of the mode
12367 parameter to scalar_int_mode.
12368 (simplify_and_const_int_1): Likewise. Update recursive call.
12370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12371 Alan Hayward <alan.hayward@arm.com>
12372 David Sherwood <david.sherwood@arm.com>
12374 * combine.c (simplify_compare_const): Check that the mode is a
12375 scalar_int_mode (rather than VOIDmode) before testing its
12377 (simplify_comparison): Move COMPARISON_P handling out of the
12378 loop and restrict the latter part of the loop to scalar_int_modes.
12379 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12380 and when considering SUBREG_REGs. Use is_int_mode instead of
12381 checking GET_MODE_CLASS against MODE_INT.
12383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12384 Alan Hayward <alan.hayward@arm.com>
12385 David Sherwood <david.sherwood@arm.com>
12387 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12388 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12389 shift_unit_mode and for modes involved in scalar shifts.
12391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12392 Alan Hayward <alan.hayward@arm.com>
12393 David Sherwood <david.sherwood@arm.com>
12395 * combine.c (force_int_to_mode): New function, split out from...
12396 (force_to_mode): ...here. Keep xmode up-to-date and use it
12397 instead of GET_MODE (x).
12399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12400 Alan Hayward <alan.hayward@arm.com>
12401 David Sherwood <david.sherwood@arm.com>
12403 * optabs-query.h (extraction_insn::struct_mode): Change type to
12404 opt_scalar_int_mode and update comment.
12405 (extraction_insn::field_mode): Change type to scalar_int_mode.
12406 (extraction_insn::pos_mode): Likewise.
12407 * combine.c (make_extraction): Update accordingly.
12408 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12409 (get_optab_extraction_insn): Likewise.
12410 * recog.c (simplify_while_replacing): Likewise.
12411 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12412 parameter to opt_scalar_int_mode.
12414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12415 Alan Hayward <alan.hayward@arm.com>
12416 David Sherwood <david.sherwood@arm.com>
12418 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12419 to a scalar_int_mode instead of a machine_mode.
12420 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12421 (get_best_mode): Return a boolean and use a pointer argument to store
12422 the selected mode. Replace the limit mode parameter with a bit limit.
12423 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12424 for the values returned by bit_field_mode_iterator::next_mode.
12425 (store_bit_field): Update call to get_best_mode.
12426 (store_fixed_bit_field): Likewise.
12427 (extract_fixed_bit_field): Likewise.
12428 * expr.c (optimize_bitfield_assignment_op): Likewise.
12429 * fold-const.c (optimize_bit_field_compare): Likewise.
12430 (fold_truth_andor_1): Likewise.
12431 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12432 Update for new type of m_mode.
12433 (get_best_mode): As above.
12435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12436 Alan Hayward <alan.hayward@arm.com>
12437 David Sherwood <david.sherwood@arm.com>
12439 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12440 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12441 (store_bit_field): Check is_a <scalar_int_mode> before calling
12442 strict_volatile_bitfield_p.
12443 (extract_bit_field): Likewise.
12445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12446 Alan Hayward <alan.hayward@arm.com>
12447 David Sherwood <david.sherwood@arm.com>
12449 * target.def (cstore_mode): Return a scalar_int_mode.
12450 * doc/tm.texi: Regenerate.
12451 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12452 * targhooks.h (default_cstore_mode): Likewise.
12453 * targhooks.c (default_cstore_mode): Likewise, using a forced
12455 * expmed.c (emit_cstore): Expect the target of the cstore to be
12458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12459 Alan Hayward <alan.hayward@arm.com>
12460 David Sherwood <david.sherwood@arm.com>
12462 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12464 (niter_desc): Likewise mode.
12465 (iv_analyze): Add a mode parameter.
12467 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12468 and change its type to scalar_int_mode.
12469 * loop-iv.c: Update commentary at head of file.
12470 (iv_constant): Pass the mode paraeter before the rtx it describes
12471 and change its type to scalar_int_mode. Remove VOIDmode handling.
12472 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12473 (iv_extend): Likewise.
12474 (shorten_into_mode): Likewise.
12475 (iv_add): Use scalar_int_mode.
12476 (iv_mult): Likewise.
12477 (iv_shift): Likewise.
12478 (canonicalize_iv_subregs): Likewise.
12479 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12480 it describes and change its mode to scalar_int_mode. Also change
12481 the type of the returned inner_mode to scalar_int_mode.
12482 (get_biv_step): Likewise, turning outer_mode from a pointer
12483 into a direct parameter. Update call to get_biv_step_1.
12484 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12485 iv_constant and get_biv_step.
12486 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12487 and change its type to scalar_int_mode. Don't initialise iv->mode
12488 to VOIDmode and remove later checks for its still being VOIDmode.
12489 Update calls to iv_analyze_op and iv_analyze_expr. Check
12490 is_a <scalar_int_mode> when changing the mode under consideration.
12491 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12492 Update call to iv_analyze_expr.
12493 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12494 inner register is not also a scalar_int_mode. Update call to
12496 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12497 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12498 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12499 separate mode class checks. Update calls to iv_analyze. Remove
12500 fix-up of VOIDmodes after iv_analyze_biv.
12501 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12502 don't have a scalar_int_mode. Update call to biv_p.
12504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12505 Alan Hayward <alan.hayward@arm.com>
12506 David Sherwood <david.sherwood@arm.com>
12508 * cfgexpand.c (convert_debug_memory_address): Use
12509 as_a <scalar_int_mode>.
12510 * combine.c (expand_compound_operation): Likewise.
12511 (make_extraction): Likewise.
12512 (change_zero_ext): Likewise.
12513 (simplify_comparison): Likewise.
12514 * cse.c (cse_insn): Likewise.
12515 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12516 (mem_loc_descriptor): Likewise.
12517 (loc_descriptor): Likewise.
12518 * expmed.c (init_expmed_one_mode): Likewise.
12519 (synth_mult): Likewise.
12520 (emit_store_flag_1): Likewise.
12521 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12522 of a comparison with size.
12523 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12524 (reduce_to_bit_field_precision): Likewise.
12525 * function.c (expand_function_end): Likewise.
12526 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12527 * loop-doloop.c (doloop_modify): Likewise.
12528 * optabs.c (expand_binop): Likewise.
12529 (expand_unop): Likewise.
12530 (expand_copysign_absneg): Likewise.
12531 (prepare_cmp_insn): Likewise.
12532 (maybe_legitimize_operand): Likewise.
12533 * recog.c (const_scalar_int_operand): Likewise.
12534 * rtlanal.c (get_address_mode): Likewise.
12535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12536 (simplify_cond_clz_ctz): Likewise.
12537 * tree-nested.c (get_nl_goto_field): Likewise.
12538 * tree.c (build_vector_type_for_mode): Likewise.
12539 * var-tracking.c (use_narrower_mode): Likewise.
12541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12542 Alan Hayward <alan.hayward@arm.com>
12543 David Sherwood <david.sherwood@arm.com>
12545 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12546 * builtins.c (expand_builtin_signbit): Use it.
12547 * cfgexpand.c (expand_debug_expr): Likewise.
12548 * dojump.c (do_jump): Likewise.
12549 (do_compare_and_jump): Likewise.
12550 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12551 * expmed.c (make_tree): Likewise.
12552 * expr.c (expand_expr_real_2): Likewise.
12553 (expand_expr_real_1): Likewise.
12554 (try_casesi): Likewise.
12555 * fold-const-call.c (fold_const_call_ss): Likewise.
12556 * fold-const.c (unextend): Likewise.
12557 (extract_muldiv_1): Likewise.
12558 (fold_single_bit_test): Likewise.
12559 (native_encode_int): Likewise.
12560 (native_encode_string): Likewise.
12561 (native_interpret_int): Likewise.
12562 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12563 * internal-fn.c (expand_addsub_overflow): Likewise.
12564 (expand_neg_overflow): Likewise.
12565 (expand_mul_overflow): Likewise.
12566 (expand_arith_overflow): Likewise.
12567 * match.pd: Likewise.
12568 * stor-layout.c (layout_type): Likewise.
12569 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12570 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12571 * tree-ssanames.c (get_range_info): Likewise.
12572 * tree-switch-conversion.c (array_value_type) Likewise.
12573 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12574 (vect_recog_divmod_pattern): Likewise.
12575 (vect_recog_mixed_size_cond_pattern): Likewise.
12576 * tree-vrp.c (extract_range_basic): Likewise.
12577 (simplify_float_conversion_using_ranges): Likewise.
12578 * tree.c (int_fits_type_p): Likewise.
12579 * ubsan.c (instrument_bool_enum_load): Likewise.
12580 * varasm.c (mergeable_string_section): Likewise.
12581 (narrowing_initializer_constant_valid_p): Likewise.
12582 (output_constant): Likewise.
12584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12585 Alan Hayward <alan.hayward@arm.com>
12586 David Sherwood <david.sherwood@arm.com>
12588 * machmode.h (NARROWEST_INT_MODE): New macro.
12589 * expr.c (alignment_for_piecewise_move): Use it instead of
12590 GET_CLASS_NARROWEST_MODE (MODE_INT).
12591 (push_block): Likewise.
12592 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12594 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12597 Alan Hayward <alan.hayward@arm.com>
12598 David Sherwood <david.sherwood@arm.com>
12600 * postreload.c (move2add_valid_value_p): Change the type of the
12601 mode parameter to scalar_int_mode.
12602 (move2add_use_add2_insn): Add a mode parameter and use it instead
12604 (move2add_use_add3_insn): Likewise.
12605 (reload_cse_move2add): Update accordingly.
12607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12608 Alan Hayward <alan.hayward@arm.com>
12609 David Sherwood <david.sherwood@arm.com>
12611 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12613 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12614 * optabs.c (expand_unop): Likewise.
12616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12617 Alan Hayward <alan.hayward@arm.com>
12618 David Sherwood <david.sherwood@arm.com>
12620 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12621 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12622 (popcount_loc_descriptor): Likewise.
12623 (bswap_loc_descriptor): Likewise.
12624 (rotate_loc_descriptor): Likewise.
12625 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12626 calling the functions above.
12628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12629 Alan Hayward <alan.hayward@arm.com>
12630 David Sherwood <david.sherwood@arm.com>
12632 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12634 (try_combine): Likewise.
12635 (simplify_if_then_else): Likewise.
12636 * cse.c (cse_insn): Likewise.
12637 * dwarf2out.c (mem_loc_descriptor): Likewise.
12638 * emit-rtl.c (gen_lowpart_common): Likewise.
12639 * simplify-rtx.c (simplify_truncation): Likewise.
12640 (simplify_binary_operation_1): Likewise.
12641 (simplify_const_relational_operation): Likewise.
12642 (simplify_ternary_operation): Likewise.
12643 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12646 Alan Hayward <alan.hayward@arm.com>
12647 David Sherwood <david.sherwood@arm.com>
12649 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12650 * reload.c (push_reload): Likewise.
12651 (find_reloads): Likewise.
12653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12654 Alan Hayward <alan.hayward@arm.com>
12655 David Sherwood <david.sherwood@arm.com>
12657 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12658 (make_compound_operation_int): Likewise.
12659 (change_zero_ext): Likewise.
12660 * expr.c (convert_move): Likewise.
12661 (convert_modes): Likewise.
12662 * fwprop.c (forward_propagate_subreg): Likewise.
12663 * loop-iv.c (get_biv_step_1): Likewise.
12664 * optabs.c (widen_operand): Likewise.
12665 * postreload.c (move2add_valid_value_p): Likewise.
12666 * recog.c (simplify_while_replacing): Likewise.
12667 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12668 (simplify_binary_operation_1): Likewise. Remove redundant
12669 mode equality check.
12671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12672 Alan Hayward <alan.hayward@arm.com>
12673 David Sherwood <david.sherwood@arm.com>
12675 * combine.c (combine_simplify_rtx): Add checks for
12676 is_a <scalar_int_mode>.
12677 (simplify_if_then_else): Likewise.
12678 (make_field_assignment): Likewise.
12679 (simplify_comparison): Likewise.
12680 * ifcvt.c (noce_try_bitop): Likewise.
12681 * loop-invariant.c (canonicalize_address_mult): Likewise.
12682 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12689 is_a <scalar_int_mode> instead of != BLKmode.
12691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12692 Alan Hayward <alan.hayward@arm.com>
12693 David Sherwood <david.sherwood@arm.com>
12695 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12696 instead of != VOIDmode.
12697 * combine.c (if_then_else_cond): Likewise.
12698 (change_zero_ext): Likewise.
12699 * dwarf2out.c (mem_loc_descriptor): Likewise.
12700 (loc_descriptor): Likewise.
12701 * rtlanal.c (canonicalize_condition): Likewise.
12702 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12705 Alan Hayward <alan.hayward@arm.com>
12706 David Sherwood <david.sherwood@arm.com>
12708 * simplify-rtx.c (simplify_binary_operation_1): Use
12709 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12712 Alan Hayward <alan.hayward@arm.com>
12713 David Sherwood <david.sherwood@arm.com>
12715 * wide-int.h (int_traits<unsigned char>) New class.
12716 (int_traits<unsigned short>) Likewise.
12717 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12718 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12720 * combine.c (set_nonzero_bits_and_sign_copies): Use
12721 is_a <scalar_int_mode>.
12722 (find_split_point): Likewise.
12723 (combine_simplify_rtx): Likewise.
12724 (simplify_logical): Likewise.
12725 (expand_compound_operation): Likewise.
12726 (expand_field_assignment): Likewise.
12727 (make_compound_operation): Likewise.
12728 (extended_count): Likewise.
12729 (change_zero_ext): Likewise.
12730 (simplify_comparison): Likewise.
12731 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12732 (ucompare_loc_descriptor): Likewise.
12733 (minmax_loc_descriptor): Likewise.
12734 (mem_loc_descriptor): Likewise.
12735 (loc_descriptor): Likewise.
12736 * expmed.c (init_expmed_one_mode): Likewise.
12737 * lra-constraints.c (lra_constraint_offset): Likewise.
12738 * optabs.c (prepare_libcall_arg): Likewise.
12739 * postreload.c (move2add_note_store): Likewise.
12740 * reload.c (operands_match_p): Likewise.
12741 * rtl.h (load_extend_op): Likewise.
12742 * rtlhooks.c (gen_lowpart_general): Likewise.
12743 * simplify-rtx.c (simplify_truncation): Likewise.
12744 (simplify_unary_operation_1): Likewise.
12745 (simplify_binary_operation_1): Likewise.
12746 (simplify_const_binary_operation): Likewise.
12747 (simplify_const_relational_operation): Likewise.
12748 (simplify_subreg): Likewise.
12749 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12750 * var-tracking.c (adjust_mems): Likewise.
12751 (prepare_call_arguments): Likewise.
12753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12754 Alan Hayward <alan.hayward@arm.com>
12755 David Sherwood <david.sherwood@arm.com>
12757 * machmode.h (is_int_mode): New fuction.
12758 * combine.c (find_split_point): Use it.
12759 (combine_simplify_rtx): Likewise.
12760 (simplify_if_then_else): Likewise.
12761 (simplify_set): Likewise.
12762 (simplify_shift_const_1): Likewise.
12763 (simplify_comparison): Likewise.
12764 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12765 * cse.c (notreg_cost): Likewise.
12766 (cse_insn): Likewise.
12767 * cselib.c (cselib_lookup_1): Likewise.
12768 * dojump.c (do_jump_1): Likewise.
12769 (do_compare_rtx_and_jump): Likewise.
12770 * dse.c (get_call_args): Likewise.
12771 * dwarf2out.c (rtl_for_decl_init): Likewise.
12772 (native_encode_initializer): Likewise.
12773 * expmed.c (emit_store_flag_1): Likewise.
12774 (emit_store_flag): Likewise.
12775 * expr.c (convert_modes): Likewise.
12776 (store_field): Likewise.
12777 (expand_expr_real_1): Likewise.
12778 * fold-const.c (fold_read_from_constant_string): Likewise.
12779 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12780 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12781 * optabs.c (expand_binop): Likewise.
12782 (expand_unop): Likewise.
12783 (expand_abs_nojump): Likewise.
12784 (expand_one_cmpl_abs_nojump): Likewise.
12785 * simplify-rtx.c (mode_signbit_p): Likewise.
12786 (val_signbit_p): Likewise.
12787 (val_signbit_known_set_p): Likewise.
12788 (val_signbit_known_clear_p): Likewise.
12789 (simplify_relational_operation_1): Likewise.
12790 * tree.c (vector_type_mode): Likewise.
12792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12793 Alan Hayward <alan.hayward@arm.com>
12794 David Sherwood <david.sherwood@arm.com>
12796 * machmode.h (smallest_mode_for_size): Fix formatting.
12797 (smallest_int_mode_for_size): New function.
12798 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12799 instead of smallest_mode_for_size.
12800 * combine.c (make_extraction): Likewise.
12801 * config/arc/arc.c (arc_expand_movmem): Likewise.
12802 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12803 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12804 * config/s390/s390.c (s390_expand_insv): Likewise.
12805 * config/sparc/sparc.c (assign_int_registers): Likewise.
12806 * config/spu/spu.c (spu_function_value): Likewise.
12807 (spu_function_arg): Likewise.
12808 * coverage.c (get_gcov_type): Likewise.
12809 (get_gcov_unsigned_t): Likewise.
12810 * dse.c (find_shift_sequence): Likewise.
12811 * expmed.c (store_bit_field_1): Likewise.
12812 * expr.c (convert_move): Likewise.
12813 (store_field): Likewise.
12814 * internal-fn.c (expand_arith_overflow): Likewise.
12815 * optabs-query.c (get_best_extraction_insn): Likewise.
12816 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12817 * stor-layout.c (layout_type): Likewise.
12818 (initialize_sizetypes): Likewise.
12819 * targhooks.c (default_get_mask_mode): Likewise.
12820 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12823 Alan Hayward <alan.hayward@arm.com>
12824 David Sherwood <david.sherwood@arm.com>
12826 * machmode.h (opt_mode::else_blk): New function.
12827 (int_mode_for_mode): Declare.
12828 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12829 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12831 * cfgexpand.c (expand_debug_expr): Likewise.
12832 * combine.c (gen_lowpart_or_truncate): Likewise.
12833 (gen_lowpart_for_combine): Likewise.
12834 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12835 * config/avr/avr.c (avr_to_int_mode): Likewise.
12836 (avr_out_plus_1): Likewise.
12837 (avr_out_plus): Likewise.
12838 (avr_out_round): Likewise.
12839 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12840 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12841 (s390_expand_vcond): Likewise.
12842 * config/spu/spu.c (spu_split_immediate): Likewise.
12843 (spu_expand_mov): Likewise.
12844 * dse.c (get_stored_val): Likewise.
12845 * expmed.c (store_bit_field_1): Likewise.
12846 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12848 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12849 (extract_low_bits): Likewise.
12850 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12851 handling rather than repeating the check.
12852 (emit_group_store): Likewise.
12853 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12854 * optabs.c (expand_absneg_bit): Likewise.
12855 (expand_copysign_absneg): Likewise.
12856 (expand_copysign_bit): Likewise.
12857 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12858 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12859 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12860 * var-tracking.c (prepare_call_arguments): Likewise.
12861 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12862 int_mode_for_mode instead of mode_for_size.
12863 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12866 Alan Hayward <alan.hayward@arm.com>
12867 David Sherwood <david.sherwood@arm.com>
12869 * machmode.h (int_mode_for_size): New function.
12870 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12871 instead of mode_for_size.
12872 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12874 * combine.c (expand_field_assignment): Use int_mode_for_size
12875 instead of mode_for_size.
12876 (make_extraction): Likewise.
12877 (simplify_shift_const_1): Likewise.
12878 (simplify_comparison): Likewise.
12879 * dojump.c (do_jump): Likewise.
12880 * dwarf2out.c (mem_loc_descriptor): Likewise.
12881 * emit-rtl.c (init_derived_machine_modes): Likewise.
12882 * expmed.c (flip_storage_order): Likewise.
12883 (convert_extracted_bit_field): Likewise.
12884 * expr.c (copy_blkmode_from_reg): Likewise.
12885 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12886 * internal-fn.c (expand_mul_overflow): Likewise.
12887 * lower-subreg.c (simple_move): Likewise.
12888 * optabs-libfuncs.c (init_optabs): Likewise.
12889 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12890 * tree.c (vector_type_mode): Likewise.
12891 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12892 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12893 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12894 * tree-vect-stmts.c (vectorizable_load): Likewise.
12895 (vectorizable_store): Likewise.
12897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12898 Alan Hayward <alan.hayward@arm.com>
12899 David Sherwood <david.sherwood@arm.com>
12901 * coretypes.h (pod_mode): New type.
12902 (scalar_int_mode_pod): New typedef.
12903 * machmode.h (pod_mode): New class.
12904 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12905 * genmodes.c (emit_mode_int_n): Update accordingly.
12906 * lower-subreg.h (target_lower_subreg): Change type to
12907 scalar_int_mode_pod.
12908 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12909 scalar_int_mode_pod.
12911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12912 Alan Hayward <alan.hayward@arm.com>
12913 David Sherwood <david.sherwood@arm.com>
12915 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12916 machine_mode to scalar_int_mode.
12917 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12918 (rs6000_option_override_internal): Remove cast to int.
12919 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12920 machine_mode to scalar_int_mode.
12921 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12922 (rs6000_option_override_internal): Remove cast to int.
12923 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12924 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12926 * config/s390/s390.c (s390_expand_builtin): Likewise.
12927 * coretypes.h (scalar_int_mode): New type.
12928 (opt_scalar_int_mode): New typedef.
12929 * machmode.h (scalar_int_mode): New class.
12930 (scalar_int_mode::includes_p): New function.
12931 (byte_mode): Change type to scalar_int_mode.
12932 (word_mode): Likewise.
12933 (ptr_mode): Likewise.
12934 * emit-rtl.c (byte_mode): Likewise.
12935 (word_mode): Likewise.
12936 (ptr_mode): Likewise.
12937 (init_derived_machine_modes): Update accordingly.
12938 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12939 and MODE_PARTIAL_INT.
12940 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12941 opt_scalar_int_mode.
12943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12944 Alan Hayward <alan.hayward@arm.com>
12945 David Sherwood <david.sherwood@arm.com>
12947 * target.def (libgcc_floating_mode_supported_p): Take a
12949 * doc/tm.texi: Regenerate.
12950 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12952 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12953 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12957 Alan Hayward <alan.hayward@arm.com>
12958 David Sherwood <david.sherwood@arm.com>
12960 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12961 * doc/tm.texi: Regenerate.
12962 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12963 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12964 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12965 * targhooks.h (default_floatn_mode): Likewise.
12966 * targhooks.c (default_floatn_mode): Likewise.
12967 * tree.c (build_common_tree_nodes): Update accordingly.
12969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12970 Alan Hayward <alan.hayward@arm.com>
12971 David Sherwood <david.sherwood@arm.com>
12973 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12974 (mode_iterator::iterate_p): Likewise.
12975 (mode_iterator::get_wider): Likewise.
12976 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12979 Alan Hayward <alan.hayward@arm.com>
12980 David Sherwood <david.sherwood@arm.com>
12982 * coretypes.h (opt_scalar_float_mode): New typedef.
12983 * machmode.h (float_mode_for_size): New function.
12984 * emit-rtl.c (double_mode): Delete.
12985 (init_emit_once): Use float_mode_for_size.
12986 * stor-layout.c (layout_type): Likewise.
12987 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12990 Alan Hayward <alan.hayward@arm.com>
12991 David Sherwood <david.sherwood@arm.com>
12993 * output.h (assemble_real): Take a scalar_float_mode.
12994 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12995 * config/arm/arm.md (consttable_4): Likewise.
12996 (consttable_8): Likewise.
12997 (consttable_16): Likewise.
12998 * config/mips/mips.md (consttable_float): Likewise.
12999 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13000 * varasm.c (assemble_real): Take a scalar_float_mode.
13001 (output_constant_pool_2): Update accordingly.
13002 (output_constant): Likewise.
13004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13005 Alan Hayward <alan.hayward@arm.com>
13006 David Sherwood <david.sherwood@arm.com>
13008 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13009 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13010 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13011 (native_encode_real): Likewise.
13012 (native_interpret_real): Likewise.
13013 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13014 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13017 Alan Hayward <alan.hayward@arm.com>
13018 David Sherwood <david.sherwood@arm.com>
13020 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13021 <scalar_float_mode>. Simplify.
13022 (gen_extend_conv_libfunc): Likewise.
13024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13025 Alan Hayward <alan.hayward@arm.com>
13026 David Sherwood <david.sherwood@arm.com>
13028 * coretypes.h (scalar_float_mode): New type.
13029 * machmode.h (mode_traits::from_int): Use machine_mode if
13030 USE_ENUM_MODES is defined.
13031 (is_a): New function.
13033 (dyn_cast): Likewise.
13034 (scalar_float_mode): New class.
13035 (scalar_float_mode::includes_p): New function.
13036 (is_float_mode): Likewise.
13037 * gdbhooks.py (MachineModePrinter): New class.
13038 (build_pretty_printer): Use it for scalar_float_mode.
13039 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13040 (format_helper::format_helper): Turn into a template.
13041 * genmodes.c (get_mode_class): New function.
13042 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13043 or machine_mode if none.
13044 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13045 as_a <scalar_float_mode>.
13046 * dwarf2out.c (mem_loc_descriptor): Likewise.
13047 (insert_float): Likewise.
13048 (add_const_value_attribute): Likewise.
13049 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13050 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13051 (expand_unop): Update accordingly.
13052 (expand_abs_nojump): Likewise.
13053 (expand_copysign_absneg): Take a scalar_float_mode.
13054 (expand_copysign_bit): Likewise.
13055 (expand_copysign): Update accordingly.
13057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13058 Alan Hayward <alan.hayward@arm.com>
13059 David Sherwood <david.sherwood@arm.com>
13061 * coretypes.h (opt_mode): New class.
13062 * machmode.h (opt_mode): Likewise.
13063 (opt_mode::else_void): New function.
13064 (opt_mode::require): Likewise.
13065 (opt_mode::exists): Likewise.
13066 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13067 (GET_MODE_2XWIDER_MODE): Likewise.
13068 (mode_iterator::get_wider): Update accordingly.
13069 (mode_iterator::get_2xwider): Likewise.
13070 (mode_iterator::get_known_wider): Likewise, turning into a template.
13071 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13072 forcing a wider mode to exist.
13073 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13074 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13075 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13076 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13077 * lower-subreg.c (init_lower_subreg): Likewise.
13078 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13079 on the final iteration.
13080 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13081 a wider mode exists before asking for a move pattern.
13082 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13083 forcing a wider mode to exist.
13084 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13085 returning false if no such mode exists.
13086 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13087 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13088 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13089 Avoid checking for a MODE_INT if we already know the mode is not a
13091 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13092 forcing a wider mode to exist.
13093 (expmed_mult_highpart_optab): Likewise.
13094 (expmed_mult_highpart): Likewise.
13095 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13097 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13098 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13099 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13100 * internal-fn.c (expand_mul_overflow): Update use of
13101 GET_MODE_2XWIDER_MODE.
13102 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13103 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13104 GET_MODE_WIDER_MODE.
13105 (convert_plusminus_to_widen): Likewise.
13106 * tree-switch-conversion.c (array_value_type): Likewise.
13107 * var-tracking.c (emit_note_insn_var_location): Likewise.
13108 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13109 Return false inside rather than outside the loop if no wider mode
13111 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13112 and GET_MODE_2XWIDER_MODE
13113 (can_compare_p): Use else_void.
13114 * gdbhooks.py (OptMachineModePrinter): New class.
13115 (build_pretty_printer): Use it for opt_mode.
13117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13118 Alan Hayward <alan.hayward@arm.com>
13119 David Sherwood <david.sherwood@arm.com>
13121 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13122 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13125 Alan Hayward <alan.hayward@arm.com>
13126 David Sherwood <david.sherwood@arm.com>
13128 * machmode.h (mode_traits): New structure.
13129 (get_narrowest_mode): New function.
13130 (mode_iterator::start): Likewise.
13131 (mode_iterator::iterate_p): Likewise.
13132 (mode_iterator::get_wider): Likewise.
13133 (mode_iterator::get_known_wider): Likewise.
13134 (mode_iterator::get_2xwider): Likewise.
13135 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13136 (FOR_EACH_MODE): Likewise.
13137 (FOR_EACH_MODE_FROM): Likewise.
13138 (FOR_EACH_MODE_UNTIL): Likewise.
13139 (FOR_EACH_WIDER_MODE): Likewise.
13140 (FOR_EACH_2XWIDER_MODE): Likewise.
13141 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13142 * combine.c (simplify_comparison): Likewise
13143 * config/i386/i386.c (type_natural_mode): Likewise.
13144 * cse.c (cse_insn): Likewise.
13145 * dse.c (find_shift_sequence): Likewise.
13146 * emit-rtl.c (init_derived_machine_modes): Likewise.
13147 (init_emit_once): Likewise.
13148 * explow.c (hard_function_value): Likewise.
13149 * expmed.c (extract_fixed_bit_field_1): Likewise.
13150 (extract_bit_field_1): Likewise.
13151 (expand_divmod): Likewise.
13152 (emit_store_flag_1): Likewise.
13153 * expr.c (init_expr_target): Likewise.
13154 (convert_move): Likewise.
13155 (alignment_for_piecewise_move): Likewise.
13156 (widest_int_mode_for_size): Likewise.
13157 (emit_block_move_via_movmem): Likewise.
13158 (copy_blkmode_to_reg): Likewise.
13159 (set_storage_via_setmem): Likewise.
13160 (compress_float_constant): Likewise.
13161 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13162 * optabs-query.c (get_best_extraction_insn): Likewise.
13163 * optabs.c (expand_binop): Likewise.
13164 (expand_twoval_unop): Likewise.
13165 (expand_twoval_binop): Likewise.
13166 (widen_leading): Likewise.
13167 (widen_bswap): Likewise.
13168 (expand_parity): Likewise.
13169 (expand_unop): Likewise.
13170 (prepare_cmp_insn): Likewise.
13171 (prepare_float_lib_cmp): Likewise.
13172 (expand_float): Likewise.
13173 (expand_fix): Likewise.
13174 (expand_sfix_optab): Likewise.
13175 * postreload.c (move2add_use_add2_insn): Likewise.
13176 * reg-stack.c (reg_to_stack): Likewise.
13177 * reginfo.c (choose_hard_reg_mode): Likewise.
13178 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13179 * stor-layout.c (mode_for_size): Likewise.
13180 (smallest_mode_for_size): Likewise.
13181 (mode_for_vector): Likewise.
13182 (finish_bitfield_representative): Likewise.
13183 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13184 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13185 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13186 * var-tracking.c (prepare_call_arguments): Likewise.
13188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13189 Alan Hayward <alan.hayward@arm.com>
13190 David Sherwood <david.sherwood@arm.com>
13192 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13193 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13194 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13195 * machmode.h (mode_size): Move earlier in file.
13196 (mode_precision): Likewise.
13197 (mode_inner): Likewise.
13198 (mode_nunits): Likewise.
13199 (mode_unit_size): Likewise.
13200 (unit_unit_precision): Likewise.
13201 (mode_wider): Likewise.
13202 (mode_2xwider): Likewise.
13203 (machine_mode): New class.
13204 (mode_to_bytes): New function.
13205 (mode_to_bits): Likewise.
13206 (mode_to_precision): Likewise.
13207 (mode_to_inner): Likewise.
13208 (mode_to_unit_size): Likewise.
13209 (mode_to_unit_precision): Likewise.
13210 (mode_to_nunits): Likewise.
13211 (GET_MODE_SIZE): Use mode_to_bytes.
13212 (GET_MODE_BITSIZE): Use mode_to_bits.
13213 (GET_MODE_PRECISION): Use mode_to_precision.
13214 (GET_MODE_INNER): Use mode_to_inner.
13215 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13216 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13217 (GET_MODE_NUNITS): Use mode_to_nunits.
13218 * system.h (ALWAYS_INLINE): New macro.
13219 * config/powerpcspe/powerpcspe-c.c
13220 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13221 int for arg1_mode and arg2_mode.
13223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13224 Alan Hayward <alan.hayward@arm.com>
13225 David Sherwood <david.sherwood@arm.com>
13227 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13228 Prefix mode names with E_ in case statements.
13229 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13230 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13231 (aarch64_split_simd_move): Likewise.
13232 (aarch64_gen_storewb_pair): Likewise.
13233 (aarch64_gen_loadwb_pair): Likewise.
13234 (aarch64_gen_store_pair): Likewise.
13235 (aarch64_gen_load_pair): Likewise.
13236 (aarch64_get_condition_code_1): Likewise.
13237 (aarch64_constant_pool_reload_icode): Likewise.
13238 (get_rsqrte_type): Likewise.
13239 (get_rsqrts_type): Likewise.
13240 (get_recpe_type): Likewise.
13241 (get_recps_type): Likewise.
13242 (aarch64_gimplify_va_arg_expr): Likewise.
13243 (aarch64_simd_container_mode): Likewise.
13244 (aarch64_emit_load_exclusive): Likewise.
13245 (aarch64_emit_store_exclusive): Likewise.
13246 (aarch64_expand_compare_and_swap): Likewise.
13247 (aarch64_gen_atomic_cas): Likewise.
13248 (aarch64_emit_bic): Likewise.
13249 (aarch64_emit_atomic_swap): Likewise.
13250 (aarch64_emit_atomic_load_op): Likewise.
13251 (aarch64_evpc_trn): Likewise.
13252 (aarch64_evpc_uzp): Likewise.
13253 (aarch64_evpc_zip): Likewise.
13254 (aarch64_evpc_ext): Likewise.
13255 (aarch64_evpc_rev): Likewise.
13256 (aarch64_evpc_dup): Likewise.
13257 (aarch64_gen_ccmp_first): Likewise.
13258 (aarch64_gen_ccmp_next): Likewise.
13259 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13260 (alpha_emit_xfloating_libcall): Likewise.
13261 (emit_insxl): Likewise.
13262 (alpha_arg_type): Likewise.
13263 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13264 (arc_preferred_simd_mode): Likewise.
13265 (arc_secondary_reload): Likewise.
13266 (get_arc_condition_code): Likewise.
13267 (arc_print_operand): Likewise.
13268 (arc_legitimate_constant_p): Likewise.
13269 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13270 * config/arc/arc.md (casesi_load): Likewise.
13271 (casesi_compact_jump): Likewise.
13272 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13273 (cc_use_register): Likewise.
13274 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13275 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13276 (arm_init_iwmmxt_builtins): Likewise.
13277 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13278 (neon_expand_vector_init): Likewise.
13279 (arm_attr_length_move_neon): Likewise.
13280 (maybe_get_arm_condition_code): Likewise.
13281 (arm_emit_vector_const): Likewise.
13282 (arm_preferred_simd_mode): Likewise.
13283 (arm_output_iwmmxt_tinsr): Likewise.
13284 (thumb1_output_casesi): Likewise.
13285 (thumb2_output_casesi): Likewise.
13286 (arm_emit_load_exclusive): Likewise.
13287 (arm_emit_store_exclusive): Likewise.
13288 (arm_expand_compare_and_swap): Likewise.
13289 (arm_evpc_neon_vuzp): Likewise.
13290 (arm_evpc_neon_vzip): Likewise.
13291 (arm_evpc_neon_vrev): Likewise.
13292 (arm_evpc_neon_vtrn): Likewise.
13293 (arm_evpc_neon_vext): Likewise.
13294 (arm_validize_comparison): Likewise.
13295 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13296 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13297 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13298 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13299 (c6x_preferred_simd_mode): Likewise.
13300 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13301 (epiphany_rtx_costs): Likewise.
13302 * config/epiphany/predicates.md (proper_comparison_operator):
13304 * config/frv/frv.c (condexec_memory_operand): Likewise.
13305 (frv_emit_move): Likewise.
13306 (output_move_single): Likewise.
13307 (output_condmove_single): Likewise.
13308 (frv_hard_regno_mode_ok): Likewise.
13309 (frv_matching_accg_mode): Likewise.
13310 * config/h8300/h8300.c (split_adds_subs): Likewise.
13311 (h8300_rtx_costs): Likewise.
13312 (h8300_print_operand): Likewise.
13313 (compute_mov_length): Likewise.
13314 (output_logical_op): Likewise.
13315 (compute_logical_op_length): Likewise.
13316 (compute_logical_op_cc): Likewise.
13317 (h8300_shift_needs_scratch_p): Likewise.
13318 (output_a_shift): Likewise.
13319 (compute_a_shift_length): Likewise.
13320 (compute_a_shift_cc): Likewise.
13321 (expand_a_rotate): Likewise.
13322 (output_a_rotate): Likewise.
13323 * config/i386/i386.c (classify_argument): Likewise.
13324 (function_arg_advance_32): Likewise.
13325 (function_arg_32): Likewise.
13326 (function_arg_64): Likewise.
13327 (function_value_64): Likewise.
13328 (ix86_gimplify_va_arg): Likewise.
13329 (ix86_legitimate_constant_p): Likewise.
13330 (put_condition_code): Likewise.
13331 (split_double_mode): Likewise.
13332 (ix86_avx256_split_vector_move_misalign): Likewise.
13333 (ix86_expand_vector_logical_operator): Likewise.
13334 (ix86_split_idivmod): Likewise.
13335 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13336 (ix86_build_const_vector): Likewise.
13337 (ix86_build_signbit_mask): Likewise.
13338 (ix86_match_ccmode): Likewise.
13339 (ix86_cc_modes_compatible): Likewise.
13340 (ix86_expand_branch): Likewise.
13341 (ix86_expand_sse_cmp): Likewise.
13342 (ix86_expand_sse_movcc): Likewise.
13343 (ix86_expand_int_sse_cmp): Likewise.
13344 (ix86_expand_vec_perm_vpermi2): Likewise.
13345 (ix86_expand_vec_perm): Likewise.
13346 (ix86_expand_sse_unpack): Likewise.
13347 (ix86_expand_int_addcc): Likewise.
13348 (ix86_split_to_parts): Likewise.
13349 (ix86_vectorize_builtin_gather): Likewise.
13350 (ix86_vectorize_builtin_scatter): Likewise.
13351 (avx_vpermilp_parallel): Likewise.
13352 (inline_memory_move_cost): Likewise.
13353 (ix86_tieable_integer_mode_p): Likewise.
13354 (x86_maybe_negate_const_int): Likewise.
13355 (ix86_expand_vector_init_duplicate): Likewise.
13356 (ix86_expand_vector_init_one_nonzero): Likewise.
13357 (ix86_expand_vector_init_one_var): Likewise.
13358 (ix86_expand_vector_init_concat): Likewise.
13359 (ix86_expand_vector_init_interleave): Likewise.
13360 (ix86_expand_vector_init_general): Likewise.
13361 (ix86_expand_vector_set): Likewise.
13362 (ix86_expand_vector_extract): Likewise.
13363 (emit_reduc_half): Likewise.
13364 (ix86_emit_i387_round): Likewise.
13365 (ix86_mangle_type): Likewise.
13366 (ix86_expand_round_sse4): Likewise.
13367 (expand_vec_perm_blend): Likewise.
13368 (canonicalize_vector_int_perm): Likewise.
13369 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13370 (expand_vec_perm_1): Likewise.
13371 (expand_vec_perm_interleave3): Likewise.
13372 (expand_vec_perm_even_odd_pack): Likewise.
13373 (expand_vec_perm_even_odd_1): Likewise.
13374 (expand_vec_perm_broadcast_1): Likewise.
13375 (ix86_vectorize_vec_perm_const_ok): Likewise.
13376 (ix86_expand_vecop_qihi): Likewise.
13377 (ix86_expand_mul_widen_hilo): Likewise.
13378 (ix86_expand_sse2_abs): Likewise.
13379 (ix86_expand_pextr): Likewise.
13380 (ix86_expand_pinsr): Likewise.
13381 (ix86_preferred_simd_mode): Likewise.
13382 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13383 * config/i386/sse.md (*andnot<mode>3): Likewise.
13384 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13385 (*<code><mode>3): Likewise.
13386 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13387 (ia64_expand_atomic_op): Likewise.
13388 (ia64_arg_type): Likewise.
13389 (ia64_mode_to_int): Likewise.
13390 (ia64_scalar_mode_supported_p): Likewise.
13391 (ia64_vector_mode_supported_p): Likewise.
13392 (expand_vec_perm_broadcast): Likewise.
13393 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13394 (iq2000_function_arg_advance): Likewise.
13395 (iq2000_function_arg): Likewise.
13396 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13397 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13398 (m68k_libcall_value): Likewise.
13399 (m68k_function_value): Likewise.
13400 (sched_attr_op_type): Likewise.
13401 * config/mcore/mcore.c (mcore_output_move): Likewise.
13402 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13404 (microblaze_function_arg): Likewise.
13405 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13406 (mips_print_operand): Likewise.
13407 (mips_mode_ok_for_mov_fmt_p): Likewise.
13408 (mips_vector_mode_supported_p): Likewise.
13409 (mips_preferred_simd_mode): Likewise.
13410 (mips_expand_vpc_loongson_even_odd): Likewise.
13411 (mips_expand_vec_unpack): Likewise.
13412 (mips_expand_vi_broadcast): Likewise.
13413 (mips_expand_vector_init): Likewise.
13414 (mips_expand_vec_reduc): Likewise.
13415 (mips_expand_msa_cmp): Likewise.
13416 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13417 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13418 (cc_flags_for_mode): Likewise.
13419 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13420 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13421 (nds32_output_casesi_pc_relative): Likewise.
13422 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13423 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13424 (nvptx_gen_unpack): Likewise.
13425 (nvptx_gen_pack): Likewise.
13426 (nvptx_gen_shuffle): Likewise.
13427 (nvptx_gen_wcast): Likewise.
13428 (nvptx_preferred_simd_mode): Likewise.
13429 * config/pa/pa.c (pa_secondary_reload): Likewise.
13430 * config/pa/predicates.md (base14_operand): Likewise.
13431 * config/powerpcspe/powerpcspe-c.c
13432 (altivec_resolve_overloaded_builtin): Likewise.
13433 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13435 (rs6000_preferred_simd_mode): Likewise.
13436 (output_vec_const_move): Likewise.
13437 (rs6000_expand_vector_extract): Likewise.
13438 (rs6000_split_vec_extract_var): Likewise.
13439 (reg_offset_addressing_ok_p): Likewise.
13440 (rs6000_legitimate_offset_address_p): Likewise.
13441 (rs6000_legitimize_address): Likewise.
13442 (rs6000_emit_set_const): Likewise.
13443 (rs6000_const_vec): Likewise.
13444 (rs6000_emit_move): Likewise.
13445 (spe_build_register_parallel): Likewise.
13446 (rs6000_darwin64_record_arg_recurse): Likewise.
13447 (swap_selector_for_mode): Likewise.
13448 (spe_init_builtins): Likewise.
13449 (paired_init_builtins): Likewise.
13450 (altivec_init_builtins): Likewise.
13451 (do_load_for_compare): Likewise.
13452 (rs6000_generate_compare): Likewise.
13453 (rs6000_expand_float128_convert): Likewise.
13454 (emit_load_locked): Likewise.
13455 (emit_store_conditional): Likewise.
13456 (rs6000_output_function_epilogue): Likewise.
13457 (rs6000_handle_altivec_attribute): Likewise.
13458 (rs6000_function_value): Likewise.
13459 (emit_fusion_gpr_load): Likewise.
13460 (emit_fusion_p9_load): Likewise.
13461 (emit_fusion_p9_store): Likewise.
13462 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13463 (fusion_gpr_mem_load): Likewise.
13464 (fusion_addis_mem_combo_load): Likewise.
13465 (fusion_addis_mem_combo_store): Likewise.
13466 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13467 (fusion_gpr_mem_load): Likewise.
13468 (fusion_addis_mem_combo_load): Likewise.
13469 (fusion_addis_mem_combo_store): Likewise.
13470 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13472 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13473 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13474 (rs6000_preferred_simd_mode): Likewise.
13475 (output_vec_const_move): Likewise.
13476 (rs6000_expand_vector_extract): Likewise.
13477 (rs6000_split_vec_extract_var): Likewise.
13478 (reg_offset_addressing_ok_p): Likewise.
13479 (rs6000_legitimate_offset_address_p): Likewise.
13480 (rs6000_legitimize_address): Likewise.
13481 (rs6000_emit_set_const): Likewise.
13482 (rs6000_const_vec): Likewise.
13483 (rs6000_emit_move): Likewise.
13484 (rs6000_darwin64_record_arg_recurse): Likewise.
13485 (swap_selector_for_mode): Likewise.
13486 (paired_init_builtins): Likewise.
13487 (altivec_init_builtins): Likewise.
13488 (rs6000_expand_float128_convert): Likewise.
13489 (emit_load_locked): Likewise.
13490 (emit_store_conditional): Likewise.
13491 (rs6000_output_function_epilogue): Likewise.
13492 (rs6000_handle_altivec_attribute): Likewise.
13493 (rs6000_function_value): Likewise.
13494 (emit_fusion_gpr_load): Likewise.
13495 (emit_fusion_p9_load): Likewise.
13496 (emit_fusion_p9_store): Likewise.
13497 * config/rx/rx.c (rx_gen_move_template): Likewise.
13498 (flags_from_mode): Likewise.
13499 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13500 (s390_slb_comparison): Likewise.
13501 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13502 (s390_vector_mode_supported_p): Likewise.
13503 (s390_cc_modes_compatible): Likewise.
13504 (s390_match_ccmode_set): Likewise.
13505 (s390_canonicalize_comparison): Likewise.
13506 (s390_emit_compare_and_swap): Likewise.
13507 (s390_branch_condition_mask): Likewise.
13508 (s390_rtx_costs): Likewise.
13509 (s390_secondary_reload): Likewise.
13510 (__SECONDARY_RELOAD_CASE): Likewise.
13511 (s390_expand_cs): Likewise.
13512 (s390_preferred_simd_mode): Likewise.
13513 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13514 * config/sh/sh.c (sh_print_operand): Likewise.
13515 (dump_table): Likewise.
13516 (sh_secondary_reload): Likewise.
13517 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13518 * config/sh/sh.md (casesi_worker_1): Likewise.
13519 (casesi_worker_2): Likewise.
13520 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13521 (fcc_comparison_operator): Likewise.
13522 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13523 (emit_soft_tfmode_cvt): Likewise.
13524 (sparc_preferred_simd_mode): Likewise.
13525 (output_cbranch): Likewise.
13526 (sparc_print_operand): Likewise.
13527 (sparc_expand_vec_perm_bmask): Likewise.
13528 (vector_init_bshuffle): Likewise.
13529 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13530 (spu_vector_mode_supported_p): Likewise.
13531 (spu_expand_insv): Likewise.
13532 (spu_emit_branch_or_set): Likewise.
13533 (spu_handle_vector_attribute): Likewise.
13534 (spu_builtin_splats): Likewise.
13535 (spu_builtin_extract): Likewise.
13536 (spu_builtin_promote): Likewise.
13537 (spu_expand_sign_extend): Likewise.
13538 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13539 (tilegx_simd_int): Likewise.
13540 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13541 (tilepro_simd_int): Likewise.
13542 * config/v850/v850.c (const_double_split): Likewise.
13543 (v850_print_operand): Likewise.
13544 (ep_memory_offset): Likewise.
13545 * config/vax/vax.c (vax_rtx_costs): Likewise.
13546 (vax_output_int_move): Likewise.
13547 (vax_output_int_add): Likewise.
13548 (vax_output_int_subtract): Likewise.
13549 * config/visium/predicates.md (visium_branch_operator): Likewise.
13550 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13551 (visium_print_operand_address): Likewise.
13552 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13553 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13554 (xtensa_expand_conditional_branch): Likewise.
13555 (xtensa_copy_incoming_a7): Likewise.
13556 (xtensa_output_literal): Likewise.
13557 * dfp.c (decimal_real_maxval): Likewise.
13558 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13561 Alan Hayward <alan.hayward@arm.com>
13562 David Sherwood <david.sherwood@arm.com>
13564 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13565 (mode_nunits_inline): Likewise.
13566 (mode_inner_inline): Likewise.
13567 (mode_unit_size_inline): Likewise.
13568 (mode_unit_precision_inline): Likewise.
13569 (emit_insn_modes_h): Likewise. Also emit a #define of the
13571 (emit_mode_wider): Add an E_ prefix to mode names.
13572 (emit_mode_complex): Likewise.
13573 (emit_mode_inner): Likewise.
13574 (emit_mode_adjustments): Likewise.
13575 (emit_mode_int_n): Likewise.
13576 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13577 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13578 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13579 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13580 (CRC32_BUILTIN, ENTRY): Likewise.
13581 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13582 (aarch64_pop_regs): Likewise.
13583 (aarch64_process_components): Likewise.
13584 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13585 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13586 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13587 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13588 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13589 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13591 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13592 (pcmpistr): Likewise.
13593 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13594 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13595 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13597 * config/rl78/rl78.c (mduc_regs): Likewise.
13598 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13599 (htm_expand_builtin): Likewise.
13600 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13601 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13602 * config/xtensa/xtensa.c (print_operand): Likewise.
13603 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13604 (NUM_MODE_VECTOR_INT): Likewise.
13605 * genoutput.c (null_operand): Likewise.
13606 (output_operand_data): Likewise.
13607 * genrecog.c (print_parameter_value): Likewise.
13608 * lra.c (debug_operand_data): Likewise.
13610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13612 * dwarf2out.c (scompare_loc_descriptor_wide)
13613 (scompare_loc_descriptor_narrow): New functions, split out from...
13614 (scompare_loc_descriptor): ...here.
13615 * expmed.c (emit_store_flag_int): New function, split out from...
13616 (emit_store_flag): ...here.
13618 2017-08-30 Richard Biener <rguenther@suse.de>
13620 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13621 (dwarf2out_early_finish): Move setting of AT_pubnames from
13622 early debug output to early finish.
13624 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13626 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13627 and -mdata-region to the assembler.
13629 2017-08-30 Richard Biener <rguenther@suse.de>
13631 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13633 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13634 (dwarf2out_early_global_decl): Only generate a DIE for the
13635 abstract origin if it doesn't already exist or is a declaration DIE.
13636 (resolve_addr): Do not add the linkage name twice when
13637 generating a stub DIE for the DW_TAG_GNU_call_site target.
13639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13641 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13642 Use machine_mode rather than int for arg1_mode.
13644 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13647 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13648 that the second argument of the built-in functions to unpack
13649 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13650 switch statement instead a lot of if statements.
13651 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13652 Allow 64-bit values to be in Altivec registers as well as
13653 traditional floating point registers.
13654 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13656 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13658 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13659 MAX_REGS_PER_ADDRESS == 1.
13661 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13663 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13664 * config/i386/i386.c (ix86_option_override_internal): Simplify
13665 setting of opts->x_flag_entry.
13667 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13668 Jakub Jelinek <jakub@redhat.com>
13669 Richard Biener <rguenther@suse.de>
13671 PR tree-optimization/81503
13672 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13673 folded constant fits in the target type; reorder tests for clarity.
13675 2017-08-29 Martin Liska <mliska@suse.cz>
13677 * passes.def: Include pass_lower_switch.
13678 * stmt.c (dump_case_nodes): Remove and move to
13679 tree-switch-conversion.
13680 (case_values_threshold): Likewise.
13681 (expand_switch_as_decision_tree_p): Likewise.
13682 (emit_case_decision_tree): Likewise.
13683 (expand_case): Likewise.
13684 (balance_case_nodes): Likewise.
13685 (node_has_low_bound): Likewise.
13686 (node_has_high_bound): Likewise.
13687 (node_is_bounded): Likewise.
13688 (emit_case_nodes): Likewise.
13689 (struct simple_case_node): New struct.
13690 (add_case_node): Remove.
13691 (emit_case_dispatch_table): Use vector instead of case_list.
13692 (reset_out_edges_aux): Remove.
13693 (compute_cases_per_edge): Likewise.
13694 (expand_case): Build list of simple_case_node.
13695 (expand_sjlj_dispatch_table): Use it.
13696 * tree-switch-conversion.c (struct case_node): Moved from
13697 stmt.c and adjusted.
13698 (emit_case_nodes): Likewise.
13699 (node_has_low_bound): Likewise.
13700 (node_has_high_bound): Likewise.
13701 (node_is_bounded): Likewise.
13702 (case_values_threshold): Likewise.
13703 (reset_out_edges_aux): Likewise.
13704 (compute_cases_per_edge): Likewise.
13705 (add_case_node): Likewise.
13706 (dump_case_nodes): Likewise.
13707 (balance_case_nodes): Likewise.
13708 (expand_switch_as_decision_tree_p): Likewise.
13709 (emit_jump): Likewise.
13710 (emit_case_decision_tree): Likewise.
13711 (try_switch_expansion): Likewise.
13712 (do_jump_if_equal): Likewise.
13713 (emit_cmp_and_jump_insns): Likewise.
13714 (fix_phi_operands_for_edge): New function.
13715 (record_phi_operand_mapping): Likewise.
13716 (class pass_lower_switch): New pass.
13717 (pass_lower_switch::execute): New function.
13718 (make_pass_lower_switch): Likewise.
13719 (conditional_probability):
13720 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13721 * tree-pass.h: Add make_pass_lower_switch.
13723 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13726 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13729 2017-08-29 Richard Biener <rguenther@suse.de>
13731 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13732 we do not add a DW_AT_inline attribute twice.
13733 (gen_subprogram_die): Remove code setting DW_AT_inline on
13734 DECL_ABSTRACT_P nodes.
13736 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13738 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13739 calls to internal functions.
13740 (gimplify_modify_expr): Likewise.
13741 * tree-call-cdce.c (use_internal_fn): Likewise.
13742 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13743 (convert_to_divmod): Set the nothrow flag.
13744 * tree-if-conv.c (predicate_mem_writes): Likewise.
13745 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13746 (vectorizable_call): Likewise.
13747 (vectorizable_store): Likewise.
13748 (vectorizable_load): Likewise.
13749 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13750 (vect_recog_mask_conversion_pattern): Likewise.
13752 2017-08-29 Martin Liska <mliska@suse.cz>
13755 * gcc.c (driver_handle_option): Add new argument.
13756 * opts-common.c (handle_option): Pass
13757 target_option_override_hook.
13758 * opts-global.c (lang_handle_option): Add new option.
13759 (set_default_handlers): Add new argument.
13760 (decode_options): Likewise.
13761 * opts.c (target_handle_option): Likewise.
13762 (common_handle_option): Call target_option_override_hook.
13763 * opts.h (struct cl_option_handler_func): Add hook for
13764 target option override.
13765 (struct cl_option_handlers): Likewise.
13766 (set_default_handlers): Add new argument.
13767 (decode_options): Likewise.
13768 (common_handle_option): Likewise.
13769 (target_handle_option): Likewise.
13770 * toplev.c (toplev::main): Pass targetm.target_option.override
13773 2017-08-29 Richard Biener <rguenther@suse.de>
13774 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13776 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13777 life to the active subtree.
13779 2017-08-28 Jeff Law <law@redhat.com>
13781 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13782 derive_equivalences.
13783 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13784 Code moved into....
13785 (edge_info::derive_equivalences): New private member function
13787 * tree-ssa-dom.c (class edge_info): Changed from a struct
13788 to a class. Add ctor/dtor, methods and data members.
13789 (edge_info::edge_info): Renamed from allocate_edge_info.
13790 Initialize additional members.
13791 (edge_info::~edge_info): New.
13792 (free_dom_edge_info): Delete the edge info.
13793 (record_edge_info): Use new class & associated member functions.
13794 Tighten forms for testing for edge equivalences.
13795 (record_temporary_equivalences): Iterate over the simple
13796 equivalences rather than assuming there's only one per edge.
13797 (cprop_into_successor_phis): Iterate over the simple
13798 equivalences rather than assuming there's only one per edge.
13799 (optimize_stmt): Use operand_equal_p rather than pointer
13800 equality for mini-DSE code.
13802 2017-08-28 Nathan Sidwell <nathan@acm.org>
13804 * gcc.c (execute): Fold SIGPIPE handling into switch
13805 statement. Adjust internal error message.
13807 2017-08-28 Richard Biener <rguenther@suse.de>
13810 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13811 Do nothing for removed DIEs.
13813 2017-08-28 Richard Biener <rguenther@suse.de>
13815 PR tree-optimization/81977
13816 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13819 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13822 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13823 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13824 using targetm.gen_mem_thread_fence.
13826 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13829 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13830 predicate to register_operand. Reorder operands.
13831 (*btr<mode>): Ditto.
13832 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13833 (*btr<mode>_mask): Ditto.
13835 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13837 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13838 * config/rs6000/xmmintrin.h: New file.
13839 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13841 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13844 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13845 parameter and_insn and return it.
13846 (recombine_lvx_pattern): Insert a copy to ensure availability of
13847 the base register of the copied masking operation at the point of
13848 the instruction replacement.
13849 (recombine_stvx_pattern): Likewise.
13851 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13853 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13854 undocumented switches.
13855 (-mpower9-dform-vector): Likewise.
13856 (-mpower9-dform): Likewise.
13857 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13858 comments to delete references to -mpower9-dform* switches.
13859 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13860 Delete reference to -mpower9-dform* switches, test for
13861 -mpower9-vector instead.
13862 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13863 (OTHER_P9_VECTOR_MASKS): Likewise.
13864 (POWERPC_MASKS): Likewise.
13865 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13866 tests against -mpower9-dform* to -mpower9-vector. Delete code
13867 that checked for -mpower9-dform* consistancy with other options.
13868 Add test for -mpower9-misc to enable other power9 switches.
13869 (rs6000_init_hard_regno_mode_ok): Likewise.
13870 (rs6000_option_override_internal): Likewise.
13871 (rs6000_emit_prologue): Likewise.
13872 (rs6000_emit_epilogue): Likewise.
13873 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13874 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13875 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13877 (emit_fusion_p9_store): Likewise.
13878 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13879 resetting these macros if the assembler does not support ISA 3.0
13881 (TARGET_P9_DFORM_VECTOR): Likewise.
13882 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13883 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13885 2017-08-25 Alan Modra <amodra@gmail.com>
13887 PR rtl-optimization/81747
13888 * cse.c (cse_extended_basic_block): Don't attempt to record
13889 equivalences for degenerate conditional branches that branch
13890 to their fall-through.
13892 2017-08-24 Martin Sebor <msebor@redhat.com>
13894 PR middle-end/81908
13895 * gimple-fold.c (size_must_be_zero_p): New function.
13896 (gimple_fold_builtin_memory_op): Call it.
13898 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13900 * config/rs6000/mm_malloc.h: New file.
13902 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13904 PR tree-optimization/81913
13905 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13906 analysis when either IVs in condition can wrap.
13908 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13910 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13911 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13913 2017-08-24 Richard Biener <rguenther@suse.de>
13916 * targhooks.c (default_target_can_inline_p): Properly
13917 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13918 is present and always compare.
13919 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13920 infer -mfpmath=sse from TARGET_SSE_P.
13921 (ix86_can_inline_p): Properly use target_option_default_node when
13922 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13924 2017-08-24 Richard Biener <rguenther@suse.de>
13927 * dwarf2out.c (output_die): Handle flag_generate_offload like
13929 (output_comp_unit): Likewise.
13930 (gen_array_type_die): Likewise.
13931 (dwarf2out_early_finish): Likewise.
13932 (note_variable_value_in_expr): Likewise.
13933 (dwarf2out_finish): Likewise. Adjust assert.
13934 * cgraphunit.c (symbol_table::compile): Move setting of
13935 flag_generate_offload earlier ...
13936 (symbol_table::finalize_compilation_unit): ... here, before
13937 early debug finalization.
13939 2017-08-24 Richard Biener <rguenther@suse.de>
13941 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13942 and ipa-fnsummary.h.
13943 (ix86_can_inline_p): When ix86_fpmath flags do not match
13944 check whether the callee uses FP math at all.
13946 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13948 PR middle-end/81931
13949 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13950 instead of TYPE_PRECISION.
13952 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13953 Alan Hayward <alan.hayward@arm.com>
13954 David Sherwood <david.sherwood@arm.com>
13956 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13958 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13960 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13962 (ix86_frame::stack_realign_allocate): New field.
13963 (struct machine_frame_state): Modify comments.
13964 (machine_frame_state::sp_realigned_fp_end): New field.
13965 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13966 layout calculation.
13967 (sp_valid_at): Add assertion to assure no attempt to access invalid
13968 offset of a realigned stack.
13969 (fp_valid_at): Likewise.
13970 (choose_baseaddr): Modify comments.
13971 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13972 ix86_expand_prologue.
13973 (ix86_expand_prologue): Modify stack realignment and allocation.
13974 (ix86_expand_epilogue): Modify comments.
13975 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13976 avx2_runtime, avx512f, and avx512f_runtime.
13978 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13980 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13981 (mstackrealign): Do not init to -1.
13982 * config/i386/i386.c (ix86_option_override_internal):
13983 Check opts_set, not opts when setting default value of
13984 opts->x_ix86_force_align_arg_pointer.
13986 2017-08-23 Richard Biener <rguenther@suse.de>
13988 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13989 lang_hooks.decl_printable_name.
13990 * print-rtl-function.c (print_rtx_function): Likewise.
13991 * tree-pretty-print.c (dump_function_header): Likewise.
13993 2017-08-23 Richard Biener <rguenther@suse.de>
13996 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13997 -g0 at compile-time.
13999 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14001 PR middle-end/19706
14002 * doc/sourcebuild.texi (Other hardware attributes):
14005 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14007 PR middle-end/19706
14008 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14009 Fix single-use check.
14011 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14013 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14015 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14017 * doc/install.texi: Modify to add more details on running selected
14020 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14022 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14023 is combined with -mabi=ms.
14024 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14027 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14029 PR tree-optimization/81488
14030 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14031 and cached_basis fields.
14032 (MAX_SPREAD): New constant.
14033 (alloc_cand_and_find_basis): Initialize new fields.
14034 (clear_visited): New function.
14035 (create_phi_basis_1): Rename from create_phi_basis, set visited
14036 and cached_basis fields.
14037 (create_phi_basis): New wrapper function.
14038 (phi_add_costs_1): Rename from phi_add_costs, add spread
14039 parameter, set visited field, short-circuit when limits reached.
14040 (phi_add_costs): New wrapper function.
14041 (record_phi_increments_1): Rename from record_phi_increments, set
14043 (record_phi_increments): New wrapper function.
14044 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14045 (phi_incr_cost): New wrapper function.
14046 (all_phi_incrs_profitable_1): Rename from
14047 all_phi_incrs_profitable, set visited field.
14048 (all_phi_incrs_profitable): New wrapper function.
14050 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14051 Alan Hayward <alan.hayward@arm.com>
14052 David Sherwood <david.sherwood@arm.com>
14054 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14055 that takes the outer and inner modes.
14056 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14057 comparison as the canonical test for a paradoxical subreg.
14058 * combine.c (simplify_set): Use paradoxical_subreg_p.
14059 (make_extraction): Likewise.
14060 (force_to_mode): Likewise.
14061 (rtx_equal_for_field_assignment_p): Likewise.
14062 (gen_lowpart_for_combine): Likewise.
14063 (simplify_comparison): Likewise.
14064 * cse.c (equiv_constant): Likewise.
14065 * expmed.c (store_bit_field_1): Likewise.
14066 * final.c (alter_subreg): Likewise.
14067 * fwprop.c (propagate_rtx): Likewise.
14068 (forward_propagate_subreg): Likewise.
14069 * ira-conflicts.c (ira_build_conflicts): Likewise.
14070 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14071 * lra-constraints.c (curr_insn_transform): Likewise.
14072 (split_reg): Likewise.
14073 * lra-eliminations.c (move_plus_up): Likewise.
14074 (lra_eliminate_regs_1): Likewise.
14075 * recog.c (general_operand): Likewise.
14076 * ree.c (combine_reaching_defs): Likewise.
14077 * reload.c (push_reload): Likewise.
14078 (find_reloads): Likewise.
14079 * reload1.c (elimination_effects): Likewise.
14080 (compute_reload_subreg_offset): Likewise.
14081 (choose_reload_regs): Likewise.
14082 * rtlanal.c (subreg_lsb_1): Likewise.
14083 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14084 (simplify_subreg): Likewise.
14085 * var-tracking.c (track_loc_p): Likewise.
14086 * emit-rtl.c (byte_lowpart_offset): Likewise.
14087 (paradoxical_subreg_p): Delete out-of-line definition.
14089 2017-08-22 Jeff Law <law@redhat.com>
14091 PR tree-optimization/81741
14092 PR tree-optimization/71947
14093 * tree-ssa-dom.c: Include tree-inline.h.
14094 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14095 equivalences if one is more expensive to compute than the other.
14096 * tree-ssa-scopedtables.h (class const_or_copies): Make
14097 record_const_or_copy_raw method private.
14098 (class avail_exprs_stack): New method simplify_binary_operation.
14099 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14100 avail_exprs_stack::simplify_binary_operation as needed.
14101 (avail_exprs_stack::simplify_binary_operation): New function.
14103 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14105 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14106 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14107 (DOT_SYMBOLS): Likewise.
14108 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14109 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14110 (RS6000_ABI_NAME): Likewise.
14111 (TARGET_CMODEL): Likewise.
14112 (TOC_SECTION_ASM_OP): Likewise.
14113 (SET_CMODEL): New macro.
14114 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14116 2017-08-22 Richard Biener <rguenther@suse.de>
14118 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14119 to allow for free-lang-data replacements similar to verify_type_variant.
14121 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14122 Alan Hayward <alan.hayward@arm.com>
14123 David Sherwood <david.sherwood@arm.com>
14125 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14126 VOIDmode for the LABEL_REF.
14128 2017-08-22 Richard Biener <rguenther@suse.de>
14130 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14132 2017-08-22 Richard Biener <rguenther@suse.de>
14134 * common.opt (feliminate-dwarf2-dups): Ignore.
14135 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14136 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14137 same_die_p_wrap, compute_section_prefix,
14138 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14139 (comdat_symbol_id, comdat_symbol_number): Likewise.
14140 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14142 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14143 (output_die): Mark unreachable path unreachable.
14144 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14145 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14146 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14147 (dwarf2out_early_finish): Likewise.
14149 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14151 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14153 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14156 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14157 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14158 (avr_attribute_table) <io, io_low, address>: Initialize
14159 .decl_required with true.
14161 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14163 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14164 undocumented debugging options.
14165 (-mvsx-scalar-double): Likewise.
14166 (-mallow-df-permute): Likewise.
14167 (-mvectorize-builtins): Likewise.
14168 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14169 (rs6000_builtin_vectorized_function): Likewise.
14170 (rs6000_builtin_md_vectorized_function): Likewise.
14171 (rs6000_opt_vars): Likewise.
14173 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14176 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14177 (*btrq_imm): Rename from *btrq.
14178 (*btcq_imm): Rename from *btcq.
14179 (btsc): New code attribute.
14180 (*<btsc><mode>): New insn pattern.
14181 (*btr<mode>): Ditto.
14182 (*<btsc><mode>_mask): New insn_and_split pattern.
14183 (*btr<mode>_mask): Ditto.
14185 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14186 Alan Hayward <alan.hayward@arm.com>
14187 David Sherwood <david.sherwood@arm.com>
14189 * function.c (pad_below): Simplify padding calculation.
14191 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14192 Alan Hayward <alan.hayward@arm.com>
14193 David Sherwood <david.sherwood@arm.com>
14195 * target.def (function_prologue): Remove frame size argument.
14196 (function_epilogue): Likewise.
14197 * doc/tm.texi: Regenerate.
14198 * final.c (final_start_function): Update call to function_prologue.
14199 (final_end_function): Update call to function_epilogue.
14200 (default_function_pro_epilogue): Remove frame size argument.
14201 * output.h (default_function_pro_epilogue): Likewise.
14202 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14203 (arm_output_function_prologue): Likewise.
14204 * config/frv/frv.c (frv_function_prologue): Likewise.
14205 (frv_function_epilogue): Likewise.
14206 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14207 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14208 (ia64_output_function_epilogue): Likewise.
14209 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14210 (m32r_output_function_epilogue): Likewise.
14211 * config/microblaze/microblaze.c (microblaze_function_prologue)
14212 (microblaze_function_epilogue): Likewise.
14213 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14214 (mips_output_function_epilogue): Likewise.
14215 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14216 (mmix_target_asm_function_epilogue): Likewise.
14217 * config/msp430/msp430.c (msp430_start_function): Likewise.
14218 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14219 (nds32_asm_function_epilogue): Likewise.
14220 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14221 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14222 (pa_output_function_epilogue): Likewise.
14223 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14224 (rs6000_output_function_epilogue): Likewise.
14225 * config/rl78/rl78.c (rl78_start_function): Likewise.
14226 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14227 (rs6000_output_function_epilogue): Likewise.
14228 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14229 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14230 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14231 (sparc_asm_function_epilogue): Likewise.
14233 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14235 * tree.h (type_has_mode_precision_p): New function.
14236 * convert.c (convert_to_integer_1): Use it.
14237 * expr.c (expand_expr_real_2): Likewise.
14238 (expand_expr_real_1): Likewise.
14239 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14240 * match.pd: Likewise.
14241 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14242 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14243 * tree-tailcall.c (process_assignment): Likewise.
14244 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14245 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14246 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14247 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14248 (vectorizable_assignment): Likewise.
14249 (vectorizable_shift): Likewise.
14250 (vectorizable_operation): Likewise.
14251 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14253 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14255 * match.pd: Add pow (C, x) simplification.
14257 2017-08-21 Richard Biener <rguenther@suse.de>
14259 PR tree-optimization/81900
14260 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14261 for blocks with abnormal predecessors.
14262 (compute_antic): Do not set visited flag prematurely.
14264 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14267 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14269 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14271 * stor-layout.h (vector_type_mode): Move to...
14272 * tree.h (vector_type_mode): ...here.
14273 * stor-layout.c (vector_type_mode): Move to...
14274 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14276 2017-08-21 Richard Biener <rguenther@suse.de>
14278 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14279 register_external_die hooks.
14280 (debug_false_tree_charstarstar_uhwistar): Declare.
14281 (debug_nothing_tree_charstar_uhwi): Likewise.
14282 * debug.c (do_nothing_debug_hooks): Adjust.
14283 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14284 (debug_nothing_tree_charstar_uhwi): Likewise.
14285 * dbxout.c (dbx_debug_hooks): Adjust.
14286 (xcoff_debug_hooks): Likewise.
14287 * sdbout.c (sdb_debug_hooks): Likewise.
14288 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14289 * dwarf2out.c (macinfo_label_base): New global.
14290 (dwarf2out_register_external_die): New function for the
14291 register_external_die hook.
14292 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14293 (dwarf2_debug_hooks): Use them.
14294 (dwarf2_lineno_debug_hooks): Adjust.
14295 (struct die_struct): Add with_offset flag.
14296 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14297 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14298 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14299 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14300 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14301 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14302 defining section names for the early LTO debug variants.
14303 (reset_indirect_string): New helper.
14304 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14305 (print_dw_val): Add support for offsetted symbol references.
14306 (get_ultimate_context): Split out from is_cxx.
14307 (is_cxx): Use get_ultimate_context.
14308 (is_fortran): Add decl overload.
14309 (compute_comp_unit_symbol): Split out worker from
14310 compute_section_prefix.
14311 (compute_section_prefix): Call compute_comp_unit_symbol and
14312 set comdat_type_p here.
14313 (output_die): Skip DIE symbol output for the LTO added one.
14314 Handle DIE symbol references with offset.
14315 (output_comp_unit): Guard section name mangling properly.
14316 For LTO debug sections emit a symbol at the section beginning
14317 which we use to refer to its DIEs.
14318 (add_abstract_origin_attribute): For DIEs registered via
14319 dwarf2out_register_external_die directly refer to the early
14320 DIE rather than indirectly through the shadow one we created.
14321 Remove obsolete call to dwarf2out_abstract_function for
14322 non-function/block origins.
14323 (gen_array_type_die): When generating early LTO debug do
14324 not emit DW_AT_string_length.
14325 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14326 late when in LTO. As suggested place a gcc_unreachable for
14327 the DECL_ABSTRACT_P case.
14328 (gen_subprogram_die): Avoid another specification DIE
14329 for early built declarations/definitions for the late LTO case.
14330 (gen_variable_die): Add type references for late duplicated VLA dies
14332 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14333 we have the abstract instance already.
14334 (process_scope_var): Adjust decl DIE contexts in LTO which
14335 first puts them in limbo.
14336 (gen_decl_die): Do not generate type DIEs late apart from
14337 types for VLAs or for decls we do not yet have a DIE. Do not
14338 call dwarf2out_abstract_function late.
14339 (dwarf2out_early_global_decl): Make sure to create DIEs
14340 for abstract instances of a decl first.
14341 (dwarf2out_late_global_decl): Adjust comment.
14342 (output_macinfo_op): With multiple macro sections use
14343 macinfo_label_base to distinguish labels.
14344 (output_macinfo): Likewise. Update macinfo_label_base.
14345 Pass in the line info label.
14346 (note_variable_value_in_expr): When generating LTO resolve
14347 all variable values here by generating DIEs as needed.
14348 (init_sections_and_labels): Add early LTO debug flag parameter
14349 and generate different sections and names if set. Add generation
14350 counter for the labels so we can have multiple of them.
14351 (reset_dies): Helper to allow DIEs to be output multiple times.
14352 (dwarf2out_finish): When outputting DIEs to the fat part of an
14353 LTO object first reset DIEs.
14354 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14355 (modified_type_die): Check for decl_ultimate_origin being self
14357 (gen_type_die_with_usage): Likewise.
14358 (gen_typedef_die): Allow decl_ultimate_origin being self.
14359 (set_decl_abstract_flags): Remove.
14360 (set_block_abstract_flags): Likewise.
14361 (dwarf2out_abstract_function): Treat the early generated DIEs
14362 as the abstract copy and only add DW_AT_inline and
14363 DW_AT_artificial here and call set_decl_origin_self.
14364 If the DIE has an abstract origin don't do anything.
14365 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14366 if we have none yet (Go fails to build one, PR78628).
14367 (variably_modified_type_p): Prevent endless recursion for Ada
14368 cyclic pointer types.
14369 * lto-streamer-in.c: Include debug.h.
14370 (dref_queue): New global.
14371 (lto_read_tree_1): Stream in DIE references.
14372 (lto_input_tree): Register DIE references.
14373 (input_function): Stream DECL_DEBUG_ARGS.
14374 * lto-streamer-out.c: Include debug.h.
14375 (lto_write_tree_1): Output DIE references.
14376 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14377 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14378 (output_function): Stream DECL_DEBUG_ARGS.
14379 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14380 Stream DECL_ABSTRACT_ORIGIN.
14381 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14382 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14383 DECL_CONTEXT for file-scope decls.
14384 * lto-streamer.h (struct dref_entry): Declare.
14385 (dref_queue): Likewise.
14386 * cfgexpand.c (pass_expand::execute): Do not call the
14387 outlining_inline_function hook here.
14388 * lto-wrapper.c (debug_obj): New global.
14389 (tool_cleanup): Unlink it if required.
14390 (debug_objcopy): New function.
14391 (run_gcc): Handle early debug sections in the IL files by
14392 extracting them to separate files, partially linkin them and
14393 feeding the result back as result to the linker.
14394 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14395 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14396 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14397 sections into a separate segment.
14398 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14400 (darwin_asm_dwarf_section): Likewise.
14401 (darwin_asm_output_dwarf_offset): Likewise.
14402 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14404 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14405 Alan Hayward <alan.hayward@arm.com>
14406 David Sherwood <david.sherwood@arm.com>
14408 * read-md.h (md_reader::record_potential_iterator_use): Replace
14409 pointer argument with an rtx and an index.
14410 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14411 (apply_mode_iterator): Likewise.
14412 (apply_code_iterator): Likewise.
14413 (apply_int_iterator): Likewise.
14414 (apply_subst_iterator): Likewise.
14415 (record_iterator_use): Likewise.
14416 (record_attribute_use): Likewise.
14417 (md_reader::record_potential_iterator_use): Likewise. Update calls
14418 to record_iterator_use and apply_iterator.
14419 (iterator_use): Replace ptr with x and index.
14420 (attribute_use): Likewise.
14421 (apply_attribute_uses): Update calls to apply_iterator.
14422 (apply_iterators): Likewise. Update initialization of iterator_use.
14423 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14424 and record_potential_iterator_use.
14425 (rtx_reader::read_rtx_operand): Likewise.
14427 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14428 Alan Hayward <alan.hayward@arm.com>
14429 David Sherwood <david.sherwood@arm.com>
14431 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14434 2017-08-21 Richard Biener <rguenther@suse.de>
14436 PR middle-end/81884
14437 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14438 at struct end conservatively when comparing common bases.
14440 2017-08-21 Richard Biener <rguenther@suse.de>
14442 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14443 (mem_ref_in_stmt): Remove.
14444 (determine_max_movement): Use ref index to get at the reference.
14445 (invariantness_dom_walker::before_dom_children): Deal with
14446 lim data already initialized.
14447 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14449 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14451 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14452 (TARGET_ISA_ROUND): Ditto.
14453 (TARGET_ROUND): Ditto.
14454 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14455 * config/i386/i386.md: Ditto.
14456 * config/i386/sse.md: Ditto.
14457 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14458 with OPTION_MASK_ISA_SSE4_1.
14460 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14463 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14464 __builtin_ia32_lzcnt_u16.
14466 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14469 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14470 (rs6000_set_current_function): Rewrite function to use it.
14472 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14475 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14476 and TYPE_WARN_IF_NOT_ALIGN.
14477 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14478 (handle_warn_if_not_align): New.
14479 (place_union_field): Call handle_warn_if_not_align.
14480 (place_field): Call handle_warn_if_not_align.
14481 Copy TYPE_WARN_IF_NOT_ALIGN.
14482 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14483 (layout_type): Likewise.
14484 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14486 (tree_decl_common): Add warn_if_not_align.
14487 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14488 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14489 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14490 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14491 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14492 * doc/extend.texi: Document warn_if_not_aligned attribute.
14493 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14495 2017-08-17 Martin Liska <mliska@suse.cz>
14498 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14499 (get_data_dependence): Use it as pointer type.
14500 (distribute_loop): Likewise.
14502 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14504 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14505 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14506 (p8_vmrgow_<mode>_direct): New define_insn.
14507 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14508 handle endianness for vmrgew and vmrgow permute patterns.
14510 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14512 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14513 * config/rs6000/rs6000-cpus.def: Remove comment.
14514 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14515 (POWERPC_MASKS): Likewise.
14516 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14517 use of TARGET_VSX_TIMODE.
14518 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14519 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14520 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14521 (rs6000_option_override_internal): Remove dead code.
14522 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14523 (rs6000_legitimize_reload_address): Likewise.
14524 (rs6000_legitimate_address_p): Likewise.
14525 (rs6000_opt_masks): Delete "vsx-timode".
14526 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14527 from function comment.
14528 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14529 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14530 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14532 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14533 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14534 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14535 (VSX_TI): Likewise.
14537 (define_peephole2): Likewise.
14539 2017-08-17 Martin Sebor <msebor@redhat.com>
14542 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14543 past the end of an array.
14544 (test_pp_format): Add test cases.
14546 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14548 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14549 missing ECF_NOTHROW flags.
14551 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14554 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14555 operands residing in integer registers.
14556 (*vsx_le_perm_load_<mode>): Likewise.
14557 (*vsx_le_perm_store_<mode>): Likewise.
14558 (define_peephole2): Add peepholes to optimize the above.
14560 2017-08-17 Marek Polacek <polacek@redhat.com>
14562 PR middle-end/81814
14563 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14564 to mimic what shorten_compare did. Change the return type to bool.
14565 (fold_cond_expr_with_comparison): Update call to
14566 operand_equal_for_comparison_p.
14567 (fold_ternary_loc): Likewise.
14569 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14571 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14573 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14575 2017-08-17 Richard Biener <rguenther@suse.de>
14577 * tree-ssa-structalias.c (solve_graph): When propagating
14578 to successors update the graphs succ edges and avoid duplicate work.
14580 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14583 * config/i386/i386.c (ix86_option_override_internal): Save target
14584 specific options after ix86_stack_protector_guard_reg was changed.
14586 2017-08-17 Richard Biener <rguenther@suse.de>
14588 PR tree-optimization/81827
14589 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14590 (new_var_info): Initialize it conservatively.
14591 (get_call_vi): Mark register vars.
14592 (new_scalar_tmp_constraint_exp): Likewise.
14593 (handle_rhs_call): Likewise.
14594 (handle_const_call): Likewise.
14595 (create_function_info_for): Likewise.
14596 (solve_constraints): Sort varinfos to separate register from
14597 non-register vars to pack points-to solution bitmaps during
14600 2017-08-17 Marek Polacek <polacek@redhat.com>
14602 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14604 2017-08-17 Richard Biener <rguenther@suse.de>
14606 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14607 to true when overflow is undefined and we saturated the result.
14609 2017-08-17 Alan Modra <amodra@gmail.com>
14612 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14613 Don't use store multiple if only one reg needs saving.
14614 (interesting_frame_related_regno): New function.
14615 (rs6000_frame_related): Don't emit frame info for regs that
14617 (rs6000_emit_epilogue): Likewise.
14619 2017-08-16 Nathan Sidwell <nathan@acm.org>
14621 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14622 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14623 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14624 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14625 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14626 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14627 (verify_type): Adjust for TYPE_BINFO move.
14628 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14629 process TYPE_BINFO directly.
14630 (hash_tree): Likewise.
14631 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14633 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14636 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14638 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14640 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14643 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14644 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14645 Add ix86_binary_operator_ok to insn constraint.
14646 (*iordi_1_bts): Ditto.
14647 (*xordi_1_btc): Ditto.
14648 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14649 Update corresponding peephole2 pattern.
14653 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14655 PR tree-optimization/81832
14656 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14657 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14659 2017-08-16 Marek Polacek <polacek@redhat.com>
14662 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14663 regarding size_zero_node.
14665 2017-08-16 Martin Liska <mliska@suse.cz>
14668 * config.gcc: Respect previously set extra_objs in case
14671 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14673 PR tree-optimization/81835
14674 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14675 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14676 not depend on the phi.
14678 2017-08-16 Alan Modra <amodra@gmail.com>
14680 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14683 2017-08-16 Alan Modra <amodra@gmail.com>
14685 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14686 (save_reg_p): ..into this. Update all callers.
14687 (first_reg_to_save): Simplify.
14689 2017-08-16 Alan Modra <amodra@gmail.com>
14691 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14694 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14698 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14699 constant count if that count is less than 32.
14701 2017-08-15 Nathan Sidwell <nathan@acm.org>
14703 * gcc.c (execute): Emit friendlier message if inferior is killed
14704 by an external cause.
14706 2017-08-15 Richard Biener <rguenther@suse.de>
14708 PR tree-optimization/81790
14709 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14710 CONSTRUCTORs from simplifying and VN.
14712 2017-08-14 Martin Sebor <msebor@redhat.com>
14714 * builtin-attrs.def: Add comments.
14716 2017-08-14 Martin Sebor <msebor@redhat.com>
14719 * doc/extend.texi (attribute nonstring): Document new attribute.
14721 2017-08-14 Martin Sebor <msebor@redhat.com>
14724 * tree-diagnostic.c (default_tree_printer): Handle %G.
14725 * gimple-pretty-print.h (percent_G_format): Declare new function.
14726 * gimple-pretty-print.c (percent_G_format): Define.
14727 * tree-pretty-print.c (percent_K_format): Add argument.
14729 2017-08-14 Martin Sebor <msebor@redhat.com>
14731 PR translation/79998
14732 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14733 Remove a stray space.
14735 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14738 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14739 (*iordi_1_bts): Ditto.
14740 (*xordi_1_btc): Ditto.
14742 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14745 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14746 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14748 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14749 quoted strings, and make more translator-friendly.
14750 (darwin_rs6000_override_options): Likewise.
14751 (rs6000_option_override_internal): Likewise.
14752 (rs6000_return_in_memory): Fix overlong line.
14753 (init_cmulative_args): Use quoted strings, and make more
14754 translator-friendly.
14755 (rs6000_pass_by_reference): Fix overlong line.
14756 (def_builtin): Use quoted strings.
14757 (altivec_expand_predicate_builtin): Use quoted strings, and make
14758 more translator-friendly.
14759 (htm_expand_builtin): Use quoted strings.
14760 (cpu_expand_builtin): Use quoted strings, and make more
14761 translator-friendly.
14762 (altivec_expand_builtin): Likewise.
14763 (paired_expand_predicate_builtin): Likewise.
14764 (rs6000_invalid_builtin): Likewise.
14765 (builtin_function_type): Use quoted strings.
14766 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14767 more translator-friendly.
14768 (rs6000_trampoline_init): Likewise.
14769 (rs6000_handle_altivec_attribute): Likewise.
14770 (rs6000_inner_target_options): Use quoted strings.
14771 (rs6000_disable_incompatible_switches): Likewise.
14772 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14773 strings, and make more translator-friendly.
14774 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14776 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14778 PR tree-optimization/81799
14779 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14780 cond_expr to simple gimple operand.
14782 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14784 PR middle-end/46932
14785 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14787 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14791 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14792 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14793 TARGET_GASISR_PROLOGUES.
14794 * config/avr/avr.c (avr_option_override): Same.
14795 (avr_pass_pre_proep::execute): Same.
14797 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14800 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14801 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14803 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14805 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14806 define_insn_and_split. Split to a memory load from 0 in
14807 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14808 using PTR mode iterator.
14809 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14810 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14811 (*add_tp_<mode>): Redefine as define_insn_and_split.
14812 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14813 address space. Merge with *add_tp_x32 using PTR mode iterator.
14814 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14815 Split to an add with a memory load from 0 in
14816 DEFAULT_TLS_SEG_REG address space.
14818 2017-08-12 Andrew Pinski <apinski@cavium.com>
14820 * config/aarch64/aarch64-option-extensions.def (rdma):
14821 Fix feature string to what Linux prints out in /proc/cpuinfo.
14823 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14826 * dwarf2out.c (modified_type_die): For C typedef types that have
14827 an ultimate origin, process the ultimate origin instead of the
14829 (gen_typedef_die): Assert that input DECLs have no ultimate
14831 (gen_type_die_with_usage): For typedef variants that have an
14832 ultimate origin, just call gen_decl_die on the original DECL.
14833 (process_scope_var): Avoid creating DIEs for local typedefs and
14834 concrete static variables.
14836 2017-08-12 Alan Modra <amodra@gmail.com>
14840 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14841 match gnu-user.h startfile.
14842 (ENDFILE_LINUX_SPEC): Similarly.
14844 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14847 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14849 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14851 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14852 * config/aarch64/aarch64.md (mov<mode>): Change.
14853 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14854 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14855 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14857 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14859 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14860 for storage order barriers.
14862 2017-08-11 Martin Liska <mliska@suse.cz>
14865 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14866 variables of void type.
14868 2017-08-11 Martin Liska <mliska@suse.cz>
14870 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14871 TARGET_SUPPORTS_ALIASES.
14872 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14873 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14874 (optimize_weakref): Likewise.
14875 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14876 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14877 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14879 2017-08-11 Martin Liska <mliska@suse.cz>
14882 * config/i386/i386.c (make_resolver_func): Do complete
14883 refactoring of the function.
14885 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14888 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14889 * config/i386/i386.c (ix86_stack_protect_guard): Use
14890 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14891 * doc/invoke.texi (x86 Options): Document
14892 -mstack-protector-guard-symbol= option.
14894 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14896 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14897 * config/i386/i386.c (ix86_split_stack_guard): New function.
14898 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14899 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14900 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14901 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14902 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14903 (split_stack_space_check): Call ix86_split_stack_guard.
14905 2017-08-10 Martin Sebor <msebor@redhat.com>
14907 * print-tree.c (print_node): Print location using the established
14909 Replace spaces with colons.
14910 (debug_raw, debug): Ditto.
14912 2017-08-10 Martin Sebor <msebor@redhat.com>
14915 * pretty-print.c (pp_format): Correct the handling of %s precision.
14917 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14920 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14922 (ix86_finalize_stack_frame_flags): This. Also clear
14923 frame_pointer_needed if -fno-omit-frame-pointer is used without
14925 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14926 with ix86_finalize_stack_frame_flags.
14927 (ix86_expand_epilogue): Likewise.
14928 (ix86_expand_split_stack_prologue): Likewise.
14929 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14931 2017-08-10 Martin Liska <mliska@suse.cz>
14934 * c-attribs.c (handle_target_attribute):
14935 Report warning for an empty string argument of target attribute.
14937 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14940 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14942 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14943 or DECL_NONLOCAL labels.
14944 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14945 or DECL_NONLOCAL labels here.
14947 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14949 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14950 to indicate when early gimple folding has been disabled.
14951 (rs6000_gimple_fold_builtin): Add debug content.
14952 (rs6000_invalid_builtin): Fix whitespace.
14953 (rs6000_expand_builtin): Fix whitespace.
14954 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14956 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14959 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14960 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14962 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14964 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14965 (qdf24xx): Likewise.
14966 * config/aarch64/aarch64.md: Include falkor.md.
14967 * config/aarch64/falkor.md: New.
14969 2017-08-09 Marek Polacek <polacek@redhat.com>
14972 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14973 * diagnostic.c (emit_diagnostic): Add a comment.
14974 (emit_diagnostic_valist): New function.
14976 2017-08-09 Marek Polacek <polacek@redhat.com>
14979 * input.c (make_location): New overload.
14980 * input.h (make_location): Declare.
14982 2017-08-08 Alan Modra <amodra@gmail.com>
14983 H.J. Lu <hongjiu.lu@intel.com>
14986 * gcc.c (NO_PIE_SPEC): Delete.
14987 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14989 (LINK_PIE_SPEC): ..to here.
14990 (LINK_COMMAND_SPEC): Support -no-pie.
14991 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14992 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14994 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14995 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14996 (ENDFILE_CRTEND_SPEC): Similarly.
14998 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15001 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15002 (mstack-protector-guard-offset=): Ditto.
15003 * config/i386/i386.c (ix86_option_override): Handle
15004 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15006 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15007 ix86_stack_protect_guard_offset variables.
15008 (TARGET_STACK_PROTECT_GUARD): Always define.
15009 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15010 and -mstack-protector-guard-offset= options.
15012 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15014 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15015 boundary case for the last candidate.
15017 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15019 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15020 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15022 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15024 PR middle-end/19706
15025 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15026 * config/aarch64/aarch64-builtins.c
15027 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15028 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15029 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15031 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15032 Andrew Pinski <pinskia@gmail.com>
15034 PR middle-end/19706
15035 * internal-fn.def (XORSIGN): New.
15036 * optabs.def (xorsign_optab): New.
15037 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15038 (convert_expand_mult_copysign): New.
15039 (pass_optimize_widening_mul::execute): Call
15040 convert_expand_mult_copysign.
15042 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15044 PR tree-optimization/81354
15045 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15046 Insert on edges rather than explicitly creating landing pads.
15047 (analyze_candidates_and_replace): Commit edge inserts.
15049 2017-08-08 Richard Biener <rguenther@suse.de>
15051 PR middle-end/81719
15052 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15053 (expand_simple_operations): Also look through ADDR_EXPRs with
15054 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15056 2017-08-08 Richard Biener <rguenther@suse.de>
15058 PR tree-optimization/81723
15059 * tree-vect-slp.c (struct bst_traits): New hash traits.
15060 (bst_fail): New global.
15061 (vect_build_slp_tree_2): New worker, split out from ...
15062 (vect_build_slp_tree): ... this now wrapping it with using
15063 bst_fail set to cache SLP tree build fails. Properly handle
15065 (vect_analyze_slp_instance): Allocate and free bst_fail.
15067 2017-08-08 Martin Liska <mliska@suse.cz>
15070 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15071 LABEL_DECLs that can be from a different function.
15073 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15075 PR tree-optimization/81744
15076 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15077 loop's number of iterations.
15079 2017-08-08 Martin Liska <mliska@suse.cz>
15081 * asan.c: Include header files.
15082 * attribs.c (build_decl_attribute_variant): New function moved
15084 (build_type_attribute_qual_variant): Likewise.
15085 (cmp_attrib_identifiers): Likewise.
15086 (simple_cst_list_equal): Likewise.
15087 (omp_declare_simd_clauses_equal): Likewise.
15088 (attribute_value_equal): Likewise.
15089 (comp_type_attributes): Likewise.
15090 (build_type_attribute_variant): Likewise.
15091 (lookup_ident_attribute): Likewise.
15092 (remove_attribute): Likewise.
15093 (merge_attributes): Likewise.
15094 (merge_type_attributes): Likewise.
15095 (merge_decl_attributes): Likewise.
15096 (merge_dllimport_decl_attributes): Likewise.
15097 (handle_dll_attribute): Likewise.
15098 (attribute_list_equal): Likewise.
15099 (attribute_list_contained): Likewise.
15100 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15101 (lookup_attribute_by_prefix): Likewise.
15102 * bb-reorder.c: Include header files.
15103 * builtins.c: Likewise.
15104 * calls.c: Likewise.
15105 * cfgexpand.c: Likewise.
15106 * cgraph.c: Likewise.
15107 * cgraphunit.c: Likewise.
15108 * convert.c: Likewise.
15109 * dwarf2out.c: Likewise.
15110 * final.c: Likewise.
15111 * fold-const.c: Likewise.
15112 * function.c: Likewise.
15113 * gimple-expr.c: Likewise.
15114 * gimple-fold.c: Likewise.
15115 * gimple-pretty-print.c: Likewise.
15116 * gimple.c: Likewise.
15117 * gimplify.c: Likewise.
15118 * hsa-common.c: Likewise.
15119 * hsa-gen.c: Likewise.
15120 * internal-fn.c: Likewise.
15121 * ipa-chkp.c: Likewise.
15122 * ipa-cp.c: Likewise.
15123 * ipa-devirt.c: Likewise.
15124 * ipa-fnsummary.c: Likewise.
15125 * ipa-inline.c: Likewise.
15126 * ipa-visibility.c: Likewise.
15128 * lto-cgraph.c: Likewise.
15129 * omp-expand.c: Likewise.
15130 * omp-general.c: Likewise.
15131 * omp-low.c: Likewise.
15132 * omp-offload.c: Likewise.
15133 * omp-simd-clone.c: Likewise.
15134 * opts-global.c: Likewise.
15135 * passes.c: Likewise.
15136 * predict.c: Likewise.
15137 * sancov.c: Likewise.
15138 * sanopt.c: Likewise.
15139 * symtab.c: Likewise.
15140 * toplev.c: Likewise.
15141 * trans-mem.c: Likewise.
15142 * tree-chkp.c: Likewise.
15143 * tree-eh.c: Likewise.
15144 * tree-into-ssa.c: Likewise.
15145 * tree-object-size.c: Likewise.
15146 * tree-parloops.c: Likewise.
15147 * tree-profile.c: Likewise.
15148 * tree-ssa-ccp.c: Likewise.
15149 * tree-ssa-live.c: Likewise.
15150 * tree-ssa-loop.c: Likewise.
15151 * tree-ssa-sccvn.c: Likewise.
15152 * tree-ssa-structalias.c: Likewise.
15153 * tree-ssa.c: Likewise.
15154 * tree-streamer-in.c: Likewise.
15155 * tree-vectorizer.c: Likewise.
15156 * tree-vrp.c: Likewise.
15157 * tsan.c: Likewise.
15158 * ubsan.c: Likewise.
15159 * varasm.c: Likewise.
15160 * varpool.c: Likewise.
15161 * tree.c: Remove functions moved to attribs.[ch].
15162 * tree.h: Likewise.
15163 * config/aarch64/aarch64.c: Add attrs.h header file.
15164 * config/alpha/alpha.c: Likewise.
15165 * config/arc/arc.c: Likewise.
15166 * config/arm/arm.c: Likewise.
15167 * config/avr/avr.c: Likewise.
15168 * config/bfin/bfin.c: Likewise.
15169 * config/c6x/c6x.c: Likewise.
15170 * config/cr16/cr16.c: Likewise.
15171 * config/cris/cris.c: Likewise.
15172 * config/darwin.c: Likewise.
15173 * config/epiphany/epiphany.c: Likewise.
15174 * config/fr30/fr30.c: Likewise.
15175 * config/frv/frv.c: Likewise.
15176 * config/ft32/ft32.c: Likewise.
15177 * config/h8300/h8300.c: Likewise.
15178 * config/i386/winnt.c: Likewise.
15179 * config/ia64/ia64.c: Likewise.
15180 * config/iq2000/iq2000.c: Likewise.
15181 * config/lm32/lm32.c: Likewise.
15182 * config/m32c/m32c.c: Likewise.
15183 * config/m32r/m32r.c: Likewise.
15184 * config/m68k/m68k.c: Likewise.
15185 * config/mcore/mcore.c: Likewise.
15186 * config/microblaze/microblaze.c: Likewise.
15187 * config/mips/mips.c: Likewise.
15188 * config/mmix/mmix.c: Likewise.
15189 * config/mn10300/mn10300.c: Likewise.
15190 * config/moxie/moxie.c: Likewise.
15191 * config/msp430/msp430.c: Likewise.
15192 * config/nds32/nds32-isr.c: Likewise.
15193 * config/nds32/nds32.c: Likewise.
15194 * config/nios2/nios2.c: Likewise.
15195 * config/nvptx/nvptx.c: Likewise.
15196 * config/pa/pa.c: Likewise.
15197 * config/pdp11/pdp11.c: Likewise.
15198 * config/powerpcspe/powerpcspe.c: Likewise.
15199 * config/riscv/riscv.c: Likewise.
15200 * config/rl78/rl78.c: Likewise.
15201 * config/rx/rx.c: Likewise.
15202 * config/s390/s390.c: Likewise.
15203 * config/sh/sh.c: Likewise.
15204 * config/sol2.c: Likewise.
15205 * config/sparc/sparc.c: Likewise.
15206 * config/spu/spu.c: Likewise.
15207 * config/stormy16/stormy16.c: Likewise.
15208 * config/tilegx/tilegx.c: Likewise.
15209 * config/tilepro/tilepro.c: Likewise.
15210 * config/v850/v850.c: Likewise.
15211 * config/vax/vax.c: Likewise.
15212 * config/visium/visium.c: Likewise.
15213 * config/xtensa/xtensa.c: Likewise.
15215 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15218 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15219 constraints since the -mupper-regs-* switches have been
15221 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15222 into a vector from a double word element that was extracted from
15223 another vector, and eliminate extra XXPERMDI instructions.
15224 (vsx_concat_<mode>_2): Likewise.
15225 (vsx_concat_<mode>_3): Likewise.
15226 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15227 concat to allow optimizing inserts from previous extracts.
15229 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15231 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15232 memory reference to a SSP offset in TLS address space.
15233 (ix86_print_operand) <case '@'>: Remove.
15234 (ix86_print_operand_punct_valid_p): Remove '@' code.
15235 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15236 UNSPEC_SP_TLS_TEST.
15237 (stack_tls_protect_set_<mode>): Remove.
15238 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15239 (stack_tls_protect_test_<mode>): Remove.
15240 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15242 2017-08-07 Olivier Hainque <hainque@adacore.com>
15245 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15247 2017-08-07 Douglas Rupp <rupp@adacore.com>
15249 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15250 variable was referenced as multidir in command.
15252 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15255 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15258 2017-08-07 Martin Liska <mliska@suse.cz>
15260 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15261 * config/rl78/rl78.c: Add include of attribs.h.
15262 * config/sh/sh.c: Likewise.
15263 * config/v850/v850.c: Likewise.
15265 2017-08-07 Tom de Vries <tom@codesourcery.com>
15267 PR middle-end/78266
15268 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15270 2017-08-07 Martin Liska <mliska@suse.cz>
15272 * config/mips/mips.c: Include attribs.h.
15274 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15277 * doc/invoke.texi: Document change in behvaior for -Ofast for
15280 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15282 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15284 (aarch64_pop_regs): Likewise.
15285 (aarch64_gen_load_pair): Likewise.
15286 (aarch64_save_callee_saves): Likewise.
15287 (aarch64_restore_callee_saves): Likewise.
15289 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15291 * config/i386/i386.c: Revert the last change.
15293 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15296 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15298 (ix86_finalize_stack_frame_flags): This. Also clear
15299 frame_pointer_needed if -fno-omit-frame-pointer is used without
15301 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15302 with ix86_finalize_stack_frame_flags.
15303 (ix86_expand_epilogue): Likewise.
15304 (ix86_expand_split_stack_prologue): Likewise.
15306 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15309 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15310 to P_AES for Westmere.
15312 2017-08-07 Jonathan Yong <10walls@gmail.com>
15314 * config/i386/mingw.opt (fset-stack-executable): Removed.
15315 * config/i386/cygming.opt (fset-stack-executable): Moved
15317 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15319 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15321 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15323 2017-08-07 Marek Polacek <polacek@redhat.com>
15325 PR middle-end/81737
15326 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15328 2017-08-07 Martin Liska <mliska@suse.cz>
15330 * attribs.h (canonicalize_attr_name): New function.
15331 (cmp_attribs): Move from c-format.c and adjusted.
15332 (is_attribute_p): Moved from tree.h.
15333 * tree-inline.c: Add new includes.
15334 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15335 (private_is_attribute_p): Remove.
15336 (private_lookup_attribute): Likewise.
15337 (private_lookup_attribute_by_prefix): Simplify.
15338 (remove_attribute): Use is_attribute_p.
15339 * tree.h: Remove removed declarations.
15341 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15343 PR middle-end/81698
15344 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15345 instead of computing it in the function. Formatting fix.
15346 (expand_case): Don't rely on default_edge being the first edge,
15347 clear it if removing it, pass default_edge to
15348 emit_case_dispatch_table.
15349 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15352 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15354 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15355 insn in the function, emit NOP after the insn.
15357 2017-08-06 Tom de Vries <tom@codesourcery.com>
15359 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15362 2017-08-06 Tom de Vries <tom@codesourcery.com>
15364 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15367 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15369 PR tree-optimization/57371
15370 * match.pd: New pattern.
15372 2017-08-04 Marek Polacek <polacek@redhat.com>
15374 PR middle-end/81695
15375 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15376 perform the computation in offset_int.
15378 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15380 PR tree-optimization/81136
15381 * tree-vectorizer.h: Include tree-hash-traits.h.
15382 (vec_base_alignments): New typedef.
15383 (vec_info): Add a base_alignments field.
15384 (vect_record_base_alignments): Declare.
15385 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15387 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15388 (create_data_ref): Add an is_conditional_in_stmt argument.
15389 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15390 the is_conditional_in_stmt field.
15391 (data_ref_loc): Add an is_conditional_in_stmt field.
15392 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15393 (find_data_references_in_stmt): Update call to create_data_ref.
15394 (graphite_find_data_references_in_stmt): Likewise.
15395 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15396 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15397 (vect_record_base_alignment): New function.
15398 (vect_record_base_alignments): Likewise.
15399 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15400 for nested statements even if we fail to compute a misalignment.
15401 Use pooled base alignments for unconditional references.
15402 (vect_find_same_alignment_drs): Compare base addresses instead
15404 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15405 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15407 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15409 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15410 Add an explicit name for the enum. Use auto_vec for slp_instances
15411 and grouped_stores.
15412 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15414 (_bb_vec_info): Add a constructor and destructor.
15415 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15416 (destroy_loop_vec_info): Delete.
15417 (vect_destroy_datarefs): Likewise.
15418 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15419 (vec_info::vec_info): New function.
15420 (vec_info::~vec_info): Likewise.
15421 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15422 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15423 destroy_loop_vec_info.
15424 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15425 (_loop_vec_info::_loop_vec_info): ...this.
15426 (destroy_loop_vec_info): Replace with...
15427 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15428 the stmt_vec_infos. Leave handling of vec_info information to its
15429 destructor. Remove explicit vector releases.
15430 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15431 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15432 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15433 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15434 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15435 (destroy_bb_vec_info): Replace with...
15436 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15437 information to its destructor.
15438 (vect_slp_analyze_bb_1): Use new and delete instead of
15439 new_bb_vec_info and destroy_bb_vec_info.
15440 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15443 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15445 * tree-data-ref.h (subscript): Add access_fn field.
15446 (data_dependence_relation): Add could_be_independent_p.
15447 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15448 (same_access_functions): Move to tree-data-ref.c.
15449 * tree-data-ref.c (ref_contains_union_access_p): New function.
15450 (access_fn_component_p): Likewise.
15451 (access_fn_components_comparable_p): Likewise.
15452 (dr_analyze_indices): Add a reference to access_fn_component_p.
15453 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15455 (constant_access_functions): Likewise.
15456 (add_other_self_distances): Likewise.
15457 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15458 (initialize_data_dependence_relation): Use XCNEW and remove
15459 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15460 of access functions that have the same type. Allow the
15461 subsequence to end with different bases in some circumstances.
15462 Record the chosen access functions in SUB_ACCESS_FN.
15463 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15464 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15465 (subscript_dependence_tester_1): Likewise dra and drb.
15466 (build_classic_dist_vector): Update calls accordingly.
15467 (subscript_dependence_tester): Likewise.
15468 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15469 DDR_COULD_BE_INDEPENDENT_P.
15470 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15471 comp_alias_ddrs instead of may_alias_ddrs.
15472 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15474 (vect_analyze_data_ref_dependence): Use it if
15475 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15476 distance vectors if that fails.
15477 (dependence_distance_ge_vf): New function.
15478 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15479 LOOP_VINFO_MAY_ALIAS_DDRS.
15481 2017-08-04 Richard Biener <rguenther@suse.de>
15483 PR middle-end/81705
15484 * fold-const.c (fold_binary_loc): Properly restrict
15485 minus_var0 && minus_var1 case when associating undefined overflow
15488 2017-08-04 Tom de Vries <tom@codesourcery.com>
15490 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15492 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15495 Don't start diagnostic messages with a capital letter.
15496 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15498 (rs6000_invalid_builtin): Likewise.
15499 (rs6000_trampoline_init): Likewise.
15501 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15504 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15505 after setting changeable df flags.
15507 2017-08-03 Richard Biener <rguenther@suse.de>
15509 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15510 up if the use is in USE - X.
15512 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15514 * toplev.c (dumpfile.h): New include.
15515 (internal_error_reentered): New static function. Use it...
15516 (internal_error_function): ...here to handle reentered internal_error.
15518 2017-08-03 Richard Biener <rguenther@suse.de>
15520 PR middle-end/81148
15521 * fold-const.c (split_tree): Add minus_var and minus_con
15522 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15523 here but always use minus_*.
15524 (associate_trees): Assert we never associate with MINUS_EXPR
15525 and NULL first operand. Do not recurse for PLUS_EXPR operands
15526 when associating as MINUS_EXPR either.
15527 (fold_binary_loc): Track minus_var and minus_con.
15529 2017-08-03 Tom de Vries <tom@codesourcery.com>
15532 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15533 ACCEL_COMPILER, apply finish_options on
15534 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15536 2017-08-03 Tom de Vries <tom@codesourcery.com>
15539 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15540 function_entry_patch_area_size > 0.
15542 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15545 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15546 instead of 10??LU, perform unit multiplication in wide_int,
15547 don't change alloc_object_size_limit if the limit is larger
15550 PR tree-optimization/81655
15551 PR tree-optimization/81588
15552 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15553 the case when ranges[i].low and high are 1 for unsigned type with
15556 PR middle-end/81052
15557 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15558 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15560 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15562 * tree-vrp.h: Add include guard.
15564 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15567 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15568 (ud2): New insn pattern.
15569 * config/i386/i386.c (ix86_expand_epilogue):
15570 For naked functions, generate ud2 instead of trap insn.
15572 2017-08-02 Marek Polacek <polacek@redhat.com>
15575 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15577 2017-08-02 Tom de Vries <tom@codesourcery.com>
15578 Cesar Philippidis <cesar@codesourcery.com>
15580 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15581 Add missing edge probabilities.
15583 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15585 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15586 Correct endianness.
15588 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15590 PR middle-end/79499
15591 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15592 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15593 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15594 prologue_seq sequences - if any.
15596 2017-08-02 Richard Biener <rguenther@suse.de>
15598 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15599 via vectors if supported, integer extracts via punning if supported
15600 or otherwise vector extracts.
15602 2017-08-02 Richard Biener <rguenther@suse.de>
15604 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15606 (bitmap_insert_into_set): ... this.
15608 2017-08-02 Richard Biener <rguenther@suse.de>
15610 PR tree-optimization/81633
15612 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15614 PR tree-optimization/71752
15615 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15617 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15619 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15620 (machine_function::call_ms2sysv_pad_out): Remove field.
15621 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15622 (ix86_compute_frame_layout): Likewise.
15624 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15627 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15628 attribute with interrupt attribute.
15630 2017-08-01 Andrew Pinski <apinski@cavium.com>
15632 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15633 BIT_INSERT_EXPR's operand 1
15634 to see if the types precision matches.
15636 2017-08-01 Martin Liska <mliska@suse.cz>
15638 PR middle-end/70140
15639 * builtins.c (expand_builtin_memcpy_args): Remove.
15640 (expand_builtin_memcpy): Call newly added function
15641 expand_builtin_memory_copy_args.
15642 (expand_builtin_memcpy_with_bounds): Likewise.
15643 (expand_builtin_mempcpy): Remove last argument.
15644 (expand_builtin_mempcpy_with_bounds): Likewise.
15645 (expand_builtin_memory_copy_args): New function created from
15646 expand_builtin_mempcpy_args with small modifications.
15647 (expand_builtin_mempcpy_args): Remove.
15648 (expand_builtin_stpcpy): Remove unused argument.
15649 (expand_builtin): Likewise.
15650 (expand_builtin_with_bounds): Likewise.
15652 2017-08-01 Martin Liska <mliska@suse.cz>
15655 Make mempcpy more optimal (PR middle-end/70140).
15657 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15660 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15661 __builtin_vec_cmpne verify both arguments are compatible vectors
15662 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15663 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15664 move computation of aligned to after checking the argument types.
15668 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15669 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15672 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15673 Doug Rupp <rupp@adacore.com>
15674 Olivier Hainque <hainque@adacore.com>
15676 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15677 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15679 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15680 for TARGET_OS_CPP_BUILTIN.
15681 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15682 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15684 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15685 passing required abi options to the assembler for EABI configurations.
15686 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15687 of .text.hot and .text.unlikely sections for kernel modules when
15688 using ARM style exceptions.
15689 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15690 options. Add EXTRA_CC1_SPEC.
15691 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15693 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15695 (ARM_TARGET2_DWARF_FORMAT): Define.
15696 * config/arm/t-vxworks: Adjust multilib control to removal of the
15697 Diab command line options.
15699 2017-08-01 Martin Liska <mliska@suse.cz>
15701 PR gcov-profile/81561
15702 * gcov.c (unblock): Make unblocking safe as we need to preserve
15703 index correspondence of blocks and block_lists.
15705 2017-08-01 Richard Biener <rguenther@suse.de>
15707 PR tree-optimization/81181
15708 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15709 (compute_antic): ... end of iteration here.
15711 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15713 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15714 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15715 (ftree-slp-vectorize): Likewise.
15716 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15717 can no longer be set independent of flag_tree_loop_vectorize.
15718 * omp-general.c (emp_max_vf): Likewise.
15719 * opts.c (enable_fdo_optimizations): Remove references to
15720 flag_tree_vectorize, these are now implicit.
15721 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15722 and leave it for the options machinery.
15724 2017-08-01 Martin Liska <mliska@suse.cz>
15726 PR middle-end/70140
15727 * builtins.c (expand_builtin_memcpy_args): Remove.
15728 (expand_builtin_memcpy): Call newly added function
15729 expand_builtin_memory_copy_args.
15730 (expand_builtin_memcpy_with_bounds): Likewise.
15731 (expand_builtin_mempcpy): Remove last argument.
15732 (expand_builtin_mempcpy_with_bounds): Likewise.
15733 (expand_builtin_memory_copy_args): New function created from
15734 expand_builtin_mempcpy_args with small modifications.
15735 (expand_builtin_mempcpy_args): Remove.
15736 (expand_builtin_stpcpy): Remove unused argument.
15737 (expand_builtin): Likewise.
15738 (expand_builtin_with_bounds): Likewise.
15740 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15743 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15744 print "ds:" only for immediates in generic address space.
15746 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15749 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15750 (ix86_function_ok_for_sibcall): Return false for naked functions.
15752 2017-08-01 Richard Biener <rguenther@suse.de>
15754 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15755 (compute_antic): Seed worklist with exit block predecessors.
15756 * cfganal.c (dfs_find_deadend): For a cycle return the source
15757 of the edge closing it.
15759 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15761 * config/aarch64/aarch64.c
15762 (aarch64_can_const_movi_rtx_p): Move 0 check.
15764 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15766 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15767 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15769 * match.pd: Ditto in address comparison pattern.
15771 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15773 PR tree-optimization/81627
15774 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15775 closed ssa form for store-store chain.
15777 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15779 PR tree-optimization/81620
15780 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15781 for store-store chain.
15783 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15785 PR tree-optimization/81588
15786 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15787 ranges[i].in_p, invert comparison code ccode. For >/>=,
15788 swap rhs1 and rhs2 and comparison code unconditionally,
15789 for </<= don't do that. Don't swap rhs1/rhs2 again if
15790 ranges[i].in_p, instead invert comparison code ccode if
15791 opcode or oe->rank is BIT_IOR_EXPR.
15794 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15795 a direct optab to conversion optab.
15796 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15797 with GET_MODE_INNER as last argument instead of optab_handler.
15798 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15799 vector extraction if possible and optab is available.
15800 * expr.c (store_constructor): Use convert_optab_handler instead
15801 of optab_handler. Use vector initialization from smaller
15802 vectors if possible and optab is available.
15803 * tree-vect-stmts.c (vectorizable_load): Likewise.
15804 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15805 now have two modes.
15806 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15807 of vec_init from half-sized vectors with the same element mode.
15808 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15809 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15810 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15811 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15812 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15813 after mode in gen_vec_extract* calls.
15814 (vec_extract<mode>): Renamed to ...
15815 (vec_extract<mode><ssescalarmodelower>): ... this.
15816 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15817 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15818 element mode after mode in gen_vec_init* calls.
15819 (VEC_INIT_HALF_MODE): New mode iterator.
15820 (vec_init<mode>): Renamed to ...
15821 (vec_init<mode><ssescalarmodelower>): ... this.
15822 (vec_init<mode><ssehalfvecmodelower>): New expander.
15823 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15824 (vec_extractv2sfsf): ... this.
15825 (vec_initv2sf): Renamed to ...
15826 (vec_initv2sfsf): ... this.
15827 (vec_extractv2si): Renamed to ...
15828 (vec_extractv2sisi): ... this.
15829 (vec_initv2si): Renamed to ...
15830 (vec_initv2sisi): ... this.
15831 (vec_extractv4hi): Renamed to ...
15832 (vec_extractv4hihi): ... this.
15833 (vec_initv4hi): Renamed to ...
15834 (vec_initv4hihi): ... this.
15835 (vec_extractv8qi): Renamed to ...
15836 (vec_extractv8qiqi): ... this.
15837 (vec_initv8qi): Renamed to ...
15838 (vec_initv8qiqi): ... this.
15839 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15840 (vec_init<mode>): Renamed to ...
15841 (vec_init<mode><VEC_base_l>): ... this.
15842 (vec_extract<mode>): Renamed to ...
15843 (vec_extract<mode><VEC_base_l>): ... this.
15844 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15845 (vec_initv2sfsf): ... this.
15846 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15847 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15848 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15849 element mode after mode in gen_vec_init* calls.
15850 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15851 (vec_init<mode><Vel>): ... this.
15852 (vec_extract<mode>): Renamed to ...
15853 (vec_extract<mode><Vel>): ... this.
15854 * config/aarch64/iterators.md (Vel): New mode attribute.
15855 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15856 Add element mode after mode in gen_vec_extract* calls.
15857 * config/s390/vector.md (non_vec_l): New mode attribute.
15858 (vec_extract<mode>): Renamed to ...
15859 (vec_extract<mode><non_vec_l>): ... this.
15860 (vec_init<mode>): Renamed to ...
15861 (vec_init<mode><non_vec_l>): ... this.
15862 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15863 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15865 * config/arm/iterators.md (V_elem_l): New mode attribute.
15866 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15867 (vec_extract<mode><V_elem_l>): ... this.
15868 (vec_extractv2di): Renamed to ...
15869 (vec_extractv2didi): ... this.
15870 (vec_init<mode>): Renamed to ...
15871 (vec_init<mode><V_elem_l>): ... this.
15872 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15873 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15874 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15875 Add element mode after gen_vec_extract* calls.
15876 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15877 (vec_init<mode><unitmode>): ... this.
15878 (vec_extract<mode>): Renamed to ...
15879 (vec_extract<mode><unitmode>): ... this.
15880 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15881 (vec_init<mode><unitmode>): ... this.
15882 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15883 (vec_initv2sfsf): ... this.
15884 (vec_extractv2sf): Renamed to ...
15885 (vec_extractv2sfsf): ... this.
15886 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15887 Add element mode after gen_vec_extract* calls.
15888 * config/mips/mips.md (unitmode): New mode iterator.
15889 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15890 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15891 * config/spu/spu.md (inner_l): New mode attribute.
15892 (vec_init<mode>): Renamed to ...
15893 (vec_init<mode><inner_l>): ... this.
15894 (vec_extract<mode>): Renamed to ...
15895 (vec_extract<mode><inner_l>): ... this.
15896 * config/sparc/sparc.md (veltmode): New mode iterator.
15897 (vec_init<VMALL:mode>): Renamed to ...
15898 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15899 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15900 (vec_initv2sisi): ... this.
15901 (vec_initv2sf): Renamed to ...
15902 (vec_initv2sfsf): ... this.
15903 (vec_extractv2sf): Renamed to ...
15904 (vec_extractv2sfsf): ... this.
15905 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15906 (vec_init<mode>): Renamed to ...
15907 (vec_init<mode><VEC_base_l>): ... this.
15908 (vec_extract<mode>): Renamed to ...
15909 (vec_extract<mode><VEC_base_l>): ... this.
15910 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15911 (vec_initv2sfsf): ... this.
15912 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15913 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15914 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15915 gen_vec_init* calls.
15917 2017-08-01 Richard Biener <rguenther@suse.de>
15919 PR tree-optimization/81297
15920 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15921 TREE_OVERFLOW from INTEGER_CSTs.
15923 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15925 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15927 2017-07-31 Carl Love <cel@us.ibm.com>
15929 * config/rs6000/rs6000-c: Add support for built-in functions
15930 vector signed char vec_xl_be (signed long long, signed char *);
15931 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15932 vector signed int vec_xl_be (signed long long, signed int *);
15933 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15934 vector signed long long vec_xl_be (signed long long, signed long long *);
15935 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15936 vector signed short vec_xl_be (signed long long, signed short *);
15937 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15938 vector double vec_xl_be (signed long long, double *);
15939 vector float vec_xl_be (signed long long, float *);
15940 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15941 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15942 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15944 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15945 (altivec_expand_builtin): Add switch statement to call
15946 altivec_expand_xl_be for each builtin.
15947 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15948 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15949 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15950 __builtin_vsx_le_be_v16qi.
15951 * doc/extend.texi: Update the built-in documentation file for the
15952 new built-in functions.
15954 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15957 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15959 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15961 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15963 * config.gcc: Add z14.
15964 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15965 CPU model numbers for z13s and z14.
15966 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15968 * config/s390/s390-opts.h (enum processor_type): Rename
15969 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15970 * config/s390/s390.c (processor_table): Add field for CPU name to
15971 be passed to Binutils.
15972 (s390_asm_output_machine_for_arch): Use the new field in
15973 processor_table for Binutils.
15974 (s390_expand_builtin): Replace arch12 with z14.
15975 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15976 (s390_get_sched_attrmask): Likewise.
15977 (s390_get_unit_mask): Likewise.
15978 * config/s390/s390.opt: Add z14 to processor_type enum.
15980 2017-07-31 Martin Jambor <mjambor@suse.cz>
15983 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15984 regardless of optimization level.
15986 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15987 Martin Liska <mliska@suse.cz>
15989 * predict.def: Remove old comment and adjust probability.
15990 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15991 PREDICT statements.
15993 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15996 * config/i386/i386.c (ix86_function_naked): New function.
15997 (ix86_can_use_return_insn_p): Return false for naked functions.
15998 (ix86_expand_prologue): Skip prologue for naked functions.
15999 (ix86_expand_epilogue): Skip epilogue for naked functions
16000 and emit trap instruction.
16001 (ix86_warn_func_return): New function.
16002 (ix86_attribute_table): Add "naked" attribute specification.
16003 (TARGET_WARN_FUNC_RETURN): Define.
16004 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16006 2017-07-31 Martin Liska <mliska@suse.cz>
16008 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16009 (dump_gimple_bb_header): Always dump BB info.
16010 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16012 2017-07-31 Martin Liska <mliska@suse.cz>
16015 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16016 also with current_function_decl non-null equality.
16018 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16021 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16022 change type to the element type, instead add eltype variable and
16023 use it where we are interested in the element type.
16025 PR tree-optimization/81603
16026 * ipa-polymorphic-call.c
16027 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16028 offset arithmetic in offset_int, bail out if the resulting bit offset
16029 doesn't fit into shwi.
16031 2017-07-31 Martin Liska <mliska@suse.cz>
16033 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16034 (gimplify_save_expr): Fix comment.
16036 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16039 * config/i386/i386.c (ix86_function_arg): Update arguments for
16041 (ix86_compute_frame_layout): Set the initial stack offset to
16042 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16043 INCOMING_FRAME_SP_OFFSET.
16044 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16045 stack before exception handler returns.
16046 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16047 the 'ERROR_CODE' for exception handler.
16049 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16051 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16052 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16053 (ASM_OUTPUT_REG_POP): Ditto.
16054 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16055 instead of asm_fprintf to output pure string.
16057 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16059 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16060 to imported_module_or_decl hook.
16061 (debug_nothing_tree_tree_tree_bool): Remove.
16062 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16063 * debug.c (do_nothing_debug_hooks): Use
16064 debug_nothing_tree_tree_tree_bool_bool instead of
16065 debug_nothing_tree_tree_tree_bool.
16066 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16067 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16068 * sdbout.c (sdb_debug_hooks): Likewise.
16069 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16070 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16072 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16073 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16074 attribute, don't add anything.
16076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16078 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16079 (fold_build2_stat_loc): Likewise.
16080 (fold_build3_stat_loc): Likewise.
16081 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16082 (fold_build1_loc): Remove macro.
16083 (fold_build2_loc): Likewise.
16084 (fold_build3_loc): Likewise.
16086 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16088 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16089 (gimple_build_debug_bind_source_stat): Likewise.
16090 * gimple.h (gimple_build_debug_bind): Remove macro.
16091 (gimple_build_debug_bind_source): Likewise.
16093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16095 * bitmap.c (bitmap_alloc): Adjust.
16096 (bitmap_gc_alloc): Likewise.
16097 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16099 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16101 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16102 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16103 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16104 (bitmap_gc_alloc_stat): Likewise.
16105 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16107 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16109 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16110 * rtl.h (shallow_copy_rtx): Remove macro.
16112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16114 * emit-rtl.c (gen_raw_REG): Adjust.
16115 * gengenrtl.c (gendef): Likewise.
16116 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16117 * rtl.h (rtx_alloc): Remove macro.
16119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16121 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16122 (build_tree_list_stat): Likewise.
16123 * tree.h (build_tree_list): Remove macro.
16124 (build_tree_list_vec): Likewise.
16126 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16128 * tree.c (make_vector_stat): Remove _stat from name.
16129 (build_vector_stat): Likewise.
16130 * tree.h (make_vector_stat): Remove macro.
16131 (build_vector_stat): Likewise.
16133 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16135 * tree.h (build_var_debug_value): Remove prototype.
16137 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16139 * tree.c (tree_cons_stat): Remove _stat from name.
16140 * tree.h (tree_cons): Remove macro.
16142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16144 * tree.c (build_vl_exp_stat): Remove _stat from name.
16145 * tree.h (build_vl_exp): Remove macro.
16147 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16149 * tree.c (build_decl_stat): Remove _stat from name.
16150 * tree.h (build_decl): Remove macro.
16152 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16154 * gimple.c (gimple_build_with_ops_stat): Adjust.
16155 (gimple_alloc_stat): Remove _stat from name.
16156 * gimple.h (gimple_alloc): Remove macro.
16158 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16160 * tree.c (make_tree_vec_stat): Remove _stat from name.
16161 (grow_tree_vec_stat): Likewise.
16162 * tree.h (make_tree_vec_stat): Adjust prototype.
16163 (grow_tree_vec_stat): Likewise.
16164 (make_tree_vec): Remove macro.
16165 (grow_tree_vec): Likewise.
16167 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16169 * fold-const.c (fold_build1_stat_loc): Adjust.
16170 (fold_build2_stat_loc): Likewise.
16171 (fold_build3_stat_loc): Likewise.
16172 * tree.c (build0_stat): Remove _stat from name.
16173 (build1_stat): Likewise.
16174 (build2_stat): Likewise.
16175 (build3_stat): Likewise.
16176 (build4_stat): Likewise.
16177 (build5_stat): Likewise.
16178 * tree.h (build1_loc): Remove macro, and rename _stat function
16180 (build2_loc): Likewise.
16181 (build3_loc): Likewise.
16182 (build4_loc): Likewise.
16183 (build5_loc): Likewise.
16185 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16187 * tree.c (make_int_cst_stat): Remove _stat from name.
16188 * tree.h (make_int_cst_stat): Adjust prototype.
16189 (make_int_cst): Remove macro.
16191 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16193 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16194 * tree.h (make_tree_binfo_stat): Adjust prototype.
16195 (make_tree_binfo): Remove.
16197 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16199 * tree.c (copy_node_stat): Rename to copy_node.
16200 (build_distinct_type_copy): Adjust.
16201 * tree.h (copy_node_stat): Adjust prototype.
16202 (copy_node): Remove macro.
16204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16206 * tree.c (make_node_stat): rename to make_node.
16207 (build_tree_list_stat): Adjust.
16208 (build0_stat): Likewise.
16209 (build2_stat): Likewise.
16210 (build3_stat): Likewise.
16211 (build4_stat): Likewise.
16212 (build5_stat): Likewise.
16213 (build_decl_stat): Likewise.
16214 * tree.h (make_node_stat): Adjust prototype.
16215 (make_node): remove macro.
16217 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16219 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16220 (PPC_FEATURE2_SCV): Likewise.
16221 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16223 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16225 * config/aarch64/aarch64.c
16226 (aarch64_internal_mov_immediate): Add new special pattern.
16227 * config/aarch64/aarch64.md (*movdi_aarch64):
16228 Add reg/32bit const mov case.
16230 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16231 Richard Sandiford <richard.sandiford@linaro.org>
16233 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16234 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16235 Add integer and movi cases.
16236 (movi-split-hf-df-sf split, fp16): New.
16237 (enabled): Added TARGET_FP_F16INST.
16238 * config/aarch64/iterators.md (GPF_HF): New.
16239 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16241 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16243 * config/aarch64/aarch64.c
16244 (aarch64_simd_container_mode): Add prototype.
16245 (aarch64_expand_mov_immediate): Add HI support.
16246 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16247 (aarch64_can_const_movi_rtx_p): New.
16248 (aarch64_preferred_reload_class):
16249 Remove restrictions of using FP registers for certain SIMD operations.
16250 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16251 (aarch64_valid_floating_const): Add integer move validation.
16252 (aarch64_simd_imm_scalar_p): Remove.
16253 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16254 (aarch64_legitimate_constant_p): Expand list of supported cases.
16255 * config/aarch64/aarch64-protos.h
16256 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16257 (aarch64_reinterpret_float_as_int): New.
16258 (aarch64_simd_imm_scalar_p): Remove.
16259 * config/aarch64/constraints.md (Uvi): New.
16260 (Dd): Split into Ds and new Dd.
16261 * config/aarch64/aarch64.md (*movsi_aarch64):
16263 (*movdi_aarch64): Add SIMD mov case.
16265 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16267 * tree-predcom.c: (struct chain): Handle store-store chain in which
16268 stores for elimination only store loop invariant values.
16269 (execute_pred_commoning_chain): Ditto.
16270 (prepare_initializers_chain_store_elim): Ditto.
16271 (prepare_finalizers): Ditto.
16272 (is_inv_store_elimination_chain): New function.
16273 (initialize_root_vars_store_elim_1): New function.
16275 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16277 * tree-predcom.c: Revise general description of the pass.
16278 (enum chain_type): New enum type for store elimination.
16279 (struct chain): New field supporting store elimination.
16280 (struct component): Ditto.
16281 (dump_chain): Dump store-stores chain.
16282 (release_chain): Release resources.
16283 (split_data_refs_to_components): Compute and create component
16284 contains only stores for elimination.
16285 (get_chain_last_ref_at): New function.
16286 (make_invariant_chain): Initialization.
16287 (make_rooted_chain): Specify chain type in parameter and record it.
16288 (add_looparound_copies): Skip for store-stores chain.
16289 (determine_roots_comp): Compute type of chain and pass it to
16291 (initialize_root_vars_store_elim_2): New function.
16292 (finalize_eliminated_stores): New function.
16293 (remove_stmt): Handle store for elimination.
16294 (execute_pred_commoning_chain): Execute predictive commoning on
16295 store-store chains.
16296 (determine_unroll_factor): Skip unroll for store-stores chain.
16297 (prepare_initializers_chain_store_elim): New function.
16298 (prepare_initializers_chain): Hanlde store-store chain.
16299 (prepare_finalizers_chain, prepare_finalizers): New function.
16300 (tree_predictive_commoning_loop): Return integer value indicating
16301 if loop is unrolled or lcssa form is corrupted.
16302 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16304 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16306 * tree-predcom.c (initialize_root): Delete.
16307 (execute_pred_commoning_chain): Initialize root vars and replace
16308 reference of non-combined chain directly, rather than call above
16311 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16313 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16314 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16316 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16318 * tree-predcom.c (struct chain): New field init_seq.
16319 (release_chain): Release init_seq.
16320 (prepare_initializers_chain): Record intialization stmts in above
16322 (insert_init_seqs): New function.
16323 (tree_predictive_commoning_loop): Call insert_init_seqs.
16325 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16327 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16329 2017-07-28 Richard Biener <rguenther@suse.de>
16331 * match.pd: Remove superfluous :c.
16332 * genmatch.c (simplify::id): Add member.
16333 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16335 (current_id): New global.
16336 (dt_node::parent): Move from ...
16337 (dt_operand::parent): ... here. Add for_id member.
16338 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16339 (decision_tree::find_node): Relax order requirement when
16340 merging DT_TRUE nodes to ones inbetween the current simplify
16341 and the one we try to merge with. Add diagnostic whenever
16342 we need to enforce pattern order by not merging.
16343 (decision_tree::insert): Set current_id.
16344 (decision_tree::print_node): Dump parent node and for_id.
16345 (parser::last_id): Add member.
16346 (parser::push_simplify): Assign unique id.
16347 (parser::parser): Initialize last_id.
16349 2017-07-28 Martin Liska <mliska@suse.cz>
16352 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16353 gimple_build_debug_bind.
16355 2017-07-28 Richard Biener <rguenther@suse.de>
16357 PR tree-optimization/81502
16358 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16360 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16362 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16363 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16364 for BIT_FIELD_REF args.
16365 * fold-const.c (make_bit_field_ref): Likewise.
16366 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16368 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16371 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16372 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16373 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16374 Or it into SANITIZER_UNDEFINED.
16375 * ubsan.c: Include gimple-fold.h and varasm.h.
16376 (ubsan_expand_ptr_ifn): New function.
16377 (instrument_pointer_overflow): New function.
16378 (maybe_instrument_pointer_overflow): New function.
16379 (instrument_object_size): Formatting fix.
16380 (pass_ubsan::execute): Call instrument_pointer_overflow
16381 and maybe_instrument_pointer_overflow.
16382 * internal-fn.c (expand_UBSAN_PTR): New function.
16383 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16384 * sanitizer.def (__ubsan_handle_pointer_overflow,
16385 __ubsan_handle_pointer_overflow_abort): New builtins.
16386 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16387 * internal-fn.def (UBSAN_PTR): New internal function.
16388 * opts.c (sanitizer_opts): Add pointer-overflow.
16389 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16390 * fold-const.c (build_range_check): Compute pointer range check in
16391 integral type if pointer arithmetics would be needed. Formatting
16394 2017-07-28 Martin Liska <mliska@suse.cz>
16397 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16398 parameters that are of a variable-length.
16400 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16402 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16404 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16405 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16406 (CRT_CALL_STATIC_FUNCTION): Likewise.
16407 (ASM_DEFAULT_SPEC): New define.
16408 (ASM_SPEC32): Likewise.
16409 (ASM_SPEC64): Likewise.
16410 (ASM_SPEC_COMMON): Likewise.
16411 (ASM_SPEC): Likewise.
16412 (INVALID_64BIT): Likewise.
16413 (LINK_OS_DEFAULT_SPEC): Likewise.
16414 (LINK_OS_SPEC32): Likewise.
16415 (LINK_OS_SPEC64): Likewise.
16416 (POWERPC_LINUX): Likewise.
16417 (PTRDIFF_TYPE): Likewise.
16418 (RESTORE_FP_PREFIX): Likewise.
16419 (RESTORE_FP_SUFFIX): Likewise.
16420 (SAVE_FP_PREFIX): Likewise.
16421 (SAVE_FP_SUFFIX): Likewise.
16422 (SIZE_TYPE): Likewise.
16423 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16424 (TARGET_64BIT): Likewise.
16425 (TARGET_64BIT): Likewise.
16426 (TARGET_AIX): Likewise.
16427 (WCHAR_TYPE_SIZE): Likewise.
16428 (WCHAR_TYPE): Undefine.
16429 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16430 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16431 (CPP_OS_RTEMS_SPEC): Delete.
16432 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16433 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16435 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16437 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16439 PR tree-optimization/81578
16440 * tree-parloops.c (build_new_reduction): Bail out if
16441 reduction_code isn't one of the standard OpenMP reductions.
16442 Move the details printing after that decision.
16444 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16446 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16447 related to reload_in_progress.
16448 (splat_input_operand): Likewise.
16449 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16451 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16453 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16454 (TARGET_INSTANTIATE_DECLS): Likewise.
16455 (legitimate_indexed_address_p): Delete reload_in_progress code.
16456 (rs6000_debug_legitimate_address_p): Likewise.
16457 (rs6000_eliminate_indexed_memrefs): Likewise.
16458 (rs6000_emit_le_vsx_store): Likewise.
16459 (rs6000_emit_move_si_sf_subreg): Likewise.
16460 (rs6000_emit_move): Likewise.
16461 (register_to_reg_type): Likewise.
16462 (rs6000_pre_atomic_barrier): Likewise.
16463 (rs6000_machopic_legitimize_pic_address): Likewise.
16464 (rs6000_allocate_stack_temp): Likewise.
16465 (rs6000_address_for_fpconvert): Likewise.
16466 (rs6000_address_for_altivec): Likewise.
16467 (rs6000_secondary_memory_needed_rtx): Delete function.
16468 (rs6000_check_sdmode): Likewise.
16469 (rs6000_alloc_sdmode_stack_slot): Likewise.
16470 (rs6000_instantiate_decls): Likewise.
16471 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16472 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16473 Delete reload_in_progress.
16474 (*vec_reload_and_plus_<mptrsize>): Likewise.
16475 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16476 (vsx_div_v2di): Likewise.
16477 (vsx_udiv_v2di): Likewise.
16479 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16481 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16482 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16483 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16484 (rs6000_debug_reg_global): Delete print of LRA status.
16485 (rs6000_option_override_internal): Delete dead LRA related code.
16486 (rs6000_lra_p): Delete function.
16487 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16489 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16491 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16492 * config/riscv/rtems.h: New file.
16494 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16495 Sudakshina Das <sudi.das@arm.com>
16497 * config/aarch64/aarch64.md
16498 (define_split for and<mode>3nr_compare): Move
16499 non aarch64_logical_operand to a register.
16500 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16501 register immediate operand to a register.
16502 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16504 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16506 PR middle-end/81564
16507 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16509 2017-07-27 Richard Biener <rguenther@suse.de>
16511 PR tree-optimization/81573
16512 PR tree-optimization/81494
16513 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16514 multi defuse cycle case.
16516 2017-07-27 Richard Biener <rguenther@suse.de>
16518 PR tree-optimization/81571
16519 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16522 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16524 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16525 earlier and only if MASK_FPU is set. Adjust formatting.
16527 2017-07-27 Martin Liska <mliska@suse.cz>
16529 * opt-functions.awk: Add validation of value of Init.
16530 * optc-gen.awk: Pass new argument.
16532 2017-07-27 Martin Liska <mliska@suse.cz>
16534 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16535 Fix wrong condition.
16537 2017-07-27 Martin Liska <mliska@suse.cz>
16539 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16540 BBs and edges seen by autoFDO.
16542 2017-07-27 Richard Biener <rguenther@suse.de>
16544 PR tree-optimization/81502
16545 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16546 with incompatible but same sized type.
16547 (execute_update_addresses_taken): Likewise.
16549 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16551 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16552 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16554 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16557 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16558 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16559 Change s_operand to memory_operand.
16561 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16563 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16564 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16565 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16566 Emit instructions rather than returning an expression. Handle TFmode
16567 and KFmode by casting to TImode.
16568 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16569 (rs6000_emit_le_vsx_store): Likewise.
16570 * config/rs6000/vsx.md (VSX_TI): New iterator.
16571 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16572 (*vsx_le_undo_permute_<mode>): Likewise.
16573 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16574 emit the split sequence.
16575 (*vsx_le_perm_store_<mode>): Likewise.
16577 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16579 PR tree-optimization/81555
16580 PR tree-optimization/81556
16581 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16582 if true, force CHANGED for the recursive invocation.
16583 (reassociate_bb): Remember original length of ops array, pass
16584 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16586 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16587 attributes for noipa attribute. For naked attribute use
16588 lookup_attribute first before lookup_attribute_spec.
16589 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16591 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16592 for functions with noipa attribute.
16593 (cgraph_externally_visible_p): Return true for functions with noipa
16595 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16596 for functions with noipa attribute.
16597 * doc/extend.texi: Document noipa function attribute.
16598 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16599 also for functions with noipa attribute.
16600 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16602 2017-07-26 Andrew Pinski <apinski@cavium.com>
16604 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16605 vec_unalign_load_cost and vec_unalign_store_cost.
16607 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16609 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16610 -mvsx-small-integer option.
16611 (ISA_3_0_MASKS_IEEE): Likewise.
16612 (OTHER_VSX_VECTOR_MASKS): Likewise.
16613 (POWERPC_MASKS): Likewise.
16614 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16615 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16616 code, only testing for DImode being allowed in non-VSX floating
16618 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16619 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16621 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16622 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16623 TARGET_P8_VECTOR test.
16624 (rs6000_secondary_reload_simple_move): Likewise.
16625 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16626 since TARGET_P9_VECTOR was already tested.
16627 (rs6000_opt_masks): Remove -mvsx-small-integer.
16628 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16629 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16631 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16632 test for TARGET_VEXTRACTUB was used, and that uses
16634 (p9 extract splitter): Likewise.
16635 (vsx_extract_<mode>_di_p9): Likewise.
16636 (vsx_extract_<mode>_store_p9): Likewise.
16637 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16638 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16639 the elimination of TARGET_VSX_SMALL_INTEGER.
16640 (vsx_extract_<mode>_p8): Likewise.
16641 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16642 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16643 (vsx_set_<mode>_p9): Likewise.
16644 (vsx_set_v4sf_p9): Likewise.
16645 (vsx_set_v4sf_p9_zero): Likewise.
16646 (vsx_insert_extract_v4sf_p9): Likewise.
16647 (vsx_insert_extract_v4sf_p9_2): Likewise.
16648 * config/rs6000/rs6000.md (sign extend splitter): Change
16649 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16650 (floatsi<mode>2_lfiwax_mem): Likewise.
16651 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16652 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16653 since a test for TARGET_P9_VECTOR was used.
16654 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16655 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16656 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16657 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16658 TARGET_P8_VECTOR test.
16659 (fix_trunc<mode>si2_stfiwx): Likewise.
16660 (fix_trunc<mode>si2_internal): Likewise.
16661 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16662 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16664 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16665 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16666 TARGET_P8_VECTOR test.
16667 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16668 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16669 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16671 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16672 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16673 since a test for TARGET_P9_VECTOR was used.
16674 (splitter for loading small constants): Likewise.
16676 2017-07-26 Andrew Pinski <apinski@cavium.com>
16678 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16681 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16684 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16685 (fp_valid_at): Likewise.
16687 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16689 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16690 (qdf24xx_addrcost_table): Likewise.
16691 (cortexa57_tunings): Update to use generic_branch_cost.
16692 (cortexa72_tunings): Likewise.
16693 (cortexa73_tunings): Likewise.
16694 (qdf24xx_tunings): Likewise.
16696 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16698 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16699 (thunderx2t99_branch_cost): Likewise.
16700 (cortexa35_tunings): Update to use generic_branch_cost.
16701 (cortexa53_tunings): Likewise.
16702 (cortexa57_tunings): Likewise.
16703 (cortexa72_tunings): Likewise.
16704 (cortexa73_tunings): Likewise.
16705 (thunderx2t99_tunings): Likewise.
16707 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16709 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16710 (sparc_option_override): Honour MASK_FSMULD.
16711 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16712 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16713 * config/sparc/sparc.opt (mfsmuld): New option.
16714 * doc/invoke.texi (mfsmuld): Document option.
16716 2017-07-26 Marek Polacek <polacek@redhat.com>
16718 PR middle-end/70992
16719 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16721 2017-07-26 Richard Biener <rguenther@suse.de>
16723 * gimple-match-head.c (do_valueize): Return OP if valueize
16725 (get_def): New helper to get at the def stmt of a SSA name
16726 if valueize allows.
16727 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16728 do_valueize to get at the def stmt.
16729 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16731 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16733 PR middle-end/46932
16734 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16736 2017-07-26 Martin Liska <mliska@suse.cz>
16739 * function.c (expand_function_start): Make expansion of
16740 nonlocal_goto_save_area after parm_birth_insn.
16742 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16744 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16745 from all CPU target flags enable members.
16747 2017-07-26 Richard Biener <rguenther@suse.de>
16749 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16750 (decision_tree::gen): Make 'type' const.
16751 (write_predicate): Likewise.
16753 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16755 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16756 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16757 (rs6000_option_override_internal): Likewise.
16758 (rs6000_expand_vector_set): Likewise.
16759 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16760 (TARGET_UPPER_REGS_SF): Likewise.
16761 (TARGET_UPPER_REGS_DI): Likewise.
16762 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16763 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16764 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16765 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16766 (Splitters for DI constants in Altivec registers): Likewise.
16767 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16768 (vsx_set_v4sf_p9): Likewise.
16769 (vsx_set_v4sf_p9_zero): Likewise.
16770 (vsx_insert_extract_v4sf_p9): Likewise.
16771 (vsx_insert_extract_v4sf_p9_2): Likewise.
16773 2017-07-25 Carl Love <cel@us.ibm.com>
16775 * doc/extend.texi: Update the built-in documentation file for the
16776 existing built-in functions
16777 vector signed char vec_cnttz (vector signed char);
16778 vector unsigned char vec_cnttz (vector unsigned char);
16779 vector signed short vec_cnttz (vector signed short);
16780 vector unsigned short vec_cnttz (vector unsigned short);
16781 vector signed int vec_cnttz (vector signed int);
16782 vector unsigned int vec_cnttz (vector unsigned int);
16783 vector signed long long vec_cnttz (vector signed long long);
16784 vector unsigned long long vec_cnttz (vector unsigned long long);
16786 2017-07-25 Andrew Pinski <apinski@cavium.com>
16788 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16789 accesses where the use is for the first operand of a BIT_INSERT.
16791 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16794 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16795 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16797 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16799 * config/i386/gstabs.h: Delete.
16800 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16802 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16804 * config/i386/i386.c (ix86_decompose_address): Do not check for
16805 register RTX when looking at index_reg or base_reg.
16806 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16808 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16810 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16811 to update EH info here.
16813 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16815 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16817 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16819 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16821 2017-07-25 Torsten Duwe <duwe@suse.de>
16823 * common.opt: Introduce -fpatchable-function-entry
16824 command line option, and its variables function_entry_patch_area_size
16825 and function_entry_patch_area_start.
16826 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16827 including a two-value parser.
16828 * target.def (print_patchable_function_entry): New target hook.
16829 * targhooks.h (default_print_patchable_function_entry): New function.
16830 * targhooks.c (default_print_patchable_function_entry): Likewise.
16831 * toplev.c (process_options): Switch off IPA-RA if
16832 patchable function entries are being generated.
16833 * varasm.c (assemble_start_function): Look at the
16834 patchable-function-entry command line switch and current
16835 function attributes and maybe generate NOP instructions by
16836 calling the print_patchable_function_entry hook.
16837 * doc/extend.texi: Document patchable_function_entry attribute.
16838 * doc/invoke.texi: Document -fpatchable_function_entry
16839 command line option.
16840 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16842 * doc/tm.texi: Re-generate.
16844 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16847 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16848 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16850 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16852 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16854 2017-07-25 Richard Biener <rguenther@suse.de>
16856 PR tree-optimization/81455
16857 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16858 not walk in cycles when looking for guards.
16860 2017-07-25 Richard Biener <rguenther@suse.de>
16862 PR tree-optimization/81529
16863 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16864 when optimizing backedge uses.
16866 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16868 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16870 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16871 to dl_section_ref. On AIX, append an expression to subtract
16872 the size of the section length to dl_section_ref.
16874 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16876 * configure.ac: If any of the config.* scripts fail, exit 1.
16877 * configure: Regenerate.
16879 2017-07-25 Richard Biener <rguenther@suse.de>
16881 PR middle-end/81546
16882 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16883 of immediate uses, be more verbose on errors.
16885 2017-07-25 Richard Biener <rguenther@suse.de>
16887 PR tree-optimization/81510
16888 * tree-vect-loop.c (vect_is_simple_reduction): When the
16889 reduction stmt is not inside the loop bail out.
16891 2017-07-25 Richard Biener <rguenther@suse.de>
16893 PR tree-optimization/81303
16894 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16895 profitability check against LOOP_VINFO_NITERSM1.
16897 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16899 * domwalk.c (cmp_bb_postorder): Simplify.
16900 (sort_bbs_postorder): New function. Use it...
16901 (dom_walker::walk): ...here to optimize common cases.
16903 2017-07-25 Martin Liska <mliska@suse.cz>
16906 * ipa-visibility.c (function_and_variable_visibility): Make the
16907 redirection just on target that supports aliasing.
16908 Fix GNU coding style.
16910 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16913 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16915 * config/arm/rtems.h: Likewise.
16916 * config/bfin/rtems.h: Likewise.
16917 * config/i386/rtemself.h: Likewise.
16918 * config/lm32/rtems.h: Likewise.
16919 * config/m32c/rtems.h: Likewise.
16920 * config/m68k/rtemself.h: Likewise.
16921 * config/microblaze/rtems.h: Likewise.
16922 * config/mips/rtems.h: Likewise.
16923 * config/moxie/rtems.h: Likewise.
16924 * config/nios2/rtems.h: Likewise.
16925 * config/powerpcspe/rtems.h: Likewise.
16926 * config/rs6000/rtems.h: Likewise.
16927 * config/rtems.h: Likewise.
16928 * config/sh/rtems.h: Likewise.
16929 * config/sh/rtemself.h: Likewise.
16930 * config/sparc/rtemself.h: Likewise.
16932 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16935 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16936 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16937 * tree-ssa-structalias.c (alias_get_name): Same.
16939 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16942 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16943 instructions if no du chain is found.
16945 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16947 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16949 2017-07-25 Richard Biener <rguenther@suse.de>
16951 PR middle-end/81505
16952 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16955 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16957 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16958 upper-regs options.
16959 (ISA_2_7_MASKS_SERVER): Likewise.
16960 (ISA_3_0_MASKS_IEEE): Likewise.
16961 (OTHER_P8_VECTOR_MASKS): Likewise.
16962 (OTHER_VSX_VECTOR_MASKS): Likewise.
16963 (POWERPC_MASKS): Likewise.
16964 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16965 duplicate list of options.
16966 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16967 explicit -mupper-regs options.
16968 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16969 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16970 alias for -mupper-regs-df.
16971 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16972 (rs6000_init_hard_regno_mode_ok): Likewise.
16973 (rs6000_option_override_internal): Likewise.
16974 (rs6000_opt_masks): Likewise.
16975 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16976 options in terms of whether -mvsx or -mpower8-vector was used.
16977 (TARGET_UPPER_REGS_DI): Likewise.
16978 (TARGET_UPPER_REGS_SF): Likewise.
16979 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16980 -mupper-regs-* options.
16982 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16984 * passes.c (emergency_dump_function): Print some empty lines and a
16985 header before the RTL dump.
16987 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16989 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16991 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16994 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16995 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16997 2017-07-24 Carl Love <cel@us.ibm.com>
16999 * config/rs6000/rs6000-c.c: Add support for built-in functions
17000 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17001 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17002 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17003 vec_extract_fp_from_shortl): Add defines for the two builtins.
17004 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17005 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17007 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17008 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17009 * doc/extend.texi: Update the built-in documentation file for the
17010 new built-in function.
17012 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17015 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17017 * doc/generic.texi: Likewise.
17018 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17019 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17021 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17023 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17024 (aarch64_mls_elt_merge<mode>): Likewise.
17026 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17028 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17029 having __cxa_atexit.
17031 2017-07-23 Michael Collison <michael.collison@arm.com>
17033 * config/arm/arm.c (arm_option_override): Deprecate
17034 use of -mstructure-size-boundary.
17035 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17036 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17038 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17041 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17042 Reduce cost estimate for direct moves.
17044 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17047 * config/i386/i386.c (ix86_option_override_internal): Disable
17048 BMI, BMI2 and TBM instructions for -m16.
17050 2017-07-21 Carl Love <cel@us.ibm.com>
17052 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17053 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17054 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17055 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17056 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17057 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17058 VMULOSW): New enum "unspec" values.
17059 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17060 altivec_vmulosw): New patterns.
17061 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17062 VMULOSW): Add definitions.
17064 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17066 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17067 (qdf24xx): Likewise.
17068 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17069 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17070 (AARCH64_FL_V8_1): Renumber.
17071 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17072 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17073 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17074 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17075 rdma to feature modifiers list.
17077 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17079 PR middle-end/56727
17080 * ipa-visibility (function_and_variable_visibility): Convert
17081 recursive PLT call to direct call if appropriate.
17083 2017-07-21 Andrew Pinski <apinski@cavium.com>
17085 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17086 operand 1 to see if the types precision matches.
17087 * fold-const.c (operand_equal_p): Likewise.
17089 2017-07-21 Richard Biener <rguenther@suse.de>
17091 PR tree-optimization/81303
17092 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17093 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17094 (vect_peeling_hash_get_lowest_cost): Adjust.
17095 (vect_enhance_data_refs_alignment): Likewise. Use
17096 vect_get_peeling_costs_all_drs to compute the penalty for no
17097 peeling to match up costs.
17099 2017-07-21 Richard Biener <rguenther@suse.de>
17101 PR tree-optimization/81500
17102 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17103 we didn't identify a reduction path.
17105 2017-07-21 Tom de Vries <tom@codesourcery.com>
17106 Cesar Philippidis <cesar@codesourcery.com>
17108 PR gcov-profile/81442
17109 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17112 2017-07-21 Tom de Vries <tom@codesourcery.com>
17115 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17117 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17118 nvptx_override_options_after_change.
17120 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17122 * dwarf2out.c (output_file_names): Avoid double testing for
17123 dwarf_version >= 5.
17125 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17127 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17129 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17131 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17133 (try_crossjump_to_edge): Do not punt on partitioned functions.
17135 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17137 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17138 Put all BBs reachable only via paths crossing cold region to cold
17140 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17142 2016-07-21 Richard Biener <rguenther@suse.de>
17144 PR tree-optimization/81303
17145 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17146 into account prologue and epilogue iterations when raising
17147 min_profitable_iters to sth at least covering one vector iteration.
17149 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17151 * config/arm/arm.c (arm_test_cpu_arch_dat):
17154 2017-07-20 Nathan Sidwell <nathan@acm.org>
17156 Remove TYPE_METHODS.
17157 * tree.h (TYPE_METHODS): Delete.
17158 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17159 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17160 (dbxout_type_methods): Scan TYPE_FIELDS.
17161 (dbxout_type): Don't check TYPE_METHODS here.
17162 * function.c (use_register_for_decl): Always ignore register for
17163 class types when not optimizing.
17164 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17165 * tree.c (free_lang_data_in_type): Stitch out member functions and
17166 templates from TYPE_FIELDS.
17167 (build_distinct_type_copy, verify_type_variant,
17168 verify_type): Member fns are on TYPE_FIELDS.
17169 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17170 * tree-pretty-print.c (dump_generic_node): Likewise.
17172 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17175 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17176 V2TImode and V4TImode.
17177 (ix86_expand_vector_extract): Likewise.
17178 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17179 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17180 (ssescalarmode): Handle V4TImode and V2TImode.
17181 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17182 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17183 (VEXTRACTI128_MODE): New mode iterator.
17184 (splitter for *vec_extractv?ti first element): New.
17185 (VEC_INIT_MODE): New mode iterator.
17186 (vec_init<mode>): Consolidate 3 expanders into one using
17187 VEC_INIT_MODE mode iterator.
17189 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17191 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17192 non_spilled_static_chain_regno_p.
17194 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17196 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17198 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17200 * bb-reorder.c (connect_traces): Allow copying of blocks within
17203 2017-07-20 Richard Biener <rguenther@suse.de>
17205 * gimple.h (gimple_phi_result): Add gphi * overload.
17206 (gimple_phi_result_ptr): Likewise.
17207 (gimple_phi_arg): Likewise. Adjust index assert to only
17208 allow actual argument accesses rather than all slots available
17210 (gimple_phi_arg_def): Add gphi * overload.
17211 * tree-phinodes.c (make_phi_node): Initialize only actual
17213 (resize_phi_node): Clear memory not covered by old node,
17214 do not initialize excess argument slots.
17215 (reserve_phi_args_for_new_edge): Initialize new argument slot
17218 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17220 PR tree-optimization/81388
17222 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17224 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17225 by removing computation of may_be_zero.
17227 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17228 Tom de Vries <tom@codesourcery.com>
17230 PR middle-end/81030
17231 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17232 when gimple level profile disagrees with what RTL expander did.
17234 2017-07-20 Richard Biener <rguenther@suse.de>
17236 PR tree-optimization/61171
17237 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17238 (vect_analyze_stmt): Add slp instance parameter.
17239 (vectorizable_reduction): Likewise.
17240 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17241 (vect_is_simple_reduction): Deal with chains not detected
17242 as SLP reduction chain, specifically not properly associated
17243 chains containing a mix of plus/minus.
17244 (get_reduction_op): Remove.
17245 (get_initial_defs_for_reduction): Simplify, pass in whether
17246 this is a reduction chain, pass in the SLP node for the PHIs.
17247 (vect_create_epilog_for_reduction): Get the SLP instance as
17249 (vectorizable_reduction): Get the SLP instance as arg.
17250 During analysis remember the SLP node with the PHIs in the
17251 instance. Simplify getting at the vectorized reduction PHIs.
17252 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17253 through SLP instance.
17254 (vect_slp_analyze_operations): Likewise.
17255 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17256 (vect_transform_stmt): Likewise.
17258 2017-07-20 Tom de Vries <tom@codesourcery.com>
17260 PR tree-optimization/81489
17261 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17262 read of phi arg location to before loop that modifies phi.
17264 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17266 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17269 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17271 PR middle-end/81331
17272 * except.c (execute): Fix ordering issue.
17274 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17276 PR rtl-optimization/81423
17277 * combine.c (make_compound_operation_int): Don't try to optimize
17278 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17280 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17282 PR rtl-optimization/81423
17283 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17284 with a constant that is -1 in the truncated to mode.
17286 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17288 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17289 (determine_unlikely_bbs): ... here.
17290 * predict.h (propagate_unlikely_bbs_forward): Declare.
17291 * cfgexpand.c (pass_expand::execute): Use it.
17292 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17294 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17295 propagate_unlikely_bbs_forward.
17297 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17299 PR middle-end/81331
17300 * except.c (maybe_add_nop_after_section_switch): New function.
17303 2017-07-19 Tom de Vries <tom@codesourcery.com>
17305 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17307 2017-07-19 Tom de Vries <tom@codesourcery.com>
17309 * gimple.h (gimple_phi_arg): Make assert more strict.
17311 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17313 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17314 * config/rs6000/mmintrin.h: New file.
17315 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17317 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17319 PR tree-optimization/81346
17320 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17322 2017-07-19 Tom de Vries <tom@codesourcery.com>
17324 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17326 2017-07-19 Tom de Vries <tom@codesourcery.com>
17328 * config/nvptx/nvptx-modes.def: Add V2DImode.
17329 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17330 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17331 (nvptx_output_mov_insn): Handle lack of mov.b128.
17332 (nvptx_print_operand): Handle 'H' and 'L' codes.
17333 (nvptx_vector_mode_supported): Allow V2DImode.
17334 (nvptx_preferred_simd_mode): New function.
17335 (nvptx_data_alignment): New function.
17336 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17337 nvptx_preferred_simd_mode.
17338 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17340 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17342 2017-07-19 Tom de Vries <tom@codesourcery.com>
17344 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17345 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17346 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17347 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17348 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17349 (mov<VECIM>_insn): New define_insn.
17350 (define_expand "mov<VECIM>): New define_expand.
17352 2017-07-19 Tom de Vries <tom@codesourcery.com>
17354 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17356 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17358 PR tree-optimization/81346
17359 * fold-const.h (fold_div_compare, range_check_type): Declare.
17360 * fold-const.c (range_check_type): New function.
17361 (build_range_check): Use range_check_type.
17362 (fold_div_compare): No longer static, rewritten into
17363 a match.pd helper function.
17364 (fold_comparison): Don't call fold_div_compare here.
17365 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17366 as helper function.
17368 2017-07-19 Nathan Sidwell <nathan@acm.org>
17370 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17371 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17372 * tree.c (find_decls_types_r, verify_type): Use
17373 TYPE_{MIN,MAX}_VALUE_RAW.
17374 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17375 (hash_tree): Likewise.
17376 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17378 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17381 2017-07-18 Tom de Vries <tom@codesourcery.com>
17383 PR middle-end/81464
17384 * omp-expand.c (expand_omp_for_static_chunk): Handle
17385 equal-argument loop exit phi.
17387 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17390 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17391 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17392 operand 2 predicate.
17393 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17394 operand 2 predicate.
17395 (ror,rol -> rorx splitters): Use const_int_operand as
17396 operand 2 predicate.
17398 2017-06-18 Richard Biener <rguenther@suse.de>
17400 PR tree-optimization/81410
17401 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17402 the gap in the ! slp_perm SLP case after each group.
17404 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17406 PR middle-end/81463
17407 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17410 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17412 PR middle-end/81462
17413 * predict.c (set_even_probabilities): Cleanup; do not affect
17414 probabilities that are already known.
17415 (combine_predictions_for_bb): Call even when count is set.
17417 2017-07-18 Nathan Sidwell <nathan@acm.org>
17419 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17422 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17425 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17426 optimization for loop niter analysis.
17428 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17431 * config/avr/avr.c (avr_optimize_casesi): Don't use
17432 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17434 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17436 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17437 body_cost_vec from _vect_peel_extended_info.
17438 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17439 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17442 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17444 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17446 2017-07-18 Richard Biener <rguenther@suse.de>
17448 PR tree-optimization/80620
17449 PR tree-optimization/81403
17450 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17451 info when re-using a VN table entry.
17453 2017-07-18 Richard Biener <rguenther@suse.de>
17455 PR tree-optimization/81418
17456 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17457 vectype_in. Verify that with lane-reducing reduction operations
17458 we have a single def-use cycle.
17460 2017-07-17 Carl Love <cel@us.ibm.com>
17462 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17464 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17465 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17466 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17468 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17469 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17470 VMULOSW): New enum "unspec" values.
17471 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17472 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17473 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17474 altivec_vmulosw): New patterns.
17475 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17476 VMULOSW): Add definitions.
17477 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17479 * config/alpha/alpha.c: Include predict.h.
17481 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17483 * tree-vrp.c (compare_assert_loc): Fix comparison function
17484 to return predictable results.
17486 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17488 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17490 (subdi3): Likewise.
17491 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17492 * doc/invoke.texi (mexpand-adddi): Update text.
17494 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17496 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17497 that also clobbers the CC register. The old expand code is moved
17499 (*arc_clzsi2): ... here.
17500 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17501 the CC register. The old expand code is moved to ...
17502 (arc_ctzsi2): ... here.
17504 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17506 * config/arc/arc.opt (mindexed-loads): Use initial value
17507 TARGET_INDEXED_LOADS_DEFAULT.
17508 (mauto-modify-reg): Use initial value
17509 TARGET_AUTO_MODIFY_REG_DEFAULT.
17510 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17511 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17512 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17513 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17515 2017-07-17 Martin Liska <mliska@suse.cz>
17518 * opts.c (finish_options): Do not allow -fgnu-tm
17519 w/ -fsanitize={kernel-,}address. Say sorry.
17521 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17524 * tree-loop-distribution.c (classify_partition): Only assert on
17525 numer of iterations.
17526 (merge_dep_scc_partitions): Delete prameter. Update function call.
17527 (distribute_loop): Remove code handling loop with unknown niters.
17528 (pass_loop_distribution::execute): Skip loop with unknown niters.
17530 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17533 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17534 function sort_partitions_by_post_order.
17536 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17538 PR tree-optimization/81374
17539 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17540 the max index of basic blocks, rather than number of basic blocks.
17542 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17544 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17546 (arc_legitimate_pic_operand_p): Likewise.
17547 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17549 (arc_needs_pcl_p): Likewise.
17550 (arc_legitimate_pc_offset_p): Likewise.
17551 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17552 function is also used in constrains.md.
17553 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17554 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17555 PLUS. Only return true/false in known cases, otherwise assert.
17556 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17557 is already called in arc_legitimate_constant_p.
17558 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17560 (LEGITIMATE_PIC_OPERAND_P): Use
17561 arc_raw_symbolic_reference_mentioned_p function.
17562 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17567 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17568 Andrew Burgess <andrew.burgess@embecosm.com>
17570 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17571 (arc_return_address_register): New function.
17572 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17573 (arc_handle_fndecl_attribute): Add naked attribute.
17574 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17575 (TARGET_WARN_FUNC_RETURN): Likewise.
17576 (arc_allocate_stack_slots_for_args): New function.
17577 (arc_warn_func_return): Likewise.
17578 (machine_function): Change type fn_type.
17579 (arc_compute_function_type): Consider new naked function type,
17580 change function return type.
17581 (arc_must_save_register): Adapt to handle new
17582 arc_compute_function_type's return type.
17583 (arc_expand_prologue): Likewise.
17584 (arc_expand_epilogue): Likewise.
17585 (arc_return_address_regs): Delete.
17586 (arc_return_address_register): New function.
17587 (arc_epilogue_uses): Use above function.
17588 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17589 (arc_function_type): Change encoding, add naked type.
17590 (ARC_INTERRUPT_P): Change to handle the new encoding.
17591 (ARC_FAST_INTERRUPT_P): Likewise.
17592 (ARC_NORMAL_P): Define.
17593 (ARC_NAKED_P): Likewise.
17594 (arc_compute_function_type): Delete prototype.
17595 * config/arc/arc.md (in_ret_delay_slot): Use
17596 arc_return_address_register function.
17597 (simple_return): Likewise.
17598 (p_return_i): Likewise.
17600 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17602 PR tree-optimization/81428
17603 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17604 can't be built for those types.
17606 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17608 Remove stuff dead since r239246.
17610 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17611 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17612 (avr_inform_devices): Remove dead stuff.
17614 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17616 * config/arm/arm_neon.h: Fix softp typo.
17618 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17620 PR tree-optimization/81365
17621 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17622 aggregate moves onto bb predecessor edges, make sure there are no
17623 loads that could alias the lhs in between the start of bb and the
17626 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17629 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17630 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17631 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17633 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17635 PR tree-optimization/81396
17636 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17637 (init_symbolic_number): Initialize it to 1.
17638 (perform_symbolic_merge): Add n_ops from both operands into the new
17640 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17641 without base_addr as useless if they need more than one operation.
17642 (bswap_replace): Handle !bswap case for NULL base_addr.
17644 2017-07-17 Tom de Vries <tom@codesourcery.com>
17647 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17650 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17652 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17653 conditional builtin define __FIX_LEON3FT_B2BST.
17655 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17657 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17658 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17661 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17663 PR rtl-optimization/81424
17664 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17665 to remove potential trapping from operands if -fnon-call-exceptions.
17667 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17669 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17670 profile_proability for scalling.
17671 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17673 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17675 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17677 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17679 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17680 fixpoint arithmetics.
17682 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17684 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17685 fixpoint arithmetics.
17687 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17689 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17690 fixpoint arithmetics.
17692 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17694 * profile-count.h (profile_probability::from_reg_br_prob_note,
17695 profile_probability::to_reg_br_prob_note): New functions.
17696 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17697 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17698 * predict.c (probability_reliable_p): Update.
17699 (edge_probability_reliable_p): Update.
17700 (br_prob_note_reliable_p): Update.
17701 (invert_br_probabilities): Update.
17702 (add_reg_br_prob_note): New function.
17703 (combine_predictions_for_insn): Update.
17704 * asan.c (asan_clear_shadow): Update.
17705 * cfgbuild.c (compute_outgoing_frequencies): Update.
17706 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17707 (update_br_prob_note): Update.
17708 (rtl_verify_edges): Update.
17709 (purge_dead_edges): Update.
17710 (fixup_reorder_chain): Update.
17711 * emit-rtl.c (try_split): Update.
17712 * ifcvt.c (cond_exec_process_insns): Update.
17713 (cond_exec_process_if_block): Update.
17714 (dead_or_predicable): Update.
17715 * internal-fn.c (expand_addsub_overflow): Update.
17716 (expand_neg_overflow): Update.
17717 (expand_mul_overflow): Update.
17718 * loop-doloop.c (doloop_modify): Update.
17719 * loop-unroll.c (compare_and_jump_seq): Update.
17720 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17721 * predict.h: Update.
17722 * reorg.c (mostly_true_jump): Update.
17724 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17725 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17726 * config/arc/arc.c: (emit_unlikely_jump): Update.
17727 * config/arm/arm.c: (emit_unlikely_jump): Update.
17728 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17729 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17730 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17731 (ix86_print_operand): Update.
17732 (ix86_split_fp_branch): Update.
17733 (predict_jump): Update.
17734 * config/ia64/ia64.c (ia64_print_operand): Update.
17735 * config/mmix/mmix.c (mmix_print_operand): Update.
17736 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17737 (rs6000_expand_split_stack_prologue): Update.
17738 * config/rs6000/rs6000.c: Update.
17739 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17740 (s390_expand_vec_movstr): Update.
17741 (s390_expand_cs_tdsi): Update.
17742 (s390_expand_split_stack_prologue): Update.
17743 * config/sh/sh.c (sh_print_operand): Update.
17744 (expand_cbranchsi4): Update.
17745 (expand_cbranchdi4): Update.
17746 * config/sparc/sparc.c (output_v9branch): Update.
17747 * config/spu/spu.c (get_branch_target): Update.
17748 (ea_load_store_inline): Update.
17749 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17750 * config/tilepro/tilepro.c: Update.
17752 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17754 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17755 (gimplify_save_expr): Likewise.
17757 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17759 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17761 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17763 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17765 * timevar.def (TV_IPA_FNSUMMARY): Define.
17767 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17769 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17770 to back store errata sensitive sequence from being generated.
17771 (sqrtdf2_fix): Likewise.
17773 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17775 * tree-ssa-threadupdate.c (compute_path_counts,
17776 update_joiner_offpath_counts): Use profile_probability.
17778 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17781 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17783 * config/arm/arm-c.c (arm_cpu_builtins): Define
17784 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17786 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17788 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17789 array entries to represent __ieee128 versions of the
17790 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17791 scalar_extract_sig, and scalar_insert_exp built-in functions.
17792 (altivec_resolve_overloaded_builtin): Add special case handling
17793 for the __builtin_scalar_insert_exp function, as represented by
17794 the P9V_BUILTIN_VEC_VSIEDP constant.
17795 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17796 exponent support for __ieee128 argument.
17797 (VSESQP): Add scalar extract signature support for __ieee128
17799 (VSTDCNQP): Add scalar test negative support for __ieee128
17801 (VSIEQP): Add scalar insert exponent support for __int128 argument
17802 with __ieee128 result.
17803 (VSIEQPF): Add scalar insert exponent support for __ieee128
17804 argument with __ieee128 result.
17805 (VSTDCQP): Add scalar test data class support for __ieee128
17807 (VSTDCNQP): Add overload support for scalar test negative with
17808 __ieee128 argument.
17809 (VSTDCQP): Add overload support for scalar test data class
17810 __ieee128 argument.
17811 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17812 UNSPEC_VSX_SXSIGDP.
17813 (UNSPEC_VSX_SIEXPQP): New constant.
17814 (xsxexpqp): New insn for VSX scalar extract exponent quad
17816 (xsxsigqp): New insn for VSX scalar extract significand quad
17818 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17819 precision with floating point argument.
17820 (xststdcqp): New expand for VSX scalar test data class quad
17822 (xststdcnegqp): New expand for VSX scalar test negative quad
17824 (xststdcqp): New insn to match expansions for VSX scalar test data
17825 class quad precision and VSX scalar test negative quad precision.
17826 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17827 special case operand checking to enforce that second operand of
17828 VSX scalar test data class with quad precision argument is a 7-bit
17830 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17831 prototypes and descriptions of __ieee128 versions of
17832 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17833 scalar_test_data_class, and scalar_test_neg built-in functions.
17835 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17837 PR tree-optimization/81162
17838 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17839 replace a negate with an add.
17841 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17843 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17845 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17847 * config/arm/arm-c.c (arm_cpu_builtins): Define
17848 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17850 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17852 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17853 (armv8-r): Set ARM Cortex-R52 as default CPU.
17854 * config/arm/arm-tables.opt: Regenerate.
17855 * config/arm/arm-tune.md: Regenerate.
17856 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17858 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17859 extension for -mcpu=cortex-r52.
17861 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17863 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17864 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17865 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17866 (fp-armv8): Define it as FP_ARMv8 only.
17867 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17868 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17870 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17871 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17872 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17873 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17874 than TARGET_FPU_ARMV8.
17875 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17876 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17877 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17879 * config/arm/neon.md (neon_vrint): Likewise.
17880 (neon_vcvt): Likewise.
17881 (neon_<fmaxmin_op><mode>): Likewise.
17882 (<fmaxmin><mode>3): Likewise.
17883 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17884 * config/arm/predicates.md (arm_cond_move_operator): Check against
17885 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17887 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17889 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17890 to top of function.
17892 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17894 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17895 loop in comment with memset.
17897 2017-07-14 Martin Liska <mliska@suse.cz>
17899 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17900 * dwarf2out.c (is_java): Remove the function.
17901 (output_pubname): Remove usage of the function.
17902 (lower_bound_default): Remove usage of DW_LANG_Java.
17903 (gen_compile_unit_die): Likewise.
17904 * gcc.c: Remove compiler defaults for .java and .zip files.
17905 * gimple-expr.c (remove_suffix): Change as there's no longer
17906 extension than 4-letter one.
17907 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17908 (gimplify_save_expr): Likewise.
17909 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17910 as it's possible even for other languages than Java.
17911 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17912 * lto-opts.c (lto_write_options): Remove reference to Java.
17913 * opts.c (strip_off_ending): Update file extension handling.
17914 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17915 * tree-eh.c (lower_resx): Likewise.
17916 * tree.c (free_lang_data_in_type): Remove dead code.
17917 (find_decls_types_r): Likewise.
17918 (build_common_builtin_nodes): Remove Java from a comment.
17919 (verify_type): Remove dead code.
17920 * varasm.c (assemble_external): Remove Java from a comment.
17922 2017-07-14 Martin Liska <mliska@suse.cz>
17924 * opts.c (finish_options): Add quotes.
17925 (common_handle_option): Likewise.
17927 2017-07-14 Martin Liska <mliska@suse.cz>
17929 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17930 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17931 Remove N_SO_PASCAL.
17932 * dwarf2out.c (lower_bound_default): Do not handle
17934 (gen_compile_unit_die): Likewise.
17935 * gcc.c: Remove default extension binding for GNU Pascal.
17936 * stmt.c: Remove Pascal language from a comment.
17937 * xcoffout.c: Likewise.
17939 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17942 * diagnostic-show-locus.c (fixit_cmp): New function.
17943 (layout::layout): Sort m_fixit_hints.
17944 (column_range::column_range): Assert that the values are valid.
17945 (struct char_span): New struct.
17946 (correction::overwrite): New method.
17947 (struct source_line): New struct.
17948 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17949 calls in terms of classes source_line and char_span, and
17950 correction::overwrite.
17951 (selftest::test_overlapped_fixit_printing_2): New function.
17952 (selftest::diagnostic_show_locus_c_tests): Call it.
17954 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17957 early if there is no lhs.
17959 2017-07-13 Martin Liska <mliska@suse.cz>
17961 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17962 (gen_reference_type_die): Likewise.
17963 * stor-layout.c: Remove Pascal-related comment.
17965 2017-07-13 Martin Liska <mliska@suse.cz>
17967 * opts.c (finish_options): Add quotes to error messages.
17968 (parse_sanitizer_options): Likewise.
17970 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17972 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17974 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17976 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17978 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17980 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17982 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17984 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17987 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17988 provides the hardware capability bits, define the macro
17989 __BUILTIN_CPU_SUPPORTS__.
17990 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17991 if GLIBC does not provide the hardware capability bits. Add a
17992 gcc_unreachable call if the built-in cpu function is neither
17993 __builtin_cpu_is nor __builtin_cpu_supports.
17994 (rs6000_get_function_versions_dispatcher): Change the warning
17995 that an old GLIBC is used which does not export the capability
17996 bits to be an error.
17997 * doc/extend.texi (target_clones attribute): Document the
17998 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17999 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18000 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18001 the macros defined by GCC if the newer GLIBC is available.
18003 2017-07-12 Jeff Law <law@redhat.com>
18005 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18006 remaining includes slightly.
18007 * config/riscv/riscv-builtins.c: Include profile-count.h.
18009 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18012 * config/avr/avr.c (avr_set_current_function): In diagnostic
18013 messages: Quote keywords and (parts of) identifiers.
18014 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18017 2017-07-12 Carl Love <cel@us.ibm.com>
18019 * config/rs6000/rs6000-c.c: Add support for built-in functions
18020 vector bool char vec_revb (vector bool char);
18021 vector bool short vec_revb (vector short char);
18022 vector bool int vec_revb (vector bool int);
18023 vector bool long long vec_revb (vector bool long long);
18024 * doc/extend.texi: Update the built-in documentation file for the
18025 new built-in functions.
18027 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18029 * config/s390/s390.md: Remove movcc splitter.
18031 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18033 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18034 load/store on condition.
18036 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18039 * config/avr/avr.c (avr_encode_section_info)
18040 [progmem && !TREE_READONLY]: Error if progmem object needs
18043 2017-07-11 Michael Collison <michael.collison@arm.com>
18045 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18048 2017-07-11 Carl Love <cel@us.ibm.com>
18050 * config/rs6000/rs6000-c.c: Add support for builtins
18051 vector unsigned int vec_parity_lsbb (vector signed int);
18052 vector unsigned int vec_parity_lsbb (vector unsigned int);
18053 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18054 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18055 vector unsigned long long vec_parity_lsbb (vector signed long long);
18056 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18057 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18058 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18059 * doc/extend.texi: Update the built-in documentation file for the
18060 new built-in functions.
18062 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18064 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18065 (layout::m_primary_loc): New field.
18066 (layout::layout): Initialize new field. Move location filtering
18067 logic from here to...
18068 (layout::maybe_add_location_range): ...this new method. Add
18069 support for filtering to just the lines already specified by other
18071 (layout::will_show_line_p): New method.
18072 (gcc_rich_location::add_location_if_nearby): New method.
18073 (selftest::test_add_location_if_nearby): New test function.
18074 (selftest::diagnostic_show_locus_c_tests): Call it.
18075 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18078 2017-07-11 Tom de Vries <tom@codesourcery.com>
18080 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18081 (bb_first_real_insn): New function.
18082 (nvptx_single): Add extra initialization of broadcasted condition
18085 2017-07-11 Nathan Sidwell <nathan@acm.org>
18087 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18089 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18091 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18092 Binutils doc as TEXI will mess them up.
18093 * doc/invoke.texi (AVR Options): Same here.
18095 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18097 * config/sparc/sparc.opt (mfix-ut700): New option.
18098 (mfix-gr712rc): Likewise.
18099 (sparc_fix_b2bst): New variable.
18100 * doc/invoke.texi (SPARC options): Document them.
18101 (ARM options): Fix warnings.
18102 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18103 instructions to prevent sequences that can trigger the store-store
18104 errata for certain LEON3FT processors.
18105 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18106 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18107 * config/sparc/sparc.md (fix_b2bst): New attribute.
18108 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18110 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18113 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18115 (*rsqrtsf2_sse): Ditto.
18117 (div<mode>3): Macroize insn from divdf3 and divsf3
18118 using MODEF mode iterator.
18120 2017-07-10 Martin Sebor <msebor@redhat.com>
18122 PR tree-optimization/80397
18123 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18124 instead of testing for equality to INTEGER_TYPE.
18126 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18128 * config.gcc: Remove uclibc from arc target spec.
18130 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18132 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18134 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18137 * lto-wrapper.c (remove_option): New function.
18138 (merge_and_complain): Merge PIC/PIE options more realistically.
18140 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18142 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18146 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18147 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18148 * config.in: Regenerate.
18149 * configure: Regenerate.
18150 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18151 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18152 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18153 (TARGET_GASISR_PROLOGUES): ...target mask.
18154 * common/config/avr/avr-common.c
18155 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18156 Set -mgas-isr-prologues.
18157 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18158 INSERT_PASS_BEFORE for it.
18159 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18160 * config/avr/avr.c (avr_option_override)
18161 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18162 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18163 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18164 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18165 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18167 (make_avr_pass_pre_proep): New function.
18168 (emit_push_sfr) <treg>: Add argument to function and use it
18169 instead of TMP_REG.
18170 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18171 and set machine->gasisr.yes.
18172 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18173 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18174 __gcc_isr.n_pushed to .L__stack_usage.
18175 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18176 (avr_asm_final_postscan_insn): ...this new static function.
18177 * config/avr/avr.h (machine_function)
18178 <is_no_gccisr, use_L__stack_usage>: New fields.
18179 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18180 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18181 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18182 (gasisr, *gasisr): New expander and insn.
18183 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18184 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18185 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18187 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18189 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18192 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18194 Move jump-tables out of .text again.
18197 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18198 (ASM_OUTPUT_ADDR_VEC): New function.
18199 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18200 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18201 INSN_ADDRESSes as asm comment.
18202 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18203 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18204 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18205 * config/avr/avr.md (*tablejump): Adjust comment.
18206 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18207 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18209 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18210 (avr_output_addr_vec): New proto.
18211 (avr_log_t) <insn_addresses>: New field.
18213 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18216 * config/i386/i386.c (ix86_function_arg_advance): Set
18217 outgoing_args_on_stack to true if there are outgoing arguments
18219 (ix86_function_arg): Likewise.
18220 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18221 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18222 * config/i386/i386.h (machine_function): Add
18223 outgoing_args_on_stack.
18225 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18227 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18228 supporting pthreds.
18229 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18231 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18233 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18234 (REAL_H): Remove $(MACHMODE_H).
18235 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18237 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18238 $(MACHMODE_H) and double-int.h.
18239 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18241 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18244 2017-07-07 Andrew Pinski <apinski@cavium.com>
18246 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18247 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18249 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18251 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18252 Add warning if GCC was not configured to link against a GLIBC that
18253 exports the hardware capability bits.
18254 (make_resolver_func): Make resolver function private and not a
18255 COMDAT function. Create the name with clone_function_name instead
18256 of make_unique_name.
18259 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18260 correct operand in doing the split.
18262 2017-07-07 Carl Love <cel@us.ibm.com>
18264 * config/rs6000/rs6000-c: Add support for built-in function
18265 vector unsigned short vec_pack_to_short_fp32 (vector float,
18267 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18268 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18269 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18270 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18271 (convert_4f32_8i16): Add define_expand.
18272 * doc/extend.texi: Update the built-in documentation file for the
18273 new built-in function.
18275 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18277 * config/sparc/m8.md: New file.
18278 * config/sparc/sparc.md: Include m8.md.
18280 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18282 * config/sparc/sparc.opt: New option -mvis4b.
18283 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18284 (sparc_option_override): Handle VIS4B.
18285 (enum sparc_builtins): Define
18286 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18287 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18288 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18289 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18290 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18291 (check_constant_argument): New function.
18292 (sparc_vis_init_builtins): Define builtins
18293 __builtin_vis_dictunpack{8,16,32},
18294 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18295 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18296 __builtin_vis_fpcmpde{8,16,32}shl and
18297 __builtin_vis_fpcmpur{8,16,32}shl.
18298 (sparc_expand_builtin): Check that the constant operands to
18299 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18300 constant and in range.
18301 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18303 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18304 (SPARC_IMM5_P): Likewise.
18305 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18306 (enabled): Handle vis4b.
18307 (UNSPEC_DICTUNPACK): New unspec.
18308 (UNSPEC_FPCMPSHL): Likewise.
18309 (UNSPEC_FPUCMPSHL): Likewise.
18310 (UNSPEC_FPCMPDESHL): Likewise.
18311 (UNSPEC_FPCMPURSHL): Likewise.
18312 (cpu_feature): New CPU feature `vis4b'.
18313 (dictunpack{8,16,32}): New insns.
18314 (FPCSMODE): New mode iterator.
18315 (fpcscond): New code iterator.
18316 (fpcsucond): Likewise.
18317 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18318 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18319 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18320 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18321 * config/sparc/constraints.md: Define constraints `q' for unsigned
18322 2-bit integer constants and `t' for unsigned 5-bit integer
18324 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18326 (imm5_operand_dictunpack16): Likewise.
18327 (imm5_operand_dictunpack32): Likewise.
18328 (imm2_operand): Likewise.
18329 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18330 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18331 ditunpack* and fpcmp*shl builtins.
18333 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18335 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18336 * config.in: Add HAVE_AS_SPARC6 define.
18337 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18339 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18341 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18342 (CPP_CPU_SPEC): Handle m8.
18343 (ASM_CPU_SPEC): Likewise.
18344 * config/sparc/sparc-opts.h (enum processor_type): Add
18346 * config/sparc/sparc.c (m8_costs): New struct.
18347 (sparc_option_override): Handle TARGET_CPU_m8.
18348 (sparc32_initialize_trampoline): Likewise.
18349 (sparc64_initialize_trampoline): Likewise.
18350 (sparc_issue_rate): Likewise.
18351 (sparc_register_move_cost): Likewise.
18352 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18353 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18354 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18355 (CPP_CPU_SPEC): Handle M8.
18356 (ASM_CPU_SPEC): Likewise.
18357 (AS_M8_FLAG): Define.
18358 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18359 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18360 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18362 * configure: Regenerate.
18363 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18366 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18368 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18370 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18371 ("*movdi_insn_sp32"): Do not set v3pipe.
18372 ("*movsi_insn"): Likewise.
18373 ("*movdi_insn_sp64"): Likewise.
18374 ("*movsf_insn"): Likewise.
18375 ("*movdf_insn_sp32"): Likewise.
18376 ("*movdf_insn_sp64"): Likewise.
18377 ("*zero_extendsidi2_insn_sp64"): Likewise.
18378 ("*sign_extendsidi2_insn"): Likewise.
18379 ("*mov<VM32:mode>_insn"): Likewise.
18380 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18381 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18382 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18383 ("<vlop:code><VL:mode>3"): Likewise.
18384 ("*not_<vlop:code><VL:mode>3"): Likewise.
18385 ("*nand<VL:mode>_vis"): Likewise.
18386 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18387 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18388 ("one_cmpl<VL:mode>2"): Likewise.
18389 ("faligndata<VM64:mode>_vis"): Likewise.
18390 ("alignaddrsi_vis"): Likewise.
18391 ("alignaddrdi_vis"): Likweise.
18392 ("alignaddrlsi_vis"): Likewise.
18393 ("alignaddrldi_vis"): Likewise.
18394 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18395 ("bmaskdi_vis"): Likewise.
18396 ("bmasksi_vis"): Likewise.
18397 ("bshuffle<VM64:mode>_vis"): Likewise.
18398 ("cmask8<P:mode>_vis"): Likewise.
18399 ("cmask16<P:mode>_vis"): Likewise.
18400 ("cmask32<P:mode>_vis"): Likewise.
18401 ("pdistn<P:mode>_vis"): Likewise.
18402 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18404 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18406 * config/sparc/sparc.md ("subtype"): New insn attribute.
18407 ("*wrgsr_sp64"): Set insn subtype.
18408 ("*rdgsr_sp64"): Likewise.
18409 ("alignaddrsi_vis"): Likewise.
18410 ("alignaddrdi_vis"): Likewise.
18411 ("alignaddrlsi_vis"): Likewise.
18412 ("alignaddrldi_vis"): Likewise.
18413 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18414 ("fexpand_vis"): Likewise.
18415 ("fpmerge_vis"): Likewise.
18416 ("faligndata<VM64:mode>_vis"): Likewise.
18417 ("bshuffle<VM64:mode>_vis"): Likewise.
18418 ("cmask8<P:mode>_vis"): Likewise.
18419 ("cmask16<P:mode>_vis"): Likewise.
18420 ("cmask32<P:mode>_vis"): Likewise.
18421 ("fchksm16_vis"): Likewise.
18422 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18423 ("fmean16_vis"): Likewise.
18424 ("fp<plusminus_insn>64_vis"): Likewise.
18425 ("<plusminus_insn>v8qi3"): Likewise.
18426 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18427 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18428 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18429 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18430 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18431 ("*movqi_insn"): Likewise.
18432 ("*movhi_insn"): Likewise.
18433 ("*movsi_insn"): Likewise.
18434 ("movsi_pic_gotdata_op"): Likewise.
18435 ("*movdi_insn_sp32"): Likewise.
18436 ("*movdi_insn_sp64"): Likewise.
18437 ("movdi_pic_gotdata_op"): Likewise.
18438 ("*movsf_insn"): Likewise.
18439 ("*movdf_insn_sp32"): Likewise.
18440 ("*movdf_insn_sp64"): Likewise.
18441 ("*zero_extendhisi2_insn"): Likewise.
18442 ("*zero_extendqihi2_insn"): Likewise.
18443 ("*zero_extendqisi2_insn"): Likewise.
18444 ("*zero_extendqidi2_insn"): Likewise.
18445 ("*zero_extendhidi2_insn"): Likewise.
18446 ("*zero_extendsidi2_insn_sp64"): Likewise.
18447 ("ldfsr"): Likewise.
18448 ("prefetch_64"): Likewise.
18449 ("prefetch_32"): Likewise.
18450 ("tie_ld32"): Likewise.
18451 ("tie_ld64"): Likewise.
18452 ("*tldo_ldub_sp32"): Likewise.
18453 ("*tldo_ldub1_sp32"): Likewise.
18454 ("*tldo_ldub2_sp32"): Likewise.
18455 ("*tldo_ldub_sp64"): Likewise.
18456 ("*tldo_ldub1_sp64"): Likewise.
18457 ("*tldo_ldub2_sp64"): Likewise.
18458 ("*tldo_ldub3_sp64"): Likewise.
18459 ("*tldo_lduh_sp32"): Likewise.
18460 ("*tldo_lduh1_sp32"): Likewise.
18461 ("*tldo_lduh_sp64"): Likewise.
18462 ("*tldo_lduh1_sp64"): Likewise.
18463 ("*tldo_lduh2_sp64"): Likewise.
18464 ("*tldo_lduw_sp32"): Likewise.
18465 ("*tldo_lduw_sp64"): Likewise.
18466 ("*tldo_lduw1_sp64"): Likewise.
18467 ("*tldo_ldx_sp64"): Likewise.
18468 ("*mov<VM32:mode>_insn"): Likewise.
18469 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18470 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18472 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18474 * config/sparc/sparc.md ("type"): New insn type viscmp.
18475 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18477 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18478 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18479 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18480 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18482 ("n7_vis_logical_11cycle"): Likewise.
18483 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18484 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18485 * config/sparc/niagara.md ("niag_vis"): Likewise.
18486 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18487 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18489 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18491 * config/sparc/sparc.md: New instruction type `bmask'.
18492 (bmaskdi_vis): Use the `bmask' type.
18493 (bmasksi_vis): Likewise.
18494 * config/sparc/ultra3.md (us3_array): Likewise.
18495 * config/sparc/niagara7.md (n7_array): Likewise.
18496 * config/sparc/niagara4.md (n4_array): Likewise.
18497 * config/sparc/niagara2.md (niag2_vis): Likewise.
18498 (niag3_vis): Likewise.
18499 * config/sparc/niagara.md (niag_vis): Likewise.
18501 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18503 * ipa-comdats.c: Remove optimize check from gate.
18504 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18505 for functions not optimized.
18506 (ipa_fn_summary_read): Skip optimize check.
18507 (ipa_fn_summary_write): Likewise.
18508 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18510 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18512 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18513 (check_callers): Give up on caller which is not optimized.
18514 (inline_small_functions): Likewise.
18515 (ipa_inline): Do not give up when not optimizing.
18516 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18517 away unoptimizes cdtors.
18518 (whole_program_function_and_variable_visibility): Do
18519 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18520 * ipa.c (process_references): Do not check optimize.
18521 (symbol_table::remove_unreachable_nodes): Update optimize check.
18522 (set_writeonly_bit): Update optimize check.
18523 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18524 (pass_ipa_single_use::gate): Remove.
18526 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18528 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18529 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18530 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18531 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18532 permute_load, permute_store, adjust_extract, adjust_splat,
18533 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18534 replace_swap_with_copy, dump_swap_insn_table,
18535 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18536 recombine_lvx_pattern, recombine_stvx_pattern,
18537 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18538 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18539 to file rs6000-p8swap.c.
18540 * config/rs6000/rs6000-p8swap.c: New file.
18541 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18542 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18543 and rs6000*-*-* targets.
18545 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18547 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18549 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18551 * lto-wrapper.c (merge_and_complain): Do not merge
18552 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18553 fsigned_zeros, ftrapping_math, fwrapv.
18554 (append_compiler_options): Do not track these options.
18555 (append_linker_options): Likewie
18557 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18559 * cgraphunit.c (cgraph_node::finalize_function): When
18560 !flag_toplevel_reorde set no_reorder flag.
18561 (varpool_node::finalize_decl): Likewise.
18562 (symbol_table::compile): Drop no toplevel reorder path.
18564 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18566 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18567 edges; zero probability is not better than uninitialized.
18569 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18571 * asan.h (asan_sanitize_allocas_p): Declare.
18572 * asan.c (asan_sanitize_allocas_p): New function.
18573 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18574 (handle_builtin_alloca): Likewise.
18575 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18576 if !asan_sanitize_allocas_p.
18577 * params.def (asan-instrument-allocas): Add new option.
18578 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18579 * opts.c (common_handle_option): Disable allocas sanitization for
18582 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18584 * asan.c: Include gimple-fold.h.
18585 (get_last_alloca_addr): New function.
18586 (handle_builtin_stackrestore): Likewise.
18587 (handle_builtin_alloca): Likewise.
18588 (asan_emit_allocas_unpoison): Likewise.
18589 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18590 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18591 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18592 (instrument_builtin_call): Pass gimple iterator to
18593 get_mem_refs_of_builtin_call.
18594 (last_alloca_addr): New global.
18595 * asan.h (asan_emit_allocas_unpoison): Declare.
18596 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18597 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18598 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18599 if function calls alloca.
18600 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18601 * gimple-fold.h (replace_call_with_value): Declare.
18602 * internal-fn.c: Include asan.h.
18603 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18604 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18606 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18608 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18609 (C_SELFTEST_FLAGS): New.
18610 (CPP_SELFTEST_FLAGS): New.
18611 (SELFTEST_DEPS): New, from deps of s-selftest.
18612 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18613 (CPP_SELFTEST_DEPS): New.
18614 (selftest): Add dependency on s-selftest-c++.
18615 (s-selftest): Rename to...
18616 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18617 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18618 than SELFTEST_FLAGS.
18619 (selftest-gdb): Rename to...
18620 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18622 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18623 (selftest-valgrind): Rename to...
18624 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18626 (selftest-valgrind): Reintroduce as an alias for
18627 selftest-c-valgrind.
18628 (s-selftest-c++): New.
18629 (selftest-c++-gdb): New.
18630 (selftest-c++-valgrind): New.
18632 2017-07-06 Olivier Hainque <hainque@adacore.com>
18634 * gcc.c (process_command): When deciding if undefined variables
18635 should be ignored when processing specs, accept "gcc -v" as well.
18637 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18639 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18640 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18642 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18644 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18645 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18646 * config/arm/arm-tables.opt: Regenerate.
18647 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18649 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18651 2017-07-06 Carl Love <cel@us.ibm.com>
18653 * ChangeLog: Clean up from mid air collision
18655 2017-07-06 Carl Love <cel@us.ibm.com>
18657 * config/rs6000/rs6000-c.c: Add support for built-in functions
18658 vector signed int vec_subc (vector signed int, vector signed int);
18659 vector signed __int128 vec_subc (vector signed __int128,
18660 vector signed __int128);
18661 vector unsigned __int128 vec_subc (vector unsigned __int128,
18662 vector unsigned __int128);
18663 vector signed int vec_sube (vector signed int, vector signed int,
18664 vector signed int);
18665 vector unsigned int vec_sube (vector unsigned int,
18666 vector unsigned int,
18667 vector unsigned int);
18668 vector signed __int128 vec_sube (vector signed __int128,
18669 vector signed __int128,
18670 vector signed__int128);
18671 vector unsigned __int128 vec_sube (vector unsigned __int128,
18672 vector unsigned __int128,
18673 vector unsigned __int128);
18674 vector signed int vec_subec (vector signed int, vector signed int,
18675 vector signed int);
18676 vector unsigned int vec_subec (vector unsigned int,
18677 vector unsigned int,
18678 vector unsigned int);
18679 vector signed __int128 vec_subec (vector signed __int128,
18680 vector signed __int128,
18681 vector signed__int128);
18682 vector unsigned __int128 vec_subec (vector unsigned __int128,
18683 vector unsigned __int128,
18684 vector unsigned __int128);
18685 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18686 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18687 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18688 BU_ALTIVEC_OVERLOAD_X definitions.
18689 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18690 * doc/extend.texi: Update the built-in documentation file for the new
18691 built-in functions.
18693 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18696 * diagnostic-show-locus.c (layout::layout): Use start and finish
18697 spelling location for the start and finish of each range.
18698 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18699 Add unused aspect param.
18700 * input.c (expand_location_1): Add "aspect" param, and use it
18701 to access the correct part of the location.
18702 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18704 (expand_location_to_spelling_point): Likewise.
18705 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18706 param, and pass it to expand_location_1.
18708 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18710 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18711 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18712 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18713 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18714 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18715 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18716 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18717 _mm_maskz_getmant_ss): New intrinsics.
18718 (__builtin_ia32_getexpss128_mask): Changed to ...
18719 __builtin_ia32_getexpss128_round ... this.
18720 (__builtin_ia32_getexpsd128_mask): Changed to ...
18721 __builtin_ia32_getexpsd128_round ... this.
18722 * config/i386/i386-builtin-types.def
18723 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18724 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18725 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18726 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18727 __builtin_ia32_getmantss_mask_round): New builtins.
18728 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18729 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18730 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18731 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18732 * config/i386/sse.md
18733 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18734 avx512f_sgetexp<mode><mask_scalar_name>
18735 <round_saeonly_scalar_name> ... this.
18736 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18737 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18738 vgetexp<ssescalarmodesuffix>
18739 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18740 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18741 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18742 avx512f_vgetmant<mode><mask_scalar_name>
18743 <round_saeonly_scalar_name> ... this.
18744 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18745 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18746 vgetmant<ssescalarmodesuffix>
18747 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18748 %0<mask_scalar_operand4>, %1, %2
18749 <round_saeonly_scalar_mask_op4>, %3} ... this.
18750 * config/i386/subst.md (mask_scalar_operand4,
18751 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18752 round_saeonly_scalar_nimm_predicate): New subst attributes.
18754 2017-07-06 Julia Koval <julia.koval@intel.com>
18756 * config/i386/i386.c (ix86_erase_embedded_rounding):
18757 Remove code for old rounding pattern.
18759 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18761 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18763 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18765 * doc/sourcebuild.texi (Test Directives, Variants of
18766 dg-require-support): Add documentation for dg-require-stack-check.
18768 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18770 * config/i386/subst.md (mask_scalar, round_scalar,
18771 round_saeonly_scalar): New meta-templates.
18772 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18773 round_scalar_mask_operand3, round_scalar_mask_op3,
18774 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18775 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18776 round_saeonly_scalar_constraint,
18777 round_saeonly_scalar_prefix): New subst attribute.
18778 * config/i386/sse.md
18779 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18780 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18781 <round_scalar_name> ... this.
18782 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18783 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18784 <round_scalar_name> ... this.
18785 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18786 <sse>_vm<code><mode>3<mask_scalar_name>
18787 <round_saeonly_scalar_name> ... this.
18788 (v<plusminus_mnemonic><ssescalarmodesuffix>
18789 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18790 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18791 v<plusminus_mnemonic><ssescalarmodesuffix>
18792 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18793 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18794 (v<multdiv_mnemonic><ssescalarmodesuffix>
18795 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18796 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18797 v<multdiv_mnemonic><ssescalarmodesuffix>
18798 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18799 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18800 (v<maxmin_float><ssescalarmodesuffix>
18801 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18802 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18803 v<maxmin_float><ssescalarmodesuffix>
18804 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18805 %0<mask_scalar_operand3>, %1, %<iptr>2
18806 <round_saeonly_scalar_mask_op3>} ... this.
18808 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18810 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18811 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18813 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18814 Alan Hayward <alan.hayward@arm.com>
18815 David Sherwood <david.sherwood@arm.com>
18817 * combine.c (simplify_if_then_else): Remove "enum" before
18819 * compare-elim.c (can_eliminate_compare): Likewise.
18820 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18822 (aarch64_lookup_simd_builtin_type): Likewise.
18823 (aarch64_simd_builtin_type): Likewise.
18824 (aarch64_init_simd_builtin_types): Likewise.
18825 (aarch64_simd_expand_args): Likewise.
18826 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18828 (aarch64_reverse_mask): Likewise.
18829 (aarch64_simd_emit_reg_reg_move): Likewise.
18830 (aarch64_gen_adjusted_ldpstp): Likewise.
18831 (aarch64_ccmp_mode_to_code): Likewise.
18832 (aarch64_operands_ok_for_ldpstp): Likewise.
18833 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18834 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18836 (aarch64_min_divisions_for_recip_mul): Likewise.
18837 (aarch64_reassociation_width): Likewise.
18838 (aarch64_get_condition_code_1): Likewise.
18839 (aarch64_simd_emit_reg_reg_move): Likewise.
18840 (aarch64_simd_attr_length_rglist): Likewise.
18841 (aarch64_reverse_mask): Likewise.
18842 (aarch64_operands_ok_for_ldpstp): Likewise.
18843 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18844 (aarch64_gen_adjusted_ldpstp): Likewise.
18845 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18847 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18848 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18849 (arm_lookup_simd_builtin_type): Likewise.
18850 (arm_simd_builtin_type): Likewise.
18851 (arm_init_simd_builtin_types): Likewise.
18852 (arm_expand_builtin_args): Likewise.
18853 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18854 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18855 (ft32_setup_incoming_varargs): Likewise.
18856 (ft32_function_arg): Likewise.
18857 (ft32_function_arg_advance): Likewise.
18858 (ft32_pass_by_reference): Likewise.
18859 (ft32_arg_partial_bytes): Likewise.
18860 (ft32_valid_pointer_mode): Likewise.
18861 (ft32_addr_space_pointer_mode): Likewise.
18862 (ft32_addr_space_legitimate_address_p): Likewise.
18863 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18865 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18866 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18867 (iamcu_alignment): Likewise.
18868 (canonicalize_vector_int_perm): Likewise.
18869 (ix86_noce_conversion_profitable_p): Likewise.
18870 (ix86_mpx_bound_mode): Likewise.
18871 (ix86_operands_ok_for_move_multiple): Likewise.
18872 * config/microblaze/microblaze-protos.h
18873 (microblaze_expand_conditional_branch_reg): Likewise.
18874 * config/microblaze/microblaze.c
18875 (microblaze_expand_conditional_branch_reg): Likewise.
18876 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18878 (rs6000_reassociation_width): Likewise.
18879 (rs6000_invalid_binary_op): Likewise.
18880 (fusion_p9_p): Likewise.
18881 (emit_fusion_p9_load): Likewise.
18882 (emit_fusion_p9_store): Likewise.
18883 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18885 (riscv_hard_regno_mode_ok_p): Likewise.
18886 (riscv_address_insns): Likewise.
18887 (riscv_split_symbol): Likewise.
18888 (riscv_legitimize_move): Likewise.
18889 (riscv_function_value): Likewise.
18890 (riscv_hard_regno_nregs): Likewise.
18891 (riscv_expand_builtin): Likewise.
18892 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18893 (riscv_build_integer): Likewise.
18894 (riscv_split_integer): Likewise.
18895 (riscv_legitimate_constant_p): Likewise.
18896 (riscv_cannot_force_const_mem): Likewise.
18897 (riscv_regno_mode_ok_for_base_p): Likewise.
18898 (riscv_valid_base_register_p): Likewise.
18899 (riscv_valid_offset_p): Likewise.
18900 (riscv_valid_lo_sum_p): Likewise.
18901 (riscv_classify_address): Likewise.
18902 (riscv_legitimate_address_p): Likewise.
18903 (riscv_address_insns): Likewise.
18904 (riscv_load_store_insns): Likewise.
18905 (riscv_force_binary): Likewise.
18906 (riscv_split_symbol): Likewise.
18907 (riscv_force_address): Likewise.
18908 (riscv_legitimize_address): Likewise.
18909 (riscv_move_integer): Likewise.
18910 (riscv_legitimize_const_move): Likewise.
18911 (riscv_legitimize_move): Likewise.
18912 (riscv_address_cost): Likewise.
18913 (riscv_subword): Likewise.
18914 (riscv_output_move): Likewise.
18915 (riscv_canonicalize_int_order_test): Likewise.
18916 (riscv_emit_int_order_test): Likewise.
18917 (riscv_function_arg_boundary): Likewise.
18918 (riscv_pass_mode_in_fpr_p): Likewise.
18919 (riscv_pass_fpr_single): Likewise.
18920 (riscv_pass_fpr_pair): Likewise.
18921 (riscv_get_arg_info): Likewise.
18922 (riscv_function_arg): Likewise.
18923 (riscv_function_arg_advance): Likewise.
18924 (riscv_arg_partial_bytes): Likewise.
18925 (riscv_function_value): Likewise.
18926 (riscv_pass_by_reference): Likewise.
18927 (riscv_setup_incoming_varargs): Likewise.
18928 (riscv_print_operand): Likewise.
18929 (riscv_elf_select_rtx_section): Likewise.
18930 (riscv_save_restore_reg): Likewise.
18931 (riscv_for_each_saved_reg): Likewise.
18932 (riscv_register_move_cost): Likewise.
18933 (riscv_hard_regno_mode_ok_p): Likewise.
18934 (riscv_hard_regno_nregs): Likewise.
18935 (riscv_class_max_nregs): Likewise.
18936 (riscv_memory_move_cost): Likewise.
18937 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18938 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18939 (rl78_addr_space_address_mode): Likewise.
18940 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18942 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18943 (rs6000_reassociation_width): Likewise.
18944 (rs6000_invalid_binary_op): Likewise.
18945 (fusion_p9_p): Likewise.
18946 (emit_fusion_p9_load): Likewise.
18947 (emit_fusion_p9_store): Likewise.
18948 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18949 (ok_for_simple_move_operands): Likewise.
18950 (ok_for_simple_move_strict_operands): Likewise.
18951 (ok_for_simple_arith_logic_operands): Likewise.
18952 (visium_legitimize_reload_address): Likewise.
18953 (visium_select_cc_mode): Likewise.
18954 (output_cbranch): Likewise.
18955 (visium_split_double_move): Likewise.
18956 (visium_expand_copysign): Likewise.
18957 (visium_expand_int_cstore): Likewise.
18958 (visium_expand_fp_cstore): Likewise.
18959 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18960 (visium_function_arg): Likewise.
18961 (visium_function_arg_advance): Likewise.
18962 (visium_libcall_value): Likewise.
18963 (visium_setup_incoming_varargs): Likewise.
18964 (visium_legitimate_constant_p): Likewise.
18965 (visium_legitimate_address_p): Likewise.
18966 (visium_legitimize_address): Likewise.
18967 (visium_secondary_reload): Likewise.
18968 (visium_register_move_cost): Likewise.
18969 (visium_memory_move_cost): Likewise.
18970 (prepare_move_operands): Likewise.
18971 (ok_for_simple_move_operands): Likewise.
18972 (ok_for_simple_move_strict_operands): Likewise.
18973 (ok_for_simple_arith_logic_operands): Likewise.
18974 (visium_function_value_1): Likewise.
18975 (rtx_ok_for_offset_p): Likewise.
18976 (visium_legitimize_reload_address): Likewise.
18977 (visium_split_double_move): Likewise.
18978 (visium_expand_copysign): Likewise.
18979 (visium_expand_int_cstore): Likewise.
18980 (visium_expand_fp_cstore): Likewise.
18981 (visium_split_cstore): Likewise.
18982 (visium_select_cc_mode): Likewise.
18983 (visium_split_cbranch): Likewise.
18984 (output_cbranch): Likewise.
18985 (visium_print_operand_address): Likewise.
18986 * expmed.c (flip_storage_order): Likewise.
18987 * expmed.h (emit_cstore): Likewise.
18988 (flip_storage_order): Likewise.
18989 * genrecog.c (validate_pattern): Likewise.
18990 * hsa-gen.c (gen_hsa_addr): Likewise.
18991 * internal-fn.c (expand_arith_overflow): Likewise.
18992 * ira-color.c (allocno_copy_cost_saving): Likewise.
18993 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18994 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18995 (process_invariant_for_inheritance): Likewise.
18996 * lra-eliminations.c (move_plus_up): Likewise.
18997 * omp-low.c (lower_oacc_reductions): Likewise.
18998 * simplify-rtx.c (simplify_subreg): Likewise.
18999 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19000 (TARGET_CHKP_BOUND_MODE): Likewise..
19001 * targhooks.c (default_chkp_bound_mode): Likewise.
19002 (default_setup_incoming_vararg_bounds): Likewise.
19003 * targhooks.h (default_chkp_bound_mode): Likewise.
19004 (default_setup_incoming_vararg_bounds): Likewise.
19005 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19006 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19007 (have_whole_vector_shift): Likewise.
19008 * tree-vect-stmts.c (vectorizable_load): Likewise.
19009 * doc/tm.texi: Regenerate.
19011 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19013 Graceful degrade if Binutils PR21472 is not available.
19016 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19017 .rodata in flash test fails.
19018 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19019 * confgure: Regenerate.
19020 * config.in: Regenerate.
19021 * config/avr/avr.c (avr_asm_named_section)
19022 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19023 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19024 (avr_asm_init_sections): Same.
19026 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19028 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19029 (fma<VH:mode>4_intrinsic): Likewise.
19030 (*fmsub<VCVTF:mode>4): Likewise.
19031 (*fmsub<VH:mode>4_intrinsic): Likewise.
19033 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19036 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19037 Don't depend on "optimize > 0".
19038 (out_movhi_r_mr, out_movqi_mr_r): Same.
19039 (out_movhi_mr_r, out_movqi_r_mr): Same.
19040 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19041 io_address_operand on "optimize > 0".
19043 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19045 * tree-loop-distribution.c: Add general explanantion on the pass.
19046 (generate_loops_for_partition): Mark distributed loop.
19047 (pg_add_dependence_edges): New parameter. Handle alias data
19048 dependence specially and record it in the parameter if asked.
19049 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19050 (init_partition_graph_vertices, add_partition_graph_edge): New.
19051 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19052 (free_partition_graph_vdata, build_partition_graph): New.
19053 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19054 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19055 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19056 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19057 (version_for_distribution_p, finalize_partitions): New.
19058 (distribute_loop): Handle alias data dependence specially. Factor
19059 out loop fusion code as functions and call these functions.
19061 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19063 * tree-loop-distribution.c (classify_partition): New parameter and
19064 better handle reduction statement.
19065 (rdg_build_partitions): Revise comment.
19066 (distribute_loop): Compute statements in all partitions and pass it
19067 to classify_partition.
19069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19071 * tree-loop-distribution.c (enum partition_type): New.
19072 (struct partition): New field type.
19073 (partition_merge_into): Add parameter. Update partition type.
19074 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19075 (build_rdg_partition_for_vertex): Compute partition type.
19076 (rdg_build_partitions): Dump partition type.
19077 (distribute_loop): Update calls to partition_merge_into.
19079 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19081 * tree-loop-distribution.c (struct ddr_hasher): New.
19082 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19084 (classify_partition): Call get_data_dependence.
19085 (pg_add_dependence_edges): Ditto.
19086 (distribute_loop): Release data dependence hash table.
19088 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19090 * tree-loop-distribution.c (ref_base_address): Delete.
19091 (similar_memory_accesses): Rename ...
19092 (share_memory_accesses): ... to this. Check if partitions access
19093 the same memory reference.
19094 (distribute_loop): Call share_memory_accesses.
19096 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19098 * tree-loop-distribution.c (struct partition): New field recording
19099 its data reference.
19100 (partition_alloc, partition_free): Init and release data refs.
19101 (partition_merge_into): Merge data refs.
19102 (build_rdg_partition_for_vertex): Collect data refs for partition.
19103 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19105 (distribute_loop): Remve data refs from vertice data of partition
19108 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19110 * tree-loop-distribution.c (params.h): Include header file.
19111 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19112 (datarefs_vec): New global var.
19113 (create_rdg_vertices): Use datarefs_vec directly.
19114 (free_rdg): Don't free data references.
19115 (build_rdg): Update use. Don't free data references.
19116 (distribute_loop): Compute global variable for data references.
19117 Bail out if there are too many data references.
19119 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19121 * tree-loop-distribution.c (loop_nest): New global var.
19122 (build_rdg): Use loop directly, rather than loop nest.
19123 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19125 (distribute_loop): Compute global variable loop nest. Update use.
19127 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19129 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19130 (partition_merge_into): New parameter. Dump reason for fusion.
19131 (distribute_loop): Update use of partition_merge_into.
19133 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19135 * tree-loop-distribution.c (bb_top_order_index): New.
19136 (bb_top_order_index_size, bb_top_order_cmp): New.
19137 (stmts_from_loop): Use topological order.
19138 (pass_loop_distribution::execute): Compute and release topological
19139 order for basic blocks.
19141 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19143 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19146 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19148 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19149 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19150 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19151 * internal-fn.def (LOOP_DIST_ALIAS): New.
19152 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19153 (fold_loop_internal_call): ... this.
19154 (vect_loop_dist_alias_call): New function.
19155 (set_uid_loop_bbs): Call fold_loop_internal_call.
19156 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19159 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19162 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19163 Require dead FLAGS_REG at the beginning of a peephole.
19165 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19168 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19169 arguments in the call to __builtin_ia32_sbb_u32.
19170 (_subborrow_u64): Swap _X and _Y arguments in the call to
19171 __builtin_ia32_sbb_u64.
19173 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19176 * tree-vrp.c (compare_assert_loc): Turn into a function template
19177 with stable template parameter. Only test if a->e is NULL,
19178 !a->e == !b->e has been verified already. Use e == NULL or
19179 e != NULL instead of e or ! e tests. If stable is true, don't use
19180 iterative_hash_expr, on the other side allow a or b or both NULL
19181 and sort the NULLs last.
19182 (process_assert_insertions): Sort using compare_assert_loc<false>
19183 instead of compare_assert_loc, later sort using
19184 compare_assert_loc<true> before calling process_assert_insertions_for
19185 in a loop. Use break instead of continue once seen NULL pointer.
19187 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19189 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19190 Cortex-R7 and Cortex-R8 processors.
19192 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19194 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19195 uninitialized while src is not.
19197 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19199 * common/config/arm/arm-common.c: Adjust include path for
19201 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19202 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19203 (arm-cpu-data.h): Likewise.
19204 (arm-cpu-cdata.h): Likewise.
19205 * config/arm/arm-cpu.h: Delete.
19206 * config/arm/arm-cpu-cdata.h: Delete.
19207 * config/arm/arm-cpu-data.h: Delete.
19209 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19211 * config/arm/arm-cpus.in (cortex-a55): New.
19212 (cortex-a75): Likewise.
19213 (cortex-a75.cortex-a55): Likewise.
19214 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19216 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19217 * config/arm/arm-cpu-cdata.h: Regenerate.
19218 * config/arm/arm-cpu-data.h: Regenerate.
19219 * config/arm/arm-cpu.h: Regenerate.
19220 * config/arm/arm-tables.opt: Regenerate.
19221 * config/arm/arm-tune.md: Regenerate.
19223 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19225 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19227 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19229 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19232 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19234 PR tree-optimization/81292
19235 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19236 full_string_p, also call adjust_related_strinfos if the adjustment
19237 is simple, otherwise invalidate related strinfos.
19239 2017-07-04 Martin Liska <mliska@suse.cz>
19242 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19243 newly created variable as DECL_IGNORED_P.
19245 2017-07-04 Martin Liska <mliska@suse.cz>
19248 * ipa-inline.c (inline_small_functions):
19249 Use xstrdup_for_dump.
19251 2017-07-04 Tom de Vries <tom@codesourcery.com>
19253 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19255 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19258 * config/darwin.c (darwin_function_switched_text_sections):
19261 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19263 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19265 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19267 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19269 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19271 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19272 min_profitable_iters, and th as inclusive lower bounds.
19273 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19274 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19275 for min_profitable_iters and min_profitable_estimate.
19276 (vect_transform_loop): Treat th as an inclusive lower bound.
19277 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19279 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19282 * config/darwin.c (darwin_function_switched_text_sections):
19283 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19284 in two pieces, and suppress the use of buf.
19286 2017-07-03 Nathan Sidwell <nathan@acm.org>
19288 * hash-table.h (hash_table_mod1): Fix indentation.
19290 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19292 PR middle-end/81290
19293 * predict.c (force_edge_cold): Be more careful about propagation
19295 * profile-count.h (profile_probability::guessed,
19296 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19298 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19300 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19302 * doc/invoke.texi (rcpc architecture extension): Document it.
19304 2017-07-03 Richard Biener <rguenther@suse.de>
19306 PR tree-optimization/60510
19307 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19308 the scalar reduction PHI and use it.
19309 (vectorizable_reduction): Properly guard the single_defuse_cycle
19310 path for non-SLP reduction chains where we cannot use it.
19311 Rework reduc_def/index and vector type deduction. Rework
19312 vector operand gathering during reduction op code-gen.
19313 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19314 chains dissolve the chain and leave it to non-SLP reduction
19317 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19319 * tree-data-ref.h (dr_alignment): Declare.
19320 * tree-data-ref.c (dr_alignment): New function.
19321 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19322 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19324 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19326 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19328 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19329 and base_misalignment fields.
19330 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19331 * tree-data-ref.c: Include builtins.h.
19332 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19333 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19334 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19335 * tree-vect-data-refs.c: Include tree-cfg.h.
19336 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19337 fields instead of calculating an alignment here.
19338 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19339 innermost_loop_behavior fields.
19341 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19343 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19345 (DR_STEP_ALIGNMENT): New macro.
19346 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19347 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19348 (create_data_ref): Print it.
19349 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19350 to tell whether the step preserves vector (mis)alignment.
19351 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19352 Move the check for an integer step and generalise to all INTEGER_CST.
19353 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19354 Print the outer step alignment.
19356 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19358 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19359 with offset_alignment.
19360 (DR_ALIGNED_TO): Delete.
19361 (DR_OFFSET_ALIGNMENT): New macro.
19362 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19363 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19364 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19365 (create_data_ref): Likewise.
19366 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19367 (vect_analyze_data_refs): Likewise.
19368 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19369 creating dummy innermost behavior.
19371 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19373 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19374 with a "innermost_loop_behavior *" and refeence tree.
19375 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19376 (create_data_ref): Update call accordingly.
19377 * tree-predcom.c (find_looparound_phi): Likewise.
19379 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19381 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19382 fields with dr_wrt_vec_loop.
19383 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19384 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19385 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19386 (vect_dr_behavior): New function.
19387 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19388 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19389 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19390 track whether the step preserves the misalignment.
19391 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19392 Use vect_dr_behavior.
19393 (vect_setup_realignment): Update call accordingly.
19394 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19395 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19396 call to vect_create_addr_base_for_vector_ref.
19397 (vect_create_cond_for_align_checks): Likewise.
19398 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19399 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19400 (vect_recog_mask_conversion_pattern): Likewise.
19401 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19402 (new_stmt_vec_info): Remove redundant zeroing.
19404 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19406 * common/config/arm/arm-common.c (arm_be8_option): New function.
19407 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19408 (ISA_ARMv6): Add isa_bit_be8.
19409 * config/arm/arm.h (arm_be8_option): Add prototype.
19410 (BE8_SPEC_FUNCTION): New define.
19411 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19412 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19413 (mlittle-endian): Similarly.
19414 (mbe8, mbe32): New options.
19415 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19416 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19418 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19420 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19422 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19424 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19425 (cleanup_tree_cfg_bb): Use it.
19426 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19428 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19430 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19433 * loop-doloop.c (add_test): Update profile.
19435 2017-07-03 Martin Liska <mliska@suse.cz>
19438 * sanopt.c (rewrite_usage_of_param): New function.
19439 (sanitize_rewrite_addressable_params): Likewise.
19440 (pass_sanopt::execute): Call rewrite_usage_of_param.
19442 2017-07-03 Richard Biener <rguenther@suse.de>
19444 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19445 back to using VIEW_CONVERT_EXPR.
19447 2017-07-03 Martin Liska <mliska@suse.cz>
19450 * doc/extend.texi: Document when a resolver function is
19451 generated for target_clones.
19453 2017-07-03 Martin Liska <mliska@suse.cz>
19455 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19456 and shadow memory of auto variables which are subject of
19457 use-after-scope sanitization.
19458 (asan_expand_mark_ifn): Add do set only when is_poison.
19460 2016-07-03 Richard Biener <rguenther@suse.de>
19462 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19464 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19465 (vectorizable_reduction): Perform reduction PHI creation when
19466 visiting a reduction PHI and adjust and simplify code generation
19467 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19468 (vect_transform_loop): Visit reduction PHIs.
19469 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19470 defs into the SLP tree.
19471 (vect_build_slp_tree): Reduction defs terminate the recursion.
19472 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19474 (vect_get_vec_defs_for_stmt_copy): Export.
19475 (vect_get_vec_defs): Likewise.
19476 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19478 (vect_get_vec_defs_for_stmt_copy): Declare.
19479 (vect_get_vec_defs): Likewise.
19481 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19483 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19484 parameter with a "loop" parameter and use it instead of the
19485 loop containing DR_STMT. Don't check simple_iv when doing
19486 BB analysis. Describe the two analysis modes in the comment.
19488 2017-07-03 Tom de Vries <tom@codesourcery.com>
19490 PR tree-optimization/69468
19491 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19492 (find_same_succ_bb): Handle ignore_edge_flags.
19494 2017-07-03 Tom de Vries <tom@codesourcery.com>
19496 PR tree-optimization/81192
19497 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19499 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19501 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19503 2017-07-03 Tom de Vries <tom@codesourcery.com>
19505 PR tree-optimization/81192
19506 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19507 BB_SAME_SUCC (bb) == NULL.
19509 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19511 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19514 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19516 * dumpfile.c: Include profile-count.h
19517 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19519 (insert_cond_bb): Update profile.
19520 * tree-cfg.h (insert_cond_bb): Update prototype.
19521 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19522 * tree-dump.c: Do not include tree-cfg.
19524 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19526 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19528 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19530 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19533 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19535 * tree-complex.c (expand_complex_div_wide): update profile.
19537 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19538 Alan Hayward <alan.hayward@arm.com>
19539 David Sherwood <david.sherwood@arm.com>
19541 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19542 (CORETYPES_H): New define.
19543 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19544 (insn-modes-inline.h, s-modes-inline-h): New rules.
19545 (generated_files): Add insn-modes-inline.h.
19546 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19547 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19548 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19549 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19550 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19551 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19552 (build/gencodes.o, build/genconditions.o): Likewise.
19553 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19554 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19555 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19556 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19557 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19558 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19559 * coretypes.h: Include everything up to real.h for generators.
19560 Include insn-modes.h first. Include wide-int-print.h after
19561 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19562 * machmode.h: Don't include insn-modes.h here.
19563 * function-tests.c: Remove includes of signop.h, machmode.h,
19564 double-int.h and wide-int.h.
19566 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19568 * optc-save-gen.awk: Likewise.
19569 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19570 * godump.c: Remove include of wide-int-print.h.
19571 * pretty-print.h: Likewise.
19572 * wide-int-print.cc: Likewise.
19573 * wide-int.cc: Likewise.
19574 * hash-map-tests.c: Remove include of signop.h.
19575 * hash-set-tests.c: Likewise.
19576 * rtl-tests.c: Likewise.
19577 * mkconfig.sh: Remove include of machmode.h.
19578 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19580 (emit_insn_modes_inline_h): ...this new function. Emit the code
19581 into an insn-modes-inline.h header file, adding appropriate
19582 include guards and end comments.
19583 (emit_insn_modes_c_header): Remove include of machmode.h.
19584 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19586 (main): Handle -i flag and call emit_insn_modes_inline_h when
19589 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19591 * tree-ssa-strlen.c (strinfo): Rename the length field to
19592 nonzero_chars. Add a full_string_p field.
19593 (compare_nonzero_chars, zero_length_string_p): New functions.
19594 (get_addr_stridx): Add an offset_out parameter.
19595 Use compare_nonzero_chars.
19596 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19597 (new_strinfo): Update after above changes to strinfo.
19598 (set_endptr_and_length): Set full_string_p.
19599 (get_string_length): Update after above changes to strinfo.
19600 (unshare_strinfo): Update call to new_strinfo.
19601 (maybe_invalidate): Likewise.
19602 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19603 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19604 as a uhwi instead of an shwi. Update after above changes to
19605 strinfo and new_strinfo.
19606 (zero_length_string): Assert that chainsi contains full strings.
19607 Use zero_length_string_p. Update call to new_strinfo.
19608 (adjust_related_strinfos): Update after above changes to strinfo.
19609 Copy full_string_p from origsi.
19610 (adjust_last_stmt): Use zero_length_string_p.
19611 (handle_builtin_strlen): Update after above changes to strinfo and
19612 new_strinfo. Install the lhs as the string length if the previous
19613 entry didn't describe a full string.
19614 (handle_builtin_strchr): Update after above changes to strinfo
19616 (handle_builtin_strcpy): Likewise.
19617 (handle_builtin_strcat): Likewise.
19618 (handle_builtin_malloc): Likewise.
19619 (handle_pointer_plus): Likewise.
19620 (handle_builtin_memcpy): Likewise. Track nonzero characters
19621 that aren't necessarily followed by a nul terminator.
19622 (handle_char_store): Likewise.
19624 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19626 PR tree-optimization/80769
19627 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19628 for malloc and calloc. Document the new invariant that all related
19629 strinfos have delayed lengths or none do.
19630 (verify_related_strinfos): Move earlier in file.
19631 (set_endptr_and_length): New function, split out from...
19632 (get_string_length): ...here. Also set the lengths of related
19634 (zero_length_string): Assert that chainsi has known (rather than
19636 (adjust_related_strinfos): Likewise.
19638 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19640 PR tree-optimization/81136
19641 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19642 assert that two references with the same misalignment have the same
19643 compile-time misalignment if those compile-time misalignments
19646 2017-07-01 Andi Kleen <ak@linux.intel.com>
19648 * print-tree.c (print_node): Print all attributes.
19650 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19652 * cfg.c (scale_bbs_frequencies): New function.
19653 * cfg.h (scale_bbs_frequencies): Declare it.
19654 * cfgloopanal.c (single_likely_exit): Cleanup.
19655 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19657 (scale_loop_profile): Likewise.
19658 (loop_version): Likewise.
19659 (create_empty_loop_on_edge): Update.
19660 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19661 scale_loop_frequencies, scale_loop_profile, loopify,
19662 loop_version): Update prototypes.
19663 * modulo-sched.c (sms_schedule): Update.
19664 * predict.c (unlikely_executed_edge_p): Also check probability.
19665 (probably_never_executed_edge_p): Fix typo.
19666 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19667 * tree-parloops.c (gen_parallel_loop): Update.
19668 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19669 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19670 * tree-ssa-loop-split.c (split_loop): Update.
19671 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19672 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19673 (vect_loop_versioning): Update.
19674 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19676 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19678 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19680 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19682 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19683 to keep profile consistent.
19685 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19687 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19688 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19689 * profile-count.h (max_safe_multiplier): Make unsigned.
19690 (profile_count::guessed_zero): New.
19692 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19694 * bb-reorder.c (fix_up_crossing_landing_pad,
19695 fix_crossing_conditional_branches): Use make_single_succ_edge
19696 to keep profile consistent.
19698 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19700 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19703 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19706 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19707 the right scopes, make sure cond_jump isn't preserved between multiple
19708 iterations. Search for fallthru edge whenever there are 3+ edges and
19709 use find_fallthru_edge for it.
19711 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19713 Patch by Alexander Monakov <amonakov@ispras.ru>
19714 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19715 probabilities consistently.
19717 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19719 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19720 emit_cmp_and_jump_insns.
19722 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19725 * tree-inline.c (expand_call_inline): Combine profile statuses.
19727 2017-06-30 Andrew Pinski <apinski@cavium.com>
19729 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19730 fold_stmt returned true.
19732 2017-06-30 Nathan Sidwell <nathan@acm.org>
19734 * ggc.h (empty_string): Delete.
19735 * cfgexpand.c (expand_asm_stmt): Use plain "".
19736 * optabs.c (expand_asm_memory_barrier): Likewise.
19737 * stringpool.c (empty_string): Delete.
19738 (digit_vector, digit_string): Delete.
19739 (ggc_alloc_string): Use plain "", don't optimize single digit
19740 strings. Use ggc_alloc_atomic.
19742 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19744 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19745 comparison set and one other set, use the cost of the non-comparison
19748 2017-06-30 Nathan Sidwell <nathan@acm.org>
19750 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19753 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19755 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19756 loops. Remove now unneeded calls to gimple_switch_set_label() that
19757 just set removed labels to NULL_TREE.
19759 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19761 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19762 (set_range_info): ...here. Only call set_range_info_raw if domain
19764 (set_nonzero_bits): Call set_range_info_raw.
19765 * tree-ssanames.h (set_range_info_raw): New.
19767 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19770 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19771 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19772 of nonimmediate_operand and <store_mask_constraint> instead of m
19773 for the input operand. For V8FI iterator, always split if input
19774 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19775 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19776 <store_mask_predicate> instead of register_operand and
19777 <store_mask_constraint> instead of v for the input operand. Make
19778 sure both operands aren't MEMs for if not <mask_applied>.
19780 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19782 * lto-wrapper.c (copy_file) Close both file descriptors before
19785 2017-06-30 Martin Liska <mliska@suse.cz>
19788 * multiple_target.c (create_dispatcher_calls): Make ifunc
19789 also for function that don't have calls or are not referenced.
19791 2017-06-30 Richard Biener <rguenther@suse.de>
19793 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19794 analyze the first scalar stmt. Move vector type computation
19795 for the BB case here from ...
19796 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19797 live operation processing in the SLP case properly.
19799 2017-06-30 Richard Biener <rguenther@suse.de>
19801 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19803 2017-06-30 Martin Liska <mliska@suse.cz>
19806 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19807 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19809 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19811 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19814 2017-06-30 Richard Biener <rguenther@suse.de>
19816 PR tree-optimization/81249
19817 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19818 condition reduction result to original scalar type.
19820 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19822 * profile-count.h (enum profile_quality): Fix typos and whitespace
19825 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19827 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19828 type for branch probabilities.
19830 2017-06-29 Julian Brown <julian@codesourcery.com>
19831 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19833 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19834 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19835 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19836 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19838 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19840 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19841 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19842 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19843 CC usage from generic code to here.
19844 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19845 CC usage into the target macros.
19847 2017-06-29 Maya Rashish <coypu@sdf.org>
19849 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19852 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19854 * arm/arm-builtins.c: Include profile-count.h
19855 * except.c (sjlj_emit_function_enter): Use
19856 profile_probability::unlikely.
19858 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19860 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19861 and tocrel_offset be pointer args rather than implicitly using
19863 (legitimate_constant_pool_address_p, rs6000_emit_move,
19864 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19865 tocrel_offset and use in toc_relative_expr_p call.
19866 (print_operand, print_operand_address): Use static tocrel_base_oac
19867 and tocrel_offset_oac.
19868 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19871 2017-06-29 Maya Rashish <coypu@sdf.org>
19873 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19875 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19877 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19878 objects, take into account only the alignment of 'op0' and 'mode1' if
19881 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19883 * ccmp.c (ccmp_tree_comparison_p): New function.
19884 (ccmp_candidate_p): Update to use above function.
19885 (get_compare_parts): New function.
19886 (expand_ccmp_next): Update to use new functions.
19887 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19889 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19890 take mode as argument.
19891 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19892 * expr.c (expand_expr_real_1): Pass mode as argument.
19894 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19896 * combine.c (combine_instructions): Print insns to dump_file, together
19899 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19901 * asan.c (asan_emit_stack_protection): Update.
19902 (create_cond_insert_point): Update.
19903 * auto-profile.c (afdo_propagate_circuit): Update.
19904 * basic-block.h (struct edge_def): Turn probability to
19905 profile_probability.
19906 (EDGE_FREQUENCY): Update.
19907 * bb-reorder.c (find_traces_1_round): Update.
19908 (better_edge_p): Update.
19909 (sanitize_hot_paths): Update.
19910 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19911 (make_single_succ_edge): Update.
19912 (check_bb_profile): Update.
19913 (dump_edge_info): Update.
19914 (update_bb_profile_for_threading): Update.
19915 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19916 probabilitycount to 0.
19917 * cfgbuild.c (compute_outgoing_frequencies): Update.
19918 * cfgcleanup.c (try_forward_edges): Update.
19919 (outgoing_edges_match): Update.
19920 (try_crossjump_to_edge): Update.
19921 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19922 (expand_gimple_tailcall): Update.
19923 (construct_init_block): Use make_single_succ_edge.
19924 (construct_exit_block): Use make_single_succ_edge.
19925 * cfghooks.c (verify_flow_info): Update.
19926 (redirect_edge_succ_nodup): Update.
19927 (split_edge): Update.
19928 (account_profile_record): Update.
19929 * cfgloopanal.c (single_likely_exit): Update.
19930 * cfgloopmanip.c (scale_loop_profile): Update.
19931 (set_zero_probability): Remove.
19932 (duplicate_loop_to_header_edge): Update.
19933 * cfgloopmanip.h (loop_version): Update prototype.
19934 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19935 (force_nonfallthru_and_redirect): Update.
19936 (update_br_prob_note): Update.
19937 (rtl_verify_edges): Update.
19938 (purge_dead_edges): Update.
19939 (rtl_lv_add_condition_to_bb): Update.
19940 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19941 * cgraphunit.c (init_lowered_empty_function): Update.
19942 (cgraph_node::expand_thunk): Update.
19943 * cilk-common.c: Include profile-count.h
19944 * dojump.c (inv): Remove.
19945 (jumpifnot): Update.
19946 (jumpifnot_1): Update.
19947 (do_jump_1): Update.
19949 (do_jump_by_parts_greater_rtx): Update.
19950 (do_compare_rtx_and_jump): Update.
19951 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19952 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19953 * dwarf2cfi.c: Include profile-count.h
19954 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19955 (sjlj_emit_dispatch_table): Likewise.
19956 * explow.c: Include profile-count.h
19957 * expmed.c (emit_store_flag_force): Update.
19958 (do_cmp_and_jump): Update.
19959 * expr.c (compare_by_pieces_d::generate): Update.
19960 (compare_by_pieces_d::finish_mode): Update.
19961 (emit_block_move_via_loop): Update.
19962 (store_expr_with_bounds): Update.
19963 (store_constructor): Update.
19964 (expand_expr_real_2): Update.
19965 (expand_expr_real_1): Update.
19966 * expr.h (try_casesi, try_tablejump): Update prototypes.
19967 * gimple-pretty-print.c (dump_probability): Update.
19968 (dump_profile): New.
19969 (dump_gimple_label): Update.
19970 (dump_gimple_bb_header): Update.
19971 * graph.c (draw_cfg_node_succ_edges): Update.
19972 * hsa-gen.c (convert_switch_statements): Update.
19973 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19974 (find_if_case_1): Update.
19975 (find_if_case_2): Update.
19976 * internal-fn.c (expand_arith_overflow_result_store): Update.
19977 (expand_addsub_overflow): Update.
19978 (expand_neg_overflow): Update.
19979 (expand_mul_overflow): Update.
19980 (expand_vector_ubsan_overflow): Update.
19981 * ipa-cp.c (good_cloning_opportunity_p): Update.
19982 * ipa-split.c (split_function): Use make_single_succ_edge.
19983 * ipa-utils.c (ipa_merge_profiles): Update.
19984 * loop-doloop.c (add_test): Update.
19985 (doloop_modify): Update.
19986 * loop-unroll.c (compare_and_jump_seq): Update.
19987 (unroll_loop_runtime_iterations): Update.
19988 * lra-constraints.c (lra_inheritance): Update.
19989 * lto-streamer-in.c (input_cfg): Update.
19990 * lto-streamer-out.c (output_cfg): Update.
19991 * mcf.c (adjust_cfg_counts): Update.
19992 * modulo-sched.c (sms_schedule): Update.
19993 * omp-expand.c (expand_omp_for_init_counts): Update.
19994 (extract_omp_for_update_vars): Update.
19995 (expand_omp_ordered_sink): Update.
19996 (expand_omp_for_ordered_loops): Update.
19997 (expand_omp_for_generic): Update.
19998 (expand_omp_for_static_nochunk): Update.
19999 (expand_omp_for_static_chunk): Update.
20000 (expand_cilk_for): Update.
20001 (expand_omp_simd): Update.
20002 (expand_omp_taskloop_for_outer): Update.
20003 (expand_omp_taskloop_for_inner): Update.
20004 * omp-simd-clone.c (simd_clone_adjust): Update.
20005 * optabs.c (expand_doubleword_shift): Update.
20006 (expand_abs): Update.
20007 (emit_cmp_and_jump_insn_1): Update.
20008 (expand_compare_and_swap_loop): Update.
20009 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20010 * predict.c (predictable_edge_p): Update.
20011 (edge_probability_reliable_p): Update.
20012 (set_even_probabilities): Update.
20013 (combine_predictions_for_insn): Update.
20014 (combine_predictions_for_bb): Update.
20015 (propagate_freq): Update.
20016 (estimate_bb_frequencies): Update.
20017 (force_edge_cold): Update.
20018 * profile-count.c (profile_count::dump): Add missing space into dump.
20019 (profile_count::debug): Add newline.
20020 (profile_count::differs_from_p): Explicitly convert to unsigned.
20021 (profile_count::stream_in): Update.
20022 (profile_probability::dump): New member function.
20023 (profile_probability::debug): New member function.
20024 (profile_probability::differs_from_p): New member function.
20025 (profile_probability::differs_lot_from_p): New member function.
20026 (profile_probability::stream_in): New member function.
20027 (profile_probability::stream_out): New member function.
20028 * profile-count.h (profile_count_quality): Rename to ...
20029 (profile_quality): ... this one.
20030 (profile_probability): New.
20031 (profile_count): Update.
20032 * profile.c (compute_branch_probabilities): Update.
20033 * recog.c (peep2_attempt): Update.
20034 * sched-ebb.c (schedule_ebbs): Update.
20035 * sched-rgn.c (find_single_block_region): Update.
20036 (compute_dom_prob_ps): Update.
20037 (schedule_region): Update.
20038 * sel-sched-ir.c (compute_succs_info): Update.
20039 * stmt.c (struct case_node): Update.
20040 (do_jump_if_equal): Update.
20041 (get_outgoing_edge_probs): Update.
20042 (conditional_probability): Update.
20043 (emit_case_dispatch_table): Update.
20044 (expand_case): Update.
20045 (expand_sjlj_dispatch_table): Update.
20046 (emit_case_nodes): Update.
20047 * targhooks.c: Update.
20048 * tracer.c (better_p): Update.
20049 (find_best_successor): Update.
20050 * trans-mem.c (expand_transaction): Update.
20051 * tree-call-cdce.c: Update.
20052 * tree-cfg.c (gimple_split_edge): Upate.
20053 (move_sese_region_to_fn): Upate.
20054 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20055 * tree-eh.c (lower_resx): Upate.
20056 (cleanup_empty_eh_move_lp): Upate.
20057 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20058 * tree-inline.c (copy_edges_for_bb): Update.
20059 (copy_cfg_body): Update.
20060 * tree-parloops.c (gen_parallel_loop): Update.
20061 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20062 (gimple_gen_time_profiler): Update.
20063 * tree-ssa-dce.c (remove_dead_stmt): Update.
20064 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20065 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20066 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20067 (unloop_loops): Update.
20068 (try_peel_loop): Update.
20069 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20070 * tree-ssa-loop-split.c (connect_loops): Update.
20071 (split_loop): Update.
20072 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20073 (hoist_guard): Update.
20074 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20075 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20076 (value_replacement): Update.
20077 * tree-ssa-reassoc.c (branch_fixup): Update.
20078 * tree-ssa-tail-merge.c (replace_block_by): Update.
20079 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20080 (create_edge_and_update_destination_phis): Update.
20081 (compute_path_counts): Update.
20082 (recompute_probabilities): Update.
20083 (update_joiner_offpath_counts): Update.
20084 (freqs_to_counts_path): Update.
20085 (duplicate_thread_path): Update.
20086 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20087 (struct switch_conv_info): Update.
20088 (gen_inbound_check): Update.
20089 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20090 (vect_do_peeling): Update.
20091 (vect_loop_versioning): Update.
20092 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20093 (optimize_mask_stores): Update.
20094 * ubsan.c (ubsan_expand_null_ifn): Update.
20095 * value-prof.c (gimple_divmod_fixed_value): Update.
20096 (gimple_divmod_fixed_value_transform): Update.
20097 (gimple_mod_pow2): Update.
20098 (gimple_mod_pow2_value_transform): Update.
20099 (gimple_mod_subtract): Update.
20100 (gimple_mod_subtract_transform): Update.
20101 (gimple_ic): Update.
20102 (gimple_stringop_fixed_value): Update.
20103 (gimple_stringops_transform): Update.
20104 * value-prof.h: Update.
20106 2017-06-29 Carl Love <cel@us.ibm.com>
20108 * config/rs6000/rs6000-c.c: Add support for built-in functions
20109 vector signed int vec_signed (vector float);
20110 vector signed long long vec_signed (vector double);
20111 vector signed int vec_signed2 (vector double, vector double);
20112 vector signed int vec_signede (vector double);
20113 vector signed int vec_signedo (vector double);
20114 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20115 instruction generator.
20116 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20117 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20118 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20120 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20121 vunsignede_v2df): Add define_expands.
20122 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20123 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20124 VEC_UNSIGNEDO): Add definitions.
20125 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20126 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20127 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20128 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20129 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20130 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20131 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20132 vec_unsignede, vec_unsignedo): Add builtin defines.
20133 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20135 * doc/extend.texi: Update the built-in documentation file for the
20136 new built-in functions.
20138 2017-06-29 Richard Biener <rguenther@suse.de>
20140 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20141 reduction chains to LOOP_VINFO_REDUCTIONS.
20142 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20143 SLP reductions after processing reduction chains.
20145 2017-06-29 Nathan Sidwell <nathan@acm.org>
20147 * builtins.c (fold_builtin_FUNCTION): Use
20148 lang_hooks.decl_printable_name.
20150 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20152 PR middle-end/81194
20153 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20154 with only one label.
20155 * stmt.c (expand_case): Assert NCASES is greater than one.
20157 2017-06-29 Richard Biener <rguenther@suse.de>
20159 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20161 (group_case_labels): Likewise.
20162 (find_taken_edge): Push sanity checking on val to workers...
20163 (find_taken_edge_cond_expr): ... here
20164 (find_taken_edge_switch_expr): ... and here, handle cases
20165 with just a default label.
20166 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20167 (group_case_labels): Likewise.
20168 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20169 group_case_labels does anything cleanup the CFG again.
20171 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20173 PR tree-optimization/81196
20174 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20175 exit condition comparing two IVs.
20177 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20179 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20180 profile to the dummy entry at the end of the list of architectures.
20181 * config/arm/arm-cpu-cdata.h: Regenerated.
20183 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20184 Michael Collison <michael.collison@arm.com>
20187 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20189 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20190 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20191 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20192 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20193 of shift when the shift amount is masked with constant equal to
20194 the size of the mode.
20195 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20198 2017-06-29 Martin Liska <mliska@suse.cz>
20200 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20202 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20205 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20208 (edge_growth_cache_entry::edge_growth_cache_entry): New
20210 (reset_edge_growth_cache): Update to use constructor.
20212 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20214 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20215 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20216 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20218 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20220 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20221 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20223 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20225 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20226 (*-linux-uclibc*): Add t-uclibc tmake_file.
20227 * config/t-musl: New.
20228 * config/t-uclibc: New.
20230 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20232 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20234 (gen_comm_data): Emit architectural setting of arch_prof.
20235 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20237 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20238 (armv8-m.base, armv8-m.main): Likewise.
20239 * arm-protos.h (arm_build_target): Add profile field.
20240 (arch_option): Likewise.
20241 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20243 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20244 arm_active_target.profile.
20246 2017-06-28 Richard Biener <rguenther@suse.de>
20248 PR middle-end/81227
20249 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20250 TYPE_OVERFLOW_WRAPS.
20251 * match.pd (negate_expr_p): Likewise.
20252 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20253 fold_build2, not fold_binary.
20255 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20257 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20258 Convert memory address to Pmode.
20259 (aarch64_print_operand): Assert MEM operands are always Pmode.
20261 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20264 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20265 Remove redundant if.
20266 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20267 * config/arm/aarch-common-protos.h
20268 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20269 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20271 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20274 * multiple_target.c (create_dispatcher_calls): Set the default
20275 clone to be static, not public.
20277 2017-06-28 Richard Biener <rguenther@suse.de>
20279 * tree-vect-loop.c (vectorizable_reduction): Move special
20280 cond reduction IV var creation ...
20281 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20282 parameter. Use STMT_VINFO_VECTYPE.
20283 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20286 2017-06-28 Martin Liska <mliska@suse.cz>
20289 * ipa-visibility.c (non_local_p): Handle visibility.
20291 2017-06-28 Martin Liska <mliska@suse.cz>
20294 * common.opt: Add IntegerRange to various options.
20295 * opt-functions.awk (integer_range_info): New function.
20296 * optc-gen.awk: Add integer_range_info to cl_options struct.
20297 * opts-common.c (decode_cmdline_option): Handle
20298 CL_ERR_INT_RANGE_ARG.
20299 (cmdline_handle_error): Likewise.
20300 * opts.c (print_filtered_help): Show valid interval in
20301 when --help is provided.
20302 * opts.h (struct cl_option): Add range_min and range_max fields.
20303 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20305 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20307 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20308 (x * C EQ/NE y * C): New transformation.
20310 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20312 * genmultilib (combination_space): Accept '+' in option names.
20314 2017-06-28 Martin Liska <mliska@suse.cz>
20317 * asan.c (instrument_derefs): Bail out inner references
20318 that are hard register variables.
20320 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20323 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20324 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20326 2017-06-28 Richard Biener <rguenther@suse.de>
20328 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20329 (vect_get_slp_defs): Adjust.
20330 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20332 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20334 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20335 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20336 (vectorizable_reduction): Adjust.
20337 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20339 (vect_get_slp_defs): Likewise.
20340 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20341 (vectorizable_bswap): Adjust.
20342 (vectorizable_call): Likewise.
20343 (vectorizable_conversion): Likewise.
20344 (vectorizable_assignment): Likewise.
20345 (vectorizable_shift): Likewise.
20346 (vectorizable_operation): Likewise.
20347 (vectorizable_store): Likewise.
20348 (vectorizable_condition): Likewise.
20349 (vectorizable_comparison): Likewise.
20351 2017-06-28 Michael Collison <michael.collison@arm.com>
20354 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20356 (arm_gen_movmemqi): Removed unused variable 'i'.
20357 Convert 'for' loop into 'while' loop.
20358 (arm_expand_prologue): Remove last unnecessary set of insn.
20359 (thumb_pop): Remove unused variable 'pushed_words'.
20360 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20362 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20364 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20365 * config/s390/s390-protos.h: Add prototype of
20366 s390_rel_address_ok_p.
20367 * config/s390/s390.c (s390_got_symbol): New function.
20368 (s390_rel_address_ok_p): New function.
20369 (legitimize_pic_address): Use s390_rel_address_ok_p.
20370 (s390_load_got): Use s390_got_symbol.
20371 (s390_option_override): Issue error if
20372 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20373 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20375 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20377 2017-06-27 Andrew Pinski <apinski@cavium.com>
20379 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20380 (X * copysign (1.0, X)): New pattern.
20381 (X * copysign (1.0, -X)): New pattern.
20382 (copysign (-1.0, CST)): New pattern.
20384 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20386 * genmultilib (combination_space): Remove variable.
20387 Validate reuse rules against regular expression for any sequence
20388 of multilib options in any order.
20390 2017-06-27 Michael Collison <michael.collison@arm.com>
20392 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20393 call aarch64_split_simd_combine.
20394 * (aarch64_combine_internal<mode>): Delete pattern.
20395 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20396 Allow register and subreg operands.
20398 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20400 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20401 specific need, just fallback on defaults.
20402 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20404 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20405 Olivier Hainque <hainque@adacore.com>
20407 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20409 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20410 targets. Pick a default if no particular attempt applied.
20411 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20414 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20416 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20417 (x86_64-wrs-vxworks7): Likewise.
20419 2017-06-27 Marek Polacek <polacek@redhat.com>
20422 * ubsan.c (instrument_null): Check get_base_address's result for null.
20424 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20426 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20428 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20430 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20431 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20432 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20433 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20434 New function types.
20435 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20436 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20437 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20438 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20439 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20440 BUILT_IN_FEUPDATEENV): New builtins.
20441 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20442 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20443 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20444 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20446 (builtin_structptr_types): Adjust size.
20447 * tree.c (builtin_structptr_types): Add four entries.
20449 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20450 Olivier Hainque <hainque@adacore.com>
20452 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20453 (TLS_SYM): New local macro, forcing reference to __tls__ on
20454 link command lines for VxWorks 7 RTPs, triggering initialization
20456 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20457 OS features TLS support, true for RTPs on VxWorks 7.
20458 * config/vxworks.c (vxworks_override_options): Setup emutls
20461 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20463 * predict.c (test_prediction_value_range): Use -1U instead of -1
20464 to avoid narrowing conversion warning.
20465 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20466 to avoid narrowing conversion warning.
20467 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20469 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20471 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20473 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20474 64bit configurations.
20475 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20476 (SIZE_TYPE): Likewise.
20477 * config/vxworks.c (vxworks_emutls_var_fields): Use
20478 long_unsigned_type_node instead of unsigned_type_node as the offset
20479 field type, which is "pointer" mode in emutls.c.
20481 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20484 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20486 PR middle-end/81207
20487 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20488 gimple_vuse copying separately from gimple_vdef copying.
20490 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20492 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20495 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20496 Olivier Hainque <hainque@adacore.com>
20498 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20499 for all vxworks7 targets.
20500 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20501 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20502 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20503 variations for VX6/VX7 and 32/64bits later on in ...
20504 (VXWORKS_LIB_SPEC): Leverage new macros.
20505 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20506 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20508 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20510 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20511 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20513 2017-06-26 Carl Love <cel@us.ibm.com>
20515 * config/rs6000/rs6000-c.c: Add support for built-in functions
20516 vector bool char vec_reve (vector bool char);
20517 vector signed char vec_reve (vector signed char);
20518 vector unsigned char vec_reve (vector unsigned char);
20519 vector bool int vec_reve (vector bool int);
20520 vector signed int vec_reve (vector signed int);
20521 vector unsigned int vec_reve (vector unsigned int);
20522 vector bool long long vec_reve (vector bool long long);
20523 vector signed long long vec_reve (vector signed long long);
20524 vector unsigned long long vec_reve (vector unsigned long long);
20525 vector bool short vec_reve (vector bool short);
20526 vector signed short vec_reve (vector signed short);
20527 vector double vec_reve (vector double);
20528 vector float vec_reve (vector float);
20529 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20530 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20531 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20532 (altivec_vreve): New pattern.
20533 * config/rs6000/altivec.h (vec_reve): New define.
20534 * doc/extend.texi (vec_rev): Update the built-in documentation file
20535 for the new built-in functions.
20537 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20539 PR tree-optimization/71815
20540 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20542 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20544 (slsr_process_phi): Likewise.
20545 (replace_uncond_cands_and_profitable_phis): Don't replace a
20546 multiply candidate with a stride of 1 (copy or cast).
20547 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20549 (lowest_cost_path): Likewise.
20550 (total_savings): Likewise.
20552 2017-06-26 Richard Biener <rguenther@suse.de>
20555 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20556 Use def_builtin_pure for all gather builtins.
20558 2017-06-26 Richard Biener <rguenther@suse.de>
20560 PR tree-optimization/81203
20561 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20562 non-dominating BBs.
20564 2017-06-26 Marek Polacek <polacek@redhat.com>
20567 * doc/invoke.texi: Document -Wmultistatement-macros.
20569 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20571 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20572 arm_neon_ok_no_float_abi effective target.
20574 2017-06-26 Richard Biener <rguenther@suse.de>
20576 PR tree-optimization/80928
20577 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20578 (copy_bbs): Set BB_DUPLICATED flag early.
20579 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20581 (execute_on_shrinking_pred): Likewise.
20582 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20583 BB_DUPLICATED blocks.
20584 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20585 iterate over all PHIs considering removal of *gsi.
20587 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20589 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20592 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20594 * config/rs6000/rs6000-string.c: (expand_block_clear,
20595 do_load_for_compare, select_block_compare_mode,
20596 compute_current_alignment, expand_block_compare,
20597 expand_strncmp_align_check, expand_strn_compare,
20598 expand_block_move, rs6000_output_load_multiple)
20599 Move functions related to string/block move/compare
20600 to a separate file.
20601 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20602 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20603 for this function which is now used in two files.
20604 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20605 * config.gcc: Add rs6000-string.o to extra_objs for
20606 targets powerpc*-*-* and rs6000*-*-*.
20608 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20611 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20612 32-bit, since indexed is not valid for DImode.
20613 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20614 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20615 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20616 (define_peephole2 for Altivec d-form store): Likewise.
20619 * multiple_target.c (create_dispatcher_calls): Only create the
20620 dispatcher call if the function is the default clone of a
20621 versioned function.
20623 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20625 PR middle-end/80902
20626 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20627 a call, force the call to not be a tail call.
20629 2017-06-23 Jeff Law <law@redhat.com>
20631 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20634 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20636 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20637 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20638 handling for early expansion of vector shifts (sl,sr,sra,rl).
20639 (builtin_function_type): Add vector shift right instructions
20640 to the unsigned argument list.
20642 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20644 rtl-optimizatoin/79286
20645 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20646 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20647 trap. PIC register plus a const unspec without offset can never trap.
20649 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20651 * tree.h (builtin_structptr_type): New type.
20652 (builtin_structptr_types): Declare new array.
20653 * tree.c (builtin_structptr_types): New array.
20654 (free_lang_data, build_common_tree_nodes): Use it.
20656 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20659 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20662 2017-06-22 Matt Turner <mattst88@gmail.com>
20664 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20665 Lake models to skylake case. Assume skylake for unknown
20666 models with clflushopt.
20668 2017-06-22 Jeff Law <law@redhat.com>
20670 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20671 frame sizes that do not satisfy aarch64_uimm12_shift.
20673 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20675 * profile-count.h (apply_probability,
20676 apply_scale, probability_in): Fix checks for zero.
20678 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20680 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20681 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20683 2016-06-22 Richard Biener <rguenther@suse.de>
20685 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20686 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20687 REDUC_MAX_EXPR support.
20688 (vectorizable_reduction): Likewise.
20689 (vect_create_epilog_for_reduction): Likewise.
20691 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20693 * match.pd (A / (1 << B) -> A >> B): New.
20694 * generic-match-head.c: Include optabs-tree.h.
20695 * gimple-match-head.c: Likewise.
20696 * optabs-tree.h (target_supports_op_p): New.
20697 * optabs-tree.c (target_supports_op_p): New.
20699 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20701 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20702 $gcc_cv_ld --help output.
20703 (gcc_cv_ld_demangle): Likewise.
20704 (gcc_cv_ld_eh_frame_hdr): Likewise.
20705 (gcc_cv_ld_pie): Likewise.
20706 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20707 (gcc_cv_ld_buildid): Likewise.
20708 (gcc_cv_ld_sysroot): Likewise.
20709 (ld_bndplt_support): Likewise.
20710 (ld_pushpopstate_support): Likewise.
20711 * configure: Regenerate.
20712 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20714 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20717 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20718 operands indexes to avoid gap between operands and match_dups.
20720 2017-06-21 Andrew Pinski <apinski@cavium.com>
20722 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20723 Increment Arith_shift and Arith_shift_reg by 1.
20724 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20726 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20727 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20728 (aarch64_strip_extend): Add new argument and test for it.
20729 (aarch64_cheap_mult_shift_p): New function.
20730 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20731 add a cost if it is true.
20732 Update calls to aarch64_strip_extend.
20733 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20735 2017-06-21 Andrew Pinski <apinski@cavium.com>
20737 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20739 (thunderxt88): Likewise.
20740 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20741 (thunderx_prefetch_tune): New variable.
20742 (thunderx2t99_prefetch_tune): Update for the correct values.
20743 (thunderxt88_tunings): New variable.
20744 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20745 generic_prefetch_tune.
20746 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20748 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20750 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20751 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20752 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20753 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20754 (aarch64_atomic_cas<mode>, GPI): Likewise.
20756 2017-06-21 Martin Liska <mliska@suse.cz>
20758 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20759 statements on cold and hot labels.
20760 * predict.c (tree_estimate_probability_bb): Remove the
20761 prediction from this place.
20763 2017-06-21 Martin Liska <mliska@suse.cz>
20765 PR tree-optimization/79489
20766 * gimplify.c (maybe_add_early_return_predict_stmt): New
20768 (gimplify_return_expr): Call the function.
20769 * predict.c (tree_estimate_probability_bb): Remove handling
20771 * predict.def: Update comment about early return predictor.
20772 * gimple-predict.h (is_gimple_predict): New function.
20773 * predict.def: Change default value of early return to 66.
20774 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20776 * passes.def: Put pass_strip_predict_hints to the beginning of
20779 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20781 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20782 FUNCTION_DECL declarations.
20783 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20785 (dwaf2out_decl): Likewise.
20786 * godump.c (go_early_global_decl): Skip call to the real debug hook
20787 for FUNCTION_DECL declarations.
20788 * passes.c (rest_of_decl_compilation): Skip call to the
20789 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20790 -fdump-go-spec is passed.
20792 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20794 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20795 Reorder for compactness.
20796 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20797 (def_builtin_pure, def_builtin_pure2): New functions.
20798 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20800 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20802 * match.pd (nop_convert): New predicate.
20803 ((A +- CST1) +- CST2): Allow some NOP conversions.
20805 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20808 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20809 with ctors/dtors if GOVD_SHARED is set.
20811 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20813 * config/aarch64/aarch64.md (movti_aarch64):
20814 Emit mov rather than orr.
20815 (movtf_aarch64): Likewise.
20816 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20817 Emit mov rather than orr.
20819 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20821 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20822 Swap alternatives, make integer dup more expensive.
20824 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20826 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20827 Return true for non-tls symbols.
20829 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20831 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20832 (cortex-a75): Likewise.
20833 (cortex-a75.cortex-a55): Likewise.
20834 * config/aarch64/aarch64-tune.md: Regenerate.
20835 * doc/invoke.texi (-mtune): Document new values for -mtune.
20837 2017-06-21 Tom de Vries <tom@codesourcery.com>
20839 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20840 stack_size feature.
20841 (Effective-Target Keywords, Other attributes): Suggest using
20842 dg-add-options stack_size feature to get stack limit in stack_size
20843 effective target documentation.
20845 2017-06-21 Julian Brown <julian@codesourcery.com>
20846 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20848 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20849 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20850 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20852 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20853 attribute type list for neon_multiply.
20854 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20855 attribute type list for neon_multiply.
20856 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20857 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20858 attribute type list for neon_multiply.
20859 * config/arm/types.md (crypto_pmull): Add.
20860 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20861 attribute type list.
20863 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20865 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20868 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20870 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20871 to make sure not to dereference a NULL cost_classes_ptr pointer.
20873 2017-06-20 Carl Love <cel@us.ibm.com>
20875 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20876 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20877 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20878 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20879 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20880 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20881 VMULOSW): New enum "unspec" values.
20882 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20883 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20884 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20885 altivec_vmulosw): New patterns.
20886 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20887 VMULOSW): Add definitions.
20889 2017-06-20 Julia Koval <julia.koval@intel.com>
20891 * config/i386/i386.c: Fix rounding expand for new pattern.
20892 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20894 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20896 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20897 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20899 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20901 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20904 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20906 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20907 architecture, then by implementer ID.
20908 * config/aarch64/aarch64-tune.md: Regenerate.
20910 2017-06-20 Richard Biener <rguenther@suse.de>
20912 PR middle-end/81097
20913 * fold-const.c (split_tree): Fold to type before negating.
20915 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20917 * diagnostic-show-locus.c
20918 (selftest::test_fixit_deletion_affecting_newline): New function.
20919 (selftest::diagnostic_show_locus_c_tests): Call it.
20921 2017-06-20 Andreas Schwab <schwab@suse.de>
20924 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20927 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20929 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20930 __ARM_FEATURE_COPROC according to support.
20932 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20934 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20935 Rewritten to avoid overflow for > 32-bit pointers.
20938 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20939 by removing enum keyword.
20940 (ubsan_type_descriptor): Likewise. Formatting fix.
20943 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20944 splitter): Require TARGET_SSE2 in the condition.
20946 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20949 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20950 for doing vector set of SFmode on ISA 3.0.
20951 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20952 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20954 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20955 SFmode value into a V4SF variable that was extracted from another
20956 V4SF variable without converting the element to double precision
20957 and back to single precision vector format.
20958 (vsx_insert_extract_v4sf_p9_2): Likewise.
20960 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20962 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20963 in UWHI to avoid undefined overflow.
20966 * ubsan.h (enum ubsan_encode_value_phase): New.
20967 (ubsan_encode_value): Change second argument to
20968 enum ubsan_encode_value_phase with default value of
20969 UBSAN_ENCODE_VALUE_GENERIC.
20970 * ubsan.c (ubsan_encode_value): Change second argument to
20971 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20972 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20973 create_tmp_var_raw instead of create_tmp_var and use a
20975 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20976 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20977 ubsan_encode_value callers.
20980 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20981 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20982 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20984 2017-06-19 Richard Biener <rguenther@suse.de>
20986 PR middle-end/81118
20987 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20988 estimates if we changed anything.
20990 2017-06-19 Richard Biener <rguenther@suse.de>
20992 PR tree-optimization/80887
20993 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20994 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20995 simplified lookups, then reset mprts_hook.
20996 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20998 (try_to_simplify): Likewise.
21000 2017-06-19 Martin Liska <mliska@suse.cz>
21003 * gimplify.c (gimplify_switch_expr):
21004 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21006 2017-06-19 Martin Liska <mliska@suse.cz>
21008 * doc/install.texi: Document that PGO runs in 4 stages.
21010 2017-06-19 Martin Liska <mliska@suse.cz>
21013 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21014 to dispatcher function name.
21015 * multiple_target.c (replace_function_decl): New function.
21016 (create_dispatcher_calls): Redirect both edges and references.
21018 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21020 * profile-count.c (profile_count::dump): Dump quality.
21021 (profile_count::differs_from_p): Update for unsigned val.
21022 * profile-count.h (profile_count_quality): New enum.
21023 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21025 2017-06-19 Richard Biener <rguenther@suse.de>
21027 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21028 struct function as arg.
21029 (estimate_numbers_of_iterations): Export overload with loop arg.
21030 (free_numbers_of_iterations_estimates_loop): Use an overload of
21031 free_numbers_of_iterations_estimates instead.
21032 * tree-cfg.c (remove_bb): Adjust.
21033 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21034 * tree-parloops.c (gen_parallel_loop): Likewise.
21035 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21037 (tree_unroll_loops_completely): Likewise.
21038 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21039 Use an overload instead and export.
21040 (estimated_loop_iterations): Adjust.
21041 (max_loop_iterations): Likewise.
21042 (likely_max_loop_iterations): Likewise.
21043 (estimate_numbers_of_iterations): Take struct function as arg
21045 (loop_exits_before_overflow): Adjust.
21046 (free_numbers_of_iterations_estimates_loop): Use an overload.
21047 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21048 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21050 2017-06-19 Richard Biener <rguenther@suse.de>
21053 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21054 RANGE_EXPR conservatively.
21056 2017-06-16 Carl Love <cel@us.ibm.com>
21058 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21059 definitions for vec_float, vec_float2, vec_floato,
21060 vec_floate built-ins.
21061 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21062 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21064 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21065 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21066 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21067 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21068 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21069 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21070 vec_floato): Add builtin defines.
21071 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21072 Update the built-in documentation file for the new built-in
21075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21077 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21078 (mthumb): Mark as the negative of -marm.
21080 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21082 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21084 (ARM Options, -mtune): Document that this accepts the same
21085 extension options as -mcpu.
21086 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21090 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21091 permitted extensions.
21093 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21095 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21096 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21097 (armv8-m.main): Add option +nodsp.
21098 * config/arm/arm-cpu-cdata.h: Regenerated.
21100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21102 * config/arm/t-fuchsia: New file.
21103 * config.gcc (arm*-*-fuchsia*): Use it.
21105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21107 * config/arm/t-symbian: Rewrite for new option infrastructure.
21109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21111 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21112 (MULTILIB_REQUIRED): Likewise.
21114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21116 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21117 (MULTILIB_RESUE): Likewise.
21118 (MULTILIB_MATCHES): Likewise.
21119 (MULTLIB_REQUIRED): Likewise.
21121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21123 * config/arm/t-rtems: Rewrite for new option framework.
21125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21127 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21128 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21129 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21130 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21131 * config/arm/t-multilib: ... here.
21132 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21133 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21134 armv7-a and armv8*-a when A-profile libraries have not been built.
21135 * config/arm/t-rmprofile: Rewrite.
21137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21139 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21140 with a backslash. Remove the backslash after substituting unescaped
21142 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21144 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21146 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21147 the driver to pass through the default setting of -mfloat-abi.
21148 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21150 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21151 (all_feat_combs): New rule.
21152 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21154 * config/arm/t-aprofile: Rewrite.
21156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21158 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21159 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21160 fpu is not specified by the user/command-line.
21161 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21162 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21163 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21164 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21165 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21166 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21170 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21171 * config/arm/t-arm-elf: Rewritten.
21173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21175 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21176 have some floating-point instructions.
21177 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21178 (TARGET_MAYBE_HARD_FLOAT): New macro.
21179 * config/arm/arm-builtins.c (arm_init_builtins): Use
21180 TARGET_MAYBE_HARD_FLOAT.
21181 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21183 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21185 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21186 (configargs.h): Include it.
21187 (arm_print_hint_for_fpu_option): New function.
21188 (arm_parse_fpu_option): New function.
21189 (candidate_extension): New class.
21190 (arm_canon_for_multilib): New function.
21191 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21192 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21193 (ARCH_CANONICAL_SPECS): New macro.
21194 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21198 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21199 are set after handling multilib fragments. Set target_cpu_default2
21202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21204 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21206 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21207 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21209 (check_arch): Likewise.
21210 * config/arm/arm.c (arm_configure_build_target): Handle
21211 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21212 options in the default.
21214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21216 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21217 when an option is an alias of another.
21218 * config/arm/parsecpu.awk (optalias): New parser token.
21219 (gen_comm_data): Mark non-alias options as such. Emit entries
21220 for extension aliases.
21221 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21222 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21223 (armv6kz, armv6zk, armv6t2): Likewise.
21224 (armv7): Make vfpv3-d16 an alias.
21225 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21227 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21228 Sort in canonical order.
21229 (armv8-a): Sort in canonical order.
21230 (armv8.1-a, armv8.2-a): Likewise.
21231 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21233 (cortex-a9): Sort in canonical order.
21234 * config/arm/arm.c (selftests.h): Include it.
21235 (arm_test_cpu_arch_data): New function.
21236 (arm_run_self_tests): New function.
21237 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21238 (targetm): Move declaration to the end of the file.
21239 * arm-cpu-cdata.h: Regenerated.
21241 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21243 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21244 call to target_mode_check describing the type of option passed.
21245 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21246 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21247 arm_parse_cpu_option_name to match parameters against list of
21249 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21250 arm_arch_core_flags data structure.
21251 * config/arm/arm-cpu_cdata.h: Regenerated.
21253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21255 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21257 (arm_print_hint_for_cpu_option): Likewise.
21258 (arm_print_hint_for_arch_option): Likewise.
21259 (arm_parse_cpu_option_name): Likewise.
21260 (arm_parse_arch_option_name): Likewise.
21261 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21262 of entries in the all_fpus list.
21263 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21264 (arm_parse_cpu_option_name): Declare.
21265 (arm_parse_arch_option_name): Declare.
21266 (arm_parse_option_features): Declare.
21267 (arm_intialize_isa): Declare.
21268 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21270 (gen_comm_data): ... here. Make definitions non-static.
21271 * config/arm/arm-cpu-data.h: Regenerated.
21272 * config/arm/arm-cpu-cdata.h: Regenerated.
21274 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21276 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21277 (cpu_arch_extension): New structure.
21278 (cpu_arch_option, arch_option, cpu_option): New structures.
21279 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21280 architecture types.
21281 (gen_data): Generate new format data tables.
21282 * config/arm/arm.c (cpu_tune): New structure.
21283 (cpu_option, processors): Delete.
21284 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21285 (arm_print_hint_for_cpu_option): ... this and ...
21286 (arm_print_hint_for_arch_option): ... this.
21287 (arm_parse_arch_cpu_name): Delete. Replace with ...
21288 (arm_parse_cpu_option_name): ... this and ...
21289 (arm_parse_arch_option_name): ... this.
21290 (arm_unrecognized_feature): Change type of target parameter to
21292 (arm_parse_arch_cpu_features): Delete. Replace with ...
21293 (arm_parse_option_features): ... this.
21294 (arm_configure_build_target): Rework to use new configuration data
21296 (arm_print_tune_info): Rework for new configuration data tables.
21297 * config/arm/arm-cpu-data.h: Regenerated.
21298 * config/arm/arm-cpu.h: Regenerated.
21300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21302 * Makefile.in (OBJS): Move sbitmap.o from here ...
21303 (OBJS-libcommon): ... to here.
21305 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21307 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21308 (ISA_ALL_CRYPTO): New macro.
21309 (ISA_ALL_SIMD): New macro
21310 (ISA_ALL_FP): New macro.
21311 * config/arm/arm.c (fpu_bitlist): Update initializer.
21312 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21314 (arm9e): Add fpu. Add option for nofp
21315 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21316 (arm926ej-s, arm1026ej-s): Likewise.
21317 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21318 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21319 neon-fp16, neon-vfpv4, nofp and nosimd.
21320 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21321 (cortex-a8): Add fpu. Add option for nofp.
21322 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21323 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21324 (cortex-r4f): Add fpu.
21325 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21326 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21328 (cortex-r8): Likewise.
21329 (cortex-m4): Add fpu. Add option for nofp.
21330 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21331 (cortex-a17.cortex-a7): Likewise.
21332 (cortex-a32): Add fpu. Add options for crypto and nofp.
21333 (cortex-a35, cortex-a53): Likewise.
21334 (cortex-a57): Add fpu. Add option for crypto.
21335 (cortex-a72, cortex-a73): Likewise.
21336 (exynos-m1): Likewise.
21337 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21338 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21339 (cortex-m33): Add fpu. Add option for nofp.
21340 * config/arm/arm-cpu-cdata.h: Regenerated
21341 * config/arm/arm-cpu-data.h: Regenerated.
21343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21345 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21346 (armv5te, armv5tej): Likewise.
21347 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21348 (armv7): Add options fp and vfpv3-d16.
21349 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21350 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21352 (armv7ve): Likewise.
21353 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21354 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21355 (armv8-a): Add nocrypto option.
21356 (armv8.1-a, armv8.2-a): Likewise.
21357 (armv8-m.main): add options fp, fp.dp and nofp.
21359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21361 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21363 (armv8-a+crc): Delete.
21364 (armv8.1-a): Add options simd, crypto and nofp.
21365 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21366 (armv8.2-a+fp16): Delete.
21367 (armv8-m.main): Add option dsp.
21368 (armv8-m.main+dsp): Delete.
21369 (cortex-a8): Add fpu. Add nofp option.
21370 (cortex-a9): Add fpu. Add nofp and nosimd options.
21371 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21372 link to main cpu and architecture data structures.
21373 (gen_comm_data): Only put isa attributes from the main architecture
21375 (option): New statement for architecture and CPU entries.
21376 * arm.c (struct cpu_option): New structure.
21377 (struct processors): Add entry for options.
21378 (arm_unrecognized_feature): New function.
21379 (arm_parse_arch_cpu_name): Ignore any characters after the first
21381 (arm_parse_arch_cpu_feature): New function.
21382 (arm_configure_build_target): Separate out any CPU and architecture
21383 features and parse separately. Don't error out if -mfpu=auto is
21384 used with only an architecture string.
21385 (arm_print_asm_arch_directives): New function.
21386 (arm_file_start): Call it.
21387 * config/arm/arm-cpu-cdata.h: Regenerated.
21388 * config/arm/arm-cpu-data.h: Likewise.
21389 * config/arm/arm-tables.opt: Likewise.
21391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21393 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21394 assembler when it is not -mfpu=auto.
21396 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21398 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21399 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21400 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21401 (ASM_CPU_SPEC): Rewrite.
21402 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21403 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21404 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21405 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21406 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21407 copied string is NUL-terminated. Also strip any characters prefixed
21409 (arm_rewrite_selected_arch): New function.
21410 (arm_rewrite_march): New function.
21412 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21414 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21415 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21416 (march, mcpu, mtune): Convert to string-based options.
21417 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21418 (arm_parse_arch_cpu_name): New function.
21419 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21420 identify selected architecture or CPU.
21421 (arm_option_save): New function.
21422 (TARGET_OPTION_SAVE): Redefine.
21423 (arm_option_restore): Restore string options.
21424 (arm_option_print): Print string options.
21426 2017-06-16 Martin Sebor <msebor@redhat.com>
21428 PR tree-optimization/80933
21429 PR tree-optimization/80934
21430 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21431 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21432 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21433 (gimple_fold_builtin): Call them.
21435 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21437 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21438 as unlikely; update profile.
21440 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21442 * predict.c (force_edge_cold): Handle declaring edges impossible
21445 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21447 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21449 (try_unroll_loop_completely): Fix reporting.
21451 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21453 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21455 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21458 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21459 if given a non-constant argument for an intrinsic which requires a
21462 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21464 * profile.c (compare_freqs): New function.
21465 (branch_prob): Sort edge list.
21466 (find_spanning_tree): Assume that the list is priority sorted.
21468 2017-06-16 Richard Biener <rguenther@suse.de>
21470 PR tree-optimization/81090
21471 * passes.def (pass_record_bounds): Remove.
21472 * tree-pass.h (make_pass_record_bounds): Likewise.
21473 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21474 make_pass_record_bounds): Likewise.
21475 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21476 not free niter estimates at the beginning but at the end.
21477 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21479 2017-06-16 Richard Biener <rguenther@suse.de>
21481 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21482 initializer to workaround ICE in host GCC 4.8.
21484 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21486 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21488 (clone_inlined_nodes): Update.
21490 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21492 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21493 prefetch settings, and enable prefetching by default at -O3.
21495 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21497 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21498 Set flag_prefetch_loop_arrays according to tuning data.
21500 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21502 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21503 New tune structure.
21504 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21505 [Unrelated to main purpose of the patch] Place the pointer field last
21506 to enable type checking errors when tune structure are wrongly merged.
21507 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21508 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21509 (thunderx2t99_prefetch_tune): New tune constants.
21510 (tune_params *_tunings): Update all tunings (no functional change).
21511 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21512 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21513 from tunings structures.
21515 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21518 * ubsan.c (instrument_null): Add T argument, use it instead
21519 of computing it based on IS_LHS.
21520 (instrument_object_size): Likewise.
21521 (pass_ubsan::execute): Adjust instrument_null and
21522 instrument_object_size callers to pass gimple_get_lhs or
21523 gimple_assign_rhs1 result to it. Use instrument_null instead of
21524 calling get_base_address and instrument_mem_ref. Handle
21525 aggregate call arguments for object-size sanitization.
21527 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21529 PR tree-optimization/81089
21530 * tree-vrp.c (is_masked_range_test): Validate operands of
21533 2017-06-15 Martin Sebor <msebor@redhat.com>
21536 * dumpfile.c (dump_register): Avoid calling memset to initialize
21537 a class with a default ctor.
21538 * gcc.c (struct compiler): Remove const qualification.
21539 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21540 * hash-table.h: Ditto.
21541 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21543 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21544 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21546 * params.h (struct param_info): Make struct members non-const.
21547 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21548 with default initialization.
21549 * vec.h (vec_copy_construct, vec_default_construct): New helper
21551 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21552 with vec_copy_construct.
21553 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21554 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21555 * doc/invoke.texi (-Wclass-memaccess): Document.
21557 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21559 * emit-rtl.h (is_leaf): Update comment about local
21560 register allocator.
21562 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21565 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21566 for a variable to have a section before checking if the section has a
21568 Set section to.persistent if persistent attribute is set.
21569 Warn if .persistent attribute is used on an automatic variable.
21571 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21573 PR rtl-optimization/80474
21574 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21576 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21578 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21581 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21583 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21584 (casesi): Emit bounds checking as RTL.
21585 (casesi_internal_mips16_<mode>): Remove bounds checking.
21587 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21589 * config/xtensa/xtensa.c (xtensa_option_override): Append
21590 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21591 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21592 xtensa_doloop_hooks): Define unconditionally.
21593 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21595 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21596 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21597 for it in xtensa_option_override.
21598 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21599 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21601 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21603 * doc/cppopts.texi: Document '-' special value to -MF.
21605 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21607 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21608 (cortex_a53_fconst): Likewise.
21609 (cortex_a53_fpmul): Likewise.
21610 (cortex_a53_f_load_64): Likewise.
21611 (cortex_a53_f_load_many): Likewise.
21612 (cortex_a53_advsimd_alu): Likewise.
21613 (cortex_a53_advsimd_alu_q): Likewise.
21614 (cortex_a53_advsimd_mul): Likewise.
21615 (cortex_a53_advsimd_mul_q): Likewise.
21616 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21617 Add missing fmul, r2f_cvt and fconst cases.
21619 2017-06-14 Richard Biener <rguenther@suse.de>
21621 PR middle-end/81088
21622 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21624 (fold_binary_loc): When associating do not treat pre-existing
21625 TREE_OVERFLOW on literal constants as a reason to allow
21626 TREE_OVERFLOW on associated literal constants.
21628 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21630 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21631 (MASK_FEATURES): New macro.
21632 * config/sparc/sparc.c (sparc_option_override): Remove the special
21633 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21635 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21637 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21638 a division of 0 if non-call exceptions are enabled.
21640 2017-06-14 Andrew Pinski <apinski@cavium.com>
21641 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21644 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21645 Improve vector initialization code gen for only variable case.
21647 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21649 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21651 2017-06-14 Richard Biener <rguenther@suse.de>
21653 PR tree-optimization/81083
21654 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21657 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21659 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21660 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21661 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21662 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21663 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21664 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21666 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21668 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21669 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21671 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21673 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21675 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21677 * config/rs6000/t-rtems: Don't handle SPE.
21679 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21681 * config/rs6000/t-linux: Don't handle SPE.
21683 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21685 * config/rs6000/eabispe.h: Delete file.
21687 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21689 * config/rs6000/t-spe: Delete file.
21691 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21693 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21694 (rs6000_legitimate_offset_address_p): Return false for anything in
21695 V2SImode or V2SFmode.
21697 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21699 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21700 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21701 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21703 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21704 (rs6000_legitimate_offset_address_p): Ditto.
21705 (rs6000_emit_move): Ditto.
21706 (rs6000_init_builtins): Remove V4HI_type_node.
21708 2017-06-13 Martin Liska <mliska@suse.cz>
21711 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21712 (gate_asan): Likewise.
21713 * asan.h (asan_no_sanitize_address_p): Remove the function.
21714 (sanitize_flags_p): New function.
21715 * builtins.def: Fix coding style.
21716 * common.opt: Use renamed enum value.
21717 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21718 * doc/extend.texi: Document no_sanitize attribute.
21719 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21720 to SANITIZE_UNDEFINED_NONDEFAULT.
21721 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21722 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21723 Use sanitize_flags_p.
21724 * gimplify.c (gimplify_function_tree): Likewise.
21725 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21726 * opts.c (parse_no_sanitize_attribute): New function.
21727 (common_handle_option): Use renamed enum value.
21728 * opts.h (parse_no_sanitize_attribute): Declare.
21729 * tree.c (sanitize_flags_p): New function.
21730 * tree.h: Declared here.
21731 * tsan.c: Use sanitize_flags_p.
21732 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21733 (instrument_mem_ref): Likewise.
21734 (instrument_bool_enum_load): Likewise.
21735 (do_ubsan_in_current_function): Remove the function.
21736 (pass_ubsan::execute): Use sanitize_flags_p.
21737 * ubsan.h: Remove do_ubsan_in_current_function
21738 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21739 (dump_function_to_file): Use it here.
21741 2017-06-13 Martin Jambor <mjambor@suse.cz>
21743 PR tree-optimization/80803
21744 PR tree-optimization/81063
21745 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21746 (propagate_subaccesses_across_link): Enqueue subtree whenever
21747 necessary instead of relying on the caller.
21749 2017-06-13 Martin Jambor <mjambor@suse.cz>
21751 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21752 that have a first_link.
21753 (sort_and_splice_var_accesses): Do not check first_link before
21755 (subtree_mark_written_and_enqueue): Likewise.
21756 (propagate_all_subaccesses): Likewise and do not stop at first
21757 parent with a first_link.
21759 2017-06-13 Martin Jambor <mjambor@suse.cz>
21761 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21764 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21766 * match.pd: New pattern.
21768 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21770 * tree-vrp.c (is_masked_range_test): New function.
21771 (register_edge_assert_for): Determine ranges for
21774 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21776 PR tree-optimization/67328
21777 * fold-const.c (maskable_range_p): New function.
21778 (build_range_check): Generate bittests if possible.
21780 2017-06-13 Martin Liska <mliska@suse.cz>
21782 * gimple-pretty-print.c (dump_probability): Add new argument.
21783 (dump_edge_probability): Dump both probability and count.
21784 (dump_gimple_label): Likewise.
21785 (dump_gimple_bb_header): Likewise.
21787 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21790 * config/avr/avr-devices.c: Fix indentation.
21791 * config/avr/gen-avr-mmcu-specs.c: Dito.
21793 2017-06-13 Richard Biener <rguenther@suse.de>
21795 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21796 instead get vector type from stmt_info.
21797 (vectorizable_reduction): Adjust. Remove dead code.
21799 2017-06-13 Richard Biener <rguenther@suse.de>
21801 PR middle-end/81065
21802 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21803 case of C * (x * C2 + C3).
21804 (fold_addr_of_array_ref_difference): Properly fold index difference.
21806 2017-06-12 David S. Miller <davem@davemloft.net>
21809 * config/sparc/sparc.md (return expander): Emit frame blockage if
21810 function uses alloca.
21812 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21814 * combine.c (make_field_assignment): Check len rather than the mode
21815 precision when calling force_to_mode.
21817 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21819 Support multilibs and devices that see flash in RAM address range.
21822 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21823 (avr_mcu_t) <flash_pm_offset>: New field.
21824 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21825 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21826 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21827 (AVR_TINY_PM_OFFSET): Remove macro.
21828 * config/avr/avr.opt (-mshort-calls): New option.
21829 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21830 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21831 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21832 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21833 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21834 instead of avr_arch->have_jmp_call.
21835 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21836 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21837 avr_arch->flash_pm_offset to define.
21838 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21839 new field flash_pm_offset. Add entry for avrxmega3.
21840 (avr_texinfo): Add entry for avrxmega3.
21841 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21842 attiny212, attiny214,
21843 attiny412, attiny414, attiny416, attiny417,
21844 attiny814, attiny816, attiny817,
21845 attiny1614, attiny1616, attiny1617,
21846 attiny3214, attiny3216, attiny3217.
21847 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21848 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21849 (avr_print_operand_address) [AVR_TINY]: Same.
21850 (avr_asm_init_sections) <readonly_data_section>: Only patch
21851 callback if avr_arch->flash_pm_offset = 0.
21852 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21853 for rodata if avr_arch->flash_pm_offset != 0.
21854 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21855 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21856 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21857 (m_options): Append opt_rcall.
21858 (m_dirnames): Append dir_rcall.
21859 * config/avr/t-multilib: Regenerate.
21861 * configure.ac [target=avr]: Check whether avrxmega3 default
21862 linker description file works as needed.
21863 * configure: Regenerate.
21864 * doc/avr-mmcu.texi: Regenerate.
21865 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21866 <__AVR_ARCH__>: Document avrxmega3 and 103.
21867 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21868 <__AVR_SHORT_CALLS__>: Document it.
21869 <__AVR_PM_BASE_ADDRESS__>: Document it.
21870 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21871 (AVR Variable Attributes) <progmem>: Document this is
21872 not needed for avrxmega3.
21873 (AVR Named Address Spaces) <__flash>: Dito.
21875 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21877 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21879 2017-06-12 Doug Rupp <rupp@adacore.com>
21881 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21882 Append vxworks-stdint.h to the tm_file list.
21883 * config/vxworks-stdint.h: New file.
21885 2017-06-12 Martin Liska <mliska@suse.cz>
21887 PR tree-optimization/81041
21888 * tree-profile.c (gimple_gen_ic_func_profiler):
21889 Create an extra BB in profile-generate
21890 (gimple_gen_time_profiler): Likewise.
21892 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21894 PR tree-optimization/81003
21895 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21896 (update_range_test): Use it instead of force_gimple_operand_gsi.
21898 2017-06-12 Richard Biener <rguenther@suse.de>
21900 PR tree-optimization/81053
21901 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21902 with backedge value not defined in loop. Simplify def stmt
21905 2017-06-11 Tom de Vries <tom@codesourcery.com>
21908 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21910 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21911 nvptx_cannot_force_const_mem.
21913 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21915 * opts.c (finish_options): Move test for flag_split_stack after
21916 it has been initialized.
21918 2017-06-11 Jason Merrill <jason@redhat.com>
21920 * tree.h (id_equal): New.
21921 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21922 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21923 instead of strcmp of IDENTIFIER_POINTER.
21925 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21927 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21928 (mark_all_inlined_calls_cdtor): Fix formating.
21929 (inline_transform): Rescale profile before inlining.
21931 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21933 * cgraph.h (cgraph_edge::clone): Update prototype.
21934 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21935 (cgraph_node::create_clone): Update.
21936 (cgraph_node::create_version_clone): Update.
21937 * tree-inline.c (copy_bb): Update.
21938 (expand_call_inline): Update.
21940 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21942 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21943 factored out from ...
21944 (rs6000_emit_prologue): ... here.
21946 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21948 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21949 factored out from ...
21950 (rs6000_emit_prologue): ... here.
21952 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21954 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21956 (handle_missing_profiles): Fix computation of tp_first_run.
21957 (counts_to_freqs): Do not touch freqs when count is 0.
21959 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21961 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21964 2017-06-10 Tom de Vries <tom@codesourcery.com>
21966 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21967 attributes): Document signal effective target.
21969 2017-06-10 Tom de Vries <tom@codesourcery.com>
21971 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21972 Document effective target stack_size.
21974 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21976 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21977 to the edit_context if they can be auto-applied.
21979 2017-06-9 Ian Lance Taylor <iant@golang.org>
21981 * opts.c (finish_options): If -fsplit-stack, disable implicit
21982 -forder-blocks-and-partition.
21983 * doc/invoke.texi (Optimize Options): Document that when using
21984 -fsplit-stack -forder-blocks-and-partition is not implicitly
21987 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21989 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21990 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21991 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21992 * builtins.def (abort, trap, unreachable): Declare cold.
21993 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21994 * tree-core.h (ECF_COLD): New.
21995 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21996 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21998 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22000 * predict.c (unlikely_executed_stmt_p): Cleanup.
22002 2017-06-09 Richard Biener <rguenther@suse.de>
22004 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22005 model if the ref is always written to.
22007 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22009 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22011 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22013 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22016 2017-06-09 Tom de Vries <tom@codesourcery.com>
22019 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22020 "target cannot support label values" when encountering LABEL_REF.
22022 2017-06-09 Martin Liska <mliska@suse.cz>
22024 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22025 (gimple_gen_ic_func_profiler): Emit direct comparison
22026 of __gcov_indirect_call_callee with NULL.
22027 (gimple_gen_time_profiler): Change probability from
22028 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22030 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22032 * profile.c (edge_gcov_counts): Turn to pointer.
22033 (compute_branch_probabilities, compute_branch_probabilities): Update.
22034 (branch_prob): Do not clear edge_gcov_count.
22035 * profile.h (edge_gcov_counts): Turn to pointer.
22036 (edge_gcov_count): Update.
22038 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22040 * gimple.h (gimple_check_failed): Mark cold.
22042 2017-06-09 Richard Biener <rguenther@suse.de>
22044 PR tree-optimization/66623
22045 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22046 refactor check_reduction into two parts, properly computing
22047 whether we have to check reduction validity for outer loop
22050 2017-06-09 Richard Biener <rguenther@suse.de>
22052 PR tree-optimization/79483
22053 * graphite-scop-detection.c (order): New global.
22054 (get_order): Compute bb to order mapping that satisfies code
22055 generation constraints.
22056 (cmp_pbbs): New helper.
22057 (build_scops): Start domwalk at entry block, sort generated
22060 2017-06-09 Richard Biener <rguenther@suse.de>
22062 PR middle-end/81007
22063 * ipa-polymorphic-call.c
22064 (ipa_polymorphic_call_context::restrict_to_inner_class):
22065 Skip FIELD_DECLs with error_mark_node type.
22066 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22069 2017-06-09 Martin Liska <mliska@suse.cz>
22071 * predict.c (struct branch_predictor): New struct.
22072 (test_prediction_value_range): New test.
22073 (predict_c_tests): New function.
22074 * selftest-run-tests.c (selftest::run_tests): Run the function.
22075 * selftest.h: Declare new tests.
22077 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22080 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22081 gen_add3_insn did not fail.
22082 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22083 r0, construct that number in a temporary reg and add that reg to r0.
22084 If asked to put the result in r0 as well, fail.
22086 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22088 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22089 for early expansion of vec_eqv.
22091 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22093 PR middle-end/81005
22094 * ubsan.c (instrument_null): Avoid pointless code temporary.
22095 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22097 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22101 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22104 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22107 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22109 * predict.c (unlikely_executed_edge_p): Move ahead.
22110 (probably_never_executed_edge_p): Use it.
22112 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22114 PR middle-end/79988
22115 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22116 gimple_call_builtin_p call.
22118 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22120 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22121 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22122 rtl_check_failed_type2, rtl_check_failed_code1,
22123 rtl_check_failed_code2, rtl_check_failed_code_mode,
22124 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22125 rtvec_check_failed_bounds, rtl_check_failed_flag,
22126 _fatal_insn_not_found, _fatal_insn): Likewise.
22127 * tree.h (tree_contains_struct_check_failed,
22128 tree_check_failed, tree_not_check_failed,
22129 tree_class_check_failed, tree_range_check_failed,
22130 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22131 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22132 tree_operand_check_failed, omp_clause_check_failed,
22133 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22136 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22138 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22139 flag_branch_probabilities.
22140 * ipa-inline.c (edge_badness): Likewise.
22141 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22142 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22143 * predict.c (maybe_hot_frequency_p): Likewise.
22144 (probably_never_executed): Likewise.
22145 * sched-ebb.c (schedule_ebbs): Likewise.
22146 * sched-rgn.c (find_single_block_region): Likewise.
22147 * tracer.c (tail_duplicate): Likewise.
22149 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22151 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22152 longer requires x_flag_profile_use.
22154 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22156 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22157 instead of flag_reorder_blocks_and_partition.
22158 * dbxout.c (dbxout_function_end): Likewise.
22159 * dwarf2out.c (gen_subprogram_die): Likewise.
22160 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22161 * hw-doloop.c (reorg_loops): Likewise.
22162 * varasm.c (assemble_start_function,
22163 assemble_end_function): Likewise.
22164 (decide_function_section): Do not check for
22165 flag_reorder_blocks_and_partition.
22167 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22169 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22171 (chkp_get_hard_register_fake_addr_expr): Ditto.
22172 (chkp_build_addr_expr): Add check for hard reg case.
22173 (chkp_parse_array_and_component_ref): Ditto.
22174 (chkp_find_bounds_1): Ditto.
22175 (chkp_process_stmt): Don't generate bounds store for
22178 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22180 * predict.c (maybe_hot_bb_p): Do not check profile status.
22181 (maybe_hot_edge_p): Likewise.
22182 (probably_never_executed): Check for zero counts even if profile
22184 (unlikely_executed_edge_p): New function.
22185 (unlikely_executed_stmt_p): New function.
22186 (unlikely_executed_bb_p): New function.
22187 (set_even_probabilities): Use unlikely predicates.
22188 (combine_predictions_for_bb): Likewise.
22189 (predict_paths_for_bb): Likewise.
22190 (predict_paths_leading_to_edge): Likewise.
22191 (determine_unlikely_bbs): New function.
22192 (estimate_bb_frequencies): Use it.
22193 (compute_function_frequency): Use zero counts even if profile is
22195 * profile-count.h: Fix typo.
22197 2017-08-08 Julia Koval <julia.koval@intel.com>
22199 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22200 _mm512_mask_cvtsepi16_storeu_epi8,
22201 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22202 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22203 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22204 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22205 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22206 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22207 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22208 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22209 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22210 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22211 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22212 __builtin_ia32_pmovuswb256mem_mask,
22213 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22214 __builtin_ia32_pmovwb512mem_mask): New builtins.
22216 2017-08-08 Julia Koval <julia.koval@intel.com>
22218 PR target/73350,80862
22219 * config/i386/subst.md (round): Fix round pattern.
22220 * config/i386/i386.c (ix86_erase_embedded_rounding):
22221 Fix erasing rounding for the fixed pattern.
22223 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22225 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22227 2017-06-08 Martin Liska <mliska@suse.cz>
22229 PR gcov-profile/80911
22230 * gcov.c (block_info::block_info): New constructor.
22232 2017-06-07 Carl Love <cel@us.ibm.com>
22234 * config/rs6000/rs6000-c: The return type of the following
22235 built-in functions was implemented as int not long long. Fix sign
22236 of return value for the unsigned version of vec_mulo and vec_mule.
22237 vector unsigned long long vec_bperm (vector unsigned long long,
22238 vector unsigned char)
22239 vector signed long long vec_mule (vector signed int,
22241 vector unsigned long long vec_mule (vector unsigned int,
22242 vector unsigned int)
22243 vector signed long long vec_mulo (vector signed int,
22245 vector unsigned long long vec_mulo (vector unsigned int,
22246 vector unsigned int)
22247 * doc/extend.texi: Fix the documentation for the built-in
22250 2017-06-07 Carl Love <cel@us.ibm.com>
22253 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22256 2017-06-07 Carl Love <cel@us.ibm.com>
22258 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22259 support, Generate doublehv for signed int/float for BE case only.
22261 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22263 * doc/invoke.texi (mcx16): Rewrite.
22265 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22267 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22268 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22269 *mov<mode>_softfloat, and an anonymous splitter): Use
22270 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22274 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22276 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22277 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22278 (rs6000_debug_reg_global): Adjust.
22279 (rs6000_init_hard_regno_mode_ok): Adjust.
22280 (rs6000_dbx_register_number): Adjust.
22281 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22282 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22283 Remove SPE_ACC and SPEFSCR.
22284 (REG_ALLOC_ORDER): Ditto.
22285 (FRAME_POINTER_REGNUM): Change to 111.
22286 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22287 (REG_CLASS_NAMES): Ditto.
22288 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22289 (REGISTER_NAMES): Ditto.
22290 (ADDITIONAL_REG_NAMES): Ditto.
22291 (rs6000_reg_names): Ditto.
22292 * config/rs6000/rs6000.md: Renumber some register number
22295 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22297 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22299 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22300 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22302 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22303 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22304 Delete the SPE high registers.
22305 (REG_ALLOC_ORDER): Ditto.
22306 (enum reg_class): Remove SPE_HIGH_REGS.
22307 (REG_CLASS_NAMES): Ditto.
22308 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22309 (REGISTER_NAMES): Ditto.
22310 (rs6000_reg_names): Ditto.
22311 * doc/tm.texi.in: Remove SPE as example.
22312 * doc/tm.texi: Regenerate.
22314 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22316 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22317 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22318 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22319 * config/rs6000/rs6000.md (type): Remove "brinc".
22321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22323 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22324 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22325 * config/rs6000/linuxspe.h: Delete file.
22326 * config/rs6000/rs6000.md: Don't include spe.md.
22327 * config/rs6000/spe.h: Delete file.
22328 * config/rs6000/spe.md: Delete file.
22329 * config/rs6000/t-rs6000: Remove spe.md.
22331 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22333 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22334 (reg_or_none500mem_operand): Delete.
22335 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22336 instead of reg_or_none500mem_operand.
22338 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22340 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22341 handling of SPE flags.
22342 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22344 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22346 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22348 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22349 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22350 paired_divv2sf3): Similar.
22351 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22352 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22353 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22355 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22356 Rename the paired_* instruction patterns.
22357 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22359 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22360 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22361 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22362 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22363 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22364 PAIRED_VECTOR_MODE.
22365 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22366 (spe_func_has_64bit_regs_p): Delete.
22367 (spe_expand_predicate_builtin): Delete.
22368 (spe_expand_evsel_builtin): Delete.
22369 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22370 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22371 (invalid_e500_subreg): Delete.
22372 (rs6000_legitimize_address): Always force_reg op2 as well, for
22373 paired single memory accesses.
22374 (rs6000_member_type_forces_blk): Delete.
22375 (rs6000_spe_function_arg): Delete.
22376 (rs6000_expand_unop_builtin): Delete SPE handling.
22377 (rs6000_expand_binop_builtin): Ditto.
22378 (spe_expand_stv_builtin): Delete.
22379 (bdesc_2arg_spe): Delete.
22380 (spe_expand_builtin): Delete.
22381 (spe_expand_predicate_builtin): Delete.
22382 (spe_expand_evsel_builtin): Delete.
22383 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22384 (spe_init_builtins): Delete.
22385 (spe_func_has_64bit_regs_p): Delete.
22386 (savres_routine_name): Delete "info" parameter. Adjust callers.
22387 (rs6000_emit_stack_reset): Ditto.
22388 (rs6000_dwarf_register_span): Delete.
22389 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22390 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22391 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22393 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22395 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22396 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22397 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22398 mulv2sf3, divv2sf3): Delete expanders.
22400 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22402 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22404 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22406 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22407 * config/rs6000/rs6000.c: Ditto.
22409 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22411 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22412 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22413 comparison_operator.
22415 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22417 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22418 * config/rs6000/rs6000.opt: Ditto.
22419 * config/rs6000/t-rtems: Ditto.
22421 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22423 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22424 TARGET_E500_SINGLE by 0, simplify.
22425 * config/rs6000/rs6000.c: Ditto.
22426 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22427 (spe_build_register_parallel): Delete.
22428 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22429 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22430 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22431 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22432 (E500_CONVERT): Delete.
22433 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22435 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22437 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22438 * config/rs6000/dfp.md: Ditto.
22439 (negdd2, *negdd2_fpr): Merge.
22440 (absdd2, *absdd2_fpr): Merge.
22441 (negtd2, *negtd2_fpr): Merge.
22442 (abstd2, *abstd2_fpr): Merge.
22443 * config/rs6000/e500.h: Delete file.
22444 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22445 TARGET_FPRS by 1 and simplify.
22446 * config/rs6000/rs6000-c.c: Ditto.
22447 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22448 TARGET_DF_SPE by 0.
22449 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22451 * config/rs6000/rs6000.md: Ditto.
22452 (floatdidf2, *floatdidf2_fpr): Merge.
22453 (move_from_CR_gt_bit): Delete.
22454 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22455 (E500_CR_IOR_COMPARE): Delete.
22456 (All patterns that require !TARGET_FPRS): Delete.
22457 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22459 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22461 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22463 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22465 * graphds.c (add_edge): Intitialize edge's attached data.
22466 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22467 pointer parameter. Call pointed function on each edge during
22468 graph traversing. Skip traversing the edge when the function
22470 (graphds_dfs, graphds_scc): Ditto.
22471 (for_each_edge): New parameter. Pass the new parameter to callback
22473 * graphds.h (skip_edge_callback): New function pointer type.
22474 (graphds_dfs, graphds_scc): New function pointer parameter.
22475 (graphds_edge_callback, for_each_edge): New parameter.
22477 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22479 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22480 out code checking if runtime alias check is possible to below ...
22481 Call the new function.
22482 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22483 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22485 2017-06-07 Marek Polacek <polacek@redhat.com>
22488 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22489 TYPE_OVERFLOW_WRAPS check.
22491 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22493 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22494 if versioning is required.
22495 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22496 peeling with the check for versioning.
22498 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22500 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22501 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22502 Set true to new parameter if new ssa variable is defined.
22503 (vect_gen_vector_loop_niters): Refactor. Set range information
22504 for the new vector loop bound variable.
22505 (vect_do_peeling): Ditto.
22507 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22509 * tree-affine.c (ssa.h): Include header file.
22510 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22511 has wrapping overflow behavior.
22513 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22515 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22517 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22519 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22520 (tree_to_aff_combination): ... here.
22522 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22524 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22525 reg_pressure model function.
22526 (ivopts_global_cost_for_size): Delete.
22527 (determine_set_costs, iv_ca_recount_cost): Call new model function
22528 ivopts_estimate_reg_pressure.
22530 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22532 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22533 expensive than udiv. Remove floating point cases from mod.
22535 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22537 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22538 Increase idiv cost.
22540 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22542 * config/aarch64/aarch64.md
22543 (copysignsf3): Fix mask generation.
22545 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22547 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22549 (class dump_manager): Add register_dumps method.
22550 * dumpfile.c: Include langhooks.h.
22551 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22552 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22553 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22554 (dump_manager::dump_register): Start with 512 entries instead of 32.
22555 (dump_manager::register_dumps): New method.
22556 * toplev.c (general_init): Instead of invoking register_dumps
22557 langhook, invoke register_dumps method on the dump manager.
22558 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22561 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22563 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22564 Say that # requires an associated define_split to exist, and that
22565 the define_split must be suitable for use after register allocation.
22567 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22569 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22570 (compute_outgoing_frequencies): Also initialize zero counts.
22571 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22572 around loops; preserve more of profile when nothing changes.
22574 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22576 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22578 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22579 * config/arm/arm-cpu-cdata.h: Regenerate.
22580 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22581 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22582 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22583 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22584 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22586 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22588 * config/arm/t-rmprofile: Likewise.
22589 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22591 2017-06-06 David S. Miller <davem@davemloft.net>
22594 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22595 blockage if function uses alloca.
22597 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22599 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22600 New "uid" fields to hold pretty-print IDs of group and ref.
22601 Memory references are now identified as <group_id>:<ref_id>
22602 instead of using [random] addresses.
22603 (dump_mem_details): Simplify, no functional change.
22604 (dump_mem_ref): Simplify and make output more concise.
22605 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22606 (find_or_create_group): Initialize group uid.
22607 (record_ref): Initialize ref uid. Improve debug output.
22608 (prune_group_by_reuse, should_issue_prefetch_p,)
22609 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22610 (mark_nontemporal_store, determine_loop_nest_reuse):
22611 Improve debug output.
22613 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22615 * dbgcnt.def (prefetch): New debug counter.
22616 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22617 (schedule_prefetches): Stop issueing prefetches if debug counter
22620 2017-06-06 Tom de Vries <tom@codesourcery.com>
22622 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22623 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22625 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22627 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22628 Use aarch64_reg_or_zero predicate for operand 4.
22629 (aarch64_compare_and_swap<mode> define_insn_and_split):
22630 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22631 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22633 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22635 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22636 (arm_compute_save_core_reg_mask): This.
22637 (thumb1_compute_save_reg_mask): Rename into ...
22638 (thumb1_compute_save_core_reg_mask): This.
22639 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22640 (arm_compute_frame_layout): Likewise.
22642 2017-06-06 Richard Biener <rguenther@suse.de>
22644 PR tree-optimization/80974
22645 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22646 keep or clear leaders SSA info.
22648 2017-06-06 Tom de Vries <tom@codesourcery.com>
22650 * config/nvptx/nvptx.c (split_mode_p): New function.
22651 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22653 2017-06-06 Tom de Vries <tom@codesourcery.com>
22655 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22657 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22660 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22663 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22665 * shrink-wrap.c (handle_simple_exit): Update profile.
22666 (try_shrink_wrapping): Upate profile.
22668 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22670 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22671 (tree_guess_outgoing_edge_probabilities): New.
22672 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22673 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22675 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22677 * ipa-split.c (split_function): Initialize return bb profile.
22679 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22681 * profile.c (compute_branch_probabilities): Also initialize
22682 EXIT_BLOCK profile.
22684 2017-06-06 Richard Biener <rguenther@suse.de>
22686 PR tree-optimization/80928
22687 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22688 (vect_analyze_loop_operations): Properly guard analysis for
22690 (vect_transform_loop): Likewise.
22691 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22692 (vect_model_induction_cost): Do not cost for pure SLP.
22693 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22694 of induction in inner loop vectorization.
22695 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22696 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22697 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22699 (vect_analyze_slp_cost_1): Cost induction.
22700 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22701 (vect_get_slp_vect_defs): Likewise.
22702 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22703 (vect_transform_stmt): Handle SLP reductions.
22704 * tree-vectorizer.h (vectorizable_induction): Adjust.
22706 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22708 * config/rs6000/rs6000.c (make_resolver_func): Update
22709 init_lowered_empty_function call.
22711 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22713 * doc/invoke.texi: Document the -fprofile-abs-path option.
22714 * common.opt (fprofile-abs-path): New option.
22715 * gcov-io.h (gcov_write_filename): Declare.
22716 * gcov-io.c (gcov_write_filename): New function.
22717 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22718 * profile.c (output_location): Likewise.
22720 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22722 * shring-wrap.c: Revert accidental commit.
22724 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22726 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22728 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22730 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22732 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22733 profile in callgraph edge.
22734 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22735 (apply_scale): Likewise.
22736 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22737 Also scale profile when inlining function with zero profile.
22738 (initialize_cfun): Update exit block profile even when it is zero.
22739 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22740 when profile is read.
22742 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22744 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22745 (CLONE_*): New constants to define the processors we can generate
22746 code for with the target_clone attribute.
22747 (rs6000_clone_map): New array to identify which clone processors
22748 the current program is running on.
22749 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22750 target_clone attribute.
22751 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22752 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22753 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22754 (cpu_expand_builtin): Add support for target_clone attribute.
22755 (rs6000_valid_attribute_p): Allow "default" attribute.
22756 (get_decl_name): New debug function to simplify printing the
22757 current function name in debugging statements.
22758 (rs6000_clone_priority): New functions to support the target_clone
22759 attribute, and be able to generate code to switch between ISA 2.05
22760 through ISA 3.0 (power6 through power9).
22761 (rs6000_compare_version_priority): Likewise.
22762 (rs6000_get_function_versions_dispatcher): Likewise.
22763 (make_resolver_func): Likewise.
22764 (add_condition_to_bb): Likewise.
22765 (dispatch_function_versions): Likewise.
22766 (rs6000_generate_version_dispatcher_body): Likewise.
22767 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22768 (fusion_gpr_load_p): Fix a spacing issue.
22769 * doc/extend.texi (Common Function Attributes): Document that the
22770 PowerPC supports the target_clone attribute.
22772 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22774 * config/arm/arm.h: explain F symbol found in description of ARM
22775 register allocation in its legend.
22777 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22779 * config/mips/frame-header-opt.c: Include profile-count.h.
22780 * config/riscv/riscv.c: Include profile-count.h
22782 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22784 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22786 (sm_set_flag_if_changed): Add bbs field.
22787 (execute_sm_if_changed_flag_set): Pass BBS.
22788 (execute_sm): Update.
22790 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22792 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22795 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22797 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22798 (peephole2): New peephole2 to emit the above.
22799 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22801 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22803 * config/aarch64/aarch64.c (define_peephole2 above
22804 *sub_<shift>_<mode>): New peephole.
22806 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22808 * config/i386/i386.c (make_resolver_func): Update.
22809 * Makefile.in: Add profile-count.h and profile-count.o
22810 * auto-profile.c (afdo_indirect_call): Update to new API.
22811 (afdo_set_bb_count): Update.
22812 (afdo_propagate_edge): Update.
22813 (afdo_propagate_circuit): Update.
22814 (afdo_calculate_branch_prob): Update.
22815 (afdo_annotate_cfg): Update.
22816 * basic-block.h: Include profile-count.h
22817 (struct edge_def): Turn count to profile_count.
22818 (struct basic_block_def): Likewie.
22819 (REG_BR_PROB_BASE): Move to profile-count.h
22820 (RDIV): Move to profile-count.h
22821 * bb-reorder.c (max_entry_count): Turn to profile_count.
22822 (find_traces): Update.
22823 (rotate_loop):Update.
22824 (connect_traces):Update.
22825 (sanitize_hot_paths):Update.
22826 * bt-load.c (migrate_btr_defs): Update.
22827 * cfg.c (RDIV): Remove.
22828 (init_flow): Use alloc_block.
22829 (alloc_block): Uninitialize count.
22830 (unchecked_make_edge): Uninitialize count.
22831 (check_bb_profile): Update.
22832 (dump_edge_info): Update.
22833 (dump_bb_info): Update.
22834 (update_bb_profile_for_threading): Update.
22835 (scale_bbs_frequencies_int): Update.
22836 (scale_bbs_frequencies_gcov_type): Update.
22837 (scale_bbs_frequencies_profile_count): New.
22838 * cfg.h (update_bb_profile_for_threading): Update.
22839 (scale_bbs_frequencies_profile_count): Declare.
22840 * cfgbuild.c (compute_outgoing_frequencies): Update.
22841 (find_many_sub_basic_blocks): Update.
22842 * cfgcleanup.c (try_forward_edges): Update.
22843 (try_crossjump_to_edge): Update.
22844 * cfgexpand.c (expand_gimple_tailcall): Update.
22845 (construct_exit_block): Update.
22846 * cfghooks.c (verify_flow_info): Update.
22847 (dump_bb_for_graph): Update.
22848 (split_edge): Update.
22849 (make_forwarder_block): Update.
22850 (duplicate_block): Update.
22851 (account_profile_record): Update.
22852 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22853 (get_estimated_loop_iterations): Update.
22854 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22855 (single_likely_exit): Update.
22856 * cfgloopmanip.c (scale_loop_profile): Update.
22858 (set_zero_probability): Update.
22859 (lv_adjust_loop_entry_edge): Update.
22860 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22861 (purge_dead_edges): Update.
22862 (rtl_account_profile_record): Update.
22863 * cgraph.c (cgraph_node::create): Uninitialize count.
22864 (symbol_table::create_edge): Uninitialize count.
22865 (cgraph_update_edges_for_call_stmt_node): Update.
22866 (cgraph_edge::dump_edge_flags): Update.
22867 (cgraph_node::dump): Update.
22868 (cgraph_edge::maybe_hot_p): Update.
22869 * cgraph.h: Include profile-count.h
22870 (create_clone), create_edge, create_indirect_edge): Update.
22871 (cgraph_node): Turn count to profile_count.
22872 (cgraph_edge0: Likewise.
22873 (make_speculative, clone): Update.
22874 (create_edge): Update.
22875 (init_lowered_empty_function): Update.
22876 * cgraphclones.c (cgraph_edge::clone): Update.
22877 (duplicate_thunk_for_node): Update.
22878 (cgraph_node::create_clone): Update.
22879 * cgraphunit.c (cgraph_node::analyze): Update.
22880 (cgraph_node::expand_thunk): Update.
22881 * final.c (dump_basic_block_info): Update.
22882 * gimple-streamer-in.c (input_bb): Update.
22883 * gimple-streamer-out.c (output_bb): Update.
22884 * graphite.c (print_global_statistics): Update.
22885 (print_graphite_scop_statistics): Update.
22886 * hsa-brig.c: Include basic-block.h.
22887 * hsa-dump.c: Include basic-block.h.
22888 * hsa-gen.c (T sum_slice): Update.
22889 (convert_switch_statements):Update.
22890 * hsa-regalloc.c: Include basic-block.h.
22891 * ipa-chkp.c (chkp_produce_thunks): Update.
22892 * ipa-cp.c (struct caller_statistics): Update.
22893 (init_caller_stats): Update.
22894 (gather_caller_stats): Update.
22895 (ipcp_cloning_candidate_p): Update.
22896 (good_cloning_opportunity_p): Update.
22897 (get_info_about_necessary_edges): Update.
22898 (dump_profile_updates): Update.
22899 (update_profiling_info): Update.
22900 (update_specialized_profile): Update.
22901 (perhaps_add_new_callers): Update.
22902 (decide_about_value): Update.
22903 (ipa_cp_c_finalize): Update.
22904 * ipa-devirt.c (struct odr_type_warn_count): Update.
22905 (struct decl_warn_count): Update.
22906 (struct final_warning_record): Update.
22907 (possible_polymorphic_call_targets): Update.
22908 (ipa_devirt): Update.
22909 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22910 * ipa-icf.c (sem_function::merge): Update.
22911 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22912 * ipa-inline.c (compute_uninlined_call_time): Update.
22913 (compute_inlined_call_time): Update.
22914 (want_inline_small_function_p): Update.
22915 (want_inline_self_recursive_call_p): Update.
22916 (edge_badness): Update.
22917 (lookup_recursive_calls): Update.
22918 (recursive_inlining): Update.
22919 (inline_small_functions): Update.
22920 (dump_overall_stats): Update.
22921 (dump_inline_stats): Update.
22922 * ipa-profile.c (ipa_profile_generate_summary): Update.
22923 (ipa_propagate_frequency): Update.
22924 (ipa_profile): Update.
22925 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22926 * ipa-utils.c (ipa_merge_profiles): Update.
22927 * loop-doloop.c (doloop_modify): Update.
22928 * loop-unroll.c (report_unroll): Update.
22929 (unroll_loop_runtime_iterations): Update.
22930 * lto-cgraph.c (lto_output_edge): Update.
22931 (lto_output_node): Update.
22932 (input_node): Update.
22933 (input_edge): Update.
22934 (merge_profile_summaries): Update.
22935 * lto-streamer-in.c (input_cfg): Update.
22936 * lto-streamer-out.c (output_cfg): Update.
22937 * mcf.c (create_fixup_graph): Update.
22938 (adjust_cfg_counts): Update.
22939 (sum_edge_counts): Update.
22940 * modulo-sched.c (sms_schedule): Update.
22941 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22942 * predict.c (maybe_hot_count_p): Update.
22943 (probably_never_executed): Update.
22944 (dump_prediction): Update.
22945 (combine_predictions_for_bb): Update.
22946 (propagate_freq): Update.
22947 (handle_missing_profiles): Update.
22948 (counts_to_freqs): Update.
22949 (rebuild_frequencies): Update.
22950 (force_edge_cold): Update.
22951 * predict.h: Include profile-count.h
22952 (maybe_hot_count_p, counts_to_freqs): UPdate.
22953 * print-rtl-function.c: Do not include cfg.h
22954 * print-rtl.c: Include basic-block.h
22955 * profile-count.c: New file.
22956 * profile-count.h: New file.
22957 * profile.c (is_edge_inconsistent): Update.
22958 (correct_negative_edge_counts): Update.
22959 (is_inconsistent): Update.
22960 (set_bb_counts): Update.
22961 (read_profile_edge_counts): Update.
22962 (compute_frequency_overlap): Update.
22963 (compute_branch_probabilities): Update; Initialize and deinitialize
22965 (branch_prob): Update.
22966 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22967 (edge_gcov_count): New.
22968 (bb_gcov_count): New.
22969 * shrink-wrap.c (try_shrink_wrapping): Update.
22970 * tracer.c (better_p): Update.
22971 * trans-mem.c (expand_transaction): Update.
22972 (ipa_tm_insert_irr_call): Update.
22973 (ipa_tm_insert_gettmclone_call): Update.
22974 * tree-call-cdce.c: Update.
22975 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22976 (gimple_duplicate_sese_tail): Update.
22977 (gimple_account_profile_record): Update.
22978 (execute_fixup_cfg): Update.
22979 * tree-inline.c (copy_bb): Update.
22980 (copy_edges_for_bb): Update.
22981 (initialize_cfun): Update.
22982 (freqs_to_counts): Update.
22983 (copy_cfg_body): Update.
22984 (expand_call_inline): Update.
22985 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22986 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22987 (try_unroll_loop_completely): Update.
22988 (try_peel_loop): Update.
22989 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22990 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22991 * tree-ssa-loop-split.c (connect_loops): Update.
22992 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22993 * tree-ssa-reassoc.c (branch_fixup): Update.
22994 * tree-ssa-tail-merge.c (replace_block_by): Update.
22995 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22996 (compute_path_counts): Update.
22997 (update_profile): Update.
22998 (recompute_probabilities): Update.
22999 (update_joiner_offpath_counts): Update.
23000 (estimated_freqs_path): Update.
23001 (freqs_to_counts_path): Update.
23002 (clear_counts_path): Update.
23003 (ssa_fix_duplicate_block_edges): Update.
23004 (duplicate_thread_path): Update.
23005 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23006 (struct switch_conv_info): Update.
23007 * tree-tailcall.c (decrease_profile): Update.
23008 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23009 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23010 * value-prof.c (check_counter): Update.
23011 (gimple_divmod_fixed_value): Update.
23012 (gimple_mod_pow2): Update.
23013 (gimple_mod_subtract): Update.
23014 (gimple_ic_transform): Update.
23015 (gimple_stringop_fixed_value): Update.
23016 * value-prof.h (gimple_ic): Update.
23018 2017-06-02 Carl Love <cel@us.ibm.com>
23020 * config/rs6000/rs6000-c: Add support for built-in functions
23021 vector double vec_doublee (vector signed int);
23022 vector double vec_doublee (vector unsigned int);
23023 vector double vec_doublee (vector float);
23024 vector double vec_doubleh (vector signed int);
23025 vector double vec_doubleh (vector unsigned int);
23026 vector double vec_doubleh (vector float);
23027 vector double vec_doublel (vector signed int);
23028 vector double vec_doublel (vector unsigned int);
23029 vector double vec_doublel (vector float);
23030 vector double vec_doubleo (vector signed int);
23031 vector double vec_doubleo (vector unsigned int);
23032 vector double vec_doubleo (vector float);.
23033 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23034 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23036 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23037 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23038 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23040 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23041 vec_doublel, vec_doubleh.
23042 * doc/extend.texi: Update the built-in documentation file for the
23043 new built-in functions.
23045 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23048 * ipa-inline-analysis.c (free_growth_caches): Set
23049 edge_removal_hook_holder to NULL after removing it.
23051 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23053 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23054 comparision with zero.
23056 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23057 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23058 for early expansion of vec_min and vec_max builtins.
23059 (builtin_function_type): Add min/max unsigned variants to those
23060 identified as having unsigned arguments.
23062 2017-06-02 Olivier Hainque <hainque@adacore.com>
23064 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23066 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23068 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23069 Use VALL_F16 iterator rather than VALL.
23071 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23073 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23074 Emit CBNZ inside loop when doing a strong exchange and comparing
23075 against zero. Generate the CC flags after the loop.
23077 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23079 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23080 (dl_section_ref): New.
23081 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23082 On AIX, append an expression to subtract the size of the
23083 section length to dl_section_ref.
23085 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23087 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23088 for early expansion of vector absolute builtins.
23090 2017-06-02 Richard Biener <rguenther@suse.de>
23092 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23093 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23095 2017-06-02 Richard Biener <rguenther@suse.de>
23097 PR tree-optimization/80948
23098 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23099 stmt order as well.
23101 2017-06-02 Richard Biener <rguenther@suse.de>
23103 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23105 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23106 for inductions as relevant.
23108 2017-06-02 Richard Biener <rguenther@suse.de>
23110 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23111 (vectorizable_induction): ... this. Remove dead code.
23113 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23115 * builtins. (expand_builtin_alloca): Remove second parameter and
23116 infer its value from the first parameter instead.
23117 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23119 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23121 PR rtl-optimization/80903
23122 * loop-doloop.c (add_test): Unshare sequence.
23124 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23126 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23128 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23130 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23132 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23133 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23134 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23135 (xlogue_layout::get_stub_rtx): Make static.
23136 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23137 (xlogue_layout::compute_stub_managed_regs): Rename to...
23138 (xlogue_layout::count_stub_managed_regs): ...this.
23139 (xlogue_layout::is_stub_managed_reg): New function.
23140 (xlogue_layout::m_stub_names): Rename to...
23141 (xlogue_layout::s_stub_names): ...this, make static.
23142 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23143 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23144 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23145 xlogue_layout::s_stub_names): Instantiate statics.
23146 (stub_managed_regs): Remove.
23147 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23148 (disable_call_ms2sysv_xlogues): Rename to...
23149 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23150 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23152 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23153 change after reload_completed.
23154 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23156 (ix86_expand_prologue): Likewise.
23157 (ix86_expand_epilogue): Likewise.
23158 (ix86_expand_split_stack_prologue): Likewise.
23159 (ix86_compute_frame_layout): Remove frame parameter ...
23160 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23161 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23163 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23164 (ix86_frame): Move from here ...
23165 * config/i386/i386.h (ix86_frame): ... to here.
23166 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23167 complete ix86_frame data structure instead. Remove some_ld_name.
23169 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23171 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23172 symbols that hold a DECL_VALUE_EXPR.
23174 2017-06-01 Martin Jambor <mjambor@suse.cz>
23176 PR tree-optimization/80898
23177 * tree-sra.c (process_subtree_disqualification): Removed.
23178 (disqualify_candidate): Do not acll
23179 process_subtree_disqualification.
23180 (subtree_mark_written_and_enqueue): New function.
23181 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23182 RHS has been disqualified and re-queue LHS if necessary. Apart
23183 from that, ignore disqualified RHS.
23185 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23187 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23188 address fetch for z10 or later.
23190 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23192 * config/arc/arc.md (tst_movb): Add guard when splitting.
23194 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23196 * config/arc/arc.c (arc_can_eliminate): Test against
23197 arc_frame_pointer_needed.
23199 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23201 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23202 to prevent store reordering.
23203 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23204 (type): Add block type.
23205 (stack_tie): Define special instruction to be used in
23208 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23210 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23211 constraint. It is not valid for the pattern.
23212 (noncommutative_binary_comparison): Likewise.
23214 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23216 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23219 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23221 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23222 be used by the reg-alloc.
23224 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23226 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23227 reg-alloc when having mul64 or mul32x16 instructions.
23228 (mulsidi3): Likewise.
23229 (umulsidi3): Likewise.
23230 (mulsi32x16): New pattern.
23231 (mulsi64): Likewise.
23232 (mulsidi64): Likewise.
23233 (umulsidi64): Likewise.
23234 (MUL32x16_REG): Define.
23235 (mul64_600): Use MUL32x16_REG.
23236 (mac64_600): Likewise.
23237 (umul64_600): Likewise.
23238 (umac64_600): Likewise.
23240 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23242 * config/arc/arc.md (mulsi3_700): Make it commutative.
23244 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23246 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23248 (*sign_extendsidi2_insn): Likewise for movstosw.
23250 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23252 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23253 the type of the input discriminant value. Convert the
23254 discriminant value of signedness vary.
23256 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23258 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23259 Add to -Wall section.
23261 2017-06-01 Richard Biener <rguenther@suse.de>
23263 PR middle-end/66313
23264 * fold-const.c (fold_plusminus_mult_expr): If the factored
23265 factor may be zero use a wrapping type for the inner operation.
23266 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23267 and handle moved defs.
23268 (process_assignment): Properly guard the unary op case. Return a
23269 tri-state indicating that moving the stmt before the call may allow
23270 to continue. Pass through to_move.
23271 (find_tail_calls): Handle moving unrelated defs before
23274 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23277 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23278 splitter result in the canonical way.
23280 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23282 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23283 also for 32bit target. Update insn attributes.
23284 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23286 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23288 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23289 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23290 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23291 (_mm_maskz_min_ss): New intrinsics.
23293 2017-05-31 Martin Liska <mliska@suse.cz>
23295 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23296 Change comment style to one we normally use.
23297 (vectorizable_reduction): Likewise.
23298 (vectorizable_induction): Likewise.
23299 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23300 (vectorizable_call): Likewise.
23301 (vectorizable_simd_clone_call): Likewise.
23302 (vectorizable_conversion): Likewise.
23303 (vectorizable_assignment): Likewise.
23304 (vectorizable_shift): Likewise.
23305 (vectorizable_operation): Likewise.
23306 (vectorizable_store): Likewise.
23307 (vectorizable_load): Likewise.
23308 * tree-vectorizer.h: Likewise.
23310 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23312 * passes.c (emergency_dump_function): New.
23313 * tree-pass.h (emergency_dump_function): Declare.
23314 * plugin.c (plugins_internal_error_function): Remove.
23315 * plugin.h (plugins_internal_error_function): Remove declaration.
23316 * toplev.c (internal_error_function): New static function. Use it...
23317 (general_init): ...here.
23319 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23321 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23322 (arc_rtx_costs): Add costs for new patterns.
23323 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23324 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23326 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23328 * tree-ssa-strlen.c (get_next_strinfo): New function.
23329 (get_stridx_plus_constant): Use it.
23330 (zero_length_string): Likewise.
23331 (adjust_related_strinfos): Likewise.
23332 (adjust_last_stmt): Likewise.
23334 2017-05-31 Richard Biener <rguenther@suse.de>
23337 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23338 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23340 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23342 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23343 loop_vinfo argument and use of dependence distance vectors.
23344 Check instead whether the two references differ only in their
23345 initial value and assume that they have the same alignment if the
23346 difference is a multiple of the vector alignment.
23347 (vect_analyze_data_refs_alignment): Update call accordingly.
23349 2017-05-31 Martin Liska <mliska@suse.cz>
23352 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23354 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23356 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23357 (create_intersect_range_checks): Move from ...
23358 * tree-data-ref.c (create_intersect_range_checks_index)
23359 (create_intersect_range_checks): ... to here.
23360 (create_runtime_alias_checks): New function factored from ...
23361 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23362 here. Call above function.
23363 * tree-data-ref.h (create_runtime_alias_checks): New function.
23365 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23367 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23368 segment length for dr_b and compute it in wide_int.
23370 2017-05-31 Richard Biener <rguenther@suse.de>
23372 PR tree-optimization/80906
23373 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23374 and pass through iv_map.
23375 (copy_bb_and_scalar_dependences): Adjust.
23376 (translate_pending_phi_nodes): Likewise.
23377 (copy_loop_close_phi_args): Handle code-generating IVs instead
23380 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23382 * diagnostic-color.c (color_dict): Add "type-diff".
23383 (parse_gcc_colors): Update comment.
23384 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23385 -fdiagnostics-show-template-tree and -fno-elide-type.
23386 (GCC_COLORS): Add type-diff to example.
23388 (-fdiagnostics-show-template-tree): New.
23389 (-fno-elide-type): New.
23390 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23391 the pp_format_decoder callback. Call any m_format_postprocessor's
23393 (pretty_printer::pretty_printer): Initialize
23394 m_format_postprocessor.
23395 (pretty_printer::~pretty_printer): Delete any
23396 m_format_postprocessor.
23397 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23398 (class format_postprocessor): New class.
23399 (struct pretty_printer::format_decoder): Document the new parameters.
23400 (struct pretty_printer::m_format_postprocessor): New field.
23401 * tree-diagnostic.c (default_tree_printer): Update for new
23402 bool and const char ** params.
23403 * tree-diagnostic.h (default_tree_printer): Likewise.
23405 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23407 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23408 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23409 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23410 rs6000_gen_cell_microcode code.
23411 (rs6000_final_prescan_insn): Delete.
23412 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23413 "warn-cell-microcode" entries.
23414 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23415 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23416 throughout. Change cc_reg_not_micro_cr0_operand to
23417 cc_reg_not_cr0_operand throughout.
23418 (*extendhi<mode>2_noload): Delete.
23419 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23420 (mwarn-cell-microcode): Delete.
23421 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23422 -mgen-cell-microcode and -mwarn-cell-microcode.
23424 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23427 * config/i386/constraints.md (Yd): New constraint.
23429 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23430 and (?Yd, r) alternatives. Update insn attributes.
23431 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23432 and (?*Yd, r) alternatives. Update insn attributes.
23433 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23435 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23437 * gimplify.c (gimplify_modify_expr): Don't create a
23438 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23441 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23443 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23445 2017-05-30 Richard Biener <rguenther@suse.de>
23447 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23448 and reduc_def fields.
23449 (STMT_VINFO_REDUC_TYPE): New define.
23450 (STMT_VINFO_REDUC_DEF): Likewise.
23451 (vect_force_simple_reduction): Adjust prototype.
23452 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23453 (vect_is_simple_reduction): Remove check_reduction argument.
23454 (vect_force_simple_reduction): Adjust and set
23455 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23456 (vectorizable_reduction): Do not re-do reduction analysis
23457 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23458 * tree-parloops.c (gather_scalar_reductions): Adjust.
23460 2017-05-30 Richard Biener <rguenther@suse.de>
23462 PR middle-end/80901
23463 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23466 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23468 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23469 Introduce unknown_misalignment parameter and remove vf.
23470 (vect_peeling_hash_get_lowest_cost):
23471 Pass unknown_misalignment parameter.
23472 (vect_enhance_data_refs_alignment):
23473 Fix unsupportable data ref treatment.
23475 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23477 * tree-vect-data-refs.c (vect_get_data_access_cost):
23478 Workaround for SLP handling.
23479 (vect_enhance_data_refs_alignment):
23480 Compute costs for doing no peeling at all, compare to the best
23481 peeling costs so far and avoid peeling if cheaper.
23483 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23485 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23486 Return peeling info and set costs to zero for unlimited cost
23488 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23489 with unknown misalignment. Compute and costs for unknown
23490 misalignment, compare them to the costs for known misalignment
23491 and choose the cheapest for peeling.
23493 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23495 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23496 (vect_get_peeling_costs_all_drs): Create function.
23497 (vect_peeling_hash_get_lowest_cost):
23498 Use vect_get_peeling_costs_all_drs.
23499 (vect_peeling_supportable): Create function.
23500 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23502 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23504 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23505 DR_HAS_NEGATIVE_STEP.
23506 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23507 (vect_enhance_data_refs_alignment): Use.
23508 (vect_duplicate_ssa_name_ptr_info): Use.
23509 * tree-vectorizer.h (dr_misalignment): Use.
23510 (known_alignment_for_access_p): Use.
23512 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23515 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23517 (has_section_name): New function.
23519 2017-05-30 Martin Liska <mliska@suse.cz>
23522 * auto-profile.c (get_function_decl_from_block): Fix
23525 2017-05-30 Richard Biener <rguenther@suse.de>
23527 PR middle-end/80876
23528 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23530 2017-05-30 Martin Liska <mliska@suse.cz>
23532 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23533 * dumpfile.h (struct dump_file_info): Remove ctors.
23535 2017-05-30 Martin Liska <mliska@suse.cz>
23537 * predict.def: Fix GNU coding style.
23539 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23541 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23542 Mark 'to' argument with ATTRIBUTE_UNUSED.
23544 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23546 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23547 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23548 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23551 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23553 * doc/install.texi (Options specification): Restore entry of
23554 --enable-sjlj-exceptions.
23556 2017-05-27 Michael Eager <eager@eagercon.com>
23559 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23561 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23563 * config/microblaze/microblaze.h
23564 (FIXED_REGISTERS): Update in macro.
23565 (CALL_USED_REGISTERS): Update in macro.
23567 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23569 * doc/install.texi: Add links to macOS binary distributions.
23571 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23575 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23577 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23579 2017-05-26 Martin Liska <mliska@suse.cz>
23581 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23583 2017-05-26 Martin Liska <mliska@suse.cz>
23585 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23586 always leading ';; '.
23587 (dump_bb_info): Likewise.
23588 (brief_dump_cfg): Likewise.
23589 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23590 * dumpfile.c: Remove usage of TDF_VERBOSE.
23591 * dumpfile.h (enum dump_kind): Likewise.
23592 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23593 * print-tree.c (debug_verbose): Remove.
23594 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23595 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23596 * tree-diagnostic.c (default_tree_printer): Replace
23597 TDF_DIAGNOSTIC with TDF_SLIM.
23599 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23601 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23602 in parameter loop, rather than loop_vinfo.
23603 (create_intersect_range_checks): Ditto.
23604 (vect_create_cond_for_alias_checks): Update call to above functions.
23606 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23608 PR tree-optimization/80815
23609 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23610 for merging runtime alias checks. Handle negative DR_STEPs.
23612 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23614 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23616 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23617 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23618 out code pruning runtime alias checks.
23619 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23620 factored out from above.
23621 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23623 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23625 (prune_runtime_alias_test_list): New decalaration.
23627 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23629 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23630 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23631 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23632 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23633 (operator==, comp_dr_with_seg_len_pair): Ditto.
23634 (vect_prune_runtime_alias_test_list): Ditto.
23636 2017-05-26 Martin Liska <mliska@suse.cz>
23639 * params.def: Bound partial-inlining-entry-probability param.
23641 2017-05-26 Marek Polacek <polacek@redhat.com>
23644 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23647 2017-05-26 Richard Biener <rguenther@suse.de>
23649 PR tree-optimization/80842
23650 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23653 2017-05-26 Richard Biener <rguenther@suse.de>
23655 PR tree-optimization/80844
23656 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23658 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23660 * doc/md.texi (Machine Constraints): Update x86 family
23661 machine constraints section to match 'config/i386/constraints.md'.
23663 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23665 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23667 2017-05-25 Nathan Sidwell <nathan@acm.org>
23669 * doc/invoke.texi (--enable-languages): Update documentation.
23671 2017-05-25 Martin Liska <mliska@suse.cz>
23673 * dumpfile.c: Add TDF_FOLDING.
23674 * dumpfile.h (enum dump_kind): Likewise.
23675 * genmatch.c (dt_simplify::gen_1): Use it.
23677 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23679 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23681 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23683 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23684 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23686 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23688 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23689 * match.pd (X == C): Rewrite it here.
23690 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23691 with_certain_nonzero_bits2): New predicates.
23692 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23694 2017-05-24 Nathan Sidwell <nathan@acm.org>
23696 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23699 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23702 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23704 * config/powerpcspe: New port. Files are copied from the rs6000
23705 port, with "rs6000" in filenames replaced by "powerpcspe".
23707 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23709 PR rtl-optimization/80754
23710 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23712 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23714 * config/sparc/sparc.md (length): Return the correct value for -mflat
23715 sibcalls to match output_sibcall.
23717 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23721 * config/rs6000/rs6000.c (struct machine_function): Add new field
23723 (rs6000_get_separate_components): Init that field, use it.
23724 (rs6000_components_for_bb): Use the field.
23726 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23728 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23730 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23732 PR middle-end/80823
23733 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23735 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23738 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23739 address against address_operand predicate.
23740 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23742 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23744 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23745 parameters passed indirectly.
23747 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23749 * config/i386/i386.md (*movdi_internal): Remove SSE4
23750 alternative 18 (?r, *v). Update insn attributes.
23751 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23752 Update insn attributes.
23753 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23754 Update insn attributes.
23755 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23756 alternative 1 (r, v). Remove isa attribute.
23757 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23758 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23759 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23761 2017-05-23 Tom de Vries <tom@codesourcery.com>
23763 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23766 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23768 * cgraphunit.c (symbol_table::process_new_functions): Update.
23769 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23770 (inline_generate_summary): Rename to ...
23771 (ipa_fn_summary_generate): ... this one.
23772 (inline_read_summary): Rename to ...
23773 (ipa_fn_summary_read): ... this one.
23774 (inline_write_summary): Rename to ...
23775 (ipa_fn_summary_write): ... this one.
23776 (inline_free_summary): Rename to ...
23777 (ipa_free_fn_summary): ... this one.
23778 (pass_data_local_fn_summary, pass_local_fn_summary,
23779 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23780 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23781 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23782 make_pass_ipa_fn_summary): New.
23783 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23784 inline_write_summary, inline_free_summary): Remove.
23785 (ipa_free_fn_summary) : New.
23786 * ipa-inline.c (ipa_inline): Update.
23787 (pass_ipa_inline): Do not generate summaries.
23788 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23790 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23791 and add pass_ipa_fn_summary.
23792 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23794 (make_pass_inline_parameters): Remove.
23796 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23798 * omp-low.c (struct omp_context): Remove "default_kind" member.
23801 * omp-offload.c (execute_oacc_device_lower): Remove the
23802 parallelism dimensions function attributes for unparallelized
23803 OpenACC kernels constructs.
23805 2017-05-23 Martin Liska <mliska@suse.cz>
23807 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23809 (cgraph_edge::make_speculative): Likewise.
23810 (cgraph_edge::resolve_speculation): Likewise.
23811 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23812 (cgraph_node::dump): Likewise.
23813 * cgraph.h: Likewise.
23814 * cgraphunit.c (analyze_functions): Likewise.
23815 (symbol_table::compile): Likewise.
23816 * ipa-cp.c (print_all_lattices): Likewise.
23817 (determine_versionability): Likewise.
23818 (initialize_node_lattices): Likewise.
23819 (ipcp_verify_propagated_values): Likewise.
23820 (estimate_local_effects): Likewise.
23821 (update_profiling_info): Likewise.
23822 (create_specialized_node): Likewise.
23823 (perhaps_add_new_callers): Likewise.
23824 (decide_about_value): Likewise.
23825 (decide_whether_version_node): Likewise.
23826 (identify_dead_nodes): Likewise.
23827 (ipcp_store_bits_results): Likewise.
23828 * ipa-devirt.c (dump_targets): Likewise.
23829 (ipa_devirt): Likewise.
23830 * ipa-icf.c (sem_item::dump): Likewise.
23831 (sem_function::equals): Likewise.
23832 (sem_variable::equals): Likewise.
23833 (sem_item_optimizer::read_section): Likewise.
23834 (sem_item_optimizer::execute): Likewise.
23835 (congruence_class::dump): Likewise.
23836 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23837 (dump_inline_summary): Likewise.
23838 (estimate_node_size_and_time): Likewise.
23839 (inline_analyze_function): Likewise.
23840 * ipa-inline-transform.c (inline_call): Likewise.
23841 * ipa-inline.c (report_inline_failed_reason): Likewise.
23842 (want_early_inline_function_p): Likewise.
23843 (edge_badness): Likewise.
23844 (update_edge_key): Likewise.
23845 (inline_small_functions): Likewise.
23846 * ipa-profile.c (ipa_profile): Likewise.
23847 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23848 (ipa_make_edge_direct_to_target): Likewise.
23849 (remove_described_reference): Likewise.
23850 (ipa_impossible_devirt_target): Likewise.
23851 (propagate_controlled_uses): Likewise.
23852 (ipa_print_node_params): Likewise.
23853 (ipcp_transform_function): Likewise.
23854 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23855 (propagate_pure_const): Likewise.
23856 * ipa-reference.c (generate_summary): Likewise.
23857 (read_write_all_from_decl): Likewise.
23858 (propagate): Likewise.
23859 (ipa_reference_read_optimization_summary): Likewise.
23860 * ipa-utils.c (ipa_merge_profiles): Likewise.
23861 * ipa.c (walk_polymorphic_call_targets): Likewise.
23862 (symbol_table::remove_unreachable_nodes): Likewise.
23863 (ipa_single_use): Likewise.
23864 * passes.c (execute_todo): Likewise.
23865 * predict.c (drop_profile): Likewise.
23866 * symtab.c (symtab_node::get_dump_name): New function.
23867 (symtab_node::dump_name): Likewise.
23868 (symtab_node::dump_asm_name): Likewise.
23869 (symtab_node::dump_references): Likewise.
23870 (symtab_node::dump_referring): Likewise.
23871 (symtab_node::dump_base): Likewise.
23872 (symtab_node::debug_symtab): Likewise.
23873 * tree-sra.c (convert_callers_for_node): Likewise.
23874 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23875 * value-prof.c (init_node_map): Likewise.
23877 2017-05-23 Martin Liska <mliska@suse.cz>
23879 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23880 and symtab_node::debug_symtab to symbol_table::debug.
23881 * cgraphunit.c (analyze_functions): Use the renamed function.
23882 (symbol_table::compile): Likewise.
23883 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23884 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23885 * passes.c (execute_todo): Likewise.
23886 * symtab.c (symbol_table::dump): New function.
23887 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23889 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23891 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23892 that nonconst implies exec.
23894 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23896 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23897 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23898 (inline_edge_summary_vec): Turn into ...
23899 (ipa_call_summaries): ... this one.
23900 (redirect_to_unreachable, edge_set_predicate,
23901 evaluate_properties_for_edge, inline_summary_alloc,
23902 reset_ipa_call_summary, reset_inline_summary,
23903 inline_summary_t::duplicate): Update.
23904 (inline_edge_duplication_hook): Turn to ...
23905 (ipa_call_summary_t::duplicate): ... this one.
23906 (inline_edge_removal_hook): Turn to ...
23907 (ipa_call_summary_t::remove): ... this one.
23908 (dump_inline_edge_summary): Turn to ...
23909 (dump_ipa_call_summary): ... this one.
23910 (estimate_function_body_sizes): Update.
23911 (inline_update_callee_summaries): Update.
23912 (remap_edge_change_prob): Update.
23913 (remap_edge_summaries): Update.
23914 (inline_merge_summary): Update.
23915 (do_estimate_edge_time): Update.
23916 (inline_generate_summary): Update.
23917 (inline_read_section): Update.
23918 (inline_read_summary): Update.
23919 (inline_free_summary): Update.
23920 * ipa-inline.c (can_inline_edge_p): Update.
23921 (compute_inlined_call_time): Update.
23922 (want_inline_small_function_p): Update.
23923 (edge_badness): Update.
23924 (early_inliner): Update.
23925 * ipa-inline.h (inline_edge_summary): Turn to ...
23926 (ipa_call_summary): ... this one.
23927 (ipa_call_summary_t): New class.
23928 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23929 (ipa_call_summaries): New.
23930 (inline_edge_summary): Remove.
23931 (estimate_edge_growth): Update.
23932 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23933 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23934 * ipa-split.c (execute_split_functions): Update.
23935 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23937 2017-05-23 Tom de Vries <tom@codesourcery.com>
23939 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23940 attributes): Document rdrand effective target.
23942 2017-05-23 Tom de Vries <tom@codesourcery.com>
23944 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23945 attributes): Sort alphabetically.
23947 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23949 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23951 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23954 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23955 V2DF/V2DI splat into two separate patterns, one that handles
23956 registers, and the other that only handles memory. Drop support
23957 for splatting from a GPR on ISA 2.07 and then splitting the
23958 splat into direct move and splat.
23959 (vsx_splat_<mode>_reg): Likewise.
23960 (vsx_splat_<mode>_mem): Likewise.
23962 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23964 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23966 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23968 PR middle-end/80809
23969 * omp-low.c (finish_taskreg_remap): New function.
23970 (finish_taskreg_scan): If unit size of ctx->record_type
23971 is non-constant, unshare the size expression and replace
23972 decls in it with possible outer var refs.
23974 PR middle-end/80809
23975 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23976 GOVD_SHARED rather than GOVD_PRIVATE with it.
23977 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23978 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23980 PR middle-end/80853
23981 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23982 as last argument to build_outer_var_ref for pointer bases of array
23983 section reductions.
23985 2017-05-19 Martin Sebor <msebor@redhat.com>
23987 * print-tree.c (print_node): Print DECL_READ_P flag.
23989 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23991 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23992 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23993 * cgraph.c: Likewise.
23994 * cgraphunit.c: Likewise.
23995 * gengtype.c: Likewise.
23996 * ipa-cp.c: Likewise.
23997 * ipa-devirt.c: Likewise.
23998 * ipa-icf.c: Likewise.
23999 * ipa-predicate.c: Likewise.
24000 * ipa-profile.c: Likewise.
24001 * ipa-prop.c: Likewise.
24002 * ipa-split.c: Likewise.
24004 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24005 edge_predicate_pool, dump_inline_hints,
24006 inline_summary::account_size_time, redirect_to_unreachable,
24007 edge_set_predicate, set_hint_predicate,
24008 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24009 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24010 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24011 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24012 ipa_call_summary_t::remove, initialize_growth_caches,
24013 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24014 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24015 mark_modified, unmodified_parm_1, unmodified_parm,
24016 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24017 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24018 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24019 will_be_nonconstant_predicate, record_modified_bb_info,
24020 get_minimal_bb, record_modified, param_change_prob,
24021 phi_result_unknown_predicate, predicate_for_phi_result,
24022 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24023 estimate_function_body_sizes, compute_inline_parameters,
24024 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24025 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24026 inline_update_callee_summaries, remap_edge_change_prob,
24027 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24028 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24029 inline_analyze_function, inline_summary_t::insert,
24030 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24031 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24032 inline_free_summary): Move to ipa-fnsummary.h
24033 (predicate_t): Remove.
24034 * ipa-fnsummary.c: New file.
24035 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24036 (enum inline_hints_vals, inline_hints, agg_position_info,
24037 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24038 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24039 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24040 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24041 inline_read_summary, inline_write_summary, inline_free_summary,
24042 inline_analyze_function, initialize_inline_failed,
24043 inline_merge_summary, inline_update_overall_summary,
24044 compute_inline_parameters): Move to ipa-fnsummary.h
24045 * ipa-fnsummary.h: New file.
24046 * ipa-inline-transform.h: Include ipa-inline.h.
24047 * ipa-inline.c: LIkewise.
24049 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24051 * ipa-inline.c (edge_badness): Use inlined_time instead of
24052 inline_summaries->get.
24054 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24056 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24058 2017-05-22 Nathan Sidwell <nathan@acm.org>
24060 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24061 (fdump-lang): Document 'raw' option.
24062 * dumpfile.h (TDI_tu): Delete.
24063 * dumpfile.c (dump_files): Remove translation-unit.
24064 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24066 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24068 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24069 command option from $(AWK) call.
24070 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24071 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24072 [FORMAT]: Remove handling of variable.
24073 * config/avr/t-multilib: Regenerate.
24075 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24077 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24079 (dump_inline_summary): Do not print self_time.
24080 (estimate_function_body_sizes): Do not set self_time.
24081 (compute_inline_parameters): Likewise.
24082 (inline_read_section, inline_write_summary): Do not stream self_time.
24083 * ipa-inline.h (inline_summary): Drop self_time.
24085 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24087 * ipa-inline-analysis.c (account_size_time): Rename to ...
24088 (inline_summary::account_size_time): ... this one.
24089 (reset_ipa_call_summary): Turn to ...
24090 (ipa_call_summary::reset): ... this one.
24091 (reset_inline_summary): Turn to ...
24092 (inline_summary::reset): ... this one.
24093 (inline_summary_t::remove): Update.
24094 (inline_summary_t::duplicate): Update.
24095 (ipa_call_summary_t::remove): Update.
24096 (dump_inline_summary): Update.
24097 (estimate_function_body_sizes): Update.
24098 (compute_inline_parameters): Update.
24099 (estimate_node_size_and_time): Update.
24100 (inline_merge_summary): Update.
24101 (inline_update_overall_summary): Update.
24102 (inline_read_section): Update.
24103 (inline_write_summary): Update.
24104 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24105 add account_size_time and reset member functions.
24106 (ipa_call_summary): Add reset function.
24107 * ipa-predicate.h (predicate::operator &): Constify.
24109 2017-05-22 Richard Biener <rguenther@suse.de>
24111 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24113 2017-05-19 Jason Merrill <jason@redhat.com>
24115 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24117 2017-05-19 Marek Polacek <polacek@redhat.com>
24120 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24121 TYPE_OVERFLOW_WRAPS checks.
24123 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24125 * tree-core.h (enum omp_clause_default_kind): Add
24126 "OMP_CLAUSE_DEFAULT_PRESENT".
24127 * tree-pretty-print.c (dump_omp_clause): Handle it.
24128 * gimplify.c (enum gimplify_omp_var_data): Add
24129 "GOVD_MAP_FORCE_PRESENT".
24130 (gimplify_adjust_omp_clauses_1): Map it to
24131 "GOMP_MAP_FORCE_PRESENT".
24132 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24134 * gimplify.c (oacc_default_clause): Clarify.
24136 2017-05-19 Nathan Sidwell <nathan@acm.org>
24138 LANG_HOOK_REGISTER_DUMPS
24139 * toplev.c (general_init): Call register dump lang hook.
24140 * doc/invoke.texi: Document -fdump-lang option family.
24141 * dumpfile.c (dump_files): Remove class dump here.
24142 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24143 * dumpfile.h (tree_dump_index): Remove TDI_class.
24144 * langhooks-def.h (lhd_register_dumps): Declare.
24145 (LANG_HOOKS_REGISTER_DUMPS): Define.
24146 (LANG_HOOKS_INITIALIZER): Add it.
24147 * langhooks.c (lhd_register_dumps): Define.
24148 * langhooks.h (struct lang_hooks): Add register_dumps.
24150 2017-05-19 Nathan Sidwell <nathan@acm.org>
24152 * context.h (context::set_passes): New.
24153 * context.c (context::context): Do not create pass manager.
24154 * toplev.c (general_init): Create pass manager here.
24156 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24158 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24159 use this splitter if two add or or instructions would also work for
24160 the constant we want to generate.
24162 2017-05-19 Richard Biener <rguenther@suse.de>
24165 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24166 predicate evaluation.
24168 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24170 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24172 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24175 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24177 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24178 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24179 (inline_edge_summary_vec): Turn into ...
24180 (ipa_call_summaries): ... this one.
24181 (redirect_to_unreachable, edge_set_predicate,
24182 evaluate_properties_for_edge, inline_summary_alloc,
24183 reset_ipa_call_summary, reset_inline_summary,
24184 inline_summary_t::duplicate): Update.
24185 (inline_edge_duplication_hook): Turn to ...
24186 (ipa_call_summary_t::duplicate): ... this one.
24187 (inline_edge_removal_hook): Turn to ...
24188 (ipa_call_summary_t::remove): ... this one.
24189 (dump_inline_edge_summary): Turn to ...
24190 (dump_ipa_call_summary): ... this one.
24191 (estimate_function_body_sizes): Update.
24192 (inline_update_callee_summaries): Update.
24193 (remap_edge_change_prob): Update.
24194 (remap_edge_summaries): Update.
24195 (inline_merge_summary): Update.
24196 (do_estimate_edge_time): Update.
24197 (inline_generate_summary): Update.
24198 (inline_read_section): Update.
24199 (inline_read_summary): Update.
24200 (inline_free_summary): Update.
24201 * ipa-inline.c (can_inline_edge_p): Update.
24202 (compute_inlined_call_time): Update.
24203 (want_inline_small_function_p): Update.
24204 (edge_badness): Update.
24205 (early_inliner): Update.
24206 * ipa-inline.h (inline_edge_summary): Turn to ...
24207 (ipa_call_summary): ... this one.
24208 (ipa_call_summary_t): New class.
24209 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24210 (ipa_call_summaries): New.
24211 (inline_edge_summary): Remove.
24212 (estimate_edge_growth): Update.
24213 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24214 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24215 * ipa-split.c (execute_split_functions): Update.
24216 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24218 2017-05-19 Richard Biener <rguenther@suse.de>
24220 PR middle-end/80764
24221 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24223 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24225 * config/rs6000/rs6000.c (struct machine_function): Add field
24226 fpr_is_wrapped_separately.
24227 (rs6000_get_separate_components): Use 64 components. Handle the
24228 new FPR components.
24229 (rs6000_components_for_bb): Handle the FPR components.
24230 (rs6000_emit_prologue_components): Handle the FPR components.
24231 (rs6000_emit_epilogue_components): Handle the FPR components.
24232 (rs6000_set_handled_components): Handle the FPR components.
24233 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24234 that are already separately shrink-wrapped.
24235 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24236 that are already separately shrink-wrapped.
24238 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24241 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24244 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24245 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24246 cases where the register allocator uses a move and an offsettable
24247 memory operation to/from a FPR register on ISA 2.06/2.07.
24248 (define_peephole2 for Altivec d-form store): Likewise.
24250 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24253 * config/i386/mmx.md (*mov<mode>_internal): Enable
24254 alternatives 11, 12, 13 and 14 also for 32bit targets.
24255 Remove alternatives 15, 16, 17 and 18.
24256 * config/i386/sse.md (vec_concatv2di): Change
24257 alternative (!x, *y) to (x, ?!*Yn).
24259 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24261 * dumpfile.h (enum dump_kind): Remove stray comma.
24263 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24265 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24266 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24267 predicate::num_conditions
24268 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24269 (CHANGED): turn into predicate::changed.
24270 (agg_position_info): Move to ipa-predicate.h
24271 (add_condition, predicate::add_clause, predicate::operator &=,
24272 predicate::or_with, predicate::evaluate, predicate::probability,
24273 dump_condition, dump_clause, predicate::dump,
24274 predicate::remap_after_duplication, predicate::remap_after_inlining,
24275 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24276 (evaluate_conditions_for_known_args): Update.
24277 (set_cond_stmt_execution_predicate): Update.
24278 * ipa-inline.h: Include ipa-predicate.h
24279 (condition, inline_param_summary, conditions, agg_position_info,
24280 predicate): Move to ipa-predicate.h
24281 * ipa-predicate.c: New file.
24282 * ipa-predicate.h: New file.
24284 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24286 * final.c (leaf_function_p): Check we are not in a sequence.
24288 2017-05-18 Martin Liska <mliska@suse.cz>
24290 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24291 * dumpfile.c (dump_register): Use new enum dump_kind.
24292 (get_dump_file_name): Likewise.
24293 (dump_enable_all): Likewise.
24294 (dump_switch_p_1): Likewise.
24295 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24296 * dumpfile.h (enum dump_kind): New enum type.
24297 (struct dump_file_info): Create constructor and
24298 format fields and comments.
24299 * passes.c (pass_manager::register_one_dump_file):
24301 * statistics.c (statistics_early_init): Likewise.
24302 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24303 TDF_TREE with TDF_SLIM.
24304 (gather_memory_references_ref): Likewise.
24306 2017-05-18 Martin Liska <mliska@suse.cz>
24308 * vec.h (struct vnull): Use it.
24310 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24312 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24313 (true_predicate, false_predicate, true_predicate_p,
24314 false_predicate_p): Remove.
24315 (single_cond_predicate, not_inlined_predicate): Turn to member function
24317 (add_condition): Update.
24318 (add_clause): Turn to...
24319 (predicate::add_clause): ... this one; update; allow passing NULL
24321 (and_predicates): Turn to ...
24322 (predicate::operator &=): ... this one.
24323 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24324 (or_predicates): Turn to ...
24325 (predicate::or_with): ... this one.
24326 (evaluate_predicate): Turn to ...
24327 (predicate::evaluate): ... this one.
24328 (predicate_probability): Turn to ...
24329 (predicate::probability): ... this one.
24330 (dump_condition): Update.
24331 (dump_predicate): Turn to ...
24332 (predicate::dump): ... this one.
24333 (account_size_time): Update.
24334 (edge_set_predicate): Update.
24335 (set_hint_predicate): UPdate.
24336 (evaluate_conditions_for_known_args): Update.
24337 (evaluate_properties_for_edge): Update.
24338 (remap_predicate_after_duplication): Turn to...
24339 (predicate::remap_after_duplication): ... this one.
24340 (remap_hint_predicate_after_duplication): Update.
24341 (inline_summary_t::duplicate): UPdate.
24342 (dump_inline_edge_summary): Update.
24343 (dump_inline_summary): Update.
24344 (set_cond_stmt_execution_predicate): Update.
24345 (set_switch_stmt_execution_predicate): Update.
24346 (compute_bb_predicates): Update.
24347 (will_be_nonconstant_expr_predicate): Update.
24348 (will_be_nonconstant_predicate): Update.
24349 (phi_result_unknown_predicate): Update.
24350 (predicate_for_phi_result): Update.
24351 (array_index_predicate): Update.
24352 (estimate_function_body_sizes): Update.
24353 (estimate_node_size_and_time): Update.
24354 (estimate_ipcp_clone_size_and_time): Update.
24355 (remap_predicate): Rename to ...
24356 (predicate::remap_after_inlining): ... this one.
24357 (remap_hint_predicate): Update.
24358 (inline_merge_summary): Update.
24359 (inline_update_overall_summary): Update.
24360 (estimate_size_after_inlining): Update.
24361 (read_predicate): Rename to ...
24362 (predicate::stream_in): ... this one.
24363 (read_inline_edge_summary): Update.
24364 (write_predicate): Rename to ...
24365 (predicate::stream_out): ... this one.
24366 (write_inline_edge_summary): Update.
24367 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24368 (clause_t): Turn to uint32_t
24369 (predicate): Turn to class; implement constructor and operators
24371 (size_time_entry): Update.
24372 (inline_summary): Update.
24373 (inline_edge_summary): Update.
24375 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24377 * fold-const.c (fold_binary_loc): Move transformation...
24378 * match.pd (C - X CMP X): ... here.
24380 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24382 * config/sparc/sparc.c (sparc_option_override): Set function
24383 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24384 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24386 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24388 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24390 2017-05-18 Marek Polacek <polacek@redhat.com>
24393 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24394 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24395 gimple_assign_load_p.
24397 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24399 PR middle-end/80692
24400 * real.c (do_compare): Give decimal_do_compare preference over
24401 comparing just the signs.
24403 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24405 * doc/md.texi (Canonicalization of Instructions): Describe the
24406 canonical form of instructions that inherently set a condition
24409 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24411 PR middle-end/80775
24412 * tree-cfg.c: Move deletion of unreachable case statements to after
24413 the merging of consecutive case labels.
24415 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24417 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24418 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24419 restoring of callee-saved registers.
24421 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24423 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24424 * config/visium/visium.c (single_set_and_flags): Likewise.
24425 * config/visium/visium.md (Substitutions): Likewise.
24427 2017-05-17 Martin Liska <mliska@suse.cz>
24429 * cfg.c: Introduce dump_flags_t type and
24430 use it instead of int type.
24432 * cfghooks.c: Likewise.
24433 * cfghooks.h (struct cfg_hooks): Likewise.
24434 * cfgrtl.c: Likewise.
24435 * cfgrtl.h: Likewise.
24436 * cgraph.c (cgraph_node::get_body): Likewise.
24437 * coretypes.h: Likewise.
24438 * domwalk.c: Likewise.
24439 * domwalk.h: Likewise.
24440 * dumpfile.c (struct dump_option_value_info): Likewise.
24441 (dump_enable_all): Likewise.
24442 (dump_switch_p_1): Likewise.
24443 (opt_info_switch_p): Likewise.
24444 * dumpfile.h (enum tree_dump_index): Likewise.
24445 (struct dump_file_info): Likewise.
24446 * genemit.c: Likewise.
24447 * generic-match-head.c: Likewise.
24448 * gengtype.c (open_base_files): Likewise.
24449 * gimple-pretty-print.c: Likewise.
24450 * gimple-pretty-print.h: Likewise.
24451 * graph.c (print_graph_cfg): Likewise.
24452 * graphite-scop-detection.c (dot_all_sese): Likewise.
24453 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24454 * loop-unroll.c (report_unroll): Likewise.
24455 * passes.c (pass_manager::register_one_dump_file): Likewise.
24456 * print-tree.c: Likewise.
24457 * statistics.c: Likewise.
24458 * tree-cfg.c: Likewise.
24459 * tree-cfg.h: Likewise.
24460 * tree-dfa.c: Likewise.
24461 * tree-dfa.h: Likewise.
24462 * tree-dump.c (dump_function): Likewise.
24463 * tree-dump.h (struct dump_info): Likewise.
24464 * tree-pretty-print.c: Likewise.
24465 * tree-pretty-print.h: Likewise.
24466 * tree-ssa-live.c: Likewise.
24467 * tree-ssa-live.h: Likewise.
24468 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24469 * tree-vect-loop.c: Likewise.
24470 * tree-vect-slp.c: Likewise.
24472 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24473 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24475 PR tree-optimization/80457
24476 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24477 of all arguments to a statement as scalar_to_vec operations.
24478 (vectorizable_call): Adjust call to vect_model_simple_cost for
24480 (vectorizable_conversion): Likewise.
24481 (vectorizable_assignment): Likewise.
24482 (vectorizable_shift): Likewise.
24483 (vectorizable_operation): Likewise.
24484 (vectorizable_comparison): Likewise.
24485 (vect_is_simple_cond): Record the def types for operands.
24486 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24487 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24488 for statement argument count.
24490 2017-05-16 Carl Love <cel@us.ibm.com>
24492 * config/rs6000/rs6000-c: Add support for built-in functions
24493 vector unsigned long long vec_bperm (vector unsigned long long,
24494 vector unsigned char)
24495 vector signed long long vec_mule (vector signed int,
24497 vector unsigned long long vec_mule (vector unsigned int,
24498 vector unsigned int)
24499 vector signed long long vec_mulo (vector signed int,
24501 vector unsigned long long vec_mulo (vector unsigned int,
24502 vector unsigned int)
24503 vector signed char vec_sldw (vector signed char,
24504 vector signed char,
24506 vector unsigned char vec_sldw (vector unsigned char,
24507 vector unsigned char,
24509 vector signed short vec_sldw (vector signed short,
24510 vector signed short,
24512 vector unsigned short vec_sldw (vector unsigned short,
24513 vector unsigned short,
24515 vector signed int vec_sldw (vector signed int,
24518 vector unsigned int vec_sldw (vector unsigned int,
24519 vector unsigned int,
24521 vector signed long long vec_sldw (vector signed long long,
24522 vector signed long long,
24524 vector unsigned long long vec_sldw (vector unsigned long long,
24525 vector unsigned long long,
24527 * config/rs6000/rs6000-c: Add support for built-in functions
24528 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24529 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24530 * doc/extend.texi: Update the built-in documentation for the
24531 new built-in functions.
24533 2017-05-16 Marek Polacek <polacek@redhat.com>
24537 * tree.c (save_expr): Don't fold the expression.
24539 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24541 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24542 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24543 and (?*y,m). Update insn attributes.
24545 2017-05-16 Martin Liska <mliska@suse.cz>
24547 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24548 flags argument of print_gimple_stmt, print_gimple_expr,
24549 print_generic_stmt and print_generic_expr.
24550 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24551 * coretypes.h: Likewise.
24552 * except.c (dump_eh_tree): Likewise.
24553 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24554 * gimple-pretty-print.h: Likewise.
24555 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24556 (backprop::push_to_worklist): Likewise.
24557 (backprop::pop_from_worklist): Likewise.
24558 (backprop::process_use): Likewise.
24559 (backprop::intersect_uses): Likewise.
24560 (note_replacement): Likewise.
24561 * gimple-ssa-store-merging.c
24562 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24563 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24564 (pass_store_merging::execute): Likewise.
24565 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24566 (ssa_base_cand_dump_callback): Likewise.
24567 (dump_incr_vec): Likewise.
24568 (replace_refs): Likewise.
24569 (replace_mult_candidate): Likewise.
24570 (create_add_on_incoming_edge): Likewise.
24571 (create_phi_basis): Likewise.
24572 (insert_initializers): Likewise.
24573 (all_phi_incrs_profitable): Likewise.
24574 (introduce_cast_before_cand): Likewise.
24575 (replace_one_candidate): Likewise.
24576 * gimplify.c (gimplify_expr): Likewise.
24577 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24578 (set_rename): Likewise.
24579 (rename_uses): Likewise.
24580 (copy_loop_phi_nodes): Likewise.
24581 (add_close_phis_to_merge_points): Likewise.
24582 (copy_loop_close_phi_args): Likewise.
24583 (copy_cond_phi_args): Likewise.
24584 (graphite_copy_stmts_from_block): Likewise.
24585 (translate_pending_phi_nodes): Likewise.
24586 * graphite-poly.c (print_pdr): Likewise.
24587 (dump_gbb_cases): Likewise.
24588 (dump_gbb_conditions): Likewise.
24589 (print_scop_params): Likewise.
24590 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24591 (build_cross_bb_scalars_use): Likewise.
24592 (gather_bbs::before_dom_children): Likewise.
24593 * hsa-dump.c (dump_hsa_immed): Likewise.
24594 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24595 (get_replacement_map): Likewise.
24596 * ipa-inline-analysis.c (dump_condition): Likewise.
24597 (estimate_function_body_sizes): Likewise.
24598 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24599 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24600 * ipa-prop.c (ipa_dump_param): Likewise.
24601 (ipa_print_node_jump_functions_for_edge): Likewise.
24602 (ipa_modify_call_arguments): Likewise.
24603 (ipa_modify_expr): Likewise.
24604 (ipa_dump_param_adjustments): Likewise.
24605 (ipa_dump_agg_replacement_values): Likewise.
24606 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24607 * ipa-pure-const.c (check_stmt): Likewise.
24608 (pass_nothrow::execute): Likewise.
24609 * ipa-split.c (execute_split_functions): Likewise.
24610 * omp-offload.c (dump_oacc_loop_part): Likewise.
24611 (dump_oacc_loop): Likewise.
24612 * trans-mem.c (tm_log_emit): Likewise.
24613 (tm_memopt_accumulate_memops): Likewise.
24614 (dump_tm_memopt_set): Likewise.
24615 (dump_tm_memopt_transform): Likewise.
24616 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24617 (print_loop): Likewise.
24618 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24619 (chkp_gather_checks_info): Likewise.
24620 (chkp_get_check_result): Likewise.
24621 (chkp_remove_check_if_pass): Likewise.
24622 (chkp_use_outer_bounds_if_possible): Likewise.
24623 (chkp_reduce_bounds_lifetime): Likewise.
24624 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24625 (chkp_mark_completed_bounds): Likewise.
24626 (chkp_register_incomplete_bounds): Likewise.
24627 (chkp_mark_invalid_bounds): Likewise.
24628 (chkp_maybe_copy_and_register_bounds): Likewise.
24629 (chkp_build_returned_bound): Likewise.
24630 (chkp_get_bound_for_parm): Likewise.
24631 (chkp_build_bndldx): Likewise.
24632 (chkp_get_bounds_by_definition): Likewise.
24633 (chkp_generate_extern_var_bounds): Likewise.
24634 (chkp_get_bounds_for_decl_addr): Likewise.
24635 * tree-chrec.c (chrec_apply): Likewise.
24636 * tree-data-ref.c (dump_data_reference): Likewise.
24637 (dump_subscript): Likewise.
24638 (dump_data_dependence_relation): Likewise.
24639 (analyze_overlapping_iterations): Likewise.
24640 * tree-inline.c (expand_call_inline): Likewise.
24641 (tree_function_versioning): Likewise.
24642 * tree-into-ssa.c (dump_defs_stack): Likewise.
24643 (dump_currdefs): Likewise.
24644 (dump_names_replaced_by): Likewise.
24645 (dump_update_ssa): Likewise.
24646 (update_ssa): Likewise.
24647 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24648 * tree-parloops.c (build_new_reduction): Likewise.
24649 (try_create_reduction_list): Likewise.
24650 (ref_conflicts_with_region): Likewise.
24651 (oacc_entry_exit_ok_1): Likewise.
24652 (oacc_entry_exit_single_gang): Likewise.
24653 * tree-pretty-print.h: Likewise.
24654 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24655 (get_scalar_evolution): Likewise.
24656 (add_to_evolution): Likewise.
24657 (get_loop_exit_condition): Likewise.
24658 (analyze_evolution_in_loop): Likewise.
24659 (analyze_initial_condition): Likewise.
24660 (analyze_scalar_evolution): Likewise.
24661 (instantiate_scev): Likewise.
24662 (number_of_latch_executions): Likewise.
24663 (gather_chrec_stats): Likewise.
24664 (final_value_replacement_loop): Likewise.
24665 (scev_const_prop): Likewise.
24666 * tree-sra.c (dump_access): Likewise.
24667 (disqualify_candidate): Likewise.
24668 (create_access): Likewise.
24669 (reject): Likewise.
24670 (maybe_add_sra_candidate): Likewise.
24671 (create_access_replacement): Likewise.
24672 (analyze_access_subtree): Likewise.
24673 (analyze_all_variable_accesses): Likewise.
24674 (sra_modify_assign): Likewise.
24675 (initialize_constant_pool_replacements): Likewise.
24676 (find_param_candidates): Likewise.
24677 (decide_one_param_reduction): Likewise.
24678 (replace_removed_params_ssa_names): Likewise.
24679 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24680 * tree-ssa-copy.c (dump_copy_of): Likewise.
24681 (copy_prop_visit_cond_stmt): Likewise.
24682 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24683 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24684 (record_equivalences_from_stmt): Likewise.
24685 * tree-ssa-dse.c (compute_trims): Likewise.
24686 (delete_dead_call): Likewise.
24687 (delete_dead_assignment): Likewise.
24688 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24689 (forward_propagate_into_cond): Likewise.
24690 (pass_forwprop::execute): Likewise.
24691 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24692 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24694 (move_computations_worker): Likewise.
24695 (execute_sm): Likewise.
24696 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24697 (remove_exits_and_undefined_stmts): Likewise.
24698 (remove_redundant_iv_tests): Likewise.
24699 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24700 (adjust_iv_update_pos): Likewise.
24701 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24702 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24703 (value_replacement): Likewise.
24704 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24705 * tree-ssa-pre.c (print_pre_expr): Likewise.
24706 (get_representative_for): Likewise.
24707 (create_expression_by_pieces): Likewise.
24708 (insert_into_preds_of_block): Likewise.
24709 (eliminate_insert): Likewise.
24710 (eliminate_dom_walker::before_dom_children): Likewise.
24711 (eliminate): Likewise.
24712 (remove_dead_inserted_code): Likewise.
24713 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24714 * tree-ssa-reassoc.c (get_rank): Likewise.
24715 (eliminate_duplicate_pair): Likewise.
24716 (eliminate_plus_minus_pair): Likewise.
24717 (eliminate_not_pairs): Likewise.
24718 (undistribute_ops_list): Likewise.
24719 (eliminate_redundant_comparison): Likewise.
24720 (update_range_test): Likewise.
24721 (optimize_range_tests_var_bound): Likewise.
24722 (optimize_vec_cond_expr): Likewise.
24723 (rewrite_expr_tree): Likewise.
24724 (rewrite_expr_tree_parallel): Likewise.
24725 (linearize_expr): Likewise.
24726 (break_up_subtract): Likewise.
24727 (linearize_expr_tree): Likewise.
24728 (attempt_builtin_powi): Likewise.
24729 (attempt_builtin_copysign): Likewise.
24730 (transform_stmt_to_copy): Likewise.
24731 (transform_stmt_to_multiply): Likewise.
24732 (dump_ops_vector): Likewise.
24733 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24734 (print_scc): Likewise.
24735 (set_ssa_val_to): Likewise.
24736 (visit_reference_op_store): Likewise.
24737 (visit_use): Likewise.
24738 (sccvn_dom_walker::before_dom_children): Likewise.
24739 (run_scc_vn): Likewise.
24740 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24742 (expr_hash_elt::print): Likewise.
24743 (const_and_copies::pop_to_marker): Likewise.
24744 (const_and_copies::record_const_or_copy_raw): Likewise.
24745 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24746 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24747 (dump_predicates): Likewise.
24748 (find_uninit_use): Likewise.
24749 (warn_uninitialized_phi): Likewise.
24750 (pass_late_warn_uninitialized::execute): Likewise.
24751 * tree-ssa.c (verify_vssa): Likewise.
24752 (verify_ssa): Likewise.
24753 (maybe_optimize_var): Likewise.
24754 * tree-vrp.c (dump_value_range): Likewise.
24755 (dump_all_value_ranges): Likewise.
24756 (dump_asserts_for): Likewise.
24757 (register_edge_assert_for_2): Likewise.
24758 (vrp_visit_cond_stmt): Likewise.
24759 (vrp_visit_switch_stmt): Likewise.
24760 (vrp_visit_stmt): Likewise.
24761 (vrp_visit_phi_node): Likewise.
24762 (simplify_cond_using_ranges_1): Likewise.
24763 (fold_predicate_in): Likewise.
24764 (evrp_dom_walker::before_dom_children): Likewise.
24765 (evrp_dom_walker::push_value_range): Likewise.
24766 (evrp_dom_walker::pop_value_range): Likewise.
24767 (execute_early_vrp): Likewise.
24769 2017-05-16 Richard Biener <rguenther@suse.de>
24771 * dwarf2out.c (loc_list_from_tree_1): Do not create
24772 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24774 2017-05-16 Richard Biener <rguenther@suse.de>
24776 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24778 (note_variable_value_in_expr): If we resolved the decl ref
24779 do not push to the stack.
24781 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24783 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24784 operations in fast-math mode.
24785 (vaddq_f16): Likewise.
24786 (vmul_f16): Likewise.
24787 (vmulq_f16): Likewise.
24788 (vsub_f16): Likewise.
24789 (vsubq_f16): Likewise.
24790 * config/arm/neon.md (add<mode>3): New.
24792 (fma:<VH:mode>3): New. Also remove outdated comment.
24795 2017-05-16 Martin Liska <mliska@suse.cz>
24799 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24800 (odr_types_equivalent_p): Likewise.
24802 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24804 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24806 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24809 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24810 non-interunit SSE move alternatives with '?'.
24811 (zero-extendsidi peephole2): New peephole to skip intermediate
24812 general register in SSE zero-extend sequence.
24814 2017-05-15 Jeff Law <law@redhat.com>
24816 * reorg.c (relax_delay_slots): Create a new variable to hold
24817 the temporary target rather than clobbering TARGET_LABEL.
24819 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24820 missing argument to extract_bit_field call.
24821 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24823 2017-05-15 Martin Liska <mliska@suse.cz>
24826 * gcc.c (process_command): Do not allow empty argument of -o option.
24828 2017-05-15 Renlin Li <renlin.li@arm.com>
24830 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24831 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24832 * config/aarch64/constraints.md (Usf): Add long call check.
24833 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24834 (call_value): Likewise.
24835 (sibcall): Likewise.
24836 (sibcall_value): Likewise.
24838 (call_value_insn): New.
24839 (sibcall_insn): Update rtx pattern.
24840 (sibcall_value_insn): Likewise.
24841 (call_internal): Remove.
24842 (call_value_internal): Likewise.
24843 (sibcall_internal): Likewise.
24844 (sibcall_value_internal): Likewise.
24845 (call_reg): Likewise.
24846 (call_symbol): Likewise.
24847 (call_value_reg): Likewise.
24848 (call_value_symbol): Likewise.
24850 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24853 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24855 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24857 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24858 compatible with CCGOCmode and with CCZmode.
24860 2017-05-14 Martin Sebor <msebor@redhat.com>
24862 PR middle-end/77671
24863 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24864 (gimple_fold_builtin_snprintf): Same.
24865 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24866 (gimple_fold_builtin_snprintf): Same.
24867 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24868 of character types.
24869 (is_call_safe): New function.
24870 (try_substitute_return_value): Call it.
24871 (try_simplify_call): New function.
24872 (pass_sprintf_length::handle_gimple_call): Call it.
24874 2017-05-14 Martin Sebor <msebor@redhat.com>
24876 PR middle-end/80669
24877 * builtins.c (expand_builtin_stpncpy): Simplify.
24879 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24881 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24882 * config/i386/i386.h
24883 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24884 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24885 (struct machine_function): Add new members call_ms2sysv,
24886 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24887 (struct machine_frame_state): New fields sp_realigned and
24888 sp_realigned_offset.
24889 * config/i386/i386.c
24890 (enum xlogue_stub): New enum.
24891 (enum xlogue_stub_sets): New enum.
24892 (class xlogue_layout): New class.
24893 (struct ix86_frame): New fields stack_realign_allocate_offset,
24894 stack_realign_offset and outlined_save_offset. Modify comments to
24895 detail stack layout when using out-of-line stubs.
24896 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24897 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24898 -mcall-ms2sysv-xlogues.
24899 (stub_managed_regs): New static variable.
24900 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24901 registers managed by out-of-line stub.
24902 (disable_call_ms2sysv_xlogues): New function.
24903 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24904 m->call_ms2sysv when appropriate and compute frame layout for
24906 (sp_valid_at, fp_valid_at): New inline functions.
24907 (choose_basereg): New function.
24908 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24910 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24911 Use align parameter of choose_baseaddr to generated aligned SSE movs
24913 (pro_epilogue_adjust_stack): Modify to track
24914 machine_frame_state::sp_realigned.
24915 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24916 (ix86_nsaved_sseregs): Likewise.
24917 (ix86_emit_save_regs): Likewise.
24918 (ix86_emit_save_regs_using_mov): Likewise.
24919 (ix86_emit_save_sse_regs_using_mov): Likewise.
24920 (get_scratch_register_on_entry): Likewise.
24921 (gen_frame_set): New function.
24922 (gen_frame_load): Likewise.
24923 (gen_frame_store): Likewise.
24924 (emit_outlined_ms2sysv_save): Likewise.
24925 (emit_outlined_ms2sysv_restore): Likewise.
24926 (ix86_expand_prologue): Modify stack re-alignment code and call
24927 emit_outlined_ms2sysv_save when appropriate.
24928 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24929 parameter rtx_insn *insn, which allows the function to be used to only
24930 generate the notes.
24931 (ix86_expand_epilogue): Modify validity checks of frame and stack
24932 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24933 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24934 * config/i386/predicates.md
24935 (save_multiple): New predicate.
24936 (restore_multiple): Likewise.
24937 * config/i386/sse.md
24938 (save_multiple<mode>): New pattern.
24939 (save_multiple_realign<mode>): Likewise.
24940 (restore_multiple<mode>): Likewise.
24941 (restore_multiple_and_return<mode>): Likewise.
24942 (restore_multiple_leave_return<mode>): Likewise.
24943 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24945 2017-05-14 Julia Koval <julia.koval@intel.com>
24947 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24948 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24949 (__builtin_ia32_xsetbv): New builtins.
24950 * config/i386/i386.c (ix86_expand_special_args_builtin):
24952 (ix86_expand_builtin): Special expand for new intrinsics.
24953 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24954 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24955 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24957 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24959 * cfganal.c (inverted_post_order_compute): Change argument type
24961 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24962 * df-core.c (rest_of_handle_df_initialize): Adjust.
24963 (rest_of_handle_df_finish): Likewise.
24964 (df_analyze_1): Likewise.
24965 (df_analyze): Likewise.
24966 (loop_inverted_post_order_compute): Change argument to be a vec *.
24967 (df_analyze_loop): Adjust.
24968 (df_get_n_blocks): Likewise.
24969 (df_get_postorder): Likewise.
24970 * df.h (struct df_d): Change field to be a vec.
24971 * lcm.c (compute_laterin): Adjust.
24972 (compute_available): Likewise.
24973 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24974 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24975 * tree-ssa-pre.c (compute_antic): Likewise.
24977 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24979 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24980 (depth_first_search::depth_first_search): Change structure init
24981 function to this constructor.
24982 (depth_first_search::add_bb): Rename function to this member.
24983 (depth_first_search::execute): Likewise.
24984 (flow_dfs_compute_reverse_finish): Adjust.
24986 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24988 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24989 (longest_simple_path): Likewise.
24990 * shrink-wrap.c (spread_components): Likewise.
24991 (disqualify_problematic_components): Likewise.
24992 (emit_common_heads_for_components): Likewise.
24993 (emit_common_tails_for_components): Likewise.
24994 (insert_prologue_epilogue_for_components): Likewise.
24996 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24998 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25001 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25003 * df-core.c (df_set_blocks): Start using auto_bitmap.
25004 (df_compact_blocks): Likewise.
25005 * df-problems.c (df_rd_confluence_n): Likewise.
25006 * df-scan.c (df_insn_rescan_all): Likewise.
25007 (df_process_deferred_rescans): Likewise.
25008 (df_update_entry_block_defs): Likewise.
25009 (df_update_exit_block_uses): Likewise.
25010 (df_entry_block_bitmap_verify): Likewise.
25011 (df_exit_block_bitmap_verify): Likewise.
25012 (df_scan_verify): Likewise.
25013 * lra-constraints.c (lra_constraints): Likewise.
25014 (undo_optional_reloads): Likewise.
25015 (lra_undo_inheritance): Likewise.
25016 * lra-remat.c (calculate_gen_cands): Likewise.
25017 (do_remat): Likewise.
25018 * lra-spills.c (assign_spill_hard_regs): Likewise.
25019 (spill_pseudos): Likewise.
25020 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25021 (bitmap_set_subtract_values): Likewise.
25023 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25025 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25026 management with auto_bitmap.
25027 (fix_inter_tick): Likewise.
25028 (fix_recovery_deps): Likewise.
25029 * ira.c (add_store_equivs): Likewise.
25030 (find_moveable_pseudos): Likewise.
25031 (split_live_ranges_for_shrink_wrap): Likewise.
25032 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25033 (rtx_reuse_manager::seen_def_p): Likewise.
25034 (rtx_reuse_manager::set_seen_def): Likewise.
25035 * print-rtl.h (class rtx_reuse_manager): Likewise.
25037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25039 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25041 (migrate_btr_def): Likewise.
25042 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25043 * df-core.c (loop_post_order_compute): Likewise.
25044 (loop_inverted_post_order_compute): Likewise.
25045 * hsa-common.h: Likewise.
25046 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25047 * init-regs.c (initialize_uninitialized_regs): Likewise.
25048 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25049 (inline_small_functions): Likewise.
25050 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25051 * ira.c (combine_and_move_insns): Likewise.
25052 (build_insn_chain): Likewise.
25053 * loop-invariant.c (find_invariants): Likewise.
25054 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25055 * predict.c (tree_predict_by_opcode): Likewise.
25056 (predict_paths_leading_to): Likewise.
25057 (predict_paths_leading_to_edge): Likewise.
25058 (estimate_loops_at_level): Likewise.
25059 (estimate_loops): Likewise.
25060 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25061 (spread_components): Likewise.
25062 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25063 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25064 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25065 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25066 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25067 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25068 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25069 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25070 (mark_threaded_blocks): Likewise.
25071 (thread_through_all_blocks): Likewise.
25072 * tree-ssa.c (verify_ssa): Likewise.
25073 (execute_update_addresses_taken): Likewise.
25074 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25076 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25078 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25080 (post_order_compute): Likewise.
25081 (inverted_post_order_compute): Likewise.
25082 (pre_and_rev_post_order_compute_fn): Likewise.
25084 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25086 * genrecog.c (int_set::int_set): Explicitly construct our
25087 auto_vec base class.
25088 * vec.h (auto_vec::auto_vec): New constructor.
25090 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25092 * bitmap.h (class auto_bitmap): New constructor taking
25093 bitmap_obstack * argument.
25095 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25097 * bitmap.h (class auto_bitmap): Change type of m_bits to
25098 bitmap_head, and adjust ctor / dtor and member operators.
25100 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25102 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25103 when returned register mode doesn't match original mode.
25105 2017-05-12 Jeff Law <law@redhat.com>
25106 Jakub Jelinek <jakub@redhat.com>
25108 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25109 we look for cc setter after the compare-elim changes.
25110 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25111 within the vector to match what compare-elim now expects.
25112 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25113 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25115 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25116 after the compare-elim changes.
25117 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25118 the vector to match what compare-elim now expects.
25119 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25120 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25121 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25122 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25123 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25125 * config/visium/visium.c (single_set_and_flags): Fix where
25126 we look for cc setter after the compare-elim changes.
25127 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25128 with the vector to match what compare-elim now expects.
25129 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25130 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25131 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25132 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25133 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25134 (neg<mode>2_insn_set_overflow): Likewise.
25136 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25138 PR middle-end/79794
25139 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25140 maybe_expand_insn call, set ops[0].target. If still set after call,
25141 set alt_rtl. Add extra arg to recursive calls.
25142 (extract_bit_field): Add alt_rtl argument. Pass to
25144 * expmed.h (extract_bit_field): Fix prototype.
25145 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25146 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25147 to extract_bit_field_calls.
25148 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25149 Pass alt_rtl to extract_bit_field calls.
25150 * calls.c (store_unaligned_arguments_into_psuedos)
25151 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25152 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25154 * optabs.h (struct expand_operand): Add target bitfield.
25156 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25158 * compare-elim.c (try_eliminate_compare): Canonicalize
25159 operation with embedded compare to
25160 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25161 (set (reg) (operation)].
25163 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25165 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25168 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25169 cost of adding a carry flag for ADC instruction.
25170 [case MINUS]: Ignore the cost of subtracting a carry flag
25171 for SBB instruction.
25173 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25175 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25177 * config/rs6000/bmiintrin.h: New file.
25178 * config/rs6000/bmi2intrin.h: New file.
25179 * config/rs6000/x86intrin.h: New file.
25181 2017-05-12 Jeff Law <law@redhat.com>
25183 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25186 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25188 PR middle-end/80707
25189 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25191 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25193 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25194 early expansion of vector divide builtins.
25195 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25196 builtins identified as having unsigned arguments.
25198 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25200 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25201 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25202 expansion of vector logical operations (and, andc, or, xor,
25205 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25207 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25208 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25210 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25212 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25213 early GIMPLE expansion of vector multiplies.
25215 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25217 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25218 TARGET_HAVE_MOVT conditional.
25219 (movt splitter): Likewise.
25221 2017-05-12 Richard Biener <rguenther@suse.de>
25223 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25224 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25225 Fold all stmts not inplace.
25227 2017-05-12 Richard Biener <rguenther@suse.de>
25229 PR tree-optimization/80713
25230 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25231 inserted_exprs bit for not removed stmts.
25233 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25235 PR middle-end/69921
25236 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25237 parallelized" attribute for parallelized OpenACC kernels.
25238 * omp-offload.c (execute_oacc_device_lower): Use it.
25240 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25241 Set "oacc kernels" attribute.
25242 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25243 parameter. Adjust all users.
25244 (oacc_fn_attrib_kernels_p): Remove function.
25245 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25246 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25247 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25248 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25249 assert "oacc kernels" attribute is set.
25251 2017-05-11 Carl Love <cel@us.ibm.com>
25253 * config/rs6000/rs6000-c: Add support for built-in functions
25254 vector unsigned char vec_popcnt (vector signed char)
25255 vector unsigned char vec_popcnt (vector unsigned char)
25256 vector unsigned short vec_popcnt (vector signed short)
25257 vector unsigned short vec_popcnt (vector unsigned short)
25258 vector unsigned int vec_popcnt (vector signed int)
25259 vector unsigned int vec_popcnt (vector unsigned int)
25260 vector unsigned long long vec_popcnt (vector signed long long)
25261 vector unsigned long long vec_popcnt (vector unsigned long long)
25262 vector signed long long vec_slo (vector signed long long,
25263 vector signed char)
25264 vector signed long long vec_slo (vector signed long long,
25265 vector unsigned char)
25266 vector unsigned long long vec_slo (vector unsigned long long,
25267 vector signed char)
25268 vector unsigned long long vec_slo (vector unsigned long long,
25269 vector unsigned char)
25270 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25271 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25272 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25273 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25274 * doc/extend.texi: Update the built-in documentation file for the
25275 new built-in functions.
25277 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25279 * attribs.h (sorted_attr_string): Move machine independent
25280 functions for target clone support from the i386 port to common
25281 code. Rename ix86_function_versions to common_function_versions.
25282 Rename make_name to make_unique_name.
25283 (common_function_versions): Likewise.
25284 (make_unique_name): Likewise.
25285 (make_dispatcher_decl): Likewise.
25286 (is_function_default_version): Likewise.
25287 * attribs.c (attr_strcmp): Likewise.
25288 (sorted_attr_string): Likewise.
25289 (common_function_versions): Likewise.
25290 (make_unique_name): Likewise.
25291 (make_dispatcher_decl): Likewise.
25292 (is_function_default_version): Likewise.
25293 * config/i386/i386.c (attr_strcmp): Likewise.
25294 (sorted_attr_string): Likewise.
25295 (ix86_function_versions): Likewise.
25296 (make_name): Likewise.
25297 (make_dispatcher_decl): Likewise.
25298 (is_function_default_version): Likewise.
25299 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25301 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25304 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25305 Account for direct move costs for vec_construct of integer
25308 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25311 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25312 (UNSPEC_STX_ATOMIC): Ditto.
25313 (loaddi_via_sse): New insn.
25314 (storedi_via_sse): Ditto.
25315 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25316 Update corresponding peephole2 patterns.
25317 (atomic_storedi_fpu): Ditto.
25319 2017-05-11 Julia Koval <julia.koval@intel.com>
25321 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25322 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25324 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25325 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25326 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25328 2017-05-11 Nathan Sidwell <nathan@acm.org>
25330 * graphite-poly.c: Include dumpfile.h.
25332 * dumpfle.h (dump_function): Declare here ...
25333 * tree-dump.h (dump_function): ... not here.
25334 * dumpfile.c: #include tree-cfg.h.
25335 (dump_function): Move here from ...
25336 * tree-dump.c (dump_function): ... here.
25337 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25338 * graphite-poly.c: Don't include tree-dump.h.
25339 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25340 * print-tree.c: Likewise.
25341 * stor-layout.c: Likewise.
25342 * tree-nested.c: Likewise.
25344 * dumpfile.c (dump_start): Use TDF_FLAGS.
25345 (dump_enable_all): Fix TDF_KIND check thinko.
25347 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25350 array entries to represent two legal parameterizations of the
25351 overloaded __builtin_cmpb function, as represented by the
25352 P6_OV_BUILTIN_CMPB constant.
25353 (altivec_resolve_overloaded_builtin): Add special case handling
25354 for the __builtin_cmpb function, as represented by the
25355 P6_OV_BUILTIN_CMPB constant.
25356 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25357 (BU_P6_64BIT_2): New macro.
25358 (BU_P6_OVERLOAD_2): New macro
25359 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25360 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25361 (CMPB): Add overload support to represent both 32-bit and 64-bit
25362 compare-bytes function.
25363 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25364 support for TARGET_CMPB.
25365 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25366 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25367 documentation of the __builtin_cmpb overloaded built-in function.
25369 2017-05-11 Richard Biener <rguenther@suse.de>
25371 PR tree-optimization/80705
25372 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25373 bases are not vectorizable.
25375 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25377 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25378 when counting register pressure.
25380 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25382 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25383 (iv_ca_more_deps): Renamed to ...
25384 (iv_ca_compare_deps): ... this.
25385 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25387 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25389 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25391 (determine_group_iv_costs): ... here.
25392 (find_inv_vars_cb): Record inv var if it's not recorded before.
25394 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25396 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25397 (get_shiftadd_cost): Ditto.
25399 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25401 * tree-ssa-address.c: Include header file.
25402 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25404 (add_to_parts): Refactor.
25405 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25406 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25409 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25411 PR tree-optimization/53090
25412 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25414 (extract_cond_operands): Detect condition with IV on both sides
25415 and return COMP_IV_EXPR_2.
25416 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25417 (rewrite_use_compare): Simplify by removing call to function
25418 extract_cond_operands.
25420 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25422 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25423 (extract_cond_operands): Detect condition comparing against non-
25424 invariant bound and return appropriate enum value.
25425 (find_interesting_uses_cond): Update use of extract_cond_operands.
25426 Handle its return value accordingly.
25427 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25429 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25431 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25432 nonlinear iv_use computation in loop invariant sensitive way.
25434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25436 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25437 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25439 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25441 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25442 (dump_cand): Support iv_cand.inv_exprs.
25443 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25445 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25448 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25450 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25452 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25453 as local function. Include necessary header files.
25454 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25456 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25458 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25462 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25463 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25464 RSHIFT_EXPR and BIT_NOT_EXPR.
25466 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25468 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25469 (adjust_setup_cost): New parameter supporting round up adjustment.
25470 (struct address_cost_data): Delete.
25471 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25472 (split_address_cost, ptr_difference_cost): Delete.
25473 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25474 (struct ainc_cost_data): New struct.
25475 (get_address_cost_ainc): New function.
25476 (get_address_cost, get_computation_cost): Reimplement.
25477 (determine_group_iv_cost_address): Record inv_expr for all uses of
25479 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25480 (iv_ca_has_deps): Reimplemented to ...
25481 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25483 (iv_ca_extend): Call iv_ca_more_deps.
25485 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25487 * tree-ssa-address.c (struct mem_address): Move to header file.
25488 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25489 * tree-ssa-address.h (struct mem_address): Move from C file.
25490 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25492 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25494 * tree-affine.h (aff_combination_type): New interface.
25495 (aff_combination_zero_p): Remove static.
25496 (aff_combination_const_p): New interface.
25497 (aff_combination_singleton_var_p): New interfaces.
25499 2017-05-11 Richard Biener <rguenther@suse.de>
25501 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25502 Skip unreachable blocks and destinations.
25503 (eliminate): Move stmt removal and fixup ...
25504 (fini_eliminate): ... here. Skip inserted exprs.
25505 (pass_pre::execute): Move fini_pre after fini_eliminate.
25506 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25507 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25508 PRE to get rid of dead code that has invalid SSA form and
25509 split critical edges again.
25511 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25513 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25515 2017-05-11 Richard Biener <rguenther@suse.de>
25517 * passes.c (execute_function_todo): Verify loops if they are
25518 said to be up-to-date.
25519 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25520 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25522 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25525 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25526 handle calling assemble_external ourself.
25529 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25530 modes with zero size. Enhance comment.
25532 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25534 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25535 built-ins for vec_xl and vec_xst with short and char pointer
25538 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25540 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25541 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25542 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25543 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25544 (_mm_maskz_min_round_ss): New intrinsics.
25545 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25546 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25547 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25548 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25549 (__builtin_ia32_minss_mask_round): New builtins.
25550 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25551 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25552 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25554 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25555 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25557 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25560 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25562 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25563 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25564 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25565 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25566 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25567 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25568 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25569 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25570 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25571 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25572 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25573 (__builtin_ia32_mulss_mask_round): New builtins.
25574 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25575 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25576 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25578 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25579 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25581 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25584 2017-05-10 Julia Koval <julia.koval@intel.com>
25586 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25587 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25588 (_mm256_setr_m128i): New intrinsics.
25590 2017-05-10 Julia Koval <julia.koval@intel.com>
25592 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25593 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25594 (_mm_maskz_rcp14_ss): New intrinsics.
25595 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25596 (__builtin_ia32_rcp14ss_mask): New builtins.
25597 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25599 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25601 PR tree-optimization/51513
25602 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25603 (assert_unreachable_fallthru_edge_p): Use it.
25604 (group_case_labels_stmt): Likewise.
25605 * tree-cfg.h: Prototype it.
25606 * stmt.c: Include cfghooks.h and tree-cfg.h.
25607 (emit_case_dispatch_table) <gap_label>: New local variable.
25608 Use it to fill dispatch table gaps.
25609 Test for default_label before updating probabilities.
25610 (expand_case) <default_label>: Remove unneeded initialization.
25611 Test for unreachable default case statement and remove its edge.
25612 Set default_label accordingly.
25613 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25615 2017-05-10 Carl Love <cel@us.ibm.com>
25617 * config/rs6000/rs6000-c: Add support for built-in functions
25618 vector signed char vec_neg (vector signed char)
25619 vector signed short int vec_neg (vector short int)
25620 vector signed int vec_neg (vector signed int)
25621 vector signed long long vec_neg (vector signed long long)
25622 vector float vec_neg (vector float)
25623 vector double vec_neg (vector double)
25624 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25626 * config/rs6000/altivec.h: Add define for vec_neg
25627 * doc/extend.texi: Update the built-in documentation for the
25628 new built-in functions.
25630 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25632 PR tree-optimization/77644
25633 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25635 2017-05-10 Nathan Sidwell <nathan@acm.org>
25637 * dumpfile.h (TDI_lang_all): New.
25638 (TDF_KIND): New. Renumber others
25639 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25641 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25643 (get_dump_file_name): Adjust suffix generation.
25644 (dump_enable_all): Use TDF_KIND.
25645 * doc/invoke.texi (-fdump-lang-all): Document.
25647 * dumpfile.h: Tabify.
25649 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25652 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25653 Move member access before delete.
25655 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25657 * tree-inline.c (expand_call_inline): Split block at stmt
25660 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25663 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25664 are now unused after splitting mov{sf,sd}_hardfloat.
25665 (f32_lr2): Likewise.
25666 (f32_lm): Likewise.
25667 (f32_lm2): Likewise.
25668 (f32_li): Likewise.
25669 (f32_li2): Likewise.
25670 (f32_lv): Likewise.
25671 (f32_sr): Likewise.
25672 (f32_sr2): Likewise.
25673 (f32_sm): Likewise.
25674 (f32_sm2): Likewise.
25675 (f32_si): Likewise.
25676 (f32_si2): Likewise.
25677 (f32_sv): Likewise.
25678 (f32_dm): Likewise.
25679 (f32_vsx): Likewise.
25680 (f32_av): Likewise.
25681 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25682 For movsf, order stores so the VSX stores occur before the GPR
25683 store which encourages the register allocator to use a traditional
25684 FPR instead of a GPR. For movsd, order the stores so that the GPR
25685 store comes before the VSX stores to allow the power6 to work.
25686 This is due to the power6 not having a 32-bit integer store
25687 instruction from a FPR.
25688 (movsf_hardfloat): Likewise.
25689 (movsd_hardfloat): Likewise.
25691 2017-05-09 Martin Sebor <msebor@redhat.com>
25693 PR translation/80280
25694 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25697 PR translation/80280
25698 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25699 data member added in r247778.
25700 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25702 2017-05-09 Nathan Sidwell <nathan@acm.org>
25704 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25706 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25709 2017-05-09 Marek Polacek <polacek@redhat.com>
25711 * doc/invoke.texi: Fix typo.
25713 2017-05-09 Richard Biener <rguenther@suse.de>
25715 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25716 (vrp_val_is_min): Likewise.
25717 (set_value_range_to_value): Likewise.
25718 (set_value_range_to_nonnegative): Likewise.
25719 (gimple_assign_nonzero_p): Likewise.
25720 (gimple_stmt_nonzero_p): Likewise.
25721 (vrp_int_const_binop): Likewise. Remove unreachable case.
25722 (adjust_range_with_scev): Adjust comments.
25723 (compare_range_with_value): Likewise.
25724 (extract_range_from_phi_node): Likewise.
25725 (test_for_singularity): Likewise.
25727 2017-05-09 Richard Biener <rguenther@suse.de>
25729 * tree-vrp.c (get_single_symbol): Add assert that we don't
25730 get overflowed constants as invariant part.
25731 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25732 checks. Use wi::cmp instead of recursing for integer constants.
25733 (compare_values): Just ignore whether we assumed undefined
25734 overflow instead of failing the compare.
25735 (extract_range_for_var_from_comparison_expr): Add comment before the
25736 TREE_NO_WARNING sets.
25737 (test_for_singularity): Likewise.
25738 (extract_range_from_comparison): Do not disable optimization
25739 when we assumed undefined overflow.
25740 (extract_range_basic): Remove init of unused var.
25742 2017-05-09 Richard Biener <rguenther@suse.de>
25744 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25745 (extract_range_from_multiplicative_op_1): Adjust.
25746 (extract_range_from_binary_expr_1): Use int_const_binop.
25748 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25751 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25752 rs6000_store_data_bypass_p in seven define_bypass directives and
25753 in several comments.
25754 * config/rs6000/rs6000-protos.h: Add prototype for
25755 rs6000_store_data_bypass_p function.
25756 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25757 function implements slightly different (rs6000-specific) semantics
25758 than store_data_bypass_p, returning false rather than aborting
25759 with assertion error when arguments do not satisfy the
25760 requirements of store data bypass.
25761 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25762 rs6000_store_data_bypass_p.
25764 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25766 * config/xtensa/xtensa-protos.h
25767 (xtensa_initial_elimination_offset): New declaration.
25768 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25769 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25770 macro definition, add case for FRAME_POINTER_REGNUM when
25771 FRAME_GROWS_DOWNWARD.
25772 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25773 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25774 xtensa_initial_elimination_offset.
25776 2017-05-08 Nathan Sidwell <nathan@acm.org>
25778 * doc/invoke.texi: Alphabetize -fdump options.
25780 2017-05-08 Martin Sebor <msebor@redhat.com>
25782 PR translation/80280
25783 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25785 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25787 * target.def (compute_frame_layout): New optional target hook.
25788 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25789 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25790 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25792 * reload1.c (verify_initial_elim_offsets): Likewise.
25793 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25794 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25795 (arm_get_frame_offsets): Split up into this ...
25796 (arm_compute_frame_layout): ... and this function.
25798 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25800 * config/aarch64/constraints.md (Usa): New constraint.
25801 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25803 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25805 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25806 with_multilib_list after it has been checked.
25808 2017-05-08 Richard Biener <rguenther@suse.de>
25810 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25811 (bitmap_set_subtract_values): Likewise.
25813 2017-05-08 Richard Biener <rguenther@suse.de>
25815 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25816 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25818 (gimple_stmt_nonzero_warnv_p): Rename to ...
25819 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25821 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25822 (extract_range_basic): Adjust, do not disable propagation on
25823 strict overflow sensitive simplification.
25824 (vrp_visit_cond_stmt): Likewise.
25826 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25828 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25829 body size unconditionally.
25831 2017-05-07 Jeff Law <law@redhat.com>
25834 2017-05-06 Jeff Law <law@redhat.com>
25835 PR tree-optimization/78496
25836 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25839 PR tree-optimization/78496
25840 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25841 (simplify_stmt_using_ranges): Call it.
25842 (vrp_dom_walker::before_dom_children): Extract equivalences
25843 from an ASSERT_EXPR with an equality comparison against a
25846 2017-05-06 Jeff Law <law@redhat.com>
25848 PR tree-optimization/78496
25849 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25852 PR tree-optimization/78496
25853 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25854 (simplify_stmt_using_ranges): Call it.
25855 (vrp_dom_walker::before_dom_children): Extract equivalences
25856 from an ASSERT_EXPR with an equality comparison against a
25859 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25861 * lra-constraints.c (lra_copy_reg_equiv): New function.
25862 (split_reg): Use it to copy equivalence information from the
25863 original register to the spill register.
25865 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25867 PR rtl-optimization/75964
25868 * simplify-rtx.c (simplify_const_relational_operation): Remove
25869 invalid handling of comparisons of integer ABS.
25871 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25873 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25874 initialize to zero.
25875 (init_regs): Remove declaration.
25876 (function_arg_advance_32): Initialize error_p as boolean variable.
25878 2017-05-05 Nathan Sidwell <nathan@acm.org>
25880 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25881 lines. Use for (;;).
25883 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25885 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25886 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25887 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25888 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25889 VF=2 that require versioning.
25891 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25893 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25896 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25898 * diagnostic.h (diagnostic_override_option_index): Convert from
25899 macro to inline function.
25901 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25903 * diagnostic.c (last_module_changed_p): New function.
25904 (set_last_module): New function.
25905 (diagnostic_report_current_module): Convert macro usage to
25906 the above functions.
25907 * diagnostic.h (diagnostic_context::last_module): Strengthen
25908 from const line_map * to const line_map_ordinary *.
25909 (diagnostic_last_module_changed): Delete macro.
25910 (diagnostic_set_last_module): Delete macro.
25912 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25914 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25915 with diagnostic_report_diagnostic.
25916 (diagnostic_n_impl_richloc): Likewise.
25917 * diagnostic.h (report_diagnostic): Delete macro.
25918 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25919 with diagnostic_report_diagnostic.
25920 * substring-locations.c (format_warning_va): Likewise.
25922 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25924 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25925 save/restor of format_spec. Move option-printing code to...
25926 (print_option_information): ...this new function, and
25927 reimplement by simply printing to the pretty_printer,
25928 rather than appending to the format string.
25930 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25932 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25933 handling logic into...
25934 (update_effective_level_from_pragmas): ...this new function.
25936 2017-05-04 Andrew Waterman <andrew@sifive.com>
25938 * config/riscv/riscv.opt (mstrict-align): New option.
25939 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25940 (SLOW_UNALIGNED_ACCESS): Define.
25941 (riscv_slow_unaligned_access): Declare.
25942 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25944 (riscv_slow_unaligned_access): New variable.
25945 (rocket_tune_info): Set slow_unaligned_access to true.
25946 (optimize_size_tune_info): Set slow_unaligned_access to false.
25947 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25948 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25949 (riscv_option_override): Set riscv_slow_unaligned_access.
25950 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25952 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25954 * config/riscv/riscv.md: Unify indentation.
25956 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25961 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25963 (extendsi<mode>2): Add support for doing sign extension via
25964 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25965 don't have ISA 3.0 instructions.
25966 (extendsi<mode>2 splitter): Likewise.
25967 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25968 generate the normal insns since SImode can now go in vector
25969 registers. Disallow the special UNSPECs needed for previous
25970 machines to hide SImode being used. Add new insns
25971 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25972 (fix_trunc<mode>si2_stfiwx): Likewise.
25973 (fix_trunc<mode>si2_internal): Likewise.
25974 (fixuns_trunc<mode>si2): Likewise.
25975 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25976 (fctiw<u>z_<mode>_smallint): Likewise.
25977 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25978 of floating point to 32-bit integer from doing a direct move to
25979 the GPR registers to do a store.
25980 (fctiwz_<mode>): Break long line.
25982 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25984 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25985 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25986 (addr_list, addr_offset_valid_p): New.
25987 (split_address_groups): Check offset validity with above function.
25988 (gt-tree-ssa-loop-ivopts.h): Include header file.
25990 2017-05-05 Nathan Sidwell <nathan@acm.org>
25992 * config.gcc (arm*-*-*): Add missing 'fi'.
25994 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25996 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25997 included in -fopt-info does not matter.
25998 * doc/optinfo.texi (-fopt-info): Fix description of default
25999 behavour. Explicitly say order of options included in -fopt-info
26002 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26004 * config.gcc: Allow combinations of aprofile and rmprofile values for
26005 --with-multilib-list.
26006 * config/arm/t-multilib: New file.
26007 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26008 variables. Remove setting of ISA and floating-point ABI in
26009 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26010 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26011 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26012 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26014 * config/arm/t-rmprofile: Likewise except for the matches changes.
26015 * doc/install.texi (--with-multilib-list): Document the combination of
26016 aprofile and rmprofile values and warn about pitfalls in doing that.
26018 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26020 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26021 (movdi_aarch64): Likewise.
26023 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26025 PR tree-optimization/80632
26026 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26028 (build_arrays): Initialize it for virtual phis.
26029 (fix_phi_nodes): Use it for virtual phis.
26031 PR tree-optimization/80558
26032 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26033 [x, y] op z into [x op, y op z] for op & or | if conditions
26036 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26037 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26040 * config/arm/arm.md (use_literal_pool): Remove.
26041 (64-bit immediate split): No longer takes cost into consideration
26042 if arm_disable_literal_pool is enabled.
26043 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26044 used when arm_disable_literal_pool is enabled.
26045 (arm_max_const_double_inline_cost): Remove use of
26046 arm_disable_literal_pool.
26047 (push_minipool_fix): Add assert.
26048 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26049 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26050 (no_literal_pool_sf_immediate): New.
26052 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26054 PR tree-optimization/80613
26055 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26056 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26058 2017-05-05 Richard Biener <rguenther@suse.de>
26060 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26062 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26064 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26065 of this flag from insn conditions due to removal from r247495.
26067 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26069 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26071 (arm_early_store_addr_dep_ptr): Likewise.
26072 * config/arm/aarch-common-protos.h
26073 (arm_early_load_addr_dep_ptr): Add prototype.
26074 (arm_early_store_addr_dep_ptr): Likewise.
26075 * config/arm/cortex-a53.md: Add new bypasses.
26077 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26079 * tree.c (next_type_uid): Change type to unsigned.
26080 (type_hash_canon): Decrement back next_type_uid if
26081 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26082 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26085 2017-05-04 Martin Sebor <msebor@redhat.com>
26087 * builtins.c: Fix a trivial typo in a comment.
26089 PR middle-end/79234
26090 * builtins.c (check_sizes): Adjust to handle reading past the end.
26091 Avoid printing excessive upper bound of ranges. Use %E to print
26092 tree nodes instead of converting them to %wu.
26093 (expand_builtin_memchr): New function.
26094 (compute_dest_size): Rename...
26095 (compute_objsize): ...to this.
26096 (expand_builtin_memcpy): Adjust.
26097 (expand_builtin_mempcpy): Adjust.
26098 (expand_builtin_strcat): Adjust.
26099 (expand_builtin_strcpy): Adjust.
26100 (check_strncat_sizes): Adjust.
26101 (expand_builtin_strncat): Adjust.
26102 (expand_builtin_strncpy): Adjust and simplify.
26103 (expand_builtin_memset): Adjust.
26104 (expand_builtin_bzero): Adjust.
26105 (expand_builtin_memcmp): Adjust.
26106 (expand_builtin): Handle memcmp.
26107 (maybe_emit_chk_warning): Check strncat just once.
26109 2017-05-04 Martin Sebor <msebor@redhat.com>
26111 PR preprocessor/79214
26112 PR middle-end/79222
26113 PR middle-end/79223
26114 * builtins.c (check_sizes): Add inlining context and issue
26115 warnings even when -Wno-system-headers is set.
26116 (check_strncat_sizes): Same.
26117 (expand_builtin_strncat): Same.
26118 (expand_builtin_memmove): New function.
26119 (expand_builtin_stpncpy): Same.
26120 (expand_builtin): Handle memmove and stpncpy.
26122 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26124 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26125 which is not used any more.
26127 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26129 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26131 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26133 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26134 (cortexa53_tunings): Likewise.
26135 (cortexa57_tunings): Likewise.
26136 (cortexa72_tunings): Likewise.
26137 (cortexa73_tunings): Likewise.
26139 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26141 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26142 Set loop alignment to 8.
26144 2017-05-04 Martin Sebor <msebor@redhat.com>
26146 PR translation/80280
26147 * builtins.c (expand_builtin_object_size): Add missing quoting to
26148 %D and like directives.
26149 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26150 (hsa_type_for_tree_type): Same.
26151 (verify_function_arguments): Same.
26152 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26153 * varasm.c (get_section): Same.
26156 2017-05-04 Martin Sebor <msebor@redhat.com>
26158 PR translation/80280
26159 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26161 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26163 * config/aarch64/aarch64.c (generic_addrcost_table):
26164 Change HI/TI mode setting.
26166 2017-05-04 Martin Jambor <mjambor@suse.cz>
26168 PR tree-optimization/80622
26169 * tree-sra.c (comes_initialized_p): New function.
26170 (build_accesses_from_assign): Only set write lazily when
26171 comes_initialized_p is false.
26172 (analyze_access_subtree): Use comes_initialized_p.
26173 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26174 instead of testing for PARM_DECL.
26176 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26178 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26179 constraint on operand 0 to allow more general addressing modes.
26180 Adjust output template.
26181 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26183 * config/aarch64/aarch64-protos.h
26184 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26185 * config/aarch64/constraints.md (Dp): New address constraint.
26186 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26189 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26191 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26192 update use of estimate_ipcp_clone_size_and_time.
26193 (estimate_local_effects): Update use of
26194 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26195 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26196 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26197 Return nonspecialized time.
26199 2017-05-04 Richard Biener <rguenther@suse.de>
26201 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26202 for the last VUSE which def dominates the PHI. Directly call
26204 (get_continuation_for_phi_1): Remove.
26206 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26208 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26209 to explain the use of truncating division. Cap the number of
26210 iterations to the maximum given by nb_iterations_upper_bound,
26213 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26215 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26216 * configure: Regenerate.
26217 * config.in: Regenerate.
26218 * config/i386/driver-mingw32.c: new file.
26219 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26220 * config.host: Link driver-mingw32.o on MinGW host.
26221 * doc/install.texi: Document new --enable-mingw-wildcard configure
26224 2017-05-04 Marek Polacek <polacek@redhat.com>
26226 PR tree-optimization/80612
26227 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26229 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26230 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26232 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26233 (movt splitter): Likewise.
26234 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26235 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26236 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26237 block for Thumb-1 with MOVT.
26238 (thumb2_legitimate_address_p): Move code block ...
26239 (can_avoid_literal_pool_for_label_p): ... into this new function.
26240 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26242 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26243 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26244 "M-profile targets with the MOVT instruction".
26246 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26248 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26249 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26250 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26252 2017-05-04 Martin Liska <mliska@suse.cz>
26254 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26255 variable cond_code.
26257 2017-05-04 Richard Biener <rguenther@suse.de>
26259 * tree.c (array_at_struct_end_p): Handle arrays at struct
26260 end with flexarrays more conservatively. Refactor and treat
26261 arrays of arrays or aggregates more strict. Fix
26262 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26263 * tree.c (array_at_struct_end_p): Adjust prototype.
26264 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26265 * gimple-fold.c (get_range_strlen): Likewise.
26266 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26268 2017-05-04 Richard Biener <rguenther@suse.de>
26270 PR tree-optimization/31130
26271 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26273 (supports_overflow_infinity): Likewise.
26274 (is_negative_overflow_infinity): Likewise.
26275 (is_positive_overflow_infinity): Likewise.
26276 (is_overflow_infinity): Likewise.
26277 (stmt_overflow_infinity): Likewise.
26278 (overflow_infinity_range_p): Likewise.
26279 (usable_range_p): Remove as always returning true.
26280 (make_overflow_infinity): Remove.
26281 (negative_overflow_infinity): Likewise.
26282 (positive_overflow_infinity): Likewise.
26283 (avoid_overflow_infinity): Likewise.
26284 (set_value_range): Adjust accordingly.
26285 (set_value_range_to_nonnegative): Likewise, remove now unused
26286 overflow_infinity arg.
26287 (vrp_operand_equal_p): Adjust.
26288 (update_value_range): Likewise.
26289 (range_int_cst_singleton_p): Likewise.
26290 (operand_less_p): Likewise.
26291 (compare_values_warnv): Likewise.
26292 (extract_range_for_var_from_comparison_expr): Likewise.
26293 (vrp_int_const_binop): Likewise.
26294 (zero_nonzero_bits_from_vr): Likewise.
26295 (extract_range_from_multiplicative_op_1): Likewise.
26296 (extract_range_from_binary_expr_1): Likewise.
26297 (extract_range_from_unary_expr): Likewise.
26298 (extract_range_from_comparison): Likewise.
26299 (extract_range_basic): Likewise.
26300 (adjust_range_with_scev): Likewise.
26301 (compare_ranges): Likewise.
26302 (compare_range_with_value): Likewise.
26303 (dump_value_range): Likewise.
26304 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26306 (simplify_cond_using_ranges): Adjust.
26308 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26310 * brig-builtins.def: Added a builtin for class_f64.
26311 * builtin-types.def: Added a builtin type needed by class_f64.
26313 2017-05-03 Jason Merrill <jason@redhat.com>
26315 * timevar.def: Add TV_CONSTEXPR.
26317 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26319 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26321 2017-05-03 Martin Jambor <mjambor@suse.cz>
26323 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26324 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26325 * ipa-cp.c (ipcp_propagate_stage): Do not call
26326 ipa_update_after_lto_read.
26327 * ipa-inline.c (ipa_inline): Likewise.
26329 2017-05-03 Martin Jambor <mjambor@suse.cz>
26331 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26332 tag. Added a default constructor and a destructor.
26333 (ipa_edge_args_sum_t): New class;
26334 (ipa_edge_args_sum): Declare.
26335 (ipa_edge_args_vector): Remove declaration.
26336 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26337 (ipa_free_edge_args_substructures): Remove declaration.
26338 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26339 (ipa_edge_args_info_available_for_edge_p): Likewise.
26340 * ipa-prop.c (ipa_edge_args_vector): Removed.
26341 (edge_removal_hook_holder): Likewise.
26342 (edge_duplication_hook_holder): Likewise.
26343 (ipa_edge_args_sum): New variable.
26344 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26345 ipa_edge_args_vector.
26346 (ipa_free_edge_args_substructures): Likewise.
26347 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26348 ipa_edge_args_vector.
26349 (ipa_edge_removal_hook): Turned into method
26350 ipa_edge_args_sum_t::remove.
26351 (ipa_edge_duplication_hook): Turned into method
26352 ipa_edge_args_sum_t::duplicate.
26353 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26354 registering edge hooks.
26355 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26356 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26357 ipa_edge_args_sum instead of ipa_edge_args_vector.
26358 * ipa-profile.c (ipa_profile): Likewise.
26360 2017-05-03 Martin Jambor <mjambor@suse.cz>
26362 * symbol-summary.h (function_summary): New method exists.
26363 (function_summary::symtab_removal): Deallocate through release.
26364 (call_summary): New class.
26365 (gt_ggc_mx): New overload.
26366 (gt_pch_nx): Likewise.
26367 (gt_pch_nx): Likewise.
26369 2017-05-03 Jeff Law <law@redhat.com>
26371 PR tree-optimization/78496
26372 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26373 from simplify_cond_using_ranges. Split off code to walk
26374 backwards through casts into ...
26375 (simplify_cond_using_ranges_2): New function.
26376 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26377 (execute_vrp): After identifying jump threads, call
26378 simplify_cond_using_ranges_2.
26380 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26383 * ipa-inline.h (inline_summary): Add ctor.
26384 (create_ggc): Do not use ggc_cleared_alloc.
26386 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26387 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26389 * gcc.c (handle_braces): Support escaping in switch matching
26391 * doc/invoke.texi (Spec Files): Document it.
26392 Remove superfluous @code markup in items.
26394 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26396 * diagnostic-show-locus.c (struct column_range): New struct.
26397 (get_affected_columns): New function.
26398 (get_printed_columns): New function.
26399 (struct correction): New struct.
26400 (correction::ensure_capacity): New function.
26401 (correction::ensure_terminated): New function.
26402 (struct line_corrections): New struct.
26403 (line_corrections::~line_corrections): New dtor.
26404 (line_corrections::add_hint): New function.
26405 (layout::print_trailing_fixits): Reimplement in terms of the new
26407 (selftest::test_overlapped_fixit_printing): New function.
26408 (selftest::diagnostic_show_locus_c_tests): Call it.
26410 2017-05-03 Nathan Sidwell <nathan@acm.org>
26412 Canonicalize canonical type hashing
26413 * tree.h (type_hash_canon_hash): Declare.
26414 * tree.c (type_hash_list, attribute_hash_list): Move into
26415 type_hash_canon_hash.
26416 (build_type_attribute_qual_variant): Break out hash code calc into
26417 type_hash_canon_hash.
26418 (type_hash_canon_hash): New. Generic type hash computation.
26419 (build_range_type_1, build_array_type_1, build_function_type,
26420 build_method_type_directly, build_offset_type, build_complex_type,
26421 make_vector_type): Call it.
26423 2017-05-03 Richard Biener <rguenther@suse.de>
26425 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26426 When all DRs have unknown misaligned do not always peel
26427 when there is a store but apply the same costing model as if
26428 there were only loads.
26430 2017-05-03 Richard Biener <rguenther@suse.de>
26433 PR tree-optimization/80492
26434 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26435 compare_base_decls returning dont-know properly.
26437 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26439 * config/arm/iterators.md (CCSI): New mode iterator.
26440 (arch): New mode attribute.
26441 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26442 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26443 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26444 code iterator for success result mode.
26445 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26446 the corresponding new insn generators.
26448 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26451 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26452 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26454 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26456 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26457 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26458 (DDR_A): Wrap DDR argument in brackets.
26459 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26460 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26461 (DDR_REVERSED_P): Likewise.
26463 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26465 PR tree-optimization/79472
26466 * tree-switch-conversion.c (struct switch_conv_info): Add
26467 contiguous_range and default_case_nonstandard fields.
26468 (collect_switch_conv_info): Compute contiguous_range and
26469 default_case_nonstandard fields, don't clear final_bb if
26470 contiguous_range and only the default case doesn't have the required
26472 (check_all_empty_except_final): Set default_case_nonstandard instead
26473 of failing if contiguous_range and the default case doesn't have empty
26475 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26476 and only the default case doesn't have the required constants. Skip
26478 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26479 if default_case_nonstandard.
26480 (build_constructors): Build constant 1 just once. Assert that default
26481 values aren't inserted in between cases if contiguous_range. Skip
26483 (build_arrays): Skip virtual phis.
26484 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26485 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26486 Handle virtual phis.
26487 (gen_inbound_check): Handle default_case_nonstandard case.
26488 (process_switch): Adjust check_final_bb caller. Call
26489 gather_default_values with the first non-default case instead of
26490 default case if default_case_nonstandard.
26492 2017-05-02 Nathan Sidwell <nathan@acm.org>
26494 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26495 check. Fix formatting.
26497 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26499 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26500 errors when comparing specialized and unspecialized times.
26502 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26504 * diagnostic-show-locus.c
26505 (layout::should_print_annotation_line_p): Make private.
26506 (layout::print_annotation_line): Make private.
26507 (layout::annotation_line_showed_range_p): Make private.
26508 (layout::show_ruler): Make private.
26509 (layout::print_source_line): Make private. Pass in line and
26510 line_width, rather than calling location_get_source_line. Drop
26512 (layout::print_leading_fixits): New method.
26513 (layout::print_any_fixits): Rename to...
26514 (layout::print_trailing_fixits): ...this, and make private.
26515 Don't print newline fixits.
26516 (diagnostic_show_locus): Move logic for printing one row into...
26517 (layout::print_line): ...this new function. Move the
26518 location_get_source_line call and error-handling from
26519 print_source_line to here. Call print_leading_fixits, and rename
26520 print_any_fixits to print_trailing_fixits.
26521 (selftest::test_fixit_insert_containing_newline): Update now that
26522 newlines are partially supported.
26523 (selftest::test_fixit_insert_containing_newline_2): New test.
26524 (selftest::test_fixit_replace_containing_newline): Update comments.
26525 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26526 * edit-context.c (class added_line): New class.
26527 (class edited_line): Describe newline handling in comment.
26528 (edited_line::actually_edited_p): New method.
26529 (edited_line::print_content): Delete redundant decl.
26530 (edited_line::m_predecessors): New field.
26531 (edited_file::print_content): Call edited_line::print_content.
26532 (edited_file::print_diff): Update to support newlines.
26533 (edited_file::print_diff_hunk): Likewise.
26534 (edited_file::print_run_of_changed_lines): New function.
26535 (edited_file::print_diff_line): Convert to...
26536 (print_diff_line): ...this.
26537 (edited_file::get_effective_line_count): New function.
26538 (edited_line::edited_line): Initialize new field m_predecessors.
26539 (edited_line::~edited_line): Clean up m_predecessors.
26540 (edited_line::apply_fixit): Handle newlines.
26541 (edited_line::get_effective_line_count): New function.
26542 (edited_line::print_content): New function.
26543 (edited_line::print_diff_lines): New function.
26544 (selftest::test_applying_fixits_insert_containing_newline): New
26546 (selftest::test_applying_fixits_replace_containing_newline): New
26548 (selftest::insert_line): New function.
26549 (selftest::test_applying_fixits_multiple_lines): Add example of
26551 (selftest::edit_context_c_tests): Call the new tests.
26553 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26555 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26556 parameter cand. Update dump information.
26557 (get_computation_cost): Update uses.
26559 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26561 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26562 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26563 (get_computation_at, rewrite_use_address): Update use of
26564 get_computation_aff.
26566 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26568 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26569 (get_computation): Delete.
26570 (get_computation_cost): Implement like get_computation_cost_at.
26571 Use get_computation_at.
26572 (get_computation_cost_at): Delete.
26573 (rewrite_use_nonlinear_expr): Use get_computation_at.
26574 (rewrite_use_compare, remove_unused_ivs): Ditto.
26576 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26578 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26580 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26582 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26583 (ivopts_global_cost_for_size): Rename parameter and update uses.
26584 (iv_ca_recount_cost): Update uses.
26585 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26586 candidates seperately in n_invs and n_cands.
26587 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26589 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26591 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26592 (find_inv_vars_cb): New.
26593 (find_depends): Renamed to ...
26594 (find_inv_vars): ... this.
26595 (add_candidate_1, force_var_cost): Call find_inv_vars.
26596 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26598 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26600 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26601 inv_vars. Add inv_exprs.
26602 (struct iv_cand): Rename depends_on to inv_vars.
26603 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26604 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26605 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26606 (dump_cand): Dump inv_vars.
26607 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26608 (record_invariant, find_depends, add_candidate_1): Ditto.
26609 (set_group_iv_cost, force_var_cost): Ditto.
26610 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26611 (get_computation_cost_at, get_computation_cost): Ditto.
26612 (determine_group_iv_cost_generic): Ditto.
26613 (determine_group_iv_cost_address): Ditto.
26614 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26615 (determine_group_iv_costs): Ditto.
26616 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26617 (iv_ca_set_remove_invariants): Renamed to ...
26618 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26619 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26620 (iv_ca_set_add_invariants): Renamed to ...
26621 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26622 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26623 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26624 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26625 (create_new_ivs): Remove useless dump.
26627 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26629 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26631 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26632 (iv_ca_set_no_cp, create_new_iv): Ditto.
26634 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26636 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26638 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26640 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26641 function tree_check2.
26643 2017-05-02 Martin Liska <mliska@suse.cz>
26645 * doc/gcov.texi: Add missing preposition.
26646 * gcov.c (function_info::function_info): Properly fill up
26647 all member variables.
26649 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26651 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26653 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26655 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26657 2017-05-02 Martin Liska <mliska@suse.cz>
26660 * lto-streamer-in.c (lto_read_tree_1): Remove
26661 LTO_STREAMER_DEBUG.
26662 * lto-streamer.c (struct tree_hash_entry): Likewise.
26663 (struct tree_entry_hasher): Likewise.
26664 (tree_entry_hasher::hash): Likewise.
26665 (tree_entry_hasher::equal): Likewise.
26666 (lto_streamer_init): Likewise.
26667 (lto_orig_address_map): Likewise.
26668 (lto_orig_address_get): Likewise.
26669 (lto_orig_address_remove): Likewise.
26670 * lto-streamer.h: Likewise.
26671 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26672 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26674 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26676 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26677 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26678 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26679 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26680 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26681 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26682 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26683 (mm_maskz_sub_ss): New intrinsics.
26684 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26685 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26686 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26687 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26688 (__builtin_ia32_subss_mask_round): New builtins.
26689 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26690 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26691 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26693 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26694 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26696 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26699 2017-05-02 Martin Jambor <mjambor@suse.cz>
26701 PR tree-optimization/78687
26702 * tree-sra.c (access): New field parent.
26703 (process_subtree_disqualification): New function.
26704 (disqualify_candidate): Call it.
26705 (build_accesses_from_assign): Reset write flag if creating an
26707 (build_access_subtree): Fill in parent field and also prpagate
26708 down grp_write flag.
26709 (create_artificial_child_access): New parameter set_grp_write, set
26710 grp_write to its value.
26711 (propagate_subaccesses_across_link): Also propagate grp_write flag
26713 (propagate_all_subaccesses): Push the closest parent back to work
26714 queue if add_access_to_work_queue returned true.
26716 2017-05-02 Richard Biener <rguenther@suse.de>
26718 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26719 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26720 -fstrict-overflow documentation.
26721 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26722 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26723 flag_strict_overflow.
26724 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26725 * lto-opts.c (lto_write_options): Do not stream it.
26726 * lto-wrapper.c (merge_and_complain): Do not handle it.
26727 * opts.c (default_options_table): Do not set -fstrict-overflow.
26728 (finish_options): Likewise do not clear it when sanitizing.
26729 * simplify-rtx.c (simplify_const_relational_operation): Do not
26730 test flag_strict_overflow.
26732 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26734 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26735 using enabled attribute.
26736 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26737 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26738 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26739 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26740 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26741 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26742 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26743 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26744 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26745 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26747 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26749 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26751 2017-05-02 Richard Biener <rguenther@suse.de>
26753 PR tree-optimization/80591
26755 2017-04-10 Richard Biener <rguenther@suse.de>
26757 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26760 2017-05-02 Richard Biener <rguenther@suse.de>
26762 PR tree-optimization/80549
26763 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26764 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26765 headers if they do not have a preheader.
26767 2017-05-02 Martin Liska <mliska@suse.cz>
26770 * common.opt: Fix typo.
26771 * doc/invoke.texi: Likewise.
26773 2017-05-01 Jan Beulich <jbeulich@suse.com>
26775 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26776 swapping, add (x,x,m,x,n) alternative.
26778 2017-05-01 Nathan Sidwell <nathan@acm.org>
26780 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26781 unnecessary unadjusted_alignment check.
26783 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26786 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26788 * gimplify.c (gimplify_cilk_detach): New function.
26789 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26790 * tree-core.h: Document EXPR_CILK_SPAWN.
26791 * tree.h (EXPR_CILK_SPAWN): Define.
26793 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26795 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26796 to use new fixit_hint representation, using the "replace" logic.
26797 (get_line_span_for_fixit_hint): Likewise.
26798 (layout::print_any_fixits): Likewise.
26799 (selftest::test_one_liner_many_fixits): Rename to...
26800 (selftest::test_one_liner_many_fixits_1): ...this, and update
26801 comment and expected output to reflect that the multiple fix-it
26802 hints are now consolidated into one insertion.
26803 (selftest::test_one_liner_many_fixits_2): New test.
26804 (selftest::test_diagnostic_show_locus_one_liner): Update for
26806 (selftest::test_fixit_consolidation): Update for fix-it API
26808 * diagnostic.c (print_parseable_fixits): Likewise.
26809 * edit-context.c (edited_line::m_line_events): Convert from
26810 auto_vec <line_event *> to auto_vec <line_event>.
26811 (class line_event): Convert from abstract base class to a concrete
26812 class, taking over the role of replace_event.
26813 (class insert_event): Delete.
26814 (class replace_event): Rename to class line_event. Convert to
26816 (edit_context::add_fixits): Reimplement.
26817 (edit_context::apply_insert): Delete.
26818 (edit_context::apply_replace): Rename to...
26819 (edit_context::apply_fixit): ...this. Convert to half-open range.
26820 (edited_file::apply_insert): Delete.
26821 (edited_file::apply_replace): Rename to...
26822 (edited_file::apply_fixit): ...this.
26823 (edited_line::~edited_line): Drop deletion of events.
26824 (edited_line::apply_insert): Delete.
26825 (edited_line::apply_replace): Rename to...
26826 (edited_line::apply_fixit): ...this. Convert to half-open range.
26827 Update for change to type of m_line_events.
26828 * edit-context.h (edit_context::apply_insert): Delete.
26829 (edit_context::apply_replace): Rename to...
26830 (edit_context::apply_fixit): ...this.
26832 2017-05-01 Martin Sebor <msebor@redhat.com>
26834 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26837 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26840 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26841 __get_cpuid_max returns 0.
26842 (__get_cpuid_count): Ditto.
26844 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26846 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26847 replacement expression is another instance of one of its arguments.
26849 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26852 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26853 check for stack push/pop autoinc.
26854 * config/i386/i386.c (ix86_agi_dependent): Return false
26855 if the only reason why modified_in_p returned true is that
26856 addr is SP based and set_insn is a push or pop.
26858 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26860 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26863 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26866 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26867 (account_size_time): Use two predicates - exec_pred and
26869 (evaluate_conditions_for_known_args): Compute both clause and
26871 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26872 (inline_summary_t::duplicate): Update.
26873 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26875 (compute_inline_parameters): Likewise.
26876 (estimate_edge_size_and_time): Update caluclation of time.
26877 (estimate_node_size_and_time): Compute both time and nonspecialized
26879 (estimate_ipcp_clone_size_and_time): Update.
26880 (inline_merge_summary): Update.
26881 (do_estimate_edge_time): Update.
26882 (do_estimate_edge_size): Update.
26883 (do_estimate_edge_hints): Update.
26884 (inline_read_section, inline_write_summary): Stream both new predicates.
26885 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26887 (compute_inlined_call_time): Cleanup.
26888 (big_speedup_p): Update.
26889 (edge_badness): Update.
26890 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26891 (size_time_entry): Replace predicate by exec_predicate and
26892 nonconst_predicate.
26893 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26894 (estimate_edge_time): Return also nonspec_time.
26895 (reset_edge_growth_cache): Update.
26897 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26899 PR rtl-optimization/80491
26900 * ifcvt.c (noce_process_if_block): When looking for x setter
26901 with missing else_bb, don't check only the insn right before
26902 cond_earliest, but look for the last insn that x is modified in
26903 within the same bb.
26905 PR rtl-optimization/80491
26906 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26908 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26910 PR tree-optimization/80487
26911 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26913 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26915 PR tree-optimization/79697
26916 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26917 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26918 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26920 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26921 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26923 2017-04-28 Martin Sebor <msebor@redhat.com>
26925 PR tree-optimization/80523
26926 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26927 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26929 (maybe_warn, format_directive, parse_directive): Use new functions.
26930 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26932 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26934 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26936 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26938 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26939 target_header_dir): Set correctly.
26940 * configure: Regenerated.
26941 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26942 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26943 instead of SYSTEM_HEADER_DIR.
26945 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26947 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26948 (estimate_local_effects): Likewise.
26949 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26950 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26951 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26952 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26953 do_estimate_edge_time, estimate_edge_time): Likewise.
26954 * ipa-inline-analysis.c (estimate_node_size_and_time,
26955 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26956 (estimate_time_after_inlining): Remove.
26958 2017-04-28 Martin Liska <mliska@suse.cz>
26960 * doc/gcov.texi: Enhance documentation of gcov.
26962 2017-04-28 Martin Liska <mliska@suse.cz>
26964 * doc/gcov.texi: Sort options in alphabetic order.
26965 * doc/gcov-dump.texi: Likewise.
26966 * doc/gcov-tool.texi: Likewise.
26967 * gcov.c (print_usage): Likewise.
26968 * gcov-dump.c (print_usage): Likewise.
26969 * gcov-tool.c (print_merge_usage_message): Likewise.
26970 (print_rewrite_usage_message): Likewise.
26971 (print_overlap_usage_message): Likewise.
26973 2017-04-28 Martin Liska <mliska@suse.cz>
26975 PR gcov-profile/53915
26976 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26978 2017-04-28 Martin Liska <mliska@suse.cz>
26980 PR gcov-profile/79891
26981 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26982 is marked by compiler as living on a line.
26983 (get_cycles_count): Remove usage of the union.
26984 (output_intermediate_file): Likewise.
26985 (find_source): Fix GNU coding style.
26986 (accumulate_line_counts): Remove old non-all block mode.
26987 (output_lines): Remove usage of the union.
26988 * profile.c (output_location): Include all BBs, even if
26989 belonging to a same line (and file) as a previous BB.
26991 2017-04-28 Martin Liska <mliska@suse.cz>
26993 * gcov.c (process_args): Handle new argument 'w'.
26994 (read_graph_file): Assign ID to BBs.
26995 (output_branch_count): Display BB # if verbose flag is set.
26996 (output_lines): Likewise for arcs.
26997 (print_usage): Add '--verbose' option help.
26998 * doc/gcov.texi: Document --verbose (-w) option.
27000 2017-04-28 Martin Liska <mliska@suse.cz>
27002 * gcov.c (struct block_location_info): New struct.
27003 (process_file): Fill up the new structure.
27004 (read_graph_file): Replace usage of encoding by the newly added
27006 (add_line_counts): Likewise.
27007 (accumulate_line_counts): Remove usage of the union.
27008 (function_info::function_info): New function.
27009 (function_info::~function_info): Likewise.
27010 (process_file): Call delete instead of release_function.
27011 (release_function): Release the function.
27012 (release_structures): Call delete instead of release_function.
27013 (solve_flow_graph): Replace usage of num_blocks.
27014 (find_exception_blocks): Likewise.
27015 (output_lines): Fix GNU coding style.
27017 2017-04-28 Martin Liska <mliska@suse.cz>
27020 * coverage.c (coverage_remove_note_file): New function.
27021 * coverage.h: Declare the function.
27022 * toplev.c (finalize): Clean if an error has been seen.
27024 2017-04-28 Martin Liska <mliska@suse.cz>
27026 PR gcov-profile/80031
27027 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27028 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27029 * gcov.c (read_graph_file): Read just number of blocks.
27030 * profile.c (branch_prob): Do not stream 0 flags per a basic
27033 2017-04-28 Martin Liska <mliska@suse.cz>
27035 * gcov-dump.c (tag_*): Add new argument to declarations.
27036 (dump_gcov_file): Likewise.
27037 (tag_blocks): Add and use new argument depth.
27038 (tag_arcs): Likewise.
27039 (tag_lines): Likewise.
27040 (tag_counters): Likewise.
27041 (tag_summary): Likewise.
27042 (dump_working_sets): Use depth to do a proper indentation.
27044 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27047 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27048 * symtab.c (symtab_node::debug_symtab): Move definition here.
27050 2017-04-28 Richard Biener <rguenther@suse.de>
27052 * lto-streamer.h (LTO_major_version): Bump to 7.
27054 2017-04-28 Richard Biener <rguenther@suse.de>
27056 * tree-vrp.c (assert_info): New struct.
27057 (add_assert_info): New helper.
27058 (register_edge_assert_for_2): Refactor to add asserts to a vector
27060 (register_edge_assert_for_1): Likewise.
27061 (register_edge_assert_for): Likewise.
27062 (finish_register_edge_assert_for): New helper actually registering
27063 asserts where live on edge.
27064 (find_conditional_asserts): Adjust.
27065 (find_switch_asserts): Likewise.
27066 (evrp_dom_walker::try_find_new_range): Generalize.
27067 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27069 2017-04-27 Marek Polacek <polacek@redhat.com>
27072 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27073 arg10 and arg11 to itype.
27075 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27077 * doc/extend.texi (Object Size Checking): Improve grammar.
27079 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27082 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27083 that the logic for permitting reciprocal estimates matches that
27086 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27089 * tree.c (type_cache_hasher::equal): Only compare
27090 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27091 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27092 non-aggregate element types.
27093 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27094 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27096 2017-04-27 Richard Biener <rguenther@suse.de>
27098 PR middle-end/80533
27099 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27100 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27101 keeping a reference to a trailing array.
27103 2017-04-27 Richard Biener <rguenther@suse.de>
27105 PR middle-end/80539
27106 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27107 being in loop-closed SSA form conservatively.
27108 (chrec_fold_multiply_poly_poly): Likewise.
27110 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27112 PR middle-end/79665
27113 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27114 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27116 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27119 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27120 (aarch64_function_arg_alignment): Return unsigned int again, but still
27121 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27122 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27123 Don't emit -Wpsabi note.
27124 (aarch64_function_arg_boundary): Likewise.
27125 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27128 2017-04-26 Nathan Sidwell <nathan@acm.org>
27130 * tree.h (crc32_unsigned_n): Declare.
27131 (crc32_unsigned, crc32_unsigned): Make inline.
27132 * tree.c (crc32_unsigned_bits): Replace with ...
27133 (crc32_unsigned_n): ... this.
27134 (crc32_unsigned, crc32_byte): Remove.
27135 (crc32_string): Remove unnecessary braces.
27137 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27139 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27140 * ipa-inline-analysis.c (MAX_TIME): Remove.
27141 (account_size_time): Use sreal for time.
27142 (dump_inline_summary): Update.
27143 (estimate_function_body_sizes): Update.
27144 (estimate_edge_size_and_time): Update.
27145 (estimate_calls_size_and_time): Update.
27146 (estimate_node_size_and_time): Update.
27147 (inline_merge_summary): Update.
27148 (inline_update_overall_summary): Update.
27149 (estimate_time_after_inlining): Update.
27150 (inline_read_section): Update.
27151 (inline_write_summary): Update.
27152 * ipa-inline.c (compute_uninlined_call_time): Update.
27153 (compute_inlined_call_time): Update.
27154 (recursive_inlining): Update.
27155 (inline_small_functions): Update.
27156 (dump_overall_stats): Update.
27157 * ipa-inline.h: Include sreal.h.
27158 (size_time_entry): Turn time to sreal.
27159 (inline_summary): Turn self_time nad time to sreal.
27161 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27163 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27165 (sreal::stream_out, sreal::stream_in): New.
27166 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27168 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27170 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27173 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27176 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27178 (dimode_scalar_chain::compute_convert_gain): Ditto.
27179 (dimode_scalar_chain::make_vector_copies): Ditto.
27180 (dimode_scalar_chain::convert_reg): Ditto.
27181 (dimode_scalar_chain::convert_insn): Ditto.
27182 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27183 (VI248_AVX512BW_1): New mode iterator.
27184 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27185 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27188 2017-04-25 Martin Sebor <msebor@redhat.com>
27190 PR tree-optimization/80497
27191 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27192 constants are representable in HOST_WIDE_INT.
27193 (parse_directive): Ditto.
27195 2017-04-25 Martin Sebor <msebor@redhat.com>
27198 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27199 (new_zero_array): Adjust signature.
27200 (dom_info::dom_init): Used unsigned rather that size_t.
27201 (dom_info::dom_info): Same.
27203 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27204 Jakub Jelinek <jakub@redhat.com>
27207 * config/arm/arm.c: Include gimple.h.
27208 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27209 returns negative, increment ncrn only if it returned positive.
27210 (arm_needs_doubleword_align): Return int instead of bool,
27211 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27212 members, but if there is any such non-FIELD_DECL
27213 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27214 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27215 returns negative, increment nregs only if it returned positive.
27216 (arm_setup_incoming_varargs): Likewise.
27217 (arm_function_arg_boundary): Emit -Wpsabi note if
27218 arm_needs_doubleword_align returns negative, return
27219 DOUBLEWORD_ALIGNMENT only if it returned positive.
27221 2017-04-25 Marek Polacek <polacek@redhat.com>
27224 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27225 first argument to type.
27227 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27230 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27231 type checks to test for compatibility instead of equality.
27233 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27234 Jakub Jelinek <jakub@redhat.com>
27237 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27239 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27240 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27241 the alignment computation, but return their maximum in warn_alignment.
27242 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27243 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27245 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27246 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27249 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27251 * config/arc/simdext.md (dmpyh): Fix typo.
27253 2017-04-25 Richard Biener <rguenther@suse.de>
27255 PR tree-optimization/80492
27256 * alias.c (compare_base_decls): Handle registers with asm
27257 specification conservatively.
27258 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27259 compare_base_decls returning dont-know properly.
27261 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27263 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27264 (legitimate_offset_address_p): New function.
27265 (arc_legitimate_address_p): Use above function.
27267 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27269 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27271 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27273 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27274 ACCH registers whenever they are available.
27276 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27278 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27279 double regs fix when not used.
27281 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27283 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27285 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27286 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27288 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27290 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27291 when emitting short ADD instructions.
27293 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27295 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27297 (cmpsi_cc_c_insn): Likewise.
27298 (cbranchsi4_scratch): Compute proper instruction length using
27299 compact_hreg_operand.
27300 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27302 2017-04-25 Richard Biener <rguenther@suse.de>
27304 PR middle-end/80509
27305 * passes.c (pass_manager::pass_manager): Initialize
27306 m_name_to_pass_map.
27308 2017-04-25 Richard Biener <rguenther@suse.de>
27310 PR tree-optimization/79201
27311 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27313 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27316 * config/s390/vector.md: Split MEM->GPR vector moves for
27317 non-s_operand addresses.
27319 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27322 * config/s390/predicates.md (reload_const_wide_int_operand): New
27324 * config/s390/s390.md ("movti"): Remove d/P alternative.
27325 ("movti_bigconst"): New pattern definition.
27327 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27330 * s390-protos.h (s390_expand_cs_hqi): Removed.
27331 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27332 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27333 modes as well as CCZ1mode and CCZmode.
27334 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27335 signature of s390_emit_compare_and_swap.
27336 (s390_expand_cs_hqi): Likewise, make static.
27337 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27338 compare-and-swap, in some cases.
27339 (s390_expand_cs): Wrapper function.
27340 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27342 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27343 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27344 patterns for small and large integers. Forbid symref memory operands.
27345 Move expander to s390.c. Require cc register.
27346 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27347 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27348 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27349 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27350 symref memory operands. Remove CC mode and call s390_match_ccmode
27352 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27354 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27356 * config/s390/s390.md (define_peephole2): New peephole to help
27357 combining the load-and-test pattern with volatile memory.
27359 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27361 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27362 with CCZmode for TARGET_Z196.
27364 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27366 PR rtl-optimization/80501
27367 * combine.c (make_compound_operation_int): Set subreg_code to SET
27368 even for AND with mask of the sign bit of mode.
27370 PR rtl-optimization/80500
27371 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27372 sum's initial value.
27374 2017-04-25 Julian Brown <julian@codesourcery.com>
27375 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27377 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27379 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27381 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27383 2017-04-25 Julian Brown <julian@codesourcery.com>
27384 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27386 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27387 (thunderx2t99_sha): New Reservation.
27389 2017-04-25 Julian Brown <julian@codesourcery.com>
27390 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27392 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27393 type for 1-element load.
27395 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27397 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27399 2017-04-24 Martin Jambor <mjambor@suse.cz>
27401 PR tree-optimization/80293
27402 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27403 char arrays not totally scalarizable if it is false.
27404 (analyze_all_variable_accesses): Pass correct value in the new
27405 parameter. Add a statistics counter.
27407 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27409 PR middle-end/79931
27410 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27412 2017-04-24 Richard Biener <rguenther@suse.de>
27414 PR tree-optimization/80494
27415 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27416 out for complex types.
27418 2017-04-24 Richard Biener <rguenther@suse.de>
27420 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27421 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27422 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27423 (DFS): Adjust and never fail.
27424 (sccvn_dom_walker::fail): Remove.
27425 (sccvn_dom_walker::before_dom_children): Adjust.
27426 (run_scc_vn): Likewise and never fail.
27427 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27428 (pass_fre::execute): Likewise.
27430 2017-04-24 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/79725
27433 * tree-ssa-sink.c (statement_sink_location): Return whether
27434 failure reason was zero uses. Move that check later.
27435 (sink_code_in_bb): Deal with zero uses by removing the stmt
27438 2017-04-24 Richard Biener <rguenther@suse.de>
27441 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27442 pointer-based references.
27444 2017-04-24 Richard Biener <rguenther@suse.de>
27447 * pass_manager.h (pass_manager::operator new): Remove.
27448 (pass_manager::operator delete): Likewise.
27449 * passes.c (pass_manager::operator new): Remove.
27450 (pass_manager::operator delete): Likewise.
27451 (pass_manager::pass_manager): Zero individual pass members.
27453 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27456 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27457 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27458 Check "XEXP (src, 1)" operand here.
27459 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27460 Check "XEXP (src, 1)" operand here.
27461 (dimode_scalar_chain::make_vector_copies): Detect count register
27462 of a shift instruction. Zero extend count register from QImode
27463 to DImode to satisfy vector shift pattern count operand predicate.
27464 Substitute vector shift count operand with a DImode copy.
27465 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27468 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27470 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27471 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27472 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27473 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27474 (UNSPEC_NOREX_MEM): Remove definition.
27476 2017-04-21 Richard Biener <rguenther@suse.de>
27478 PR tree-optimization/79547
27479 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27480 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27481 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27482 without any constraints.
27484 2017-04-21 Richard Biener <rguenther@suse.de>
27486 PR tree-optimization/78847
27487 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27489 2017-04-21 Richard Biener <rguenther@suse.de>
27491 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27492 (build_distinct_type_copy): Likewise.
27493 (build_variant_type_copy): Likewise.
27494 * tree.c (build_qualified_type): Pass down mem-stat info.
27495 (build_distinct_type_copy): Likewise.
27496 (build_variant_type_copy): Likewise.
27498 2017-04-21 Richard Biener <rguenther@suse.de>
27500 PR tree-optimization/80237
27501 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27503 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27504 for a simplified result.
27506 2016-04-21 Richard Biener <rguenther@suse.de>
27508 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27509 sth as strict as a simple_iv but a chrec without symbols and an
27510 operand defined in the loop we are peeling (and not some subloop).
27511 (propagate_constants_for_unrolling): Propagate all constants.
27513 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27516 * config/i386/i386.c (print_reg): Remove assert for disalowed
27517 regno values, call output_operand_lossage instead.
27519 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27522 * config/i386/constraints.md (Yc): New register constraint.
27523 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27524 Use Yc constraint for alternative 2 of operand 0. Remove
27525 preferred_for_speed attribute.
27527 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27529 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27530 lastprivate clauses in SIMT case.
27532 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27534 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27536 2017-04-20 Richard Biener <rguenther@suse.de>
27538 PR tree-optimization/57796
27539 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27540 as N scalar stores.
27541 (vect_model_load_cost): Cost gathers as N scalar loads.
27543 2017-04-20 Richard Biener <rguenther@suse.de>
27545 * ggc-page.c (ggc_allocated_p): Rename to ...
27546 (safe_lookup_page_table_entry): ... this and return the lookup
27548 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27550 2017-04-20 Richard Biener <rguenther@suse.de>
27552 PR tree-optimization/80453
27553 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27554 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27555 from the conditions.
27556 (vn_phi_eq): Pass them down.
27557 (vn_phi_lookup): Record them.
27558 (vn_phi_insert): Likewise.
27560 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27562 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27563 uninitialized variable warning to avoid buffer overrun.
27565 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27568 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27569 is suppressed for '{ 0 }' in C.
27571 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27573 * BASE-VER: Set to 8.0.0.
27575 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27577 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27578 priority .init_array and .fini_array section with SECTION_NOTYPE
27581 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27583 PR middle-end/80423
27584 * tree.h (build_array_type): Add typeless_storage default argument.
27585 * tree.c (type_cache_hasher::equal): Also compare
27586 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27587 (build_array_type): Add typeless_storage argument, set
27588 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27590 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27591 (build_array_type): Likewise. Add typeless_storage argument.
27593 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27594 Jakub Jelinek <jakub@redhat.com>
27596 PR tree-optimization/80426
27597 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27598 operation on symbolic operands, also compute the overflow for the
27599 invariant part when the operation degenerates into a negation.
27601 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27604 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27605 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27608 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27610 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27613 * config/avr/avr.c (tree.h): Include it.
27614 (cgraph.h): Include it.
27615 (avr_encode_section_info): Don't warn for uninitialized progmem
27616 variable if it's just an alias.
27618 2017-04-19 Richard Biener <rguenther@suse.de>
27621 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27622 when needed by AutoPGO.
27624 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27627 * doc/lto.texi: Remove an extra 'that'.
27629 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27631 PR rtl-optimization/80429
27632 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27633 are only used in debug insns.
27635 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27636 Vladimir Makarov <vmakarov@redhat.com>
27638 * config/sparc/predicates.md (input_operand): Add comment. Return
27639 true for any memory operand when LRA is in progress.
27640 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27642 2017-04-18 Jeff Law <law@redhat.com>
27645 * mips.md ({return,simple_return}_internal): Do not overwrite
27648 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27650 PR tree-optimization/80443
27651 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27652 instead of adding 1, subtract -1 and similarly instead of subtracting
27655 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27657 PR rtl-optimization/80357
27658 * haifa-sched.c (tmp_bitmap): New variable.
27659 (model_recompute): Handle duplicate use records.
27660 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27661 (free_global_sched_pressure_data): Free it.
27663 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27666 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27667 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27668 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27669 instead of SYSTEM_HEADER_DIR.
27671 2017-04-18 Jeff Law <law@redhat.com>
27673 PR middle-end/80422
27674 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27675 predecessors after walking up the insn chain.
27677 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27680 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27681 sizetype type into debug info.
27683 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27686 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27687 unneeded test for TARGET_UPPER_REGS_SF.
27688 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27690 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27693 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27694 instead of gsi_after_labels.
27696 2017-04-18 Jeff Law <law@redhat.com>
27698 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27702 2017-04-13 Jeff Law <law@redhat.com>
27703 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27704 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27706 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27709 * config/avr/avr.c (intl.h): Include it.
27710 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27712 2017-04-18 Martin Liska <mliska@suse.cz>
27714 PR gcov-profile/78783
27715 * gcov-tool.c (gcov_output_files): Validate that destination
27716 file is either removed by the tool or by a user.
27718 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27719 Guy Benyei <guybe@mellanox.com>
27721 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27722 block, and do not negate it, the stored id is already negative.
27724 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27726 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27728 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27731 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27732 masks of options that should be turned off if the VSX vector
27733 options are turned off.
27734 (OTHER_P8_VECTOR_MASKS): Likewise.
27735 (OTHER_VSX_VECTOR_MASKS): Likewise.
27736 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27737 rs6000_disable_incompatible_switches to validate no type switches
27739 (rs6000_incompatible_switch): New function to disallow turning on
27740 other vector options if -mno-vsx, -mno-power8-vector, or
27741 -mno-power9-vector are specified.
27743 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27745 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27747 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27749 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27750 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27751 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27752 (ARG_POINTER_CFA_OFFSET): Likewise.
27754 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27756 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27757 conditions to take advantage of various optimizations.
27759 2017-04-13 Jeff Law <law@redhat.com>
27761 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27762 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27763 (zero_extendsidi2_dext): Likewise.
27765 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27768 * fold-const.c (fold_ternary_loc): Revert
27769 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27772 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27774 PR rtl-optimization/80343
27775 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27776 new scratch pseudo.
27778 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27781 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27782 to ubsan_encode_value.
27784 2017-04-13 Jeff Law <law@redhat.com>
27786 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27787 appearing in DEBUG_INSNs.
27789 2017-04-13 Martin Liska <mliska@suse.cz>
27791 PR gcov-profile/80413
27792 * gcov-io.c (gcov_write_string): Copy to buffer just when
27793 allocated size is greater than zero.
27795 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27798 * dwarf2out.c (decls_for_scope): Ignore declarations of
27799 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27801 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27804 * ipa-visibility.c (non_local_p): Fix typos.
27805 (localize_node): When localizing symbol in same comdat group,
27806 dissolve the group only when we know external symbols are going
27808 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27810 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27812 PR tree-optimization/79390
27813 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27814 order does not result in usable sequence, retry with reversed operand
27820 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27821 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27822 op0 instead of fold_convert_loc (loc, type, arg0).
27824 2017-04-12 Jeff Law <law@redhat.com>
27826 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27827 has a delay slot in the generated code.
27829 * config/cris/cris.md (cris_preferred_reload_class): Return
27830 GENNONACR_REGS rather than GENERAL_REGS.
27832 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27835 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27836 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27837 signedness of the result type.
27839 2017-04-12 Richard Biener <rguenther@suse.de>
27840 Jeff Law <law@redhat.com>
27842 PR tree-optimization/80359
27843 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27844 trim stores to TARGET_MEM_REFs.
27846 2017-04-12 Richard Biener <rguenther@suse.de>
27848 PR tree-optimization/79390
27849 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27850 threading case even more.
27852 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27855 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27856 for quad_address_p for TImode, instead of just not indexed_address.
27858 2017-04-12 Richard Biener <rguenther@suse.de>
27859 Bernd Edlinger <bernd.edlinger@hotmail.de>
27861 PR middle-end/79671
27862 * alias.c (component_uses_parent_alias_set_from): Handle
27863 TYPE_TYPELESS_STORAGE.
27864 (get_alias_set): Likewise.
27865 * tree-core.h (tree_type_common): Add typeless_storage flag.
27866 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27867 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27868 for types containing members with TYPE_TYPELESS_STORAGE.
27869 (place_field): Likewise.
27870 (layout_type): Likewise for ARRAY_TYPE.
27871 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27872 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27873 TYPE_TYPELESS_STORAGE.
27874 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27876 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27879 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27880 first argument to type.
27882 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27886 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27887 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27888 (rs6000_expand_binop_builtin): Likewise.
27889 (rs6000_expand_ternop_builtin): Likewise; also add missing
27890 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27891 vshasigma built-ins.
27892 * doc/extend.texi: Document that vec_xxpermdi's third argument
27893 must be a constant.
27895 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27897 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27898 Use shift_const cost parameter when calculating gain of STV shifts.
27900 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27902 PR rtl-optimization/70478
27903 * lra-constraints.c (process_alt_operands): Check memory for
27904 disfavoring memory insn operand.
27906 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27908 PR middle-end/80100
27909 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27910 left shift in unsigned HOST_WIDE_INT type.
27912 PR rtl-optimization/80385
27913 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27914 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27917 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27918 if they have any depend clauses.
27920 2017-04-11 Martin Liska <mliska@suse.cz>
27923 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27924 * ipa-split.c (split_function): Create a local comdat symbol
27925 if caller is in a comdat group.
27927 2017-04-11 Martin Liska <mliska@suse.cz>
27930 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27933 2017-04-11 Martin Sebor <msebor@redhat.com>
27935 PR middle-end/80364
27936 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27937 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27939 (directive::set_width, directive::set_precision, format_character):
27941 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27944 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27947 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27948 conflict, set target->arch_name instead of target->cpu_name.
27950 2017-04-11 Richard Biener <rguenther@suse.de>
27952 PR tree-optimization/80374
27953 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27954 build_zero_cst, remove fold_convertible_p check again.
27956 2017-04-11 Martin Liska <mliska@suse.cz>
27959 * ubsan.c (instrument_object_size): Do not instrument register
27962 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27965 * config/i386/i386-builtin-types.def
27966 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27967 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27968 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27969 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27970 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27971 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27972 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27973 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27974 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27975 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27976 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27977 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27978 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27979 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27980 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27981 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27982 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27983 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27984 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27985 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27986 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27987 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27988 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27989 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27990 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27991 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27992 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27993 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27994 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27995 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27996 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27997 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27998 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27999 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28000 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28001 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28002 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28003 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28004 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28005 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28006 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28007 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28008 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28009 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28010 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28011 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28013 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28014 flag to second_arg_count, handle 4 argument function type _COUNT
28015 aliases, handle second_arg_count on second argument rather than last.
28017 2017-04-10 Jeff Law <law@redhat.com>
28019 PR tree-optimization/80374
28020 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28021 record anything if we can not convert integer_zero_node to the
28024 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28027 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28028 Enhance special handling given to the TARGET_P9_MINMAX option in
28029 relation to certain other options.
28031 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28033 PR tree-optimization/80153
28034 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28035 remove POINTER_PLUS_EXPR's base part directly, rather than through
28038 2017-04-10 Richard Biener <rguenther@suse.de>
28039 Bin Cheng <bin.cheng@arm.com>
28041 PR tree-optimization/80153
28042 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28043 the first element of pointer type aff_tree. Build result expr in
28045 (add_elt_to_tree): Convert to type unconditionally. Remove other
28046 fold_convert calls.
28047 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28048 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28050 2017-04-10 Richard Biener <rguenther@suse.de>
28052 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28055 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28057 PR rtl-optimization/70478
28058 * lra-constraints.c (curr_small_class_check): New.
28059 (update_and_check_small_class_inputs): New.
28060 (process_alt_operands): Update curr_small_class_check. Disfavor
28061 alternative insn memory operands. Check available regs for small
28064 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28067 * config/mips/mips.opt (-mvirt): Update description.
28068 * doc/invoke.texi (-mvirt): Likewise.
28070 2017-04-10 Richard Biener <rguenther@suse.de>
28072 PR middle-end/80362
28073 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28074 looking for NEGATE_EXPR in -A / -B to A / B folding.
28076 2017-04-10 Martin Liska <mliska@suse.cz>
28078 PR gcov-profile/80224
28079 * gcov.c (print_usage): Fix usage string.
28080 (get_gcov_intermediate_filename): Remove.
28081 (output_gcov_file): Use both for normal and intermediate format.
28082 (generate_results): Do not initialize special file for
28083 intermediate format.
28085 2017-04-10 Richard Biener <rguenther@suse.de>
28087 PR tree-optimization/80304
28088 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28091 2017-04-10 Nathan Sidwell <nathan@acm.org>
28094 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28095 (rs6000_init_builtins): Use it.
28097 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28099 * config/arm/arm.md (<mrc>): Add mode to SET source.
28100 (<mrrc>): Likewise.
28102 2017-04-10 Richard Biener <rguenther@suse.de>
28104 PR middle-end/80344
28105 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28107 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28110 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28111 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28112 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28113 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28114 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28115 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28116 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28117 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28118 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28119 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28120 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28121 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28122 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28123 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28124 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28125 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28126 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28127 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28128 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28129 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28130 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28131 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28132 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28134 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28136 PR rtl-optimization/70478
28137 * lra-constraints.c: Reverse the last patch.
28139 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28141 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28142 Add comment for WCHAR_T.
28144 2017-04-08 Martin Liska <mliska@suse.cz>
28147 2017-04-07 Martin Liska <mliska@suse.cz>
28150 * ipa-split.c (split_function): Add function part to a same comdat
28153 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28156 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28158 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28160 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28161 (vec_pairing): ...to this.
28162 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28163 (rs6000_sched_init): Adjust for name change.
28164 (struct rs6000_sched_context): Likewise.
28165 (rs6000_init_sched_context): Likewise.
28166 (rs6000_set_sched_context): Likewise.
28168 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28174 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28176 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28177 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28178 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28180 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28182 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28184 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28186 PR rtl-optimization/70703
28187 * ira-color.c (update_conflict_hard_regno_costs): Use
28188 int64_t instead of HOST_WIDE_INT.
28190 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28192 PR rtl-optimization/70478
28193 * lra-constraints.c (process_alt_operands): Disfavor alternative
28194 insn memory operands.
28196 2017-04-07 Jeff Law <law@redhat.com>
28198 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28199 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28201 2017-04-07 Martin Liska <mliska@suse.cz>
28204 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28205 Show error message instead of an ICE.
28207 2017-04-07 Martin Liska <mliska@suse.cz>
28210 * ipa-split.c (split_function): Add function part to a same comdat
28213 2017-04-07 Richard Biener <rguenther@suse.de>
28215 PR middle-end/80341
28216 * tree.c (get_unwidened): Also handle ! for_type case for
28218 * convert.c (do_narrow): Split out from ...
28219 (convert_to_integer_1): ... here. Do not pass final truncation
28220 type to get_unwidened for TRUNC_DIV_EXPR.
28222 2017-04-07 Richard Biener <rguenther@suse.de>
28224 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28226 (aff_combination_const): Adjust.
28227 (aff_combination_scale): Likewise.
28228 (aff_combination_add_elt): Likewise.
28229 (aff_combination_add_cst): Likewise.
28230 (aff_combination_convert): Likewise.
28231 (add_elt_to_tree): Likewise. Remove unused argument.
28232 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28234 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28236 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28238 * config/arm/arm.c (arm_default_short_enums): Use
28239 ARM_DEFAULT_SHORT_ENUMS.
28240 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28242 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28245 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28246 members with redundant out-of-class redeclaration.
28248 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28251 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28252 * config/i386/i386.md (*zero_extendsidi2):
28253 Add (?*x,*x) and (?*v,*v) alternatives.
28255 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28258 * config/i386/i386.c (ix86_expand_builtin)
28259 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28260 mode from insn data. Convert operands to insn operand mode.
28261 Copy operands that don't satisfy insn predicate to a register.
28263 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28265 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28268 2017-04-06 Richard Biener <rguenther@suse.de>
28270 PR tree-optimization/80334
28271 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28272 preserve alignment of accesses.
28274 2017-04-06 Richard Biener <rguenther@suse.de>
28276 PR tree-optimization/80262
28277 * tree-sra.c (build_ref_for_offset): Preserve address-space
28279 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28280 Drop useless address-space information on MEM_REF offsets.
28282 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28284 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28286 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28288 PR rtl-optimization/70703
28289 * ira-color.c (update_conflict_hard_regno_costs): Use
28290 HOST_WIDE_INT instead of long.
28292 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28295 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28296 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28298 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28299 for x86_64 target. Handle -m3dnowa option.
28301 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28303 PR rtl-optimization/70703
28304 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28305 (update_conflict_hard_regno_costs): Use long instead of unsigned
28306 arithmetic for cost calculation.
28308 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28309 Bernd Edlinger <bernd.edlinger@hotmail.de>
28312 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28315 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28318 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28319 ptr_mode with Pmode throughout.
28320 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28321 into probe_stack_range and use DImode.
28323 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28326 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28327 call_eh_return is true.
28329 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28331 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28332 Initialize last_match_fntype_index.
28334 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28337 * tree-nvr.c: Include internal-fn.h.
28338 (pass_return_slot::execute): Ignore internal calls without
28341 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28342 Richard Biener <rguenther@suse.de>
28345 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28346 captures used multiple times, except for the last use.
28347 * generic-match-head.c: Include gimplify.h.
28349 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28351 PR tree-optimization/79390
28352 * target.h (struct noce_if_info): Declare.
28353 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28354 * target.def (noce_conversion_profitable_p): New target hook.
28355 * ifcvt.h (struct noce_if_info): New type, moved from ...
28356 * ifcvt.c (struct noce_if_info): ... here.
28357 (noce_conversion_profitable_p): Renamed to ...
28358 (default_noce_conversion_profitable_p): ... this. No longer
28360 (noce_try_store_flag_constants, noce_try_addcc,
28361 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28362 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28363 instead of noce_conversion_profitable_p.
28364 * config/i386/i386.c: Include ifcvt.h.
28365 (ix86_option_override_internal): Don't override
28366 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28367 (ix86_noce_conversion_profitable_p): New function.
28368 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28369 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28370 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28371 * doc/tm.texi: Regenerated.
28373 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28375 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28378 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28381 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28382 instructions for small multiply cores.
28384 2017-04-04 Jeff Law <law@redhat.com>
28386 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28388 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28389 that are not set by the loop over the elements.
28391 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28394 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28395 int mode, convert_modes it to mode as unsigned, otherwise use
28396 lowpart_subreg to mode rather than SImode.
28397 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28398 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28399 Use DImode instead of SImode for the shift count operand.
28400 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28403 2017-04-04 Richard Biener <rguenther@suse.de>
28405 PR middle-end/80281
28406 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28407 arithmetic done for the negate or the plus. Simplify.
28408 (A - (-B) -> A + B): Likewise.
28409 * fold-const.c (split_tree): Make sure to not negate pointers.
28411 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28413 PR rtl-optimization/60818
28414 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28415 a compare of comparisons with the thing compared if this results
28416 in a different machine mode.
28418 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28420 * alias.c (base_alias_check): Fix typo in comment.
28421 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28422 * cgraphunit.c (symbol_table::compile): Likewise.
28423 * collect2.c (maybe_run_lto_and_relink): Likewise.
28424 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28425 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28426 * config/avr/avr.c (avr_map_op_t): Likewise.
28427 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28428 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28429 * config/epiphany/epiphany.md (movcc): Likewise.
28430 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28431 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28433 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28434 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28435 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28436 * config/sh/sh.c (sh_rtx_costs): Likewise.
28437 * fold-const.c (fold_truth_andor): Likewise.
28438 * genautomata.c (collapse_flag): Likewise.
28439 * gengtype.h (struct type::u::s): Likewise.
28440 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28441 * input.c (FORMAT_AMOUNT): Likewise.
28442 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28443 (known_aggs_to_agg_replacement_list): Likewise.
28444 * ipa-inline-analysis.c: Likewise.
28445 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28446 * ipa-polymorphic-call.c
28447 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28448 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28449 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28451 * modulo-sched.c (apply_reg_moves): Likewise.
28452 * omp-expand.c (build_omp_regions_1): Likewise.
28453 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28454 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28455 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28456 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28457 * value-prof.c: Likewise.
28458 * var-tracking.c (val_reset): Likewise.
28460 2017-04-03 Richard Biener <rguenther@suse.de>
28462 PR tree-optimization/80275
28463 * fold-const.c (split_address_to_core_and_offset): Handle
28466 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28468 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28469 descriptors is at least equal to that of functions.
28471 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28473 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28475 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28478 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28479 (mov<IMOD4:mode>): New expander.
28480 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28482 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28484 PR rtl-optimization/79405
28485 * fwprop.c (propagations_left): New variable.
28486 (forward_propagate_into): Decrement it.
28487 (fwprop_init): Initialize it.
28488 (fw_prop): If the variable has reached zero, stop propagating.
28489 (fwprop_addr): Ditto.
28491 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28494 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28495 a FUNCTION_DECL, pass it as decl instead of origin to
28498 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28500 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28503 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28506 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28507 TARGET_VSX_SMALL_INTEGER.
28509 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28511 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28512 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28514 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28516 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28517 extraction from odd-numbered MSA register.
28519 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28521 PR middle-end/80173
28522 * expmed.c (store_bit_field_1): Don't attempt to create
28523 a word subreg out of hard registers wider than word if they
28524 have HARD_REGNO_NREGS of 1 for their mode.
28526 PR middle-end/80163
28527 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28528 conversions to integer types wider than word and pointer.
28531 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28532 (rtx_equal_for_cselib_p): Pass 0 to it.
28533 * cselib.c (cselib_hasher::equal): Likewise.
28534 (rtx_equal_for_cselib_1): Add depth argument. If depth
28535 is 128, don't look up VALUE locs and punt. Increment
28536 depth in recursive calls when walking VALUE locs.
28538 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28540 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28541 (make_gcov_file_name): Use the canonical path name for generating
28543 (read_line): Fix handling of files with ascii null bytes.
28545 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28547 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28548 to initialise a vector register instead
28549 of using a const_int.
28551 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28553 PR translation/80189
28554 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28555 diagnostic messages.
28557 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28560 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28561 (dfp_diex_<mode>): Update mode of operand 1.
28562 * doc/extend.texi (dxex, dxexq): Document change to return type.
28563 (diex, diexq): Document change to argument type.
28565 2017-03-30 Martin Jambor <mjambor@suse.cz>
28568 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28569 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28570 it reflects the signature changes performed at the callee side.
28571 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28572 to cgraph_build_function_type_skip_args.
28573 (build_function_decl_skip_args): Adjust call to the above function.
28575 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28578 * config/i386/sse.md
28579 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28580 register as dest whenever it is a MEM not rtx_equal_p to the
28581 corresponding dup operand, and when forcing into reg move the
28582 reg into the memory afterwards.
28583 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28584 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28585 for the force_reg mode.
28586 (avx512vl_vextractf128<mode>): Use register as dest either
28587 always when a MEM, or when it is a MEM not rtx_equal_p to the
28588 corresponding dup operand, or even not when it is a CONST_VECTOR
28589 depending on the mode and lo vs. hi.
28590 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28592 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28593 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28594 Likewise. Require that operands[2] is even.
28595 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28596 Remove extraneous parens. Require that operands[2] is a multiple
28598 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28599 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28600 disallow memory then.
28602 2017-03-30 Richard Biener <rguenther@suse.de>
28604 PR tree-optimization/77498
28605 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28606 to non-constants over backedges.
28608 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28610 PR rtl-optimization/80233
28611 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28612 as last_combined_insn. Do not test for BARRIER_P separately.
28614 2017-03-29 Andreas Schwab <schwab@suse.de>
28617 * calls.c (prepare_call_address): Convert funexp to Pmode before
28618 copying to temp reg.
28620 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28622 PR tree-optimization/80158
28623 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28624 Handle possible future case of more than one alternate
28626 (replace_rhs_if_not_dup): Likewise.
28627 (replace_one_candidate): Likewise.
28629 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28631 PR rtl-optimization/80193
28632 * ira.c (ira): Do not check allocation for LRA.
28634 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28636 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28637 (nvptx_output_simt_exit): Declare.
28638 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28639 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28640 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28641 (nvptx_declare_function_name): ...here. Emit declaration of local
28642 memory space buffer for omp_simt_enter insn.
28643 (nvptx_output_unisimt_switch): New.
28644 (nvptx_output_softstack_switch): New.
28645 (nvptx_output_simt_enter): New.
28646 (nvptx_output_simt_exit): New.
28647 * config/nvptx/nvptx.h (struct machine_function): New fields
28648 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28649 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28650 (UNSPECV_SIMT_EXIT): Ditto.
28651 (omp_simt_enter_insn): New insn.
28652 (omp_simt_enter): New expansion.
28653 (omp_simt_exit): New insn.
28654 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28656 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28657 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28658 (expand_GOMP_SIMT_EXIT): New.
28659 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28660 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28661 (GOMP_SIMT_EXIT): Ditto.
28662 * target-insns.def (omp_simt_enter): New insn.
28663 (omp_simt_exit): Ditto.
28664 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28666 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28667 (lower_rec_input_clauses): Likewise.
28668 (lower_lastprivate_clauses): Handle SIMT privatization.
28670 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28671 (ompdevlow_adjust_simt_enter): New.
28672 (find_simtpriv_var_op): New.
28673 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28674 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28676 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28677 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28678 (copy_decl_for_dup_finish): Ditto.
28680 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28682 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28685 * config/i386/i386.c (ix86_option_override_internal): Always
28686 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28688 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28690 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28692 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28694 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28695 mark new edge's irreducible flag accordign to it.
28696 (vect_do_peeling): Check loop preheader edge's irreducible flag
28697 and pass it to function slpeel_add_loop_guard.
28699 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28701 PR tree-optimization/80218
28702 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28703 Update block frequencies and counts.
28705 2017-03-28 Richard Biener <rguenther@suse.de>
28707 PR tree-optimization/78644
28708 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28709 of a simplification result we may not use it at all.
28711 2017-03-28 Richard Biener <rguenther@suse.de>
28714 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28715 without arguments, generate default definition of a SSA name.
28717 2017-03-28 Richard Biener <rguenther@suse.de>
28719 PR middle-end/80222
28720 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28721 TYPE_REF_CAN_ALIAS_ALL references.
28722 * fold-const.c (fold_indirect_ref_1): Likewise.
28724 2017-03-28 Martin Liska <mliska@suse.cz>
28727 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28728 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28730 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28731 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28733 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28734 (EXTRA_SPECS): Define.
28735 (SUBTARGET_EXTRA_SPECS): Likewise.
28736 (SUBTARGET_CPP_SPEC): Likewise.
28737 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28738 SUBTARGET_EXTRA_SPECS.
28739 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28741 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28743 * config/arc/simdext.md (vst64_insn): Update pattern.
28744 (vld32wh_insn): Likewise.
28745 (vld32wl_insn): Likewise.
28746 (vld64_insn): Likewise.
28747 (vld32_insn): Likewise.
28749 2017-03-28 Marek Polacek <polacek@redhat.com>
28752 * fold-const.c (fold_comparison): Use protected_set_expr_location
28753 instead of SET_EXPR_LOCATION.
28755 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28757 * tree.c (add_expr): Avoid name lookup warning.
28759 2017-03-27 Jeff Law <law@redhat.com>
28761 PR tree-optimization/80216
28762 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28763 function name. Limit recursion depth.
28764 (record_temporary_equivalences): Corresponding changes.
28766 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28768 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28771 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28774 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28776 * cfgcleanup.c (reg_note_cfa_p): New array.
28777 (insns_have_identical_cfa_notes): New function.
28778 (old_insns_match_p): Don't cross-jump in between /f
28779 and non-/f instructions. If both i1 and i2 are frame related,
28780 verify all CFA notes, their order and content.
28782 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28785 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28786 HImode and SImode with zero extend to DImode to one insn.
28787 (bswap<mode>2_extenddi): Likewise.
28788 (bswapsi2_extenddi): Likewise.
28789 (bswaphi2_extendsi): Likewise.
28790 (bswaphi2): Combine bswap HImode and SImode into one insn.
28791 Separate memory insns from swapping register.
28792 (bswapsi2): Likewise.
28793 (bswap<mode>2): Likewise.
28794 (bswaphi2_internal): Delete, no longer used.
28795 (bswapsi2_internal): Likewise.
28796 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28797 store, and gpr<-gpr swap insns.
28798 (bswap<mode>2_store): Likewise.
28799 (bswaphi2_reg): Register only splitter, combine with the splitter.
28800 (bswaphi2 splitter): Likewise.
28801 (bswapsi2_reg): Likewise.
28802 (bswapsi2 splitter): Likewise.
28803 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28804 the insns into load, store, and register/register insns.
28805 (bswapdi2_ldbrx): Likewise.
28806 (bswapdi2_load): Likewise.
28807 (bswapdi2_store): Likewise.
28808 (bswapdi2_reg): Likewise.
28810 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28812 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28813 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28815 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28818 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28820 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28821 special handling for target option conflicts between dform
28822 options (-mpower9-dform, -mpower9-dform-vector,
28823 -mpower9-dform-scalar) and -mno-direct-move.
28825 2017-03-27 Richard Biener <rguenther@suse.de>
28827 PR tree-optimization/80181
28828 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28830 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28832 * config/arc/predicates.md (move_double_src_operand): Replace the
28833 call to move_double_src_operand with a call to address_operand.
28835 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28837 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28838 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28839 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28841 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28843 * config/arc/predicates.md (long_immediate_loadstore_operand):
28844 Consider scaled addresses cases.
28846 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28848 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28849 restored when in interrupt.
28850 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28851 doesn't have delay slot.
28853 2017-03-27 Richard Biener <rguenther@suse.de>
28856 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28857 inlined thunk clones.
28859 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28862 * asan.c (instrument_derefs): Copy over last operand from
28863 original COMPONENT_REF to the new COMPONENT_REF with
28864 DECL_BIT_FIELD_REPRESENTATIVE.
28865 * ubsan.c (instrument_object_size): Likewise.
28867 2017-03-27 Richard Biener <rguenther@suse.de>
28869 PR tree-optimization/80170
28870 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28871 sure DR/SCEV didnt fold in constants we do not see when looking
28872 at the reference base alignment.
28874 2017-03-27 Richard Biener <rguenther@suse.de>
28876 PR middle-end/80171
28877 * gimple-fold.c (fold_ctor_reference): Properly guard against
28878 NULL return value from canonicalize_constructor_val.
28880 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28883 * config/i386/i386.c (ix86_expand_builtin)
28884 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28885 flags reg setting and flags reg using instructions.
28886 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28887 clobbering instructions to zero extend op2.
28889 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28891 * doc/install.texi (Configuration) <--with-aix-soname>:
28892 Update link to AIX ld.
28894 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28896 PR rtl-optimization/80160
28897 PR rtl-optimization/80159
28898 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28899 reg_alternate_class into account.
28901 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28904 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28905 to consider in curr_insn_transform.
28907 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28909 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28910 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28911 and emit_mode_inner.
28913 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28915 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28916 argument to the overloaded builtin variants. Use the new flag to
28917 deprecate certain builtin variants.
28918 * config/s390/s390-builtin-types.def: Add new builtin types.
28919 * config/s390/s390-builtins.h: Support new flags field for
28920 overloaded builtins.
28921 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28922 (s390_macro_to_expand): Enable vector float data type.
28923 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28924 builtins by incrementing the __VEC__ version number.
28925 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28927 (s390_resolve_overloaded_builtin): Emit error messages depending
28928 on the builtin flags.
28929 * config/s390/s390.c (s390_expand_builtin): Support additional
28930 flags argument. Change error message to match the messages
28931 emitted in s390-c.c.
28932 * config/s390/s390.md: New UNSPEC_* constants.
28933 (op_type): Add new instruction types.
28934 * config/s390/vecintrin.h: Add new builtins and test data class
28936 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28937 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28938 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28939 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28940 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28941 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28943 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28944 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28945 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28946 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28948 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28949 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28950 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28951 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28953 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28954 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28955 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28956 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28957 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28958 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28959 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28961 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28962 ("vec_scatter_element<V_HW_4:mode>_DI")
28963 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28964 ("vec_fpint<mode>", "vflls")
28965 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28966 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28967 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28968 ("*vec_cmphe<mode>_cc"): ... these.
28970 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28971 mode constant instead of magic value.
28973 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28975 * config/s390/s390.c (s390_expand_vec_compare): Support other
28976 vector floating point modes than just V2DF.
28977 (s390_expand_vcond): Likewise.
28978 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28979 (s390_cannot_change_mode_class): Prevent mode changes between TF
28980 and V1TF in vector registers.
28981 * config/s390/s390.md (DF, SF): New mode attributes.
28982 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28983 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28984 SFmode support for VRs.
28985 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28987 (VFT, VF_HW): New mode iterators.
28988 (vw, sdx): New mode attributes.
28989 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28990 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28991 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28992 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28993 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28994 also the new vector floating point modes. Renaming to ...
28996 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28997 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28998 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28999 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29000 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29001 ("vec_unordered<mode>"): ... these.
29003 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29004 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29005 ("*vec_extendv2df"): New insn definitions.
29007 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29009 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29010 ("mulditi3_2", "*muldi3_sign"): New patterns.
29011 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29012 rename the pattern definition.
29014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29016 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29018 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29020 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29022 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29023 instruction if possible.
29024 * config/s390/vector.md (vec_halfnumelts): New mode
29026 ("*vec_vllezlf<mode>"): New pattern.
29028 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29030 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29031 ("popcountv4si2", "popcountv2di2"): Rename to ...
29032 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29033 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29035 ("popcount<mode>2_vxe"): New pattern.
29037 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29039 * common/config/s390/s390-common.c (processor_flags_table): Add
29041 * config.gcc: Add arch12.
29042 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29043 Default to arch12 for unknown CPU model numbers.
29044 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29045 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29046 PROCESSOR_max sanity check.
29047 * config/s390/s390-opts.h (enum processor_type): Add
29049 * config/s390/s390.c (processor_table): Add arch12.
29050 (s390_expand_builtin): Add check for B_VXE flag.
29051 (s390_issue_rate): Add PROCESSOR_ARCH12.
29052 (s390_get_sched_attrmask): Likewise.
29053 (s390_get_unit_mask): Likewise.
29054 (s390_sched_score): Enable z13 scheduling for arch12.
29055 (s390_sched_reorder): Likewise.
29056 (s390_sched_variable_issue): Likewise.
29057 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29059 (s390_tune_attr): Use z13 scheduling also for arch12.
29060 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29061 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29062 (TARGET_VXE_P): New macros.
29063 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29064 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29065 * config/s390/s390.opt: Add arch12 as processor_type.
29067 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29069 * config/s390/s390.md
29070 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29071 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29072 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29074 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29075 Rename expanders to ...
29077 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29078 ("fixuns_truncdddi2_emu"): ... these.
29080 ("fixuns_trunc<mode>si2_emu"): New expander.
29082 ("*fixuns_truncdfdi2_z13"): Rename to ...
29083 ("*fixuns_truncdfdi2_vx"): ... this.
29085 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29087 * config/s390/2964.md: Remove the single element vector compare
29088 instructions which are no longer used.
29089 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29091 (s390_canonicalize_comparison): Remove handling of DFmode
29093 (s390_expand_vec_compare_scalar): Remove function.
29094 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29095 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29097 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29099 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29101 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29103 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29104 will anyway by matched by mov<mode>_64dfp.
29106 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29108 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29109 vlef/vstef. Add missing operand to vleif.
29111 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29113 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29114 pair for all vector types with 64 bit elements.
29115 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29116 * config/s390/vector.md (V_HW_64): ... here.
29117 (V_128_NOSINGLE): New mode iterator.
29118 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29119 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29120 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29121 ("*vec_load_pairv2di"): Change to ...
29122 ("*vec_load_pair<mode>"): ... this one.
29124 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29126 * config/s390/constraints.md: Add comments.
29127 (jKK): Reject element sizes > 8 bytes.
29128 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29130 * config/s390/s390.md: Add the s_operand checks formerly in
29131 s390_split_ok_p to various splitters where they are still
29133 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29134 for 128 bit vectors. Plus two splitters.
29136 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29138 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29141 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29144 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29145 error if the boundary argument is not constant.
29147 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29149 PR rtl-optimization/80112
29150 * loop-doloop.c (doloop_condition_get): Don't check condition
29151 if cmp isn't SET with IF_THEN_ELSE src.
29153 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29155 PR tree-optimization/80158
29156 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29157 replacing a candidate statement, also replace it for the
29158 candidate's alternate interpretation.
29159 (replace_rhs_if_not_dup): Likewise.
29160 (replace_one_candidate): Likewise.
29162 2017-03-24 Richard Biener <rguenther@suse.de>
29164 PR tree-optimization/80167
29165 * graphite-isl-ast-to-gimple.c
29166 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29168 (translate_isl_ast_to_gimple::get_rename): Likewise.
29170 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29172 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29173 handling of certain combinations of target options, including the
29174 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29175 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29177 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29180 * config/arm/arm.md (*load_multiple): Add reload_completed to
29181 matching condition.
29183 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29184 Richard Biener <rguenth@suse.de>
29186 PR tree-optimization/79908
29187 PR tree-optimization/80136
29188 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29189 been cast away, gimplify_and_add suffices.
29191 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29193 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29195 2017-03-23 Richard Biener <rguenther@suse.de>
29197 PR tree-optimization/80032
29198 * gimplify.c (gimple_push_cleanup): Forced unconditional
29199 cleanups still have to go to the conditional_cleanups
29202 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29204 PR tree-optimization/80072
29205 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29207 (next_operand_entry_id): Change type to unsigned int.
29208 (sort_by_operand_rank): Make sure to return the right return value
29209 even if unsigned fields are bigger than INT_MAX.
29210 (struct oecount): Change cnt and id type to unsigned int.
29211 (oecount_hasher::equal): Formatting fix.
29212 (oecount_cmp): Make sure to return the right return value
29213 even if unsigned fields are bigger than INT_MAX.
29214 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29217 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29218 TREE_READONLY on result if writing it more than once.
29221 * doc/invoke.texi (-fsanitize=thread): Document that with
29222 -fnon-call-exceptions atomics are not able to throw
29226 * tsan.c: Include tree-eh.h.
29227 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29228 maybe_clean_or_replace_eh_stmt where needed.
29229 (instrument_memory_accesses): Add cfg_changed argument.
29230 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29231 if it returned true.
29232 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29234 PR rtl-optimization/63191
29235 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29236 wrapper function, moved the whole old content into ...
29237 (ix86_delegitimize_address_1): ... this. New inline function.
29238 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29239 true as last argument instead of ix86_delegitimize_address.
29241 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29243 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29244 cortexa57_branch_cost.
29246 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29248 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29250 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29253 * doc/md.texi (Constraints): Document wA constraint.
29254 * config/rs6000/constraints.md (wA): New.
29255 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29256 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29257 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29258 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29260 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29263 * gimplify.c (is_oacc_declared): New function.
29264 (oacc_default_clause): Use it to set default flags for acc declared
29265 variables inside parallel regions.
29266 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29267 declared variables.
29268 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29269 declare attribute to any decl as necessary.
29271 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29274 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29275 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29276 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29277 (arm_arch_lpae): This.
29278 * config/arm/arm.c (arm_arch7ve): Rename into ...
29279 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29280 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29283 2017-03-22 Martin Liska <mliska@suse.cz>
29286 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29287 error message instead of an ICE.
29289 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29291 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29293 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29295 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29297 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29300 2017-03-21 Martin Sebor <msebor@redhat.com>
29302 * doc/extend.texi: Use "cannot" instead of "can't."
29303 * doc/hostconfig.texi: Same.
29304 * doc/install.texi: Same.
29305 * doc/invoke.texi: Same.
29306 * doc/loop.texi: Same.
29307 * doc/md.texi: Same.
29308 * doc/objc.texi: Same.
29309 * doc/rtl.texi: Same.
29310 * doc/tm.texi: Same.
29311 * doc/tm.texi.in: Same.
29312 * doc/trouble.texi: Same.
29314 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29317 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29318 (collect_checksum_attributes): Set it.
29319 (die_checksum_ordered): Use it.
29321 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29323 PR tree-optimization/79908
29324 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29325 change: For a VA_ARG whose LHS has been cast away, use
29326 force_gimple_operand to construct the side effects.
29328 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29330 PR translation/80001
29331 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29332 more amenable to translation.
29333 (oacc_loop_auto_partitions): Likewise.
29335 2017-03-21 Marek Polacek <polacek@redhat.com>
29336 Martin Sebor <msebor@redhat.com>
29338 PR tree-optimization/80109
29339 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29340 on INTEGRAL_TYPE_P.
29342 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29343 Segher Boessenkool <segher@kernel.crashing.org>
29346 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29347 check reg_used_between_p between insn and one of succ or succ2
29348 depending on if succ is artificial insn not inserted into insn
29351 2017-03-21 Martin Liska <mliska@suse.cz>
29353 PR gcov-profile/80081
29354 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29355 * doc/gcc.texi: Include gcov-dump stuff.
29356 * doc/gcov-dump.texi: New file.
29358 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29360 PR rtl-optimization/79150
29361 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29362 conditional jump, if the jump is the last insn of the loop.
29364 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29365 Richard Biener <rguenth@suse.de>
29367 PR tree-optimization/79908
29368 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29369 been cast away, use force_gimple_operand to construct the side
29372 2017-03-21 Martin Liska <mliska@suse.cz>
29374 PR libfortran/79956
29375 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29378 2017-03-21 Brad Spengler <spender@grsecurity.net>
29381 * plugin.c (htab_hash_plugin): New function.
29382 (add_new_plugin): Use it and adjust.
29383 (parse_plugin_arg_opt): Adjust.
29384 (init_one_plugin): Likewise.
29386 2017-03-21 Richard Biener <rguenther@suse.de>
29388 PR tree-optimization/80032
29389 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29390 if set force the cleanup to happen unconditionally.
29391 (gimplify_target_expr): Push inserted clobbers with force_uncond
29392 to avoid them being removed by control-dependent DCE.
29394 2017-03-21 Richard Biener <rguenther@suse.de>
29396 PR tree-optimization/80122
29397 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29398 va_arg_pack_len when the inlined call stmt requires pack
29400 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29402 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29405 * tsan.c (instrument_builtin_call): If the memory model argument
29406 is not a constant, assume it is valid.
29409 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29412 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29414 PR rtl-optimization/79910
29415 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29416 if its dest is used by an insn before I2 (other than the combined
29417 insns themselves, which are properly handled already).
29419 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29422 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29424 * combine.c (record_used_regs): New static function.
29425 (try_combine): Handle situations where there is an additional
29426 instruction between I2 and I3 which needs to have a LOG_LINK
29430 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29432 * combine.c (try_combine): Delete redundant i1 test. Call
29433 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29435 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29438 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29439 alternatives 13/14.
29441 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29443 PR tree-optimization/80054
29444 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29445 the optimization if a PHI or any of its arguments is not dominated
29446 by the candidate's basis. Use gphi* rather than gimple* as
29448 (replace_profitable_candidates): Clean up a gimple* variable that
29449 should be a gphi* variable.
29451 2017-03-20 Martin Sebor <msebor@redhat.com>
29454 * doc/extend.texi (attribute constructor): Document present limitation.
29456 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29459 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29460 __POWER9_VECTOR__ #ifdef control, change template definition to
29461 use Power9-specific built-in function.
29462 (vec_any_eq): Likewise.
29463 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29464 to control outcomes from this test.
29465 (vector_ae_<mode>p): For VEC_F modes, likewise.
29467 2017-03-20 Ian Lance Taylor <iant@google.com>
29469 * config/i386/i386.c (ix86_function_regparm): Save an extra
29470 register for -fsplit-stack with DECL_STATIC_CHAIN.
29472 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29475 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29476 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29478 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29480 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29482 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29485 2017-03-20 Marek Polacek <polacek@redhat.com>
29488 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29490 2017-03-20 Richard Biener <rguenther@suse.de>
29492 PR tree-optimization/80113
29493 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29494 allocate extra SSA name for PHI def.
29495 (add_close_phis_to_outer_loops): Likewise.
29496 (add_close_phis_to_merge_points): Likewise.
29497 (copy_loop_close_phi_args): Likewise.
29498 (copy_cond_phi_nodes): Likewise.
29500 2017-03-20 Martin Liska <mliska@suse.cz>
29502 PR middle-end/79753
29503 * tree-chkp.c (chkp_build_returned_bound): Do not build
29504 returned bounds for a LHS that's not a BOUNDED_P type.
29506 2017-03-20 Martin Liska <mliska@suse.cz>
29510 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29511 COMPLEX_CST and VECTOR_CST.
29513 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29516 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29517 target operand. A new splitter adds the clobber statement in case
29518 the target operand is dead anyway.
29520 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29522 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29523 to age-old versions of binutils and glibc.
29525 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29527 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29529 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29531 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29533 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29535 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29536 requirement for binutils 2.13.
29538 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29540 * combine.c (try_combine): Delete redundant i1 test. Call
29541 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29543 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29545 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29546 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29548 <riscv64-*-elf>: Re-arrange section
29549 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29550 <riscv32-*-linux>: Likewise.
29551 <riscv64-*-elf>: Likewise
29552 <riscv64-*-linux>: Likewise.
29554 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29557 * aarch64.opt(verbose-cost-dump): Fix typo.
29559 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29562 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29563 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29565 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29567 * reload.c (find_reloads): When reloading a nonoffsettable address,
29568 use RELOAD_OTHER for it and its address reloads.
29570 PR rtl-optimization/79910
29571 * combine.c (record_used_regs): New static function.
29572 (try_combine): Handle situations where there is an additional
29573 instruction between I2 and I3 which needs to have a LOG_LINK
29576 2017-03-17 Jeff Law <law@redhat.com>
29578 PR tree-optimization/71437
29579 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29580 conditional in the hash table first.
29581 (vrp_dom_walker::before_dom_children): Extract condition from
29582 ASSERT_EXPR. Record condition, its inverion and any implied
29583 conditions as well.
29585 2017-03-17 Marek Polacek <polacek@redhat.com>
29586 Markus Trippelsdorf <markus@trippelsdorf.de>
29588 PR tree-optimization/80079
29589 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29592 2017-03-17 Richard Biener <rguenther@suse.de>
29594 PR middle-end/80075
29595 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29596 Properly verify the LHS before the RHS possibly claims to be
29598 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29601 2017-03-17 Martin Jambor <mjambor@suse.cz>
29603 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29604 (List of -O2 options): Likewise.
29605 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29608 2017-03-17 Tom de Vries <tom@codesourcery.com>
29610 * gcov-dump.c (print_usage): Print bug_report_url.
29612 2017-03-17 Richard Biener <rguenther@suse.de>
29614 PR middle-end/80050
29615 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29616 (parser::peek): Likewise.
29618 2017-03-17 Richard Biener <rguenther@suse.de>
29620 PR tree-optimization/80048
29621 * sese.c (free_sese_info): Properly release rename_map and
29622 copied_bb_map elements.
29624 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29626 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29627 Add linked-list forward and backlinks. Insert on
29628 construction, remove on destruction.
29629 (class pass_store_merging): Add m_stores_head field.
29630 (pass_store_merging::terminate_and_process_all_chains):
29631 Iterate over m_stores_head list.
29632 (pass_store_merging::terminate_all_aliasing_chains):
29634 (pass_store_merging::execute): Check for debug stmts first.
29635 Push new chains onto the m_stores_head stack.
29637 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29640 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29641 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29642 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29644 2017-03-16 Jeff Law <law@redhat.com>
29646 PR tree-optimization/71437
29647 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29648 member function. Implementation moved into after_dom_children
29649 member function and into the threader's thread_outgoing_edges
29651 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29652 some code into new thread_outgoing_edges.
29653 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29654 definition. Simplify marker handling (do it here). Assume we always
29655 have the available expression and the const/copies tables.
29656 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29658 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29659 * tree-vrp.c (equiv_stack): No longer file scoped.
29660 (vrp_dom_walker): New class.
29661 (vrp_dom_walker::before_dom_children): New member function.
29662 (vrp_dom_walker::after_dom_children): Likewise.
29663 (identify_jump_threads): Setup domwalker. Use it rather than
29664 walking edges in a random order by hand. Simplify setup/finalization.
29665 (finalize_jump_threads): Remove.
29666 (vrp_finalize): Do not call identify_jump_threads here.
29667 (execute_vrp): Do it here instead and call thread_through_all_blocks
29670 PR tree-optimization/71437
29671 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29673 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29675 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29676 (dom_opt_dom_walker::thread_across_edge): Remove
29677 handle_dominating_asserts argument. All callers changed.
29678 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29679 changes. Remove calls to lhs_of_dominating_assert. Other
29680 uses of handle_dominating_asserts turn into unconditional code
29681 (simplify_control_stmt_condition_1): Likewise.
29682 (simplify_control_stmt_condition): Likewise.
29683 (thread_through_normal_block, thread_across_edge): Likewise.
29684 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29685 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29686 object if it is not an SSA_NAME.
29687 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29688 before calling into the VRP specific simplifiers.
29689 (identify_jump_threads): Remove handle_dominating_asserts
29692 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29695 * tree-diagnostic.c (default_tree_printer): No longer static.
29696 * tree-diagnostic.h (default_tree_printer): New prototype.
29698 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29700 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29701 Change ins into fmov.
29703 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29705 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29706 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29707 Use h_con constraint for operand 1.
29708 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29709 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29711 2017-03-15 Jeff Law <law@redhat.com>
29713 PR tree-optimization/71437
29714 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29715 (record_temporary_equivalences): Use it.
29717 PR tree-optimization/71437
29718 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29719 tree-ssa-scopedtables.
29720 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29721 (record_conditions, record_cond, vuse_eq): Likewise.
29722 (record_edge_info): Adjust to API tweak of record_conditions.
29723 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29724 (record_temporary_equivalences, optimize_stmt): Likewise.
29725 (eliminate_redundant_computations): Likewise.
29726 (record_equivalences_from_stmt): Likewise.
29727 * tree-ssa-scopedtables.c: Include options.h and params.h.
29728 (vuse_eq): New function, moved from tree-ssa-dom.c
29729 (build_and_record_new_cond): Likewise.
29730 (record_conditions): Likewise. Accept vector of conditions rather
29731 than edge_equivalence structure for first argument.
29732 for the first argument.
29733 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29734 from tree-ssa-dom.c.
29735 (avail_exprs_stack::record_cond): Likewise.
29736 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29737 from tree-ssa-dom.c.
29738 (avail_exprs_stack): Add new member functions lookup_avail_expr
29740 (record_conditions): Declare.
29742 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29745 * lra-constraints.c (process_alt_operands): Increase reject for
29746 reloading an input/output operand.
29748 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29751 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29752 insns to convert from signed/unsigned char/short to IEEE 128-bit
29754 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29756 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29759 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29760 subreg of inner mode for values already in registers.
29762 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29764 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29765 iteration reg is used after the loop.
29767 2017-03-14 Martin Sebor <msebor@redhat.com>
29769 PR tree-optimization/79800
29770 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29771 precision in negative-positive range.
29772 (format_floating): Call non-const overload with adjusted precision.
29774 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29777 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29780 2017-03-14 Martin Sebor <msebor@redhat.com>
29782 PR middle-end/80020
29783 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29784 * builtins.def (aligned_alloc): Use it.
29787 * Makefile.in (GTFILES): Add calls.c.
29788 * calls.c: Include "gt-calls.h".
29790 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29792 PR rtl-optimization/79728
29793 * regs.h (struct target_regs): New field
29794 x_contains_allocatable_regs_of_mode.
29795 (contains_allocatable_regs_of_mode): New macro.
29796 * reginfo.c (init_reg_sets_1): Initialize it, and change
29797 contains_reg_of_mode so it includes global regs as well.
29798 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29799 rather than contains_regs_of_mode.
29801 2017-03-14 Martin Liska <mliska@suse.cz>
29803 * doc/invoke.texi: Document options that can't be combined with
29804 -fcheck-pointer-bounds.
29806 2017-03-14 Martin Liska <mliska@suse.cz>
29808 PR middle-end/79831
29809 * doc/invoke.texi (-Wchkp): Document the option.
29811 2017-03-14 Martin Liska <mliska@suse.cz>
29813 * Makefile.in: Install gcov-dump.
29815 2017-03-14 Martin Liska <mliska@suse.cz>
29817 * multiple_target.c (expand_target_clones): Bail out for
29818 an invalid attribute.
29820 2017-03-14 Richard Biener <rguenther@suse.de>
29822 * alias.c (struct alias_set_entry): Pack properly.
29823 * cfgloop.h (struct loop): Likewise.
29824 * cse.c (struct set): Likewise.
29825 * ipa-utils.c (struct searchc_env): Likewise.
29826 * loop-invariant.c (struct invariant): Likewise.
29827 * lra-remat.c (struct cand): Likewise.
29828 * recog.c (struct change_t): Likewise.
29829 * rtl.h (struct address_info): Likewise.
29830 * symbol-summary.h (function_summary): Likewise.
29831 * tree-loop-distribution.c (struct partition): Likewise.
29832 * tree-object-size.c (struct object_size_info): Likewise.
29833 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29834 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29835 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29836 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29837 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29838 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29839 (struct _stmt_vec_info): Likewise.
29841 2017-03-14 Martin Liska <mliska@suse.cz>
29844 * multiple_target.c (create_dispatcher_calls): Check that
29845 a target can create a function dispatcher.
29847 2017-03-14 Martin Liska <mliska@suse.cz>
29850 * multiple_target.c (expand_target_clones): Drop local.local
29851 flag for default implementation.
29853 2017-03-14 Richard Biener <rguenther@suse.de>
29855 PR tree-optimization/80030
29856 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29858 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29860 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29861 gcc_fallthrough() instead of __attribute__((fallthrough));
29863 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29865 * doc/gcc.texi: Remove "up" link to (DIR).
29866 * doc/gccint.texi: Ditto.
29868 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29870 * doc/install.texi (Specific) <avr>: Remove reference to
29873 2017-03-13 Jeff Law <law@redhat.com>
29875 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29876 attribute rather than comments.
29878 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29879 match_scratch operand is highest.
29881 2017-03-13 Martin Liska <mliska@suse.cz>
29883 PR middle-end/78339
29884 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29885 is a CHKP clone, use original declaration.
29887 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29889 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29890 (arc_conditional_register_usage): Use a different allocation order
29891 when optimizing for size.
29892 * common/config/arc/arc-common.c (arc_option_optimization_table):
29893 Section anchors default on when optimizing for size.
29895 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29897 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29899 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29901 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29902 * config/arc/arc.md (cpu_facility): Add cd variant.
29903 (*movqi_insn): Add code density variant.
29904 (*movhi_insn): Likewise.
29905 (*movqi_insn): Likewise.
29906 (*addsi3_mixed): Likewise.
29907 (subsi3_insn): Likewise.
29909 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29911 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29913 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29915 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29916 expressions with MINUS and UNARY ops.
29918 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29921 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29923 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29924 between vec_select and vector argument.
29925 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29926 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29927 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29928 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29929 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29930 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29932 2017-03-13 Richard Biener <rguenther@suse.de>
29935 * params.def (vect-max-peeling-for-alignment): Fix typo.
29937 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29939 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29940 issue that only occurred with binutils below 2.18.
29942 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29944 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29945 refer to binutils 2.11/2.12 minimum.
29947 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29949 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29950 ftp.kernel.org and simplify binutils requirement.
29952 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29954 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29956 (Optimize Options): Ditto. Also remove redundancy.
29958 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29960 PR translation/79848
29961 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29963 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29964 to G_ to avoid double translation.
29966 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29968 PR translation/79923
29969 * auto-profile.c (get_combined_location): Convert leading
29970 character of diagnostics to lower case and remove trailing period.
29971 (read_profile): Likewise for various diagnostics.
29972 * config/arm/arm.c (arm_option_override): Remove trailing period
29973 from various diagnostics.
29974 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29975 (msp430_expand_delay_cycles): Likewise.
29977 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29980 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29981 full command-line argument, rather than just "str".
29982 (aarch64_validate_march): Likewise.
29983 (aarch64_validate_mtune): Likewise.
29985 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29987 PR rtl-optimization/78911
29988 * lra-assigns.c (must_not_spill_p): New function.
29989 (spill_for): Use it.
29991 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29993 PR tree-optimization/79981
29994 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29995 ATOMIC_COMPARE_EXCHANGE ifn result.
29996 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29997 IFN_ATOMIC_COMPARE_EXCHANGE.
29999 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30002 * opts.c (parse_sanitizer_options): Add missing question mark to
30003 "did you mean" message.
30005 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30007 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30009 (VMULEUH_UNS): Likewise.
30010 (VMULOUB_UNS): Likewise.
30011 (VMULOUH_UNS): Likewise.
30012 * config/rs6000/rs6000.c (builtin_function_type): Remove
30013 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30015 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30018 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30019 x with result of extra_parsing_for_operand_code_0.
30020 (function_reader::extra_parsing_for_operand_code_0): Convert
30021 return type from void to rtx, returning x. When reading
30022 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30023 larger size containing struct block_symbol.
30025 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30027 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30028 -mfloat128-hardware without -m64.
30030 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30033 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30034 entries to the case statement that marks unsigned arguments to
30035 overloaded functions.
30037 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30040 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30042 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30045 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30046 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30048 2017-03-10 Martin Liska <mliska@suse.cz>
30052 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30053 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30056 2017-03-10 Olivier Hainque <hainque@adacore.com>
30058 * tree-switch-conversion (array_value_type): Start by resetting
30059 candidate type to it's main variant.
30061 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30063 PR rtl-optimization/79909
30064 * combine.c (try_combine): Use simplify_replace_rtx on individual
30065 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30066 of the whole CALL_INSN_FUNCTION_USAGE.
30068 PR tree-optimization/79972
30069 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30070 get_range_info on SSA_NAMEs. Formatting fixes.
30072 2017-03-10 Richard Biener <rguenther@suse.de>
30073 Jakub Jelinek <jakub@redhat.com>
30075 PR tree-optimization/77975
30076 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30077 edge to be constant.
30078 (get_val_for): For constant x return it. Formatting fix.
30079 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30080 would use the same bases as the current one.
30082 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30084 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30085 instead of vec_select for V1TImode.
30086 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30088 (VSX_LE_128): Add V1TI to this mode iterator.
30089 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30090 (*vsx_le_perm_store_<mode>): Likewise.
30091 (pre-reload splitter for VSX stores): Likewise.
30092 (post-reload splitter for VSX stores): Likewise.
30093 (*vsx_xxpermdi2_le_<mode>): Likewise.
30094 (*vsx_lxvd2x2_le_<mode>): Likewise.
30095 (*vsx_stxvd2x2_le_<mode>): Likewise.
30097 2017-03-09 Michael Eager <eager@eagercon.com>
30099 Correct failures with --enable-checking=yes,rtl.
30101 * config/microblaze/microblaze.c (microblaze_expand_shift):
30102 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30103 test for const0_rtx.
30104 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30105 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30107 2017-03-09 Richard Biener <rguenther@suse.de>
30109 PR tree-optimization/79977
30110 * graphite-scop-detection.c (scop_detection::merge_sese):
30111 Handle the case of extra exits to blocks dominating the entry.
30113 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30115 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30118 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30120 PR rtl-optimization/79949
30121 * lra-constraints.c (process_alt_operands): Check memory when
30122 trying to predict a cycle. Print about the overall increase.
30124 2017-03-09 Richard Biener <rguenther@suse.de>
30126 PR middle-end/79971
30127 * gimple-expr.c (useless_type_conversion_p): Preserve
30128 TYPE_SATURATING for fixed-point types.
30130 2017-03-09 Richard Biener <rguenther@suse.de>
30133 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30134 alignment of BLKmode params.
30136 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30139 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30140 (VALL_NO_V2Q): Likewise.
30142 * config/aarch64/aarch64-simd.md
30143 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30145 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30146 VALL_NO_V2Q mode iterator.
30147 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30149 2017-03-09 Martin Liska <mliska@suse.cz>
30151 PR tree-optimization/79631
30152 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30153 tree_int_cst_sign_bit just for INTEGER constants.
30155 2017-03-09 Martin Liska <mliska@suse.cz>
30159 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30162 2017-03-09 Marek Polacek <polacek@redhat.com>
30165 * tree.c (inchash::add_expr): Handle TREE_VEC.
30167 2017-03-09 Martin Liska <mliska@suse.cz>
30170 (chkp_narrow_size_and_offset): New function.
30171 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30172 (void chkp_parse_bit_field_ref): New function.
30173 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30174 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30176 2017-03-09 Martin Liska <mliska@suse.cz>
30179 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30180 (chkp_find_bounds_1): Remove gcc_unreachable.
30182 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30185 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30186 BUILT_IN_SYNC*, determine the access type from the size suffix and
30187 always build a MEM_REF with that type. Handle forgotten
30188 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30191 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30192 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30193 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30194 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30195 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30196 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30197 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30198 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30199 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30200 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30201 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30202 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30203 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30204 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30205 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30206 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30207 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30208 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30209 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30210 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30211 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30212 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30213 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30214 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30215 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30216 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30217 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30218 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30219 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30220 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30221 definitions outside of __OPTIMIZE__ guarded section.
30224 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30225 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30226 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30227 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30230 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30232 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30233 ("vfenez<mode>"): Add missing constraints.
30235 2017-03-08 Martin Sebor <msebor@redhat.com>
30238 * config/nds32/nds32.c (nds32_option_override):
30239 Fix misspelled diagnostic.
30241 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30244 * gimplify.c (gimplify_omp_for): Replace index var in outer
30245 taskloop statement with an artificial variable and add
30246 OMP_CLAUSE_PRIVATE clause for it.
30248 2017-03-08 Richard Biener <rguenther@suse.de>
30250 PR tree-optimization/79955
30251 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30252 for accesses that are completely outside of the variable.
30254 2017-03-08 Andrew Haley <aph@redhat.com>
30256 PR tree-optimization/79943
30257 * tree-ssa-loop-split.c (compute_new_first_bound): When
30258 calculating the new upper bound, (END-BEG) should be added, not
30261 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30263 * config/avr/avr.md (setmemhi): Make sure match_dup
30264 operand number comes before match_scratch.
30266 2017-03-08 Richard Biener <rguenther@suse.de>
30268 PR tree-optimization/79920
30269 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30270 with ncopies == 1 to ...
30271 (vect_transform_slp_perm_load): ... here. Properly compute
30272 all element loads by iterating VF times over the group. Do
30273 not handle ncopies (computed in a broken way) in
30274 vect_create_mask_and_perm.
30276 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30279 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30280 is a uniform vector, use uniform_vector_p return value instead of
30281 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30283 2017-03-07 Marek Polacek <polacek@redhat.com>
30285 PR middle-end/79809
30286 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30287 (alloca_call_type): Likewise.
30289 2017-03-07 Martin Liska <mliska@suse.cz>
30291 * gcov.c (process_args): Put comment to correct location.
30293 2017-03-07 Martin Liska <mliska@suse.cz>
30295 PR middle-end/68270
30296 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30297 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30298 (chkp_narrow_bounds_for_field): Likewise.
30299 (chkp_parse_array_and_component_ref): Pass one more argument to
30302 2017-03-07 Richard Biener <rguenther@suse.de>
30304 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30307 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30309 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30310 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30312 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30315 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30316 to end of description.
30317 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30319 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30321 PR rtl-optimization/79901
30322 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30324 (*avx512f_<code><mode>3<mask_name>): ... this.
30325 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30326 iterator instead of VI8_AVX2_AVX512BW.
30328 PR rtl-optimization/79901
30329 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30330 min/max expander, expand it using expand_vec_cond_expr.
30333 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30336 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30339 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30340 to void * for PCH reasons.
30341 * dwarf2out.c (output_loc_operands, output_die): Cast
30342 v.val_vec.array to unsigned char *.
30344 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30347 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30350 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30352 PR rtl-optimization/79571
30353 * lra-constraints.c (process_alt_operands): Calculate static
30354 reject and subtract it from overall when only addresses will be
30357 2017-03-06 Julia Koval <julia.koval@intel.com>
30360 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30361 incoming stack boundary to 128 for 64-bit targets.
30363 2017-03-06 Richard Biener <rguenther@suse.de>
30365 PR tree-optimization/79894
30366 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30367 to NULL after folding it.
30369 2017-03-06 Richard Biener <rguenther@suse.de>
30371 PR tree-optimization/79824
30372 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30373 check disabling peeling for gaps.
30375 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30377 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30378 attributes): Document gettimeofday.
30380 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30382 * config/s390/s390.c (s390_option_override_internal): Set
30383 PARAM_MIN_VECT_LOOP_BOUND
30385 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30387 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30388 * config/s390/s390.md: Likewise.
30390 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30393 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30394 (<avx2_avx512>_perm<mode>): Rename to ...
30395 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30397 (<avx512>_perm<mode>_mask): Rename to ...
30398 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30400 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30401 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30402 instead of VI8F_256_512.
30403 (avx512f_perm<mode>): New define_expand.
30404 (avx512f_perm<mode>_mask): Likewise.
30405 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30406 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30408 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30410 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30411 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30413 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30415 2017-03-06 Martin Liska <mliska@suse.cz>
30418 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30419 when having a SSA NAME w/o VAR_DECL assigned to it.
30421 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30423 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30424 msa_dpsub_<su>_d): Fix MODE for vec_select.
30426 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30428 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30430 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30432 2017-03-06 Richard Biener <rguenther@suse.de>
30434 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30435 * plugin.c (register_plugin_info): Likewise.
30436 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30438 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30440 * config/i386/sse.md (sse_storehps, sse_storelps,
30441 avx_<castmode><avxsizesuffix>_<castmode>,
30442 avx512f_<castmode><avxsizesuffix>_<castmode>,
30443 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30444 in condition that at least one operand is not a MEM.
30446 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30448 PR middle-end/79805
30449 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30450 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30452 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30453 gimple_call_nothrow_p flag based on whether original builtin can throw.
30454 If it can, emit following stmts on the fallthrough edge.
30455 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30456 don't create new bb if inserting just debug stmts on the edge, try to
30457 insert them on the fallthru bb or just reset debug stmts.
30459 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30462 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30463 restore recog_data (including the operand rtxes inside it) around
30464 the call to get_insn_template.
30466 2017-03-03 Martin Sebor <msebor@redhat.com>
30468 PR tree-optimization/79699
30469 * context.c (context::~context): Free MPFR caches to avoid
30470 a memory leak on program exit.
30472 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30474 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30475 Use wide_int::ulow () instead of .elt (0).
30477 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30479 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30480 (*pushxf): Limit oF constraint to 32bit targets and add oC
30481 constraint for 64bit targets.
30482 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30483 (*pushdf): Change rmF constraint to rmC.
30485 2017-03-03 Martin Liska <mliska@suse.cz>
30487 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30488 Remove unused variable.
30490 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30493 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30494 is a memory operand, increase num_memory.
30495 (ix86_expand_args_builtin): Likewise.
30497 2017-03-03 Jan Hubicka <jh@suse.cz>
30500 * ipa-devirt.c (maybe_record_node): Properly handle
30501 __cxa_pure_virtual visibility.
30503 2017-03-03 Martin Liska <mliska@suse.cz>
30505 PR tree-optimization/79803
30506 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30508 (pass_loop_prefetch::execute): Disabled optimization if an
30509 assumption about L1 cache size is not met.
30511 2017-03-03 Martin Liska <mliska@suse.cz>
30513 PR rtl-optimization/79574
30514 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30515 (hash_scan_set): Likewise.
30516 (dump_hash_table): Likewise.
30517 (hoist_code): Likewise.
30519 2017-03-03 Richard Biener <rguenther@suse.de>
30521 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30522 in place of uhigh ().
30523 (fixed_convert_from_real): Likewise.
30525 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30528 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30530 2017-03-03 Richard Biener <rguenther@suse.de>
30532 PR middle-end/79818
30533 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30534 TYPE_OVERFLOW_UNDEFINED check.
30536 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30538 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30540 (vector_ae_<mode>_p): Likewise.
30541 (vector_nez_<mode>_p): Likewise.
30542 (vector_ne_v2di_p): Likewise.
30543 (vector_ae_v2di_p): Likewise.
30544 (vector_ne_<mode>_p): Likewise.
30545 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30547 (vsx_tsqrt<mode>2_fe): Likewise.
30549 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30552 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30554 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30556 PR rtl-optimization/79780
30557 * cprop.c (one_cprop_pass): When second and further conditional trap
30558 in a single basic block is turned into an unconditional trap, turn it
30559 into a deleted note to avoid RTL verification failures.
30561 2017-03-02 Richard Biener <rguenther@suse.de>
30563 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30565 2017-03-02 Richard Biener <rguenther@suse.de>
30567 PR tree-optimization/79345
30569 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30570 param and abort the walk, returning -1 if it is hit.
30571 (walk_aliased_vdefs): Take a limit param and pass it on.
30572 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30573 defaulting to 0 and return a signed int.
30574 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30575 (check_defs): New helper.
30576 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30577 about uninitialized memory.
30578 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30579 bogus uninitialized warning.
30580 (fixed_convert_from_real): Likewise.
30582 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30584 PR tree-optimization/66768
30585 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30586 iv_use if base object can't be determined.
30588 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30590 PR tree-optimization/79345
30591 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30592 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30593 (get_pattern_stats): Initialize it.
30594 * genemit.c (gen_expand): Verify match_scratch numbers come after
30595 match_operand/match_dup numbers.
30596 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30597 match_scratch numbers.
30598 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30600 * config/s390/s390.md (trunctdsd2): Likewise.
30602 2017-03-02 Richard Biener <rguenther@suse.de>
30604 * wide-int.h (wide_int_storage::operator=): Implement in terms
30607 2017-03-02 Richard Biener <rguenther@suse.de>
30609 PR tree-optimization/79777
30610 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30611 the to insert expression to sth existing.
30613 2017-03-01 Martin Sebor <msebor@redhat.com>
30615 PR middle-end/79692
30616 * gimple-ssa-sprintf.c
30617 (directive::known_width_and_precision): New function.
30618 (format_integer): Use it.
30619 (get_mpfr_format_length): Consider the full range of precision
30620 when computing %g output with the # flag. Set the likely byte
30621 count to 3 rather than 1 when precision is indeterminate.
30622 (format_floating): Correct the lower bound of precision.
30624 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30626 * doc/invoke.texi: Document default code model for 64-bit Linux.
30628 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30631 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30632 udiv rather than div since input pattern is unsigned.
30634 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30636 * config/i386/i386.c (print_reg): Warn for values of
30637 unsupported size in integer register.
30639 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30642 * config/rs6000/predicates.md (current_file_function_operand): Do
30643 not allow self calls to be local if the function is replaceable.
30645 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30648 * config/rs6000/altivec.h (vec_ctz and others): Change the
30649 preprocessor macro that controls conditional compilation from
30650 _ARCH_PWR9 to __POWER9_VECTOR__.
30651 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30652 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30653 control (instead of _ARCH_PWR9 control) so that template
30654 definition uses power9-specific function.
30655 (vec_any_eq): Likewise.
30656 (vec_all_ne): Change macro definition to use a power9-specific
30657 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30658 _ARCH_PWR9 control).
30659 (vec_any_eq) Likewise.
30660 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30661 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30662 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30663 support for xvcmpnedp instruction.
30664 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30665 macro expansion so that Power9 implementation of vec_all_ne does
30666 not use the AltiVec predicate framework.
30667 (VCMPNEH_P): Likewise.
30668 (VCMPNEW_P): Likewise.
30669 (VCMPNED_P): Likewise.
30670 (VCMPNEFP_P): Likewise.
30671 (VCMPNEDP_P): Likewise.
30672 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30673 implementation of vec_any_eq to not use AltiVec predicate
30675 (VCMPAEH_P): Likewise.
30676 (VCMPAEW_P): Likewise.
30677 (VCMPAED_P): Likewise.
30678 (VCMPAEFP_P): Likewise.
30679 (VCMPAEDP_P): Likewise.
30680 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30681 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30682 not use the AltiVec predicate framework.
30683 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30684 of vec_any_eq to not use AltiVec predicate framework.
30685 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30686 support for predefined __POWER9_VECTOR__ macro to indicate that
30687 Power9 instruction selection is enabled.
30688 (altivec_overloaded_builtins): Remove extraneous
30689 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30690 function argument types RS6000_BTI_bool_V16QI and
30691 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30692 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30693 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30694 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30695 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30696 Power9 for implementations of vec_cmpne. Change the signature for
30697 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30698 (representing vec_all_ne) to remove the previously described first
30699 argument of type RS6000_BTI_INTSI, as this was an artifact of
30700 reliance on the AltiVec predicate framework, which is no longer
30701 used in the implementation of these functions. Add
30702 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30703 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30704 since, unlike the AltiVec predicate framework implementation, we
30705 do not share function descriptors between vec_alle and vec_anyeq.
30706 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30707 set of modes that receive special treatment even when
30708 TARGET_P9_VECTOR is true. The special treatment emits code that
30709 does not depend on Power9 instructions.
30710 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30711 define_expand to not rely on AltiVec predicate framework.
30712 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30714 (vector_ne_v2di_p): Change this define_expand to not rely on
30715 AltiVec predicate framework.
30716 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30718 (vector_ne_<mode>_p): Change this define_expand to not rely on
30719 AltiVec predicate framework.
30720 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30722 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30723 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30724 define_insn pattern.
30725 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30726 define_insn pattern because the xvcmpne<VSs>. instruction is not
30728 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30729 instruction is not supported.
30731 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30733 * config/nvptx/nvptx.c: Include intl.h.
30735 2017-03-01 Martin Jambor <mjambor@suse.cz>
30738 * ipa-prop.h (ipa_bits): Removed field known.
30739 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30740 to pointers. Adjusted their comments to warn about their sharing.
30741 (ipcp_transformation_summary): Change bits to a vector of pointers.
30742 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30743 (ipa_get_ipa_bits_for_value): Declare.
30744 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30745 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30746 (ipa_bits_hash_table): Likewise.
30747 (ipa_vr_ggc_hash_traits): Likewise.
30748 (ipa_vr_hash_table): Likewise.
30749 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30750 being pointers and vr_known being removed.
30751 (ipa_set_jf_unknown): Likewise.
30752 (ipa_get_ipa_bits_for_value): New function.
30753 (ipa_set_jfunc_bits): Likewise.
30754 (ipa_get_value_range): New overloaded functions.
30755 (ipa_set_jfunc_vr): Likewise.
30756 (ipa_compute_jump_functions_for_edge): Use the above functions to
30757 construct bits and vr parts of jump functions.
30758 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30759 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30761 (ipcp_grow_transformations_if_necessary): Also allocate
30762 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30764 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30765 them. Fix too long lines.
30766 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30767 vr_known being removed.
30768 (ipa_read_jump_function): Use new setter functions to construct bits
30769 and vr parts of jump functions or set them to NULL.
30770 (write_ipcp_transformation_info): Adjust for bits being pointers.
30771 (read_ipcp_transformation_info): Likewise.
30772 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30774 Include gt-ipa-prop.h.
30775 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30777 (ipcp_store_bits_results): Likewise.
30778 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30779 Do not write to existing jump functions but use a temporary instead.
30781 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30784 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30785 attempt to use its first operand as BIT_FIELD_REF base.
30787 2017-03-01 Richard Biener <rguenther@suse.de>
30789 PR middle-end/79721
30790 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30791 interpolating formula in wrapping arithmetic.
30792 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30794 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30796 PR tree-optimization/79734
30797 * tree-vect-generic.c (expand_vector_condition): Optimize
30798 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30799 Handle VEC_COND_EXPR where comparison has different inner width from
30800 type's inner width.
30802 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30804 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30805 markup, and similar issues. Remove @opindex entries for things
30806 that aren't options. Add missing -mmpy-option entries.
30808 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30810 PR tree-optimization/79737
30811 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30812 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30813 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30814 instead of byte_size. Formatting fix.
30815 (shift_bytes_in_array_right): Formatting fix.
30817 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30820 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30821 condition on optimize for the leaf function test.
30823 2017-02-28 Martin Liska <mliska@suse.cz>
30826 * read-rtl-function.c (function_reader::handle_unknown_directive):
30827 Bail out when one uses -flto.
30829 2017-02-28 Martin Liska <mliska@suse.cz>
30831 * common.opt: Replace space with tabular for options of <number>
30833 * config/i386/i386.opt: Show <number> value for
30834 -mlarge-data-threshold.
30835 * opts.c (print_filtered_help): Do not display number in hexadecimal
30838 2017-02-28 Martin Liska <mliska@suse.cz>
30840 * common.opt: Fix --help=option -Q for options which are of
30843 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30845 * config/i386/i386.c (print_reg): Error out for values
30846 of 8-bit size in invalid integer register.
30848 2017-02-28 Martin Sebor <msebor@redhat.com>
30850 PR tree-optimization/79691
30851 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30853 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30856 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30857 gcc_unreachable with output_operand_lossage.
30859 2017-02-28 Richard Biener <rguenther@suse.de>
30861 PR tree-optimization/79740
30862 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30864 (visit_nary_op): Insert the nary into the hashtable if we
30865 pattern-matched sth.
30866 * tree-ssa-pre.c (eliminate_insert): Robustify.
30868 2017-02-28 Richard Biener <rguenther@suse.de>
30870 PR middle-end/79731
30871 * fold-const.c (decode_field_reference): Reject out-of-bound
30874 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30876 * config/i386/i386.c: Include intl.h.
30877 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30878 instead of just cond ? "..." : "...".
30879 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30880 * coverage.c (read_counts_file): Likewise.
30881 * omp-offload.c: Include intl.h.
30882 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30883 of just cond ? "..." : "...".
30884 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30885 of just cond ? "..." : "...".
30887 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30890 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30892 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30893 'tune for' CPU name.
30894 * config/arm/arm-cpu-data.h: Regenerated.
30896 2017-02-28 Richard Biener <rguenther@suse.de>
30898 PR tree-optimization/79732
30899 * tree-inline.c (expand_call_inline): Do not shadow var.
30901 2017-02-28 Richard Biener <rguenther@suse.de>
30903 PR tree-optimization/79723
30904 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30905 address-space properly.
30907 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30909 * doc/optinfo.texi (Optimization groups): Fix option used for
30911 * doc/invoke.texi (-fopt-info): Document "omp".
30912 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30913 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30914 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30915 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30916 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30918 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30920 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30921 OPTGROUP_OMP with "omp".
30923 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30926 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30927 for arithmetic shift of unsigned V2DI.
30929 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30931 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30932 arc/linux.h headers.
30933 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30934 (LINK_SPEC): Likewise.
30935 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30936 (EXTRA_SPECS): Likewise.
30937 (STARTFILE_SPEC): Likewise.
30938 (ENDFILE_SPEC): Likewise.
30939 (LIB_SPEC): Likewise.
30940 (TARGET_SDATA_DEFAULT): Likewise.
30941 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30942 (MULTILIB_DEFAULTS): Likewise.
30943 (DWARF2_UNWIND_INFO): Likewise.
30944 * config/arc/big.h: New file.
30945 * config/arc/elf.h: Likewise.
30946 * config/arc/linux.h: Likewise.
30947 * config/arc/t-uClibc: Remove.
30949 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30951 PR tree-optimization/77536
30952 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30953 (tree_transform_and_unroll_loop): Use above function to compute the
30954 estimated niter of unrolled loop and use it when scaling profile.
30955 Also use count info rather than frequency if it's non-zero.
30956 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30957 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30958 (vect_transform_loop): Call above function.
30960 2017-02-27 Richard Biener <rguenther@suse.de>
30962 PR tree-optimization/45397
30963 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30964 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30965 (visit_nary_op): Add pattern matching for CSEing sign-changed
30966 or truncated operations with wider ones.
30968 2017-02-27 Richard Biener <rguenther@suse.de>
30970 PR tree-optimization/79690
30971 * tree-vect-stmts.c (vectorizable_store): Use vector type
30972 built from the DR with address-space.
30974 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30976 * doc/invoke.texi (Optimize Options): Refine the description
30977 of asan-use-after-return.
30979 2017-02-25 Alan Modra <amodra@gmail.com>
30981 PR rtl-optimization/79584
30982 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30983 base, not ad->base_term, the reg within base. Remove assertion
30984 that ad->base == ad->base_term. Replace gen_int_mode using
30985 bogus mode with const0_rtx.
30987 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30989 PR middle-end/79396
30990 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30991 FMA_EXPR like tcc_binary or tcc_unary.
30993 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30996 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30998 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30999 (output_loc_operands): Handle DW_OP_call_ref and
31000 DW_OP_GNU_variable_value.
31001 (struct variable_value_struct): New type.
31002 (struct variable_value_hasher): Likewise.
31003 (variable_value_hash): New variable.
31004 (string_types): Remove.
31005 (copy_loc_descr): New function.
31006 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31007 (prepend_loc_descr_to_each): New function.
31008 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31009 instead of add_loc_descr_to_each if the first argument is single
31010 location list and the second has multiple.
31011 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31012 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31013 when looking for variable value which doesn't have other location info.
31014 (loc_list_from_tree): Formatting fix.
31015 (gen_array_type_die): Simplify DW_AT_string_length handling.
31016 (adjust_string_types): Remove.
31017 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31018 string_types. Call resolve_variable_values.
31019 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31020 (resolve_addr_in_expr): Likewise. Add A argument.
31021 (copy_deref_exprloc): Remove deref argument. Adjust for the
31022 original expression being DW_OP_GNU_variable_value with optionally
31023 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31024 optionally after it.
31025 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31026 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31027 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31028 (variable_value_hasher::hash, variable_value_hasher::equal): New
31030 (resolve_variable_value_in_expr, resolve_variable_value,
31031 resolve_variable_values, note_variable_value_in_expr,
31032 note_variable_value): New functions.
31033 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31036 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31039 * opts.h (handle_generated_option): Add GENERATED_P argument.
31040 * opts-common.c (handle_option): Adjust function comment.
31041 (handle_generated_option): Add GENERATED_P argument, pass it to
31043 (control_warning_option): Pass false to handle_generated_option
31045 * opts.c (maybe_default_option): Pass true to handle_generated_option
31047 * optc-gen.awk: Likewise.
31049 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31051 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31052 a REG, look at the REG it is a SUBREG of.
31053 (splitter for cmpeqsi_t): Ditto.
31055 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31057 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31058 the special USEs with the pattern of the insn, not the insn itself.
31060 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31063 * doc/invoke.texi: Document -mload-store-pairs.
31065 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31066 Sandra Loosemore <sandra@codesourcery.com>
31068 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31069 argument isn't a CONST_INT.
31070 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31071 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31072 (nios2_validate_compare): Bypass alternate compare logic if *op2
31073 is not a CONST_INT.
31074 (ldstwm_operation_p): Return false if first_base is not a REG or
31075 if first_offset is not a CONST_INT.
31077 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31079 * config/cris/cris.md: Use correct operand in a define_peephole2.
31081 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31083 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31085 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31087 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31088 this_insn if it is an INSN or JUMP_INSN.
31089 (force_offsettable): Look at base, not at addr.
31090 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31091 on things that aren't necessarily CONST_INTs.
31093 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31095 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31096 -mfpmath=sse is the default also for x86-32 targets with SSE2
31097 instruction set when @option{-ffast-math} is enabled
31099 2017-02-24 Jeff Law <law@redhat.com>
31101 PR rtl-optimizatoin/79286
31102 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31105 2017-02-24 Richard Biener <rguenther@suse.de>
31107 PR tree-optimization/79389
31108 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31111 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31113 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31114 function comment to reflect reality.
31115 (loop_exits_before_overflow): Fix typo in function description.
31117 2017-02-24 Richard Biener <rguenther@suse.de>
31119 PR tree-optimization/79389
31120 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31121 properly that a threading opportunity exists. Detect conditional
31122 copy/constant propagation opportunities.
31124 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31126 * config/visium/visium.md (type): Add trap.
31127 (b): New mode attribute.
31128 (*btst): Rename into...
31129 (*btst<mode>): ...this and adjust.
31130 (*cbranchsi4_btst_insn): Rename into...
31131 (*cbranch<mode>4_btst_insn): ...this and adjust.
31132 (trap): New define_insn.
31134 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31136 PR tree-optimization/79389
31137 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31138 (noce_reversed_cond_code): New function.
31139 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31140 reversed_comparison_code. Formatting fix.
31141 (noce_try_store_flag): Test rev_cond != NULL in addition to
31142 reversed_comparison_code.
31143 (noce_try_store_flag_constants): Likewise.
31144 (noce_try_store_flag_mask): Likewise.
31145 (noce_try_addcc): Use rev_cond if non-NULL instead of
31146 reversed_comparison_code.
31147 (noce_try_cmove_arith): Likewise. Formatting fixes.
31148 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31149 (noce_find_if_block): Initialize rev_cond.
31150 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31151 instead of false as last argument never attempt to reverse it
31154 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31156 PR tree-optimization/79663
31157 * tree-predcom.c (combine_chains): Process refs in reverse order
31158 only for ZERO length chains, and add explaining comment.
31160 2017-02-23 Jeff Law <law@redhat.com>
31162 PR tree-optimization/79578
31163 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31164 in call to operand_equal_p.
31166 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31169 * config/i386/cpuid.h: Fix another undefined behavior.
31171 2017-02-23 Richard Biener <rguenther@suse.de>
31173 PR tree-optimization/79683
31174 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31175 vector types for data-refs.
31177 2017-02-23 Martin Liska <mliska@suse.cz>
31179 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31181 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31183 PR middle-end/79665
31184 * internal-fn.c (get_range_pos_neg): Moved to ...
31185 * tree.c (get_range_pos_neg): ... here. No longer static.
31186 * tree.h (get_range_pos_neg): New prototype.
31187 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31188 are known to be in between 0 and signed maximum inclusive, try to
31189 expand both unsigned and signed divmod and use the cheaper one from
31192 2017-02-22 Jeff Law <law@redhat.com>
31194 PR tree-optimization/79578
31195 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31196 to compare base operands.
31198 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31201 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31202 gpc_reg_operand instead of fpr_reg_operand.
31204 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31206 * config/mips/mips.c (mips_return_in_memory): Force FP
31207 vector types to be returned in memory for o32 ABI.
31209 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31211 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31212 instead of DW_TAG_member for static data member declarations and don't
31213 set no_linkage_name for static inline data members.
31214 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31217 2017-02-22 Martin Liska <mliska@suse.cz>
31219 * doc/invoke.texi: Replace inequality signs with square brackets
31222 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31225 * lra-constraints.c (simplify_operand_subreg): Handle
31226 WORD_REGISTER_OPERATIONS targets.
31228 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31231 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31232 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31233 elimination by swapping fld*.
31235 2017-02-22 Richard Biener <rguenther@suse.de>
31237 PR tree-optimization/79673
31238 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31239 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31240 irrelevant address-space qualifiers and avoiding a
31241 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31243 2017-02-22 Richard Biener <rguenther@suse.de>
31245 PR tree-optimization/79666
31246 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31247 to not symbolically negate if that may introduce undefined
31250 2017-02-22 Martin Liska <mliska@suse.cz>
31253 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31254 * data-streamer-out.c (streamer_write_gcov_count_stream):
31256 * value-prof.c (stream_out_histogram_value): Make assert more
31257 precise based on type of counter.
31259 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31262 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31263 Use nonimmediate_operand instead of memory_operand for operand 1.
31264 (float-extend standard_x87sse_constant_load splitter): Ditto.
31266 2017-02-21 Jeff Law <law@redhat.com>
31268 PR tree-optimization/79621
31269 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31270 blocks with edges to themselves.
31272 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31275 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31276 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31277 Use gimple_call_builtin_p.
31280 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31281 on temporarily removed DEBUG_INSNs.
31283 PR tree-optimization/79649
31284 * tree-loop-distribution.c (classify_partition): Give up on
31285 non-generic address space loads/stores.
31287 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31289 * doc/loop.texi (Loop manipulation): Remove nonexistent
31290 tree_ssa_loop_version from the documentation.
31291 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31293 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31296 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31297 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31298 * config/rs6000/rs6000.c: Include except.h.
31299 (rs6000_expand_split_stack_prologue): Call
31300 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31302 2017-02-21 Martin Jambor <mjambor@suse.cz>
31305 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31306 have been analyzed.
31308 2017-02-21 Martin Jambor <mjambor@suse.cz>
31310 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31311 for backward compatibility only.
31312 * doc/invoke.texi (Option Summary): Remove all references to
31313 -fipa-cp-alignment.
31315 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31319 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31321 * lra-constraints.c (curr_insn_transform): Handle
31322 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31324 2017-02-21 Martin Liska <mliska@suse.cz>
31326 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31328 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31330 PR translation/79638
31331 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31333 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31336 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31337 (arm_function_ok_for_sibcall): Return false for an indirect call by
31338 descriptor if all the argument registers are used.
31339 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31340 alignment of the function.
31342 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31344 PR tree-optimization/61441
31345 * simplify-rtx.c (simplify_const_unary_operation): For
31346 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31347 the sNaN unmodified.
31349 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31351 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31352 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31353 instead of SYSTEM_HEADER_DIR.
31355 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31356 Martin Liška <mliska@suse.cz>
31358 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31359 Fix typos and grammar, use active voice, and clarify.
31361 2017-02-20 Marek Polacek <polacek@redhat.com>
31363 PR middle-end/79537
31364 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31367 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31369 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31372 * config/i386/i386.c (ix86_expand_builtin): Handle
31373 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31374 ix86_builtins_isa[fcode].isa as a requirement of those
31375 flags and any other flag in the bitmask.
31376 (ix86_init_mmx_sse_builtins): Use 0 instead of
31377 ~OPTION_MASK_ISA_64BIT as mask.
31378 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31379 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31380 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31381 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31383 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31386 * lra-constraints.c (split_reg): Check requested split mode
31387 is supported by the register.
31389 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31391 * lra-constraints.c (simplify_operand_subreg): Remove early
31394 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31397 * lra-constraints.c (curr_insn_transform): Tighten condition
31398 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31400 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31403 * lra-constraints.c (curr_insn_transform): Handle
31404 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31406 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31409 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31411 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31413 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31416 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31419 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31421 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31424 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31426 * final.c (last_columnnum, override_columnnum): New variables.
31427 (final_start_function): Set last_columnnum, pass it to begin_prologue
31428 hook and pass 0 to dwarf2out_begin_prologue.
31429 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31430 to source_line debug hook.
31431 (notice_source_line): Compute last_columnnum and for debug_column_info
31432 return true on column changes.
31433 * debug.h (struct gcc_debug_hooks): Add column argument to
31434 source_line and begin_prologue hooks.
31435 (debug_nothing_int_charstar_int_bool): Remove prototype.
31436 (debug_nothing_int_int_charstar,
31437 debug_nothing_int_int_charstar_int_bool): New prototypes.
31438 (dwarf2out_begin_prologue): Add column argument.
31439 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31440 begin_prologue hooks.
31441 (debug_nothing_int_charstar_int_bool): Remove.
31442 (debug_nothing_int_int_charstar,
31443 debug_nothing_int_int_charstar_int_bool): New functions.
31444 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31445 through to dwarf2out_source_line.
31446 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31447 (dwarf2out_source_line): Add column argument, emit it if requested.
31448 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31450 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31451 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31452 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31453 through to dwarf2out_begin_prologue.
31454 (vmsdbgout_source_line): Add column argument, pass it through to
31455 dwarf2out_source_line.
31456 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31457 dbxout_source_line caller.
31458 (dbxout_source_line): Add column argument.
31460 * common.opt (gno-column-info, gcolumn-info): New options.
31461 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31462 (check_die): Also test for multiple DW_AT_decl_column attributes.
31463 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31464 DW_AT_decl_column if requested.
31465 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31467 (gen_variable_die): Likewise.
31468 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31469 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31472 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31473 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31474 (ix86_handle_option): Handle OPT_m3dnowa.
31475 * doc/invoke.texi (-m3dnowa): Document.
31476 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31477 -m3dnowa instead of -m3dnow -march=athlon.
31480 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31481 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31483 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31486 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31487 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31488 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31489 generator for vsx_xxpermdi_<mode>_be.
31490 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31491 force big-endian semantics.
31492 (vsx_xxpermdi_<mode>_be): New define_expand with same
31493 implementation as previous version of vsx_xxpermdi_<mode>.
31495 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31497 PR tree-optimization/79327
31498 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31499 variable, its initialization and use.
31501 2017-02-17 Julia Koval <julia.koval@intel.com>
31503 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31504 (OPTION_MASK_ISA_PKU_UNSET): New.
31505 (ix86_handle_option): Handle -mrdpid.
31506 * config/i386/cpuid.h (bit_RDPID): New.
31507 * config/i386/driver-i386.c (host_detect_local_cpu):
31508 Detect RDPID feature.
31509 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31510 * config/i386/i386-c.c (ix86_target_macros_internal):
31512 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31513 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31514 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31515 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31516 * config/i386/i386.md (define_insn "rdpid"): New.
31517 * config/i386/i386.opt Add -mrdpid.
31518 * config/i386/immintrin.h (_rdpid_u32): New.
31520 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31522 PR rtl-optimization/79541
31523 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31524 instead of transforming it into USE.
31526 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31528 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31529 If HONOR_SNANS (SFmode) force the input to a register.
31530 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31531 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31532 an frsp or similar insn.
31534 2017-02-17 Martin Liska <mliska@suse.cz>
31536 PR rtl-optimization/79577
31537 * params.def (selsched-max-sched-times): Increase minimum to 1.
31539 2017-02-17 Martin Liska <mliska@suse.cz>
31541 PR rtl-optimization/79574
31542 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31544 2017-02-17 Martin Liska <mliska@suse.cz>
31546 PR tree-optimization/79529
31547 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31548 ssa_defined_default_def_p to handle cases which are implicitly
31550 * tree-ssa.c (ssa_defined_default_def_p): New function.
31551 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31552 which are implicitly defined.
31553 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31555 2017-02-17 Richard Biener <rguenther@suse.de>
31557 PR middle-end/79576
31558 * params.def (max-ssa-name-query-depth): Limit to 10.
31560 2017-02-17 Richard Biener <rguenther@suse.de>
31562 PR tree-optimization/79552
31563 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31566 2017-02-17 Richard Biener <rguenther@suse.de>
31569 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31571 2017-02-17 Marek Polacek <polacek@redhat.com>
31573 PR middle-end/79536
31574 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31575 (fold_negate_expr): New wrapper.
31577 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31579 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31580 Correct terminology and de-emphasize pre-standard behavior.
31582 2017-02-16 Alan Modra <amodra@gmail.com>
31584 PR rtl-optimization/79286
31585 * ira.c (def_dominates_uses): New function.
31586 (update_equiv_regs): Don't create an equivalence for insns that
31587 may trap where the register def does not dominate the use.
31589 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31591 PR rtl-optimization/78127
31592 * lra.c (lra): Call lra_eliminate before finish the loop after
31595 2017-02-16 Richard Biener <rguenther@suse.de>
31597 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31599 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31600 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31601 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31602 (isl_val_int_from_wi): New function.
31603 (extract_affine_gmp): Rename to ...
31604 (extract_affine_wi): ... this, take a widest_int.
31605 (extract_affine_int): Just wrap extract_affine_wi.
31606 (add_param_constraints): Use isl_val_int_from_wi.
31607 (add_loop_constraints): Likewise, and extract_affine_wi.
31609 2017-02-15 Jeff Law <law@redhat.com>
31611 PR middle-end/79521
31612 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31613 ira_init_register_move_cost_if_necessary.
31615 2017-02-15 Martin Sebor <msebor@redhat.com>
31617 PR middle-end/32003
31618 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31619 removed in a prior commit.
31621 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31623 PR tree-optimization/79347
31624 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31625 counters during peeling.
31627 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31629 * Makefile.in (site.exp): Remove "set ISLVER".
31631 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31634 * real.c (real_from_integer): Call real_convert even for decimal.
31636 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31639 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31641 2017-02-14 Andrew Pinski <apinski@cavium.com>
31643 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31644 cores and change the partno/implementer to be correct.
31645 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31646 the 'B" as the implementer.
31647 * config/aarch64/aarch64-tune.md: Regenerate.
31649 2017-02-14 Carl Love <cel@us.ibm.com>
31651 * config/rs6000/rs6000.c: Add case statement entry to make the
31652 xvcvuxdsp built-in argument unsigned.
31653 * config/rs6000/vsx.md: Fix the source and return operand types so they
31654 match the instruction definitions from the ISA document. Fix typo
31655 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31658 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31661 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31662 member early_clobber_alts.
31663 * lra-lives.c (reg_early_clobber_p): New.
31664 (process_bb_lives): Use it.
31665 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31666 (debug_operand_data): Initialize early_clobber_alts.
31667 (setup_operand_alternative): Set up early_clobber_alts.
31668 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31669 alternatives to new_insn_reg.
31670 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31672 (lra_update_insn_regno_info): Pass the new arg.
31674 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31676 PR middle-end/79505
31677 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31678 (new_oacc_loop_raw): Don't clear already cleared fields.
31681 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31682 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31683 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31685 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31688 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31690 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31693 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31694 the extra instruction to the right place to store 128-bit constant
31697 2017-02-14 Martin Sebor <msebor@redhat.com>
31699 PR middle-end/79448
31700 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31701 warning for strings of unknown length.
31703 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31705 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31707 2017-02-14 Jeff Law <law@redhat.com>
31710 * ira-costs.c (scan_one_insn): Initialize register move costs
31711 for pseudos seen in USE/CLOBBER insns.
31713 PR tree-optimization/79095
31714 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31715 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31716 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31717 if the operands are known to be not equal, then the resulting range
31719 (intersect_ranges): If the new range is ~[0,0] and the old range is
31720 wide, then prefer ~[0,0].
31721 * tree-vrp.c (overflow_comparison_p_1): New function.
31722 (overflow_comparison_p): New function.
31723 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31724 if NAME is used in an overflow test.
31725 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31726 overflow check that can be expressed as an equality test, then adjust
31727 ops to be that equality test.
31729 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31731 * config/s390/s390-builtin-types.def: Remove flags argument.
31732 * config/s390/s390.c (s390_init_builtins): Likewise.
31734 2017-02-14 Martin Liska <mliska@suse.cz>
31736 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31737 vector. Fix trailing white spaces.
31739 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31741 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31744 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31746 PR rtl-optimization/68664
31747 * config/arm/arm.c (arm_sched_can_speculate_insn):
31748 New function. Declare prototype.
31749 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31751 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31753 PR rtl-optimization/68664
31754 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31756 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31758 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31760 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31761 max skip bytes for function, loop and jump.
31763 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31765 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31766 ABS_EXPR for gimple dump.
31768 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31771 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31773 PR tree-optimization/79408
31774 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31775 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31776 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31777 also if rhs1 is INTEGER_CST.
31779 2017-02-14 Richard Biener <rguenther@suse.de>
31781 PR middle-end/79432
31782 * tree-into-ssa.c (insert_phi_nodes): When the function can
31783 have abnormal edges rewrite SSA names with broken use-def
31784 dominance out of SSA and register them for PHI insertion.
31786 2017-02-13 Martin Sebor <msebor@redhat.com>
31788 PR middle-end/79496
31789 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31790 clearing info.nowrite flag when snprintf size argument is a range.
31792 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31794 * cprop.c (cprop_jump): Add missing space in string literal.
31795 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31796 (get_constraint_for_component_ref): Likewise.
31797 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31798 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31799 * lra-constraints.c (process_alt_operands): Likewise.
31800 * ipa-inline.c (inline_small_functions): Likewise.
31801 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31802 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31803 * trans-mem.c (diagnose_tm_1_op): Likewise.
31804 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31805 (grid_parallel_clauses_gridifiable): Likewise.
31807 * config/nvptx/mkoffload.c (process): Add space in between
31810 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31811 "MOD4_SSE_REGS" and "ALL_REGS".
31813 * spellcheck.c (test_data): Add , in between "foo" and "food".
31815 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31818 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31819 boundary crossing check and subsequent code generation agree.
31821 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31823 * config/aarch64/aarch64.c (has_memory_op): Delete.
31824 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31827 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31829 PR rtl-optimization/79388
31830 PR rtl-optimization/79450
31831 * combine.c (distribute_notes): When removing TEM_INSN for which
31832 corresponding dest has last value recorded, invalidate that last
31835 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31837 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31838 of explicit '@'. Add missing assembly comment marker on branch costs
31841 2017-02-13 Nathan Sidwell <nathan@acm.org>
31843 * gengtype-lex.l (<in_struct>): Add '/'.
31845 2017-02-13 Martin Liska <mliska@suse.cz>
31848 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31850 2017-02-13 Richard Biener <rguenther@suse.de>
31852 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31854 * configure: Re-generate.
31855 * config.in: Likewise.
31856 * graphite-dependences.c: Simplify as if
31857 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31858 * graphite-isl-ast-to-gimple.c: Likewise.
31859 * graphite-optimize-isl.c: Likewise.
31860 * graphite-poly.c: Likewise.
31861 * graphite-sese-to-poly.c: Likewise.
31862 * graphite.h: Likewise.
31863 * toplev.c: Include isl/version.h and use isl_version () for
31864 printing the ISL version.
31865 * doc/install.texi: Update ISL requirement.
31867 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31869 * doc/standards.texi (Standards): Update reference to
31872 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31874 * doc/extend.texi (Named Address Spaces): sourceware.org now
31876 * doc/install.texi (Binaries): Ditto.
31879 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31881 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31882 standard terminology "stringize"/"stringizing" throughout.
31883 * doc/cppinternals.texi: Likewise.
31885 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31887 * doc/extend.texi: Fix some spelling mistakes and typos.
31888 * doc/invoke.texi: Likewise.
31890 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31893 * params.def (inline-min-speedup) Change from 10 to 8.
31895 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31897 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31900 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31903 * ipa-inline-analysis.c (get_minimal_bb): New function.
31904 (record_modified): Use it.
31905 (remap_edge_change_prob): Handle also ancestor functions.
31907 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31909 * doc/contrib.texi (Contributors): Remove broken link into
31910 the Mauve CVS repository.
31912 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31914 PR middle-end/79454
31915 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31916 result computation whenever lhs doesn't have vector mode, not
31917 just when it has BLKmode.
31919 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31921 * doc/makefile.texi (profiledbootstrap): Refer to the
31922 installation instructions only in textual form.
31924 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31927 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31929 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31931 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31932 (Specific): Update mingw-w64 reference.
31934 (Specific): Remove broken link to Renesas RX processor.
31936 2017-02-10 Richard Biener <rguenther@suse.de>
31938 * toplev.c (process_options): Do not mention obsolete graphite
31939 options when printing sorry message about missing graphite support.
31940 Mention -floop-nest-optimize.
31942 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31944 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31945 (vtst_p16): Likewise.
31946 (vtstq_p8): Likewise.
31947 (vtstq_p16): Likewise.
31949 (vtstq_p64): Likewise.
31950 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31951 (vset_lane_p64): New.
31952 (vsetq_lane_p64): New.
31954 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31956 PR tree-optimization/79411
31957 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31958 stmt operands are SSA_NAMEs used in abnormal phis.
31959 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31962 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31965 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31968 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31970 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31972 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31974 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31978 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31979 not arbitrary TREE_CONSTANT.
31982 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31983 "omp declare target link" attribute unless is_global_var.
31984 * omp-offload.c (find_link_var_op): Likewise.
31986 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31987 Chung-Lin Tang <cltang@codesourcery.com>
31989 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31991 (gimplify_adjust_omp_clauses): Don't delete TILE.
31992 (gimplify_omp_for): Deal with TILE.
31993 * internal-fn.c (expand_GOACC_TILE): New function.
31994 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31996 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31997 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31999 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32000 avoid DIV for outermost collapse var.
32001 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32002 Remove out of date comments, fix whitespace.
32003 * omp-general.c (omp_extract_for_data): Deal with tiling.
32004 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32005 adjust OLF_DIM_BASE value.
32006 (struct omp_for_data): Add tiling field.
32007 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32008 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32009 for auto loops. Remove default auto determining, moved to
32010 oacc_loop_fixed_partitions.
32011 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32012 stmts, add e_mask field.
32013 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32014 (oacc_thread_numbers): Use oacc_dim_call.
32015 (oacc_xform_tile): New.
32016 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32017 (finish_oacc_loop): Adjust for ifns vector.
32018 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32019 add case for GOACC_TILE fns.
32020 (oacc_loop_xform_loop): Delete.
32021 (oacc_loop_process): Iterate over call list directly, and add
32022 handling for GOACC_TILE fns.
32023 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32025 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32026 vector partitioning to outer loops. Assign 2 partitions to loops
32027 when available. Add TILE handling.
32028 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32029 (execite_oacc_device_lower): Process GOACC_TILE fns,
32030 ignore unknown specs.
32031 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32032 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32033 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32035 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32037 * configure.ac (ACX_BUGURL): Update.
32038 * configure: Regenerate.
32040 2017-02-09 Richard Biener <rguenther@suse.de>
32042 PR tree-optimization/69823
32043 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32044 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32046 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32048 * config/arc/arc-c.def: Add __NPS400__ definition.
32049 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32050 (TARGET_NPS400): Define.
32052 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32054 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32056 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32057 pointer, arch_info.
32058 (arc_cpu_types): Fill the arch_info field with a pointer into the
32059 arc_arch_types table.
32060 (arc_selected_cpu): Declare.
32061 * config/arc/arc.c (arc_selected_cpu): Make global.
32062 (arc_selected_arch): Delete.
32063 (arc_base_cpu): Delete.
32064 (arc_override_options): Remove references to deleted variables,
32065 update access to arch information.
32066 (ARC_OPT): Update access to arch information.
32067 (ARC_OPTX): Likewise.
32068 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32069 (TARGET_ARC600): Update access to arch information.
32070 (TARGET_ARC601): Likewise.
32071 (TARGET_ARC700): Likewise.
32072 (TARGET_EM): Likewise.
32073 (TARGET_HS): Likewise.
32074 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32077 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32080 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32081 condition/operands for integer GE/LE/GEU/LEU operations.
32083 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32085 PR translation/79397
32086 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32089 2017-02-08 Martin Jambor <mjambor@suse.cz>
32092 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32093 whether allocation happened.
32094 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32095 nothing was allocated.
32097 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32099 PR tree-optimization/79408
32100 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32101 constant, but SSA_NAME with a known integer range, use the minimum
32102 of that range instead of op1 to determine if modulo can be replaced
32103 with its first operand.
32105 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32107 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32109 2017-02-08 Richard Biener <rguenther@suse.de>
32111 PR tree-optimization/71824
32112 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32113 Check all loops contained in the merged region.
32115 2017-02-07 Andrew Pinski <apinski@cavium.com>
32117 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32119 2017-02-07 Andrew Pinski <apinski@cavium.com>
32121 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32122 (thunderxt88): Likewise.
32123 (thunderxt81): Disable LSE and change v8.1 to v8.
32124 (thunderxt83): Likewise.
32126 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32127 Richard Biener <rguenther@suse.de>
32129 PR middle-end/79399
32130 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32131 type from int to size_t.
32132 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32134 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32136 PR rtl-optimization/79386
32137 * cprop.c (bypass_conditional_jumps): Initialize
32138 bypass_last_basic_block already before splitting bbs after
32139 unconditional traps...
32140 (bypass_conditional_jumps): ... rather than here.
32143 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32144 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32145 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32146 fix -masm=intel patterns.
32148 2017-02-07 Richard Biener <rguenther@suse.de>
32150 PR tree-optimization/79256
32151 PR middle-end/79278
32152 * builtins.c (get_object_alignment_2): Use min_align_of_type
32153 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32154 and ADJUST_FIELD_ALIGN.
32156 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32158 * doc/tm.texi: Regenerate.
32159 * stor-layout.c (layout_decl): Adjust.
32160 (update_alignment_for_field): Likewise.
32161 (place_field): Likewise.
32162 (min_align_of_type): Likewise.
32163 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32164 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32165 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32166 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32167 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32168 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32169 * config/i386/i386.c (x86_field_alignment): Likewise.
32170 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32171 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32172 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32173 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32174 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32175 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32179 2017-01-30 Richard Biener <rguenther@suse.de>
32181 PR tree-optimization/79256
32182 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32183 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32186 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32188 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32189 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32190 builtins to SImode and emit a zero-extend, if necessary.
32192 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32194 * docs/invoke.texi (RISC-V Options): Alphabetize.
32196 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32198 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32201 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32203 * config/riscv/riscv.c: New file.
32204 * common/config/riscv/riscv-common.c: Likewise.
32205 * config.gcc: Likewise.
32206 * config/riscv/constraints.md: Likewise.
32207 * config/riscv/elf.h: Likewise.
32208 * config/riscv/generic.md: Likewise.
32209 * config/riscv/linux.h: Likewise.
32210 * config/riscv/multilib-generator: Likewise.
32211 * config/riscv/peephole.md: Likewise.
32212 * config/riscv/pic.md: Likewise.
32213 * config/riscv/predicates.md: Likewise.
32214 * config/riscv/riscv-builtins.c: Likewise.
32215 * config/riscv/riscv-c.c: Likewise.
32216 * config/riscv/riscv-ftypes.def: Likewise.
32217 * config/riscv/riscv-modes.def: Likewise.
32218 * config/riscv/riscv-opts.h: Likewise.
32219 * config/riscv/riscv-protos.h: Likewise.
32220 * config/riscv/riscv.h: Likewise.
32221 * config/riscv/riscv.md: Likewise.
32222 * config/riscv/riscv.opt: Likewise.
32223 * config/riscv/sync.md: Likewise.
32224 * config/riscv/t-elf-multilib: Likewise.
32225 * config/riscv/t-linux: Likewise.
32226 * config/riscv/t-linux-multilib: Likewise.
32227 * config/riscv/t-riscv: Likewise.
32228 * configure.ac: Likewise.
32229 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32230 Waterman as RISC-V maintainers.
32231 * doc/install.texi: Add RISC-V entries.
32232 * doc/invoke.texi: Add RISC-V options section.
32233 * doc/md.texi: Add RISC-V constraints section.
32234 * configure: Regenerated.
32236 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32239 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32240 false values to be constant vectors with all 0 or all 1 bits set.
32241 (vcondu<mode><mode>): Likewise.
32242 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32244 (fpmask_comparison_operator): Update comment.
32245 (vecint_comparison_operator): New predicate.
32246 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32247 vector conditionals when the true and false values are constant
32248 vectors with all 0 bits or all 1 bits set.
32250 2017-02-06 Martin Sebor <msebor@redhat.com>
32252 PR tree-optimization/79376
32253 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32255 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32257 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32258 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32259 to simplify split condition.
32261 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32263 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32264 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32267 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32269 PR rtl-optimization/68664
32270 * target.def (can_speculate_insn): New hook.
32271 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32272 * doc/tm.texi: Regenerate.
32273 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32274 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32275 (rs6000_sched_can_speculate_insn): New function.
32277 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32279 PR tree-optimization/79284
32280 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32281 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32282 vectorizable_mask_load_store, vectorizable_operation,
32283 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32284 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32285 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32286 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32287 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32288 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32289 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32290 is_gimple_assign (stmt). Replace another such test with
32291 is_gimple_assign (stmt).
32293 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32296 * config/avr/avr.c (rtl-iter.h): Include it.
32297 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32298 (avr_legitimate_combined_insn): ...and implementation.
32300 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32302 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32303 * config/s390/s390.c (s390_const_operand_ok)
32304 (s390_canonicalize_comparison, s390_extract_part)
32305 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32306 (s390_contiguous_bitmask_p, s390_rtx_costs)
32307 (legitimize_pic_address): Likewise.
32308 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32309 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32310 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32311 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32312 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32314 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32316 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32317 REGNO($0) == REGNO($1).
32319 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32321 * config/s390/linux.h(SIZE_TYPE): Add comment.
32323 2017-02-06 Julian Brown <julian@codesourcery.com>
32324 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32325 Virendra Pathak <virendra.pathak@broadcom.com>
32327 * config/aarch64/aarch64-cores.def: Change the scheduler
32329 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32330 * config/aarch64/thunderx2t99.md: New file.
32332 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32334 * doc/standards.texi (Go Language): Update link to language
32337 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32339 * tree-eh.c (lower_resx): Sanitize profile.
32340 (cleanup_empty_eh_move_lp): Likewise.
32342 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32345 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32347 * cfgloopmanip.h (loop_version): Update prototype.
32348 * modulo-sched.c (sms_schedule): Update call of loop_version.
32349 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32350 * tree-parloops.c (gen_parallel_loop): Likewise.
32351 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32352 * tree-ssa-loop-split.c (split_loop): Likewise.
32353 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32354 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32356 2017-02-05 Martin Liska <mliska@suse.cz>
32359 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32361 (print_operand_address): Initialize a struct to zero.
32363 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32365 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32366 garbage collector only in textual form.
32368 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32370 * doc/extend.texi (x86 specific memory model extensions for
32371 transactional memory): Simplify a phrase.
32373 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32376 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32377 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32378 (atomic_storedi_1): Likewise.
32380 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32382 PR tree-optimization/79338
32383 * tree-parloops.c (gather_scalar_reductions): Don't call
32384 vect_analyze_loop_form for loop->inner before destroying loop's
32387 2017-02-03 Martin Sebor <msebor@redhat.com>
32389 PR tree-optimization/79327
32390 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32391 when precision has resulted in leading zeros.
32392 (format_integer): Adjust the likely counter to assume an unknown
32393 argument that may be zero is non-zero.
32395 2017-02-03 Jason Merrill <jason@redhat.com>
32398 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32399 avoid copying non-taken branch.
32401 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32403 PR tree-optimization/79340
32404 * tree-vect-loop.c (vectorizable_reduction): Release
32405 vec_defs elements after safe_splicing them into other vectors.
32408 PR tree-optimization/79327
32409 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32410 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32412 (format_integer): Use wide_int_to_tree instead of build_int_cst
32413 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32414 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32415 of shortest and longest sequence.
32417 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32419 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32420 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32422 2017-02-03 Walter Lee <walt@tilera.com>
32425 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32426 after initial stackframe link reg save.
32427 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32429 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32432 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32433 wu for stxssp alternative.
32435 2017-02-03 Martin Sebor <msebor@redhat.com>
32437 PR tree-optimization/79352
32438 * gimple-fold.c (get_range_strlen): Add argument.
32439 (get_range_strlen): Change return type to bool.
32440 (get_maxval_strlen): Pass in a dummy argument.
32441 * gimple-fold.h (get_range_strlen): Change return type to bool.
32442 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32443 * tree.h (array_at_struct_end_p): Add argument.
32444 * tree.c (array_at_struct_end_p): Handle it.
32446 2017-02-03 Martin Liska <mliska@suse.cz>
32449 * multiple_target.c (create_dispatcher_calls): Redirect edge
32450 from a caller of a dispatcher.
32451 (expand_target_clones): Make the clones local.
32452 (ipa_target_clone): Do both target clones and resolvers.
32453 (ipa_dispatcher_calls): Remove the pass.
32454 (pass_dispatcher_calls::gate): Likewise.
32455 (make_pass_dispatcher_calls): Likewise.
32456 * passes.def (pass_target_clone): Put as very first IPA early
32459 2017-02-03 Martin Liska <mliska@suse.cz>
32461 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32462 in case of a function with ifunc attribute.
32464 2017-02-03 Martin Liska <mliska@suse.cz>
32466 * cgraph.c (cgraph_node::dump): Dump function version info.
32467 * symtab.c (symtab_node::dump_base): Add missing new line.
32469 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32471 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32472 (ifcombine_ifandif): Use it.
32474 2017-02-03 Martin Liska <mliska@suse.cz>
32476 * doc/invoke.texi: Document default value for
32477 use-after-scope-direct-emission-threshold.
32479 2017-02-03 Martin Liska <mliska@suse.cz>
32481 PR tree-optimization/79339
32482 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32483 (format_floating): Likewise.
32485 2017-02-03 Martin Liska <mliska@suse.cz>
32488 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32490 (ipa_node_params_t::remove): Likewise.
32491 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32492 initialization from removed ipa_node_params_t::insert.
32493 (ipa_node_params::~ipa_node_params): Move from removed
32494 ipa_node_params_t::release.
32495 * symbol-summary.h (symbol_summary::m_released): New member.
32496 Do not release a summary twice. Do not allow to call finalizer
32497 for types of a summary that live in GGC memory.
32499 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32501 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32504 2017-02-02 Martin Sebor <msebor@redhat.com>
32506 PR middle-end/79275
32507 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32508 (format_string): Tighten up the range of output for non-constant
32509 strings and correct the expected range for wide non-constant strings.
32511 2017-02-02 Martin Sebor <msebor@redhat.com>
32513 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32515 PR middle-end/32003
32516 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32518 (-fdump-tree-@var): Add to index and document how to come up
32519 with pass-specific option and dump file names.
32520 (-fdump-passes): Clarify where to look for output.
32522 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32524 PR middle-end/77445
32525 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32526 statistics of the analyzed path; allow threading for speed when
32527 any of BBs along the path are optimized for speed.
32529 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32531 PR middle-end/78468
32532 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32533 settings of the virtual registers.
32536 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32538 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32539 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32542 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32544 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32545 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32547 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32549 * config/s390/s390.md: Add missing comments with the expanded
32551 * config/s390/vector.md: Likewise.
32552 * config/s390/vx-builtins.md: Likewise.
32554 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32557 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32558 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32559 conditions on a single line.
32561 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32563 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32564 __S390_VX__ to __VX__.
32566 2017-02-01 Andrew Pinski <apinski@cavium.com>
32568 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32569 stmt_info to record_stmt_cost.
32570 (vect_get_known_peeling_cost): Pass stmt_info if known to
32572 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32573 cpu_vector_cost field into
32574 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32575 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32576 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32577 splitting of scalar_stmt_cost and vec_stmt_cost.
32578 (thunderx_vector_cost): Likewise.
32579 (cortexa57_vector_cost): LIkewise.
32580 (exynosm1_vector_cost): Likewise.
32581 (xgene1_vector_cost): Likewise.
32582 (thunderx2t99_vector_cost): Improve after the splitting of the two
32584 (aarch64_builtin_vectorization_cost): Update for the splitting of
32585 scalar_stmt_cost and vec_stmt_cost.
32587 2017-02-01 Torvald Riegel <triegel@redhat.com>
32588 Richard Henderson <rth@redhat.com>
32590 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32591 conditional on existance of a fast atomic load.
32592 * optabs-query.c (can_atomic_load_p): New function.
32593 * optabs-query.h (can_atomic_load_p): Declare it.
32594 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32595 no fast atomic load is available for the particular size of access.
32596 (expand_atomic_compare_and_swap): Likewise.
32597 (expand_atomic_load): Likewise.
32598 (expand_atomic_store): Likewise.
32599 (expand_atomic_fetch_op): Likewise.
32600 * testsuite/lib/target-supports.exp
32601 (check_effective_target_sync_int_128): Remove x86 because it provides
32602 no fast atomic load.
32603 (check_effective_target_sync_int_128_runtime): Likewise.
32605 2017-02-01 Richard Biener <rguenther@suse.de>
32607 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32608 (graphite_transform_loops): Provide opt-info for optimized nests.
32609 * tree-parloop.c (parallelize_loops): Provide opt-info for
32610 parallelized loops.
32612 2017-02-01 Richard Biener <rguenther@suse.de>
32614 PR middle-end/79315
32615 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32616 was not set before.
32618 2017-02-01 Richard Biener <rguenther@suse.de>
32620 PR tree-optimization/71824
32621 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32622 Verify the loops are valid in the merged SESE region.
32623 (scop_detection::can_represent_loop_1): Check analyzing the
32624 evolution of the number of iterations in the region succeeds.
32626 2017-01-31 Ian Lance Taylor <iant@golang.org>
32628 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32629 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32631 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32633 PR preprocessor/79210
32634 * input.c (get_substring_ranges_for_loc): Replace line_width
32635 assertion with error-handling.
32637 2017-01-31 Richard Biener <rguenther@suse.de>
32639 PR tree-optimization/77318
32640 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32641 (create_pw_aff_from_tree): Take loop parameter.
32642 (add_condition_to_pbb): Pass loop of the condition to
32643 create_pw_aff_from_tree.
32645 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32647 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32648 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32650 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32654 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32656 (convert_int_to_float128): Likewise.
32657 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32658 (convert_int_to_float128): Likewise.
32659 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32660 (UNSPEC_IEEE128_CONVERT): Likewise.
32661 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32662 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32663 Use local variables for IBM extended format.
32664 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32665 (fix_trunc<mode>si2_fprs): Likewise.
32666 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32667 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32668 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32669 to know that we can now have integers of all sizes in vector
32671 (fix<uns>_<mode>di2_hw): Likewise.
32672 (float<uns>_<mode>si2_hw): Likewise.
32673 (fix_<mode>si2_hw): Likewise.
32674 (fixuns_<mode>si2_hw): Likewise.
32675 (float<uns>_<mode>di2_hw): Likewise.
32676 (float_<mode>di2_hw): Likewise.
32677 (float_<mode>si2_hw): Likewise.
32678 (floatuns_<mode>di2_hw): Likewise.
32679 (floatuns_<mode>si2_hw): Likewise.
32680 (xscvqp<su>wz_<mode>): Delete, no longer used.
32681 (xscvqp<su>dz_<mode>): Likewise.
32682 (xscv<su>dqp_<mode>): Likewise.
32683 (ieee128_mfvsrd_64bit): Likewise.
32684 (ieee128_mfvsrd_32bit): Likewise.
32685 (ieee128_mfvsrwz): Likewise.
32686 (ieee128_mtvsrw): Likewise.
32687 (ieee128_mtvsrd_64bit): Likewise.
32688 (ieee128_mtvsrd_32bit): Likewise.
32690 2017-01-31 Martin Liska <mliska@suse.cz>
32693 * ipa-prop.c (ipa_free_all_node_params): Call release method
32694 instead of ~sumbol_summary to not to trigger double times
32697 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32699 PR tree-optimization/71691
32700 * bitmap.h (class auto_bitmap): New.
32701 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32702 is_maybe_undefined instead of ssa_undefined_value_p.
32704 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32706 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32707 __S390_ARCH_LEVEL__ to __ARCH__.
32709 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32711 PR tree-optimization/79267
32712 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32713 if should_remove_lhs_p is true.
32715 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32718 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32719 (add_alignment_attribute): New.
32720 (base_type_die): Add alignment attribute.
32721 (subrange_type_die): Likewise.
32722 (modified_type_die): Likewise.
32723 (gen_array_type_die): Likewise.
32724 (gen_descr_array_type_die: Likewise.
32725 (gen_enumeration_type_die): Likewise.
32726 (gen_subprogram_die): Likewise.
32727 (gen_variable_die): Likewise.
32728 (gen_field_die): Likewise.
32729 (gen_ptr_to_mbr_type_die): Likewise.
32730 (gen_struct_or_union_type_die): Likewise.
32731 (gen_subroutine_type_die): Likewise.
32732 (gen_typedef_die): Likewise.
32733 (base_type_cmp): Compare alignment attribute.
32735 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32738 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32739 (setb_unsigned) New pattern for setb with CCUNS.
32740 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32741 subfc./subfe sequence to avoid overflow problems. Generate a
32742 shorter sequence with cmpld/setb for power9.
32743 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32744 for generating subfc. instruction.
32745 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32746 now uses this instruction.
32748 2017-01-30 Ian Lance Taylor <iant@google.com>
32751 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32752 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32754 2017-01-30 Martin Sebor <msebor@redhat.com>
32756 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32757 Move constant to the right of a relational operator.
32758 (get_mpfr_format_length, format_character, format_string): Ditto.
32759 (should_warn_p, maybe_warn): Same.
32761 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32763 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32766 * asan.c (get_translation_unit_decl): Remove function.
32767 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32769 2017-01-30 Martin Liska <mliska@suse.cz>
32771 PR gcov-profile/79259
32772 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32773 -fprofile-generate.
32775 2017-01-30 Martin Liska <mliska@suse.cz>
32778 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32779 Initialize variables with NULL value.
32781 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32784 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32786 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32788 2017-01-30 Richard Biener <rguenther@suse.de>
32790 PR tree-optimization/79276
32791 * tree-vrp.c (process_assert_insertions): Properly adjust common
32792 when removing a duplicate.
32794 2017-01-30 Richard Biener <rguenther@suse.de>
32796 PR tree-optimization/79256
32797 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32798 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32800 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32802 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32805 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32806 ("*r<noxa>sbg_<mode>_sll_bitmask")
32807 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32808 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32809 Use contiguous_bitmask_nowrap_operand.
32811 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32814 * config/rs6000/altivec.h (vec_xl): Revise #define.
32815 (vec_xst): Likewise.
32817 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32819 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32821 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32823 PR rtl-optimization/79194
32824 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32825 traps before call to bypass_conditional_jumps.
32827 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32829 PR tree-optimization/71374
32830 * lra-constraints.c (check_conflict_input_operands): New.
32831 (match_reload): Use it.
32833 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32836 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32837 account to calculate conflict_set.
32839 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32841 PR rtl-optimization/78559
32842 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32843 other_insn in combine.
32845 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32847 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32848 uint16_type_node for BT_UINT16.
32850 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32852 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32853 "RTL Tests" to menu.
32854 (GIMPLE Tests): New node.
32855 (RTL Tests): New node.
32857 2017-01-27 Richard Biener <rguenther@suse.de>
32859 PR tree-optimization/79245
32860 * tree-loop-distribution.c (distribute_loop): Apply cost
32861 modeling also to detected patterns.
32863 2017-01-27 Richard Biener <rguenther@suse.de>
32865 PR tree-optimization/71433
32866 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32867 (compare_assert_loc): New function.
32868 (process_assert_insertions): Sort and optimize assert locations
32869 to remove duplicates and push down identical assertions on
32870 edges to their destination block.
32872 2017-01-27 Richard Biener <rguenther@suse.de>
32874 PR tree-optimization/79244
32875 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32876 out SSA names even if abnormal.
32878 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32880 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32881 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32882 instead of MPFR_RNDN.
32884 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32887 * arm.c (arm_option_override): Don't call build_target_option_node
32888 until after doing all option overrides.
32889 (arm_valid_target_attribute_tree): Likewise.
32891 2017-01-27 Martin Liska <mliska@suse.cz>
32893 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32894 for {cd}tors and C++ {cd}tors.
32896 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32898 * config/s390/s390.md ("*setmem_long_and")
32899 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32901 2017-01-26 Martin Sebor <msebor@redhat.com>
32903 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32906 2017-01-26 Martin Sebor <msebor@redhat.com>
32908 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32909 HAVE_DFmode before using XFmode or DFmode.
32910 (parse_directive): Avoid using the z length modifier to avoid
32911 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32913 PR middle-end/78703
32914 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32915 to accept adjustment as an array.
32916 (get_int_range): New function.
32917 (struct directive): Make width and prec arrays.
32918 (directive::set_width, directive::set_precision): Call get_int_range.
32919 (format_integer, format_floating): Handle width and precision ranges.
32920 (format_string, parse_directive): Same.
32922 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32925 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32926 comdat_type_p set, just clone them, but keep the children in the
32930 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32931 which have direct callers with -fvar-tracking-assignments enabled
32933 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32934 inside of type units.
32936 2017-01-26 Martin Sebor <msebor@redhat.com>
32938 PR middle-end/78703
32939 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32941 (struct format_result): Replace number_chars, number_chars_min,
32942 and number_chars_max with a single member of struct result_range.
32944 (format_result::operator+=): Adjust.
32945 (struct fmtresult): Remove bounded. Handle likely and unlikely
32947 (fmtresult::adjust_for_width_or_precision): New function.
32948 (fmtresult:type_max_digits): New function.
32949 (bytes_remaining): Handle likely and unlikely counters.
32950 (min_bytes_remaining): Remove.
32951 (format_percent): Simplify.
32952 (format_integer, format_floating): Set likely and unlikely counters.
32953 (get_string_length, format_character, format_string): Same.
32954 (format_plain, should_warn_p): New function.
32955 (maybe_warn): Call should_warn_p. Update diagnostic messages
32956 and handle those for all directives, including plain strings.
32957 (format_directive): Handle likely and unlikely counters.
32958 Remove unnecessary quoting from diagnostics. Add an informational
32960 (add_bytes): Remove.
32961 (pass_sprintf_length::compute_format_length): Simplify.
32962 (try_substitute_return_value): Handle likely and unlikely counters.
32964 2017-01-26 Carl Love <cel@us.ibm.com>
32966 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32967 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32969 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32972 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32973 endianess for subregs into account.
32974 * lra-constraints.c (lra_constraints): Do risky transformations
32975 always on the first iteration.
32976 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32977 last_call_used_reg_set.
32978 (process_bb_lives): Define and use last_call_used_reg_set.
32979 * lra.c (lra): Always continue after lra_constraints on the first
32982 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32984 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32986 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32988 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32990 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32991 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32992 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32993 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32994 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32995 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32996 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32997 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32998 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33000 2017-01-26 Marek Polacek <polacek@redhat.com>
33003 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33004 for the third operand.
33006 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33008 PR middle-end/79236
33009 * omp-low.c (struct omp_context): Add simt_stmt field.
33010 (scan_omp_for): Return omp_context *.
33011 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33012 context to the _simt_ SIMD stmt.
33013 (lower_omp_for): For combined SIMD with sibling _simt_
33014 SIMD, make sure to use the same decls in _looptemp_
33015 clauses as in the sibling.
33017 2017-01-26 David Sherwood <david.sherwood@arm.com>
33019 PR middle-end/79212
33020 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33023 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33026 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33027 emit fld b; fld a; if possible.
33029 * brig-builtins.def: Update copyright years.
33030 * config/arm/arm_acle_builtins.def: Update copyright years.
33032 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33035 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33036 constraint instead of o for the stxsd instruction.
33038 2017-01-25 Carl Love <cel@us.ibm.com>
33040 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33041 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33043 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33045 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33047 2017-01-25 Richard Biener <rguenther@suse.de>
33049 PR tree-optimization/69264
33050 * target.def (vector_alignment_reachable): Improve documentation.
33051 * doc/tm.texi: Regenerate.
33052 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33054 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33055 earlier changes with respect to TYPE_USER_ALIGN.
33056 (vector_alignment_reachable_p): Likewise. Improve dumping.
33058 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33061 * config/arm/arm.md (xordi3): Force constant operand into a register
33064 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33066 * doc/invoke.texi (-fstore-merging): Correct default optimization
33067 levels at which it is enabled.
33068 (-O): Move -fstore-merging from list to...
33071 2017-01-25 Richard Biener <rguenther@suse.de>
33074 * omp-expand.c: Include debug.h.
33075 (expand_omp_taskreg): Make sure to generate early debug before
33076 outlining anything from a function.
33077 (expand_omp_target): Likewise.
33078 (grid_expand_target_grid_body): Likewise.
33080 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33083 * asan.c (get_translation_unit_decl): New function.
33084 (asan_add_global): Extract modules file name from globals
33085 TRANSLATION_UNIT_DECL name.
33087 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33090 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33091 for long calls with APCS frame and VFP.
33093 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33095 * cfg.c (original_copy_tables_initialized_p): New function.
33096 * cfg.h (original_copy_tables_initialized_p): New decl.
33097 * cfgrtl.c (relink_block_chain): Guard the call to
33098 free_original_copy_tables with a call to
33099 original_copy_tables_initialized_p.
33100 * cgraph.h (symtab_node::native_rtl_p): New decl.
33101 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33102 (symtab_node::needed_p): Don't assert for early assembly output
33103 for __RTL functions.
33104 (cgraph_node::finalize_function): Set "force_output" for __RTL
33106 (cgraph_node::analyze): Bail out early for __RTL functions.
33107 (analyze_functions): Update assertion to support __RTL functions.
33108 (cgraph_node::expand): Bail out early for __RTL functions.
33109 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33111 * function.h (struct function): Update comment for field
33113 * gimple-expr.c: Include "tree-pass.h".
33114 (gimple_has_body_p): Return false for __RTL functions.
33115 * Makefile.in (OBJS): Add run-rtl-passes.o.
33116 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33118 (gcc::pass_manager::get_clean_slate): New accessor.
33119 * passes.c: Include "insn-addr.h".
33120 (should_skip_pass_p): Add logging. Update logic for running
33121 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33122 property-provider override so it is only done for gimple passes.
33124 (skip_pass): New function.
33125 (execute_one_pass): Call skip_pass when skipping passes.
33126 * read-md.c (md_reader::read_char): Support filtering
33127 the input to a subset of line numbers.
33128 (md_reader::md_reader): Initialize fields
33129 m_first_line and m_last_line.
33130 (md_reader::read_file_fragment): New function.
33131 * read-md.h (md_reader::read_file_fragment): New decl.
33132 (md_reader::m_first_line): New field.
33133 (md_reader::m_last_line): New field.
33134 * read-rtl-function.c (function_reader::create_function): Only
33135 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33136 curr_properties. Set DECL_INITIAL to a dummy block.
33137 (read_rtl_function_body_from_file_range): New function.
33138 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33140 * run-rtl-passes.c: New file.
33141 * run-rtl-passes.h: New file.
33143 2017-01-24 Jeff Law <law@redhat.com>
33145 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33148 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33150 PR tree-optimization/79159
33151 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33152 (record_nonwrapping_iv): Improve boundary using above function if no
33153 value range information.
33155 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33156 Martin Jambor <mjambor@suse.cz>
33158 * brig-builtins.def: New file.
33159 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33160 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33161 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33162 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33163 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33164 * builtin-types.def (BT_INT8): New.
33165 (BT_INT16): Likewise.
33166 (BT_UINT8): Likewise.
33167 (BT_UINT16): Likewise.
33168 (BT_FN_ULONG): Likewise.
33169 (BT_FN_UINT_INT): Likewise.
33170 (BT_FN_UINT_ULONG): Likewise.
33171 (BT_FN_UINT_LONG): Likewise.
33172 (BT_FN_UINT_PTR): Likewise.
33173 (BT_FN_ULONG_PTR): Likewise.
33174 (BT_FN_INT8_FLOAT): Likewise.
33175 (BT_FN_INT16_FLOAT): Likewise.
33176 (BT_FN_UINT32_FLOAT): Likewise.
33177 (BT_FN_UINT16_FLOAT): Likewise.
33178 (BT_FN_UINT8_FLOAT): Likewise.
33179 (BT_FN_UINT64_FLOAT): Likewise.
33180 (BT_FN_UINT16_UINT32): Likewise.
33181 (BT_FN_UINT32_UINT16): Likewise.
33182 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33183 (BT_FN_INT_PTR_INT): Likewise.
33184 (BT_FN_UINT_PTR_UINT): Likewise.
33185 (BT_FN_LONG_PTR_LONG): Likewise.
33186 (BT_FN_ULONG_PTR_ULONG): Likewise.
33187 (BT_FN_VOID_UINT64_UINT64): Likewise.
33188 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33189 (BT_FN_INT8_INT8_INT8): Likewise.
33190 (BT_FN_INT16_INT16_INT16): Likewise.
33191 (BT_FN_INT_INT_INT): Likewise.
33192 (BT_FN_UINT_FLOAT_UINT): Likewise.
33193 (BT_FN_FLOAT_UINT_UINT): Likewise.
33194 (BT_FN_ULONG_UINT_UINT): Likewise.
33195 (BT_FN_ULONG_UINT_PTR): Likewise.
33196 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33197 (BT_FN_UINT_UINT_UINT): Likewise.
33198 (BT_FN_VOID_UINT_PTR): Likewise.
33199 (BT_FN_UINT_UINT_PTR: Likewise.
33200 (BT_FN_UINT32_UINT64_PTR): Likewise.
33201 (BT_FN_INT_INT_UINT_UINT): Likewise.
33202 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33203 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33204 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33205 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33206 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33207 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33208 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33209 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33210 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33211 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33212 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33213 * doc/frontends.texi: List BRIG FE.
33214 * doc/install.texi (Testing): Add BRIG tesring requirements.
33215 * doc/invoke.texi (Overall Options): Mention BRIG.
33216 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33218 2017-01-24 Richard Biener <rguenther@suse.de>
33220 PR translation/79208
33221 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33223 2017-01-24 Martin Jambor <mjambor@suse.cz>
33226 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33227 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33228 and known_contexts.
33230 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33232 PR middle-end/79123
33233 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33234 casts from signed to unsigned really don't have a range.
33236 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33238 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33239 GMP_RNDx for compatiblity.
33241 2017-01-24 Martin Liska <mliska@suse.cz>
33244 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33245 that would prevent us to call alloca with -1 as argument.
33247 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33249 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33250 Avoid -Wformat-security warning.
33252 2017-01-23 Andrew Pinski <apinski@cavium.com>
33254 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33257 2017-01-23 Martin Sebor <msebor@redhat.com>
33259 PR middle-end/78703
33260 * gimple-ssa-sprintf.c (warn_level): New global.
33261 (format_integer): Use it here and throughout the rest of the file.
33262 Use the same switch to compute sign as base.
33263 (maybe_warn): New function.
33264 (format_directive): Factor out warnings into maybe_warn.
33265 Add debugging output. Use warn_level.
33266 (add_bytes): Use warn_level.
33267 (pass_sprintf_length::compute_format_length): Add debugging output.
33268 (try_substitute_return_value): Same.
33269 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33271 PR middle-end/78703
33272 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33273 (struct fmtresult, format_integer, format_floating): Adjust.
33274 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33275 (get_string_length, format_string,format_directive): Same.
33276 (pass_sprintf_length::compute_format_length): Same.
33277 (try_substitute_return_value): Simplify slightly.
33279 PR middle-end/78703
33280 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33281 (fmtresult::operator+=): Outlined.
33282 (struct fmtresult): Add ctors.
33283 (struct conversion_spec): Rename...
33284 (struct directive): ...to this. Add and remove data members.
33285 (directive::set_width, directive::set_precision): New functions.
33286 (format_percent): Use fmtresult ctor.
33287 (get_width_and_precision): Remove.
33288 (format_integer): Make naming changes. Avoid computing width and
33290 (format_floating): Same. Adjust indentation.
33291 (format_character, format_none): New functions.
33292 (format_string): Moved character handling to format_character.
33293 (format_directive): Remove arguments, change return type.
33294 (parse_directive): New function.
33295 (pass_sprintf_length::compute_format_length): Move directive
33296 parsing to parse_directive.
33298 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33300 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33301 (assign_assembler_name_if_needed): ... this.
33302 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33303 (assign_assembler_name_if_needed): ... this.
33304 (free_lang_data_in_cgraph): Adjust callers.
33305 * cgraphunit.c (cgraph_node::analyze): Likewise.
33306 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33309 2017-01-23 Richard Biener <rguenther@suse.de>
33311 PR tree-optimization/79088
33312 PR tree-optimization/79188
33313 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33314 resetting loop bounds after last path deletion. Reset loop
33315 bounds of the target loop, make code match the comments.
33316 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33317 Make sure loops need no fixups.
33319 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33321 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33322 exponent support with double type for first argument.
33323 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33324 type returned by __builtin_vec_extract_sig,
33325 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33326 functions from "vector int" to "vector unsigned int" or from
33327 "vector long long int" to "vector unsigned long long int".
33328 Changed type returned by __builtin_vec_extract_exp,
33329 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33330 functions from "vector int" to "vector unsigned int" or from
33331 "vector long long int" to "vector unsigned long long int".
33332 Changed return type of __builtin_vec_test_data_class,
33333 __builtin_vec_test_data_class_sp, and
33334 __builtin_vec_test_data_class_dp from "vector int" to
33335 "vector bool int" or from "vector long long int" to "vector bool
33336 long long int" and changed second argument type from "unsigned
33337 int" to "int". Added new overloaded function forms "vector float
33338 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33339 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33340 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33341 double, vector unsigned long long int)" and "vector double
33342 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33343 long int)". Changed return type of
33344 __builtin_scalar_test_data_class and
33345 __builtin_scalar_test_data_class_sp and
33346 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33347 int" and changed second argument from "unsigned int" to "int".
33348 Changed type returned by __builtin_scalar_test_neg,
33349 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33350 from "int" to "bool int". Added new overloaded function form
33351 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33352 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33353 exponent double-precision with floating point first argument.
33354 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33355 documentation of scalar_test_data_class, scalar_test_neg,
33356 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33357 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33358 vec_test_data_class built-in functions to reflect refinements in
33359 their type signatures.
33361 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33363 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33365 (aarch64_elf_asm_destructor): Likewise.
33367 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33369 PR rtl-optimization/78634
33370 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33371 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33372 * ifcvt.c (noce_try_cmove): Add missing cost check.
33374 PR rtl-optimization/71724
33375 * combine.c (if_then_else_cond): Look for situations where it is
33376 beneficial to undo the work of one of the recursive calls.
33378 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33380 PR tree-optimization/70754
33381 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33382 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33383 combined stmt before it if not NULL.
33384 (combine_chains): Process refs reversely and compute dominance point
33387 2017-01-23 Martin Liska <mliska@suse.cz>
33389 PR tree-optimization/79196
33390 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33391 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33393 (strlen_optimize_stmt): Call the renamed function.
33395 2017-01-23 Michael Matz <matz@suse.de>
33397 PR tree-optimization/78384
33398 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33400 2017-01-23 Richard Biener <rguenther@suse.de>
33402 PR tree-optimization/79186
33403 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33404 both incoming edges before moving an assert.
33406 2017-01-23 Martin Jambor <mjambor@suse.cz>
33408 * ipa-prop.c (load_from_param_1): Removed.
33409 (load_from_unmodified_param): Bits from load_from_param_1 put back
33411 (load_from_param): Removed.
33412 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33413 with stmt. Reverted back to use of load_from_unmodified_param.
33415 2017-01-23 Martin Jambor <mjambor@suse.cz>
33418 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33419 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33420 field a pointer to garbage collected vector, mark lattices and
33421 ipcp_orig_node with GTY((skip)).
33422 (ipa_get_param_count): Adjust to descriptors being a pointer.
33423 (ipa_get_param): Likewise.
33424 (ipa_get_type): Likewise.
33425 (ipa_get_param_move_cost): Likewise.
33426 (ipa_set_param_used): Likewise.
33427 (ipa_get_controlled_uses): Likewise.
33428 (ipa_set_controlled_uses): Likewise.
33429 (ipa_is_param_used): Likewise.
33430 (ipa_node_params_t): Move into garbage collector. New methods insert
33432 (ipa_node_params_sum): Annotate wth GTY(()).
33433 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33435 (ipa_load_from_parm_agg): Adjust declaration.
33436 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33437 * ipa-profile.c (ipa_profile): Likewise.
33438 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33439 (ipa_populate_param_decls): Make descriptors parameter garbage
33441 (ipa_dump_param): Adjust to descriptors being a pointer.
33442 (ipa_alloc_node_params): Likewise.
33443 (ipa_initialize_node_params): Likewise.
33444 (load_from_param_1): Make descriptors parameter garbage collected.
33445 (load_from_unmodified_param): Likewise.
33446 (load_from_param): Likewise.
33447 (ipa_load_from_parm_agg): Likewise.
33448 (ipa_node_params::~ipa_node_params): Removed.
33449 (ipa_free_all_node_params): Remove call to delete operator.
33450 (ipa_node_params_t::insert): New.
33451 (ipa_node_params_t::remove): Likewise.
33452 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33453 copy known_csts and known_contexts vectors.
33454 (ipa_read_node_info): Adjust to descriptors being a pointer.
33455 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33457 (ipcp_transform_function): Make descriptors variable garbage
33460 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33462 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33463 * config/i386/avx512dqintrin.h: Ditto.
33464 * config/i386/avx512fintrin.h: Ditto.
33465 * config/i386/i386.c: Handle new builtins.
33466 * config/i386/i386-builtin.def: Add new builtins.
33467 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33468 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33470 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33471 Martin Liska <mliska@suse.cz>
33473 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33474 * asan.c (asan_expand_poison_ifn): Support stores and use
33475 appropriate ASAN report function.
33476 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33477 * internal-fn.def (ASAN_POISON_USE): Declare.
33478 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33479 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33480 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33481 ASAN_POISON calls w/o LHS.
33482 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33483 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33484 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33485 * gimplify.c (asan_poison_variables): Add attribute
33486 use_after_scope_memory to variables that really needs to live
33488 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33489 having the attribute.
33491 2017-01-23 Martin Liska <mliska@suse.cz>
33493 * asan.c (create_asan_shadow_var): New function.
33494 (asan_expand_poison_ifn): Likewise.
33495 * asan.h (asan_expand_poison_ifn): New declaration.
33496 * internal-fn.c (expand_ASAN_POISON): Likewise.
33497 * internal-fn.def (ASAN_POISON): New builtin.
33498 * sanopt.c (pass_sanopt::execute): Expand
33499 asan_expand_poison_ifn.
33500 * tree-inline.c (copy_decl_for_dup_finish): Make function
33502 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33503 * tree-ssa.c (is_asan_mark_p): New function.
33504 (execute_update_addresses_taken): Rewrite local variables
33505 (identified just by use-after-scope as addressable) into SSA.
33507 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33509 * doc/install.texi (Specific): opensource.apple.com uses https
33510 now. Remove trailing slash.
33512 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33514 * README.Portability: Remove note on an Irix compatibility issue.
33516 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33518 * gcov.c (INCLUDE_ALGORITHM): Define.
33519 (INCLUDE_VECTOR): Define.
33520 No longer include <vector> and <algorithm> directly.
33522 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33524 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33526 * doc/invoke.texi (Code Gen Options): Ditto.
33528 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33531 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33533 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33535 rtl-optimization/79125
33536 * cprop.c (local_cprop_pass): Handle cases where we make an
33537 unconditional trap.
33539 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33543 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33544 read from, for big endian.
33546 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33548 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33549 register pauth builtins for LP64 only.
33551 2017-01-20 Marek Polacek <polacek@redhat.com>
33554 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33557 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33559 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33561 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33562 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33563 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33565 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33568 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33569 in big-endian mode when they are not a single duplicated value.
33571 2017-01-20 Richard Biener <rguenther@suse.de>
33573 * BASE-VER: Bump to 7.0.1.
33575 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33577 * omp-low.c (omplow_simd_context): New struct. Use it...
33578 (lower_rec_simd_input_clauses): ...here and...
33579 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33580 references to idx, lane, max_vf, is_simt.
33582 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33584 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33587 2017-01-20 Martin Jambor <mjambor@suse.cz>
33589 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33590 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33592 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33593 (GTFILES): Rename hsa.c to hsa-common.c.
33594 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33595 * hsa-dump.c: Likewise.
33596 * hsa-gen.c: Likewise.
33597 * hsa-regalloc.c: Likewise.
33598 * ipa-hsa.c: Likewise.
33599 * omp-expand.c: Likewise.
33600 * omp-low.c: Likewise.
33601 * toplev.c: Likewise.
33603 2017-01-20 Marek Polacek <polacek@redhat.com>
33606 * doc/invoke.texi: Document -Wduplicated-branches.
33607 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33608 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33609 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33610 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33611 return 0 only when not OEP_LEXICOGRAPHIC.
33612 (fold_build_cleanup_point_expr): Use the expression
33613 location when building CLEANUP_POINT_EXPR.
33614 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33615 * tree.c (add_expr): Handle error_mark_node.
33617 2017-01-20 Martin Liska <mliska@suse.cz>
33620 * tree-profile.c (init_ic_make_global_vars): Do not call
33622 (gimple_init_gcov_profiler): Likewise.
33624 2017-01-20 Martin Liska <mliska@suse.cz>
33627 * cgraph.h (maybe_create_reference): Remove argument and
33629 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33631 * ipa-cp.c (create_specialized_node): Likewise.
33632 * symtab.c (symtab_node::maybe_create_reference): Handle
33633 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33635 2017-01-20 Martin Liska <mliska@suse.cz>
33637 * read-rtl-function.c (function_reader::create_function): Use
33638 build_decl instread of build_decl_stat.
33640 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33642 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33643 * config/i386/avx512dqintrin.h: Ditto.
33644 * config/i386/avx512fintrin.h: Ditto.
33645 * config/i386/i386-builtin-types.def: Add new types.
33646 * config/i386/i386.c: Handle new types.
33647 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33648 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33649 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33650 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33651 (__builtin_ia32_kshiftridi): New.
33652 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33654 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33658 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33659 define to rs6000_init_stack_protect_guard.
33660 (rs6000_init_stack_protect_guard): New function.
33662 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33663 Yunqiang Su <yunqiang.su@imgtec.com>
33665 * config.gcc (supported_defaults): Add madd4.
33666 (with_madd4): Add validation.
33667 (all_defaults): Add madd4.
33668 * config/mips/mips.opt (mmadd4): New option.
33669 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33671 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33673 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33674 (ISA_HAS_FUSED_MADD4): Likewise.
33675 * doc/invoke.texi (-mmadd4): Document the new option.
33676 * doc/install.texi (--with-madd4): Document the new option.
33678 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33680 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33681 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33682 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33683 (aarch64_init_pauth_hint_builtins): New.
33684 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33685 (aarch64_expand_builtin): Expand new builtins.
33687 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33689 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33690 * combine-stack-adj.c (no_unhandled_cfa): Handle
33691 REG_CFA_TOGGLE_RA_MANGLE.
33692 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33693 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33694 info for return address signing.
33695 (aarch64_expand_epilogue): Likewise.
33697 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33699 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33700 * config/aarch64/aarch64-protos.h
33701 (aarch64_return_address_signing_enabled): New declaration.
33702 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33704 (aarch64_expand_prologue): Sign return address before it's pushed onto
33706 (aarch64_expand_epilogue): Authenticate return address fetched from
33708 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33709 (aarch64_attributes): New function attributes for "sign-return-address".
33710 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33711 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33712 ("*do_return"): Generate combined instructions according to key index.
33713 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33714 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33716 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33717 * config/aarch64/aarch64.opt (msign-return-address=): New.
33718 * doc/extend.texi (AArch64 Function Attributes): Documents
33719 "sign-return-address=".
33720 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33722 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33724 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33725 overall option summary.
33727 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33729 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33730 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33731 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33732 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33734 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33736 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33737 -mpower9-minmax by default for -mcpu=power9.
33738 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33739 128-bit floating point.
33741 2017-01-20 Alan Modra <amodra@gmail.com>
33743 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33744 optimizing for size.
33746 2017-01-20 Alan Modra <amodra@gmail.com>
33749 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33750 for strcmp and strncmp from corresponding builtin decl.
33752 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33754 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33755 instead of i386/rtems-64.h.
33756 * config/i386/rtems-64.h: Remove.
33758 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33762 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33764 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33766 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33768 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33769 Change int to HOST_WIDE_INT.
33770 * config/aarch64/aarch64-protos.h
33771 (aarch64_simd_gen_const_vector_dup): Likewise.
33772 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33774 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33776 * langhooks-def.h (lhd_type_for_size): New decl.
33777 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33778 * langhooks.c (lhd_type_for_size): New function, taken from
33781 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33783 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33784 define_bypass for CR latency.
33785 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33786 (power9-alu2): Add define_bypass for CR latency.
33788 (power9-mul): Update insn latency.
33789 (power9-mul-compare): Update insn latency, bypass latency and remove
33792 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33794 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33796 * config/aarch64/aarch64.md
33797 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33798 aarch64_nopcrelative_literal_loads.
33799 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33801 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33803 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33804 TARGET_LOONGSON_3A.
33805 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33807 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33810 * config.gcc (supported_defaults): Add lxc1-sxc1.
33811 (with_lxc1_sxc1): Add validation.
33812 (all_defaults): Add lxc1-sxc1.
33813 * config/mips/mips.opt (mlxc1-sxc1): New option.
33814 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33816 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33817 __mips_no_lxc1_sxc1.
33818 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33819 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33820 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33822 2017-01-19 Richard Biener <rguenther@suse.de>
33824 PR tree-optimization/72488
33825 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33826 sure to restore SSA info.
33827 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33829 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33831 PR rtl-optimization/79121
33832 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33833 of the inner type when shifting an extended value.
33835 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33838 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33839 interposable aliases.
33841 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33844 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33845 Use the evmergelohi instruction.
33846 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33847 (mov_sitf_e500_subreg8_2_be): Likewise.
33848 (mov_sitf_e500_subreg12_2_le): Likewise.
33849 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33850 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33851 (mov_sitf_e500_subreg8_2_le): Likewise.
33852 (mov_sitf_e500_subreg12_2_be): Likewise.
33854 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33856 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33857 attribute from vecsimple to vecperm.
33858 (altivec_vbpermq2): Likewise.
33860 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33863 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33865 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33866 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33867 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33868 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33869 case where N arg is SIZE_MAX.
33870 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33871 (cmpstrsi): Add pattern.
33873 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33875 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33876 __builtin_vec_revb builtins.
33877 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33878 built-in functions to support generation of the ISA 3.0 XXBR<x>
33879 vector byte reverse instructions.
33880 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33881 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33882 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33883 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33884 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33885 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33886 (P9V_BUILTIN_VEC_REVB): Likewise.
33887 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33888 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33889 (p9_xxbrq_v16qi): Likewise.
33890 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33891 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33892 (p9_xxbrh_v8hi): Likewise.
33893 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33894 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33895 vec_revb built-in functions.
33897 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33899 PR rtl-optimization/78952
33900 * config/i386/i386.md (any_extract): New code iterator.
33901 (*insvqi_2): Use any_extract for source operand.
33902 (*insvqi_3): Use any_shiftrt for source operand.
33904 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33906 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33908 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33910 2017-01-18 Matthias Klose <doko@ubuntu.com>
33912 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33914 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33916 * config/rs6000/altivec.h (vec_bperm): Change #define.
33917 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33918 (altivec_vbpermq2): New define_insn.
33919 (altivec_vbpermd): Likewise.
33920 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33921 function interface.
33922 (VBPERMD): Likewise.
33923 (VBPERM): New polymorphic function interface.
33924 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33925 Add entries for P9V_BUILTIN_VEC_VBPERM.
33926 * doc/extend.texi: Add interfaces for vec_bperm.
33928 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33930 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33931 first letter of error messages.
33932 (s390_resolve_overloaded_builtin): Likewise.
33933 * config/s390/s390.c (s390_expand_builtin): Likewise.
33934 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33935 (s390_valid_target_attribute_inner_p): Likewise.
33936 * config/s390/s390.md ("tabort"): Likewise.
33938 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33940 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33941 (ISA_AVOID_DIV_HILO): New macro.
33942 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33943 (ISA_HAS_DDIV): Likewise.
33945 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33947 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33949 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33951 * doc/invoke.texi (fabi-version): Spelling fix.
33953 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33956 * doc/invoke.texi (fabi-version): Mention mangling fix for
33959 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33962 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33964 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33967 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33968 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33970 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33972 (stack_protect_test): Ditto.
33973 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33974 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33976 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33977 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33978 -mstack-protector-guard-offset=.
33979 (RS/6000 and PowerPC Options): Ditto.
33981 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33983 * config/i386/i386.h (MASK_CLASS_P): New define.
33984 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33985 there are no registers from different register sets also when
33986 mask registers are used. Update function comment.
33987 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33988 to (*k/*r) and (*k/*km) alternatives.
33990 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33992 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33993 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33994 (EH_RETURN_HANDLER_RTX): New define.
33995 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33996 Force frame pointer in EH return functions.
33997 (aarch64_expand_epilogue): Add barrier for eh_return.
33998 (aarch64_final_eh_return_addr): Remove.
33999 (aarch64_eh_return_handler_rtx): New function.
34000 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34002 (aarch64_eh_return_handler_rtx): New prototype.
34004 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34006 * config/rs6000/altivec.h (vec_rlmi): New #define.
34007 (vec_vrlnm): Likewise.
34008 (vec_rlnm): Likewise.
34009 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34010 (UNSPEC_VRLNM): Likewise.
34011 (VIlong): New mode iterator.
34012 (altivec_vrl<VI_char>mi): New define_insn.
34013 (altivec_vrl<VI_char>nm): Likewise.
34014 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34016 (VRLDNM): Likewise.
34017 (RLNM): New polymorphic function entry.
34018 (VRLWMI): New monomorphic function entry.
34019 (VRLDMI): Likewise.
34020 (RLMI): New polymorphic function entry.
34021 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34022 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34023 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34026 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34029 * dwarf2out.c (field_byte_offset): Restore the
34030 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34031 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34032 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34035 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34038 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34040 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34043 * dwarf2out.c (add_data_member_location_attribute): For constant
34044 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34045 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34046 DW_AT_byte_size attributes.
34048 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34050 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34051 after forcing to constant memory when the code model is medium.
34053 2017-01-17 Julia Koval <julia.koval@intel.com>
34056 * config/i386/avx512fintrin.h
34057 (_mm512_i32gather_ps): Change __addr type to void const*.
34058 (_mm512_mask_i32gather_ps): Ditto.
34059 (_mm512_i32gather_pd): Ditto.
34060 (_mm512_mask_i32gather_pd): Ditto.
34061 (_mm512_i64gather_ps): Ditto.
34062 (_mm512_mask_i64gather_ps): Ditto.
34063 (_mm512_i64gather_pd): Ditto.
34064 (_mm512_mask_i64gather_pd): Ditto.
34065 (_mm512_i32gather_epi32): Ditto.
34066 (_mm512_mask_i32gather_epi32): Ditto.
34067 (_mm512_i32gather_epi64): Ditto.
34068 (_mm512_mask_i32gather_epi64): Ditto.
34069 (_mm512_i64gather_epi32): Ditto.
34070 (_mm512_mask_i64gather_epi32): Ditto.
34071 (_mm512_i64gather_epi64): Ditto.
34072 (_mm512_mask_i64gather_epi64): Ditto.
34073 (_mm512_i32scatter_ps): Change __addr type to void*.
34074 (_mm512_mask_i32scatter_ps): Ditto.
34075 (_mm512_i32scatter_pd): Ditto.
34076 (_mm512_mask_i32scatter_pd): Ditto.
34077 (_mm512_i64scatter_ps): Ditto.
34078 (_mm512_mask_i64scatter_ps): Ditto.
34079 (_mm512_i64scatter_pd): Ditto.
34080 (_mm512_mask_i64scatter_pd): Ditto.
34081 (_mm512_i32scatter_epi32): Ditto.
34082 (_mm512_mask_i32scatter_epi32): Ditto.
34083 (_mm512_i32scatter_epi64): Ditto.
34084 (_mm512_mask_i32scatter_epi64): Ditto.
34085 (_mm512_i64scatter_epi32): Ditto.
34086 (_mm512_mask_i64scatter_epi32): Ditto.
34087 (_mm512_i64scatter_epi64): Ditto.
34088 (_mm512_mask_i64scatter_epi64): Ditto.
34089 * config/i386/avx512pfintrin.h
34090 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34091 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34092 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34093 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34094 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34095 (_mm512_prefetch_i32scatter_ps): Ditto.
34096 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34097 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34098 (_mm512_prefetch_i64scatter_pd): Ditto.
34099 (_mm512_prefetch_i64scatter_ps): Ditto.
34100 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34101 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34102 * config/i386/avx512vlintrin.h
34103 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34104 (_mm_mmask_i32gather_ps): Ditto.
34105 (_mm256_mmask_i32gather_pd): Ditto.
34106 (_mm_mmask_i32gather_pd): Ditto.
34107 (_mm256_mmask_i64gather_ps): Ditto.
34108 (_mm_mmask_i64gather_ps): Ditto.
34109 (_mm256_mmask_i64gather_pd): Ditto.
34110 (_mm_mmask_i64gather_pd): Ditto.
34111 (_mm256_mmask_i32gather_epi32): Ditto.
34112 (_mm_mmask_i32gather_epi32): Ditto.
34113 (_mm256_mmask_i32gather_epi64): Ditto.
34114 (_mm_mmask_i32gather_epi64): Ditto.
34115 (_mm256_mmask_i64gather_epi32): Ditto.
34116 (_mm_mmask_i64gather_epi32): Ditto.
34117 (_mm256_mmask_i64gather_epi64): Ditto.
34118 (_mm_mmask_i64gather_epi64): Ditto.
34119 (_mm256_i32scatter_ps): Change __addr type to void*.
34120 (_mm256_mask_i32scatter_ps): Ditto.
34121 (_mm_i32scatter_ps): Ditto.
34122 (_mm_mask_i32scatter_ps): Ditto.
34123 (_mm256_i32scatter_pd): Ditto.
34124 (_mm256_mask_i32scatter_pd): Ditto.
34125 (_mm_i32scatter_pd): Ditto.
34126 (_mm_mask_i32scatter_pd): Ditto.
34127 (_mm256_i64scatter_ps): Ditto.
34128 (_mm256_mask_i64scatter_ps): Ditto.
34129 (_mm_i64scatter_ps): Ditto.
34130 (_mm_mask_i64scatter_ps): Ditto.
34131 (_mm256_i64scatter_pd): Ditto.
34132 (_mm256_mask_i64scatter_pd): Ditto.
34133 (_mm_i64scatter_pd): Ditto.
34134 (_mm_mask_i64scatter_pd): Ditto.
34135 (_mm256_i32scatter_epi32): Ditto.
34136 (_mm256_mask_i32scatter_epi32): Ditto.
34137 (_mm_i32scatter_epi32): Ditto.
34138 (_mm_mask_i32scatter_epi32): Ditto.
34139 (_mm256_i32scatter_epi64): Ditto.
34140 (_mm256_mask_i32scatter_epi64): Ditto.
34141 (_mm_i32scatter_epi64): Ditto.
34142 (_mm_mask_i32scatter_epi64): Ditto.
34143 (_mm256_i64scatter_epi32): Ditto.
34144 (_mm256_mask_i64scatter_epi32): Ditto.
34145 (_mm_i64scatter_epi32): Ditto.
34146 (_mm_mask_i64scatter_epi32): Ditto.
34147 (_mm256_i64scatter_epi64): Ditto.
34148 (_mm256_mask_i64scatter_epi64): Ditto.
34149 (_mm_i64scatter_epi64): Ditto.
34150 (_mm_mask_i64scatter_epi64): Ditto.
34151 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34152 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34153 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34154 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34155 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34156 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34157 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34158 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34159 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34160 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34161 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34162 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34163 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34164 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34165 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34166 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34167 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34168 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34169 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34170 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34171 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34172 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34173 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34174 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34175 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34176 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34177 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34178 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34179 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34180 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34181 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34182 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34183 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34184 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34185 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34186 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34187 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34188 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34189 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34190 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34191 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34192 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34193 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34194 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34195 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34196 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34197 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34198 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34199 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34200 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34201 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34202 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34203 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34204 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34205 definitions accordingly.
34207 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34208 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34211 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34214 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34217 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34218 hard regs for inner regno.
34220 2017-01-17 Martin Liska <mliska@suse.cz>
34223 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34224 assumption and add comment.
34226 2017-01-17 Nathan Sidwell <nathan@acm.org>
34228 * ipa-visibility.c (localize_node): New function, broken out of ...
34229 (function_and_variable_visibility): ... here. Call it.
34231 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34233 PR middle-end/77445
34234 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34235 correctly set frequency of oudgoing edge.
34236 (duplicate_thread_path): Fix profile updating.
34238 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34241 * configure.ac: Add GCC_BASE_VER.
34242 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34243 version from BASE-VER file.
34244 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34245 (gcc.o): Depend on $(BASEVER).
34246 * common.opt (dumpfullversion): New option.
34247 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34248 * doc/invoke.texi: Document -dumpfullversion.
34249 * doc/install.texi: Document --with-gcc-major-version-only.
34250 * configure: Regenerated.
34252 2017-01-17 Richard Biener <rguenther@suse.de>
34254 PR tree-optimization/71433
34255 * tree-vrp.c (register_new_assert_for): Merge same asserts
34256 on all incoming edges.
34257 (process_assert_insertions_for): Handle insertions at the
34260 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34262 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34263 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34265 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34268 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34271 2017-01-17 Alan Modra <amodra@gmail.com>
34274 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34275 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34276 symbolic stack limit when pic.
34278 2017-01-16 Martin Sebor <msebor@redhat.com>
34280 PR tree-optimization/78608
34281 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34283 2017-01-16 Jeff Law <law@redhat.com>
34286 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34287 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34288 for several include directories that may be relative to sysroot.
34289 * config/i386/x-mingw32 (gplus_includedir): Define.
34290 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34291 (native_system_includedir): Likewise.
34292 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34293 override if TARGET_SYSTEM_ROOT is defined.
34294 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34296 PR tree-optimization/79090
34297 PR tree-optimization/33562
34298 PR tree-optimization/61912
34299 PR tree-optimization/77485
34300 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34301 and computed trims into the dump file.
34303 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34305 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34307 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34310 * gimplify.c (gimplify_init_constructor): If want_value and
34311 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34315 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34316 sequence. Formatting fixes.
34317 (doloop_optimize): Formatting fixes.
34320 * gcc.c (debug_level_greater_than_spec_func): New function.
34321 (static_spec_functions): Add debug-level-gt spec function.
34322 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34324 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34325 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34326 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34327 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34328 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34329 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34331 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34333 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34334 QImode fixups to general and mask registers only.
34336 2017-01-16 Carl Love <cel@us.ibm.com>
34338 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34339 for built-in functions
34340 vector signed char vec_nabs (vector signed char)
34341 vector signed short vec_nabs (vector signed short)
34342 vector signed int vec_nabs (vector signed int)
34343 vector signed long long vec_nabs (vector signed long long)
34344 vector float vec_nabs (vector float)
34345 vector double vec_nabs (vector double)
34346 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34348 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34349 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34350 * doc/extend.texi: Update the documentation file for the new built-in
34353 2017-01-16 Martin Sebor <msebor@redhat.com>
34355 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34358 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34360 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34361 UNSPEC_VSX__XXSPLTD to require special splat handling.
34363 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34366 * system.h: Poison strndup.
34368 2017-01-16 Alan Modra <amodra@gmail.com>
34371 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34374 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34376 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34378 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34380 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34381 call recog here. Assert that INSN_CODE (insn) is non-negative.
34383 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34386 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34388 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34389 in the currently scheduled RTL fragment.
34391 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34393 PR rtl-optimization/78751
34394 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34397 2017-01-14 Jeff Law <law@redhat.com>
34399 PR tree-optimization/79090
34400 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34401 variable length stores.
34402 (compute_trims): Delete dead assignment to *trim_tail.
34403 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34406 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34408 PR rtl-optimization/78626
34409 PR rtl-optimization/78727
34410 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34411 of a block, and split such blocks after everything else is finished.
34413 2017-01-14 Alan Modra <amodra@gmail.com>
34416 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34417 target legitimate_combined_insn.
34418 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34419 (rs6000_legitimate_combined_insn): New function.
34420 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34422 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34423 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34424 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34426 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34428 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34430 2017-01-13 Jeff Law <law@redhat.com>
34432 PR tree-optimization/33562
34433 PR tree-optimization/61912
34434 PR tree-optimization/77485
34435 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34437 (delete_dead_assignment): Likewise.
34438 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34439 statement to delete_dead_call and delete_dead_assignment.
34441 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34444 * substring-locations.c (format_warning_va): Strengthen case 1 so
34445 that both endpoints of the substring must be within the format
34446 range for just the substring to be printed.
34448 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34450 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34451 * config/i386/i386.c (ix86_target_string): Add missing options
34452 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34453 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34454 flags_other and ix86_target_other to flags2_other. Display unknown
34456 (ix86_valid_target_attribute_inner_p): Add missing options and
34457 reorder options by implied ISAs, as in ix86_target_string.
34459 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34461 * hash-table.h (hash_table::too_empty_p): New function.
34462 (hash_table::expand): Use it.
34463 (hash_table::traverse): Likewise.
34464 (hash_table::empty_slot): Use sizeof (value_type) instead of
34465 sizeof (PTR) to convert bytes to elements. Shrink the table
34466 if the current size is excessive for the current number of
34469 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34471 * ira-costs.c (record_reg_classes): Break from the inner loop
34472 early once alt_fail is known to be true. Update outer loop
34473 handling accordingly.
34475 2017-01-13 Jeff Law <law@redhat.com>
34477 * tree-ssa-dse.c (decrement_count): New function.
34478 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34479 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34480 when we know the partially dead statement is a mem* function.
34482 PR tree-optimization/61912
34483 PR tree-optimization/77485
34484 * tree-ssa-dse.c: Include expr.h.
34485 (maybe_trim_constructor_store): New function.
34486 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34488 PR tree-optimization/33562
34489 PR tree-optimization/61912
34490 PR tree-optimization/77485
34491 * doc/invoke.texi: Document new dse-max-object-size param.
34492 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34493 * tree-ssa-dse.c: Include params.h.
34494 (dse_store_status): New enum.
34495 (initialize_ao_ref_for_dse): New, partially extracted from
34497 (valid_ao_ref_for_dse, normalize_ref): New.
34498 (setup_live_bytes_from_ref, compute_trims): Likewise.
34499 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34500 (maybe_trim_partially_dead_store): Likewise.
34501 (maybe_trim_complex_store): Likewise.
34502 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34503 Track what bytes live from the original store. Return tri-state
34504 for dead, partially dead or live.
34505 (dse_dom_walker): Add constructor, destructor and new private members.
34506 (delete_dead_call, delete_dead_assignment): New extracted from
34508 (dse_optimize_stmt): Make a member of dse_dom_walker.
34509 Use initialize_ao_ref_for_dse.
34511 PR tree-optimization/33562
34512 PR tree-optimization/61912
34513 PR tree-optimization/77485
34514 * sbitmap.h (bitmap_count_bits): Prototype.
34515 (bitmap_clear_range, bitmap_set_range): Likewise.
34516 * sbitmap.c (bitmap_clear_range): New function.
34517 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34519 2017-01-13 Martin Liska <mliska@suse.cz>
34522 * function.c (set_cfun): Add new argument force.
34523 * function.h (set_cfun): Likewise.
34524 * ipa-inline-transform.c (inline_call): Use the function when
34525 strict alising from is dropped for function we inline to.
34527 2017-01-13 Richard Biener <rguenther@suse.de>
34529 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34530 for dumping GIMPLE INTEGER_CSTs.
34532 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34534 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34535 to 201112L since C++17.
34537 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34540 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34541 if -fsanitize=kernel-address is present.
34543 2017-01-13 Richard Biener <rguenther@suse.de>
34545 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34546 as _Literal ( type ) number in case usual suffixes do not
34547 preserve all information.
34549 2017-01-13 Richard Biener <rguenther@suse.de>
34551 PR tree-optimization/77283
34552 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34553 and ssa-iterators.h.
34554 (is_feasible_trace): Implement a cost model based on joiner
34557 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34560 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34561 char or short to __float128/_Float128 directly.
34563 2017-01-12 Martin Sebor <msebor@redhat.com>
34565 to -Wformat-overflow.
34566 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34567 (min_bytes_remaining): Same.
34568 (get_string_length): Same.
34569 (format_string): Same.
34570 (format_directive): Same.
34572 (pass_sprintf_length::handle_gimple_call): Same.
34574 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34576 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34577 info.nowrite calls with no lhs that can't throw. Return bool
34578 whether gsi_remove has been called or not.
34579 (pass_sprintf_length::handle_gimple_call): Return bool whether
34580 try_substitute_return_value called gsi_remove. Formatting fix.
34581 (pass_sprintf_length::execute): Don't use gsi_remove if
34582 handle_gimple_call returned true.
34585 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34586 be removed due to side-effects, don't remove following barrier nor
34587 turn the successor edge into fallthru edge.
34589 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34592 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34593 element-reversing loads and stores as not swappable.
34595 2017-01-12 Nathan Sidwell <nathan@acm.org>
34596 Nicolai Stange <nicstange@gmail.com>
34598 * combine.c (try_combine): Don't ignore result of overlap checking
34599 loop. Combine overlap & asm check into single loop.
34601 2017-01-12 Richard Biener <rguenther@suse.de>
34603 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34604 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34606 2017-01-12 Richard Biener <rguenther@suse.de>
34608 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34609 and TS_TARGET_OPTION directly derive from TS_BASE.
34610 * tree-core.h (tree_optimization_option): Derive from tree_base.
34611 (tree_target_option): Likewise.
34613 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34615 * config/i386/i386.c (memory_address_length): Increase len
34616 only when rip_relative_addr_p returns false.
34618 2017-01-11 Julia Koval <julia.koval@intel.com>
34620 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34621 (OPTION_MASK_ISA_SGX_SET): New.
34622 (ix86_handle_option): Handle OPT_msgx.
34623 * config.gcc: Added sgxintrin.h.
34624 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34625 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34626 * config/i386/i386.c (ix86_target_string): Add -msgx.
34628 (ix86_option_override_internal): Handle new options.
34629 (ix86_valid_target_attribute_inner_p): Add sgx.
34630 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34631 * config/i386/i386.opt: Add msgx.
34632 * config/i386/sgxintrin.h: New file.
34633 * config/i386/x86intrin.h: Add sgxintrin.h.
34635 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34638 * fold-const.c (maybe_nonzero_address): Return 1 for function
34640 (tree_single_nonzero_warnv_p): Don't handle function local objects
34644 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34647 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34650 * opts.c: Include "spellcheck.h"
34651 (struct string_fragment): New struct.
34652 (struct edit_distance_traits<const string_fragment &>): New
34654 (get_closest_sanitizer_option): New function.
34655 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34657 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34659 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34661 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34662 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34663 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34664 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34665 for initial die_offset if dwarf_split_debug_info.
34666 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34667 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34669 (output_skeleton_debug_sections): Formatting fix. Use
34670 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34671 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34673 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34675 * config/arm/cortex-a53.md: Add bypasses for
34676 cortex_a53_r2f_cvt.
34677 (cortex_a53_r2f): Only use for transfers.
34678 (cortex_a53_f2r): Likewise.
34679 (cortex_a53_r2f_cvt): Add reservation for conversions.
34680 (cortex_a53_f2r_cvt): Likewise.
34682 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34684 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34685 to all inlined functions, change static to extern.
34687 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34690 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34692 (arm_assemble_integer): Likewise.
34694 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34696 * config.gcc: Use new awk script to check CPU, FPU and architecture
34697 parameters for --with-... options.
34698 * config/arm/parsecpu.awk: New file
34699 * config/arm/arm-cpus.in: New file.
34700 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34702 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34704 * config/arm/t-arm: Update dependency rules.
34705 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34706 of processing .def files.
34707 * config/arm/genopt.sh: Deleted.
34708 * config/arm/gentune.sh: Deleted.
34709 * config/arm/arm-cores.def: Deleted.
34710 * config/arm/arm-arches.def: Deleted.
34711 * config/arm/arm-fpus.def: Deleted.
34712 * config/arm/arm-tune.md: Regenerated.
34713 * config/arm/arm-tables.opt: Regenerated.
34714 * config/arm/arm-cpu.h: New generated file.
34715 * config/arm/arm-cpu-data.h: New generated file.
34716 * config/arm/arm-cpu-cdata.h: New generated file.
34718 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34721 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34723 (input_varpool_node): Unpack dynamically_initialized bit.
34725 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34727 PR rtl-optimization/79032
34728 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34729 the alignment of the adjusted memory reference against that of MODE,
34730 instead of the alignment of the original memory reference.
34732 2017-01-11 Martin Jambor <mjambor@suse.cz>
34734 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34736 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34737 decorated functions.
34739 2017-01-11 Richard Biener <rguenther@suse.de>
34741 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34742 set range/nonnull info for PHI results. Do not set it on
34743 stmts marked for removal.
34745 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34747 * expr.c (store_field): In the bitfield case, fetch the return value
34748 from the registers before applying a single big-endian adjustment.
34749 Always do a final load for a BLKmode value not larger than a word.
34751 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34754 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34755 that we correctly handle column numbers greater than
34756 LINE_MAP_MAX_COLUMN_NUMBER.
34758 2017-01-10 Martin Sebor <msebor@redhat.com>
34760 PR middle-end/78245
34761 * gimple-ssa-sprintf.c (get_destination_size): Call
34762 {init,fini}object_sizes.
34763 * tree-object-size.c (addr_object_size): Adjust.
34764 (pass_through_call): Adjust.
34765 (pass_object_sizes::execute): Adjust.
34766 * tree-object-size.h (fini_object_sizes): Declare.
34768 2017-01-10 Martin Sebor <msebor@redhat.com>
34770 PR tree-optimization/78775
34771 * builtins.c (get_size_range): Move...
34772 * calls.c: ...to here.
34773 (alloc_max_size): Accept zero argument.
34774 (operand_signed_p): Remove.
34775 (maybe_warn_alloc_args_overflow): Call get_size_range.
34776 * calls.h (get_size_range): Declare.
34778 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34780 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34781 from TI's devices.csv file as of September 2016.
34782 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34784 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34786 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34787 * doc/invoke.texi: Likewise.
34788 * doc/md.texi: Likewise.
34789 * doc/objc.texi: Likewise.
34791 2017-01-10 Joshua Conner <joshconner@google.com>
34793 * config/arm/fuchsia-elf.h: New file.
34794 * config/fuchsia.h: New file.
34795 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34796 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34798 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34800 2016-01-10 Richard Biener <rguenther@suse.de>
34802 PR tree-optimization/79034
34803 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34804 Propagate out degenerate PHIs in the joiner.
34806 2017-01-10 Martin Liska <mliska@suse.cz>
34808 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34809 (sort_congruence_classes_by_decl_uid): Likewise.
34810 (sort_congruence_class_groups_by_decl_uid): Likewise.
34811 (sem_item_optimizer::merge_classes): Sort class, groups in these
34812 classes and members in the groups by DECL_UID of declarations.
34813 This would make merge operations stable.
34815 2017-01-10 Martin Liska <mliska@suse.cz>
34817 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34818 usage of m_classes_vec.
34819 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34820 (sem_item_optimizer::get_group_by_hash): Likewise.
34821 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34822 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34823 (sem_item_optimizer::verify_classes): Likewise.
34824 (sem_item_optimizer::process_cong_reduction): Likewise.
34825 (sem_item_optimizer::dump_cong_classes): Likewise.
34826 (sem_item_optimizer::merge_classes): Likewise.
34827 * ipa-icf.h (congruence_class_hash): Rename from
34828 congruence_class_group_hash. Remove declaration of m_classes_vec.
34830 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34832 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34833 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34834 * config.gcc: Add avx512vpopcntdqintrin.h.
34835 * config/i386/avx512vpopcntdqintrin.h: New.
34836 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34837 * config/i386/i386-builtin-types.def: Add new types.
34838 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34839 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34840 __builtin_ia32_vpopcountq_v8di_mask): New.
34841 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34842 __AVX512VPOPCNTDQ__.
34843 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34844 (PTA_AVX512VPOPCNTDQ): Define.
34845 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34846 TARGET_AVX512VPOPCNTDQ_P): Define.
34847 * config/i386/i386.opt: Add mavx512vpopcntdq.
34848 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34849 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34851 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34853 PR middle-end/77484
34854 * predict.def (PRED_CALL): Set to 67.
34856 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34858 * expr.c (store_field): In the bitfield case, if the value comes from
34859 a function call and is of an aggregate type returned in registers, do
34860 not modify the field mode; extract the value in all cases if the mode
34861 is BLKmode and the size is not larger than a word.
34863 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34866 * config/i386/cpuid.h: Fix undefined behavior.
34868 2017-01-04 Jeff Law <law@redhat.com>
34870 PR tree-optimization/79007
34871 PR tree-optimization/67955
34872 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34873 conservative for pt.null when flag_non_call_exceptions is on.
34875 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34877 PR translation/79019
34878 PR translation/79020
34879 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34880 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34881 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34883 * config/avr/avr.opt (maccumulate-args): Likewise.
34884 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34885 * common.opt (freport-bug): Likewise.
34886 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34887 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34888 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34889 translatable string.
34890 * config/i386/i386.c (function_value_32): Likewise.
34891 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34892 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34894 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34895 * common/config/msp430/msp430-common.c (msp430_handle_option):
34897 * symtab.c (symtab_node::verify_base): Likewise.
34898 * opts.c (set_debug_level): Likewise.
34899 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34900 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34901 missing whitespace to translatable strings.
34902 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34903 * config/sh/superh.h: Likewise.
34904 * config/i386/xopintrin.h: Likewise.
34905 * config/i386/znver1.md: Likewise.
34906 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34907 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34908 * double-int.h (struct double_int): Likewise.
34909 * double-int.c (div_and_round_double): Likewise.
34910 * wide-int.cc: Likewise.
34911 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34912 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34913 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34914 (crossjumps_occurred): ... this.
34915 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34918 PR tree-optimization/78899
34919 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34920 returning bool return struct loop *, NULL for failure and the new
34922 (versionable_outer_loop_p): Don't version outer loop if it has
34923 dont_vectorized bit set.
34924 (tree_if_conversion): When versioning outer loop, ensure
34925 tree_if_conversion is performed also on the inner loop of the
34926 non-vectorizable outer loop copy.
34927 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34928 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34929 prevent vectorization of it.
34930 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34931 the outer loop vectorization of the non-scalar version is attempted
34932 before vectorization of the inner loop in scalar version. If
34933 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34934 vectorization of its inner loop.
34935 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34936 has 2 inner loops, rename also on edges from bb whose single pred
34937 is outer_loop->header. Fix typo in function comment.
34939 2017-01-09 Martin Sebor <msebor@redhat.com>
34942 * asan.c (asan_emit_stack_protection): Increase local buffer size
34943 to avoid snprintf truncation warning.
34945 2017-01-09 Andrew Pinski <apinski@cavium.com>
34947 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34948 to reference thunderx2t99 for the tuning structure
34949 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34951 (thunderx2t99_extra_costs): This.
34952 * config/aarch64/aarch64-tune.md: Regenerate.
34953 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34954 (vulcan_addrcost_table): This.
34955 (vulcan_regmove_cost): Rename to ...
34956 (thunderx2t99_regmove_cost): This.
34957 (vulcan_vector_cost): Rename to ...
34958 (thunderx2t99_vector_cost): this.
34959 (vulcan_branch_cost): Rename to ...
34960 (thunderx2t99_branch_cost): This.
34961 (vulcan_tunings): Rename to ...
34962 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34963 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34965 2017-01-09 Martin Jambor <mjambor@suse.cz>
34969 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34970 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34971 (propagate_vr_accross_jump_function): Use the above function for all
34972 value range computations for pass-through jump functions and type
34973 converasion from explicit value range values.
34974 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34975 parameters from TYPE_ARG_TYPES.
34976 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34977 (ipa_write_node_info): Stream type of the actual argument.
34978 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34980 2017-01-09 Martin Liska <mliska@suse.cz>
34983 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34984 (lookup_compiler): Do not show error message with have_E.
34986 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34988 PR tree-optimization/78938
34989 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34990 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34991 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34992 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34995 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34997 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35000 2017-01-09 Richard Biener <rguenther@suse.de>
35002 PR tree-optimization/78997
35003 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35004 name condition properly.
35006 2017-01-09 Richard Biener <rguenther@suse.de>
35009 * dwarf2out.c (is_cxx): New overload with context.
35010 (is_naming_typedef_decl): Use it.
35012 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35014 * invoke.texi (Option Summary): Correct spacing in option lists
35015 and add line breaks to fix over-long lines.
35017 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35019 PR middle-end/17660
35021 * extend.texi (Common Variable Attributes): Add xref to GCC
35022 Internals manual to explain mode attribute keywords.
35024 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35027 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35028 and Preprocessor Options.
35029 (Options for Linking): Document -pthread here....
35030 (RS/6000 and PowerPC Options): ...not here.
35031 (Solaris 2 Options): ...or here.
35032 * doc/cppopts.texi: Document -pthread.
35034 2017-01-08 Martin Sebor <msebor@redhat.com>
35036 PR middle-end/77708
35037 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35038 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35039 New member functions.
35040 (format_directive): Used them.
35042 (pass_sprintf_length::handle_gimple_call): Same.
35043 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35044 to avoid truncation for any argument.
35045 (extract_affine_mul): Same.
35046 * tree.c (get_file_function_name): Same.
35048 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35050 PR middle-end/77484
35051 * predict.def (PRED_INDIR_CALL): Set to 86.
35053 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35055 PR preprocessor/54124
35056 * doc/cppopts.texi: Reformat -d subtable to list the full name
35057 of the options. Add cross-reference to the docs for the general
35058 compiler -d options.
35059 * doc/invoke.texi (Developer Options): Add cross-reference to the
35060 preprocessor-specific -d option documentation.
35062 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35064 PR preprocessor/13498
35065 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35066 redudant material, and reflect new command-line options.
35067 (System Headers): Likewise.
35069 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35071 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35072 -isystem, and -idirafter. Copy-edit.
35073 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35074 default for -ftrack-macro-expansion. Delete obsolete and
35075 badly-formatted implementation details about -fdebug-cpp output.
35076 * doc/cppwarnopts.texi: Copy-edit.
35078 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35081 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35082 that the transition from a max line width >= 1<<10 to narrower
35083 lines works correctly.
35085 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35087 * doc/options.texi (PerFunction): New.
35088 * opt-functions.awk (switch_flags): Map both Optimization and
35089 PerFunction to CL_OPTIMIZATION.
35090 * opth-gen.awk: Test for PerFunction flag along with
35092 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35093 it only when the latter is present. Skip those that don't in
35094 the hash function generator.
35095 * common.opt (fvar-tracking): Mark as PerFunction instead of
35097 (fvar-tracking-assignments): Likewise.
35098 (fvar-tracking-assignments-toggle): Likewise.
35099 (fvar-tracking-uninit): Likewise.
35101 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35103 PR translation/79018
35104 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35107 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35110 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35111 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35112 TARGET_LONG_JUMP_TABLE_OFFSETS.
35113 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35114 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35115 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35116 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35117 * config/m68k/m68k.md (tablejump expander): Likewise.
35118 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35119 TARGET_LONG_JUMP_TABLE_OFFSETS.
35120 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35121 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35123 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35124 David Holsgrove <david.holsgrove@xilinx.com>
35126 * common/config/microblaze/microblaze-common.c
35127 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35128 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35130 * config/microblaze/microblaze.c (microblaze_must_save_register)
35131 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35133 (microblaze_eh_return): New function.
35134 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35135 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35136 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35137 * config/microblaze/microblaze.md (eh_return): New pattern.
35139 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35141 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35142 GCC_DIAGNOSTIC_STRINGIFY): Define.
35144 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35146 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35148 * config/arm/arm.md (<mcrr>): New.
35150 * config/arm/arm.c (arm_arch5te): New.
35151 (arm_option_override): Set arm_arch5te.
35152 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35154 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35155 (arm_mcrr_qualifiers): ... this. New.
35156 (MRRC_QUALIFIERS): Define to...
35157 (arm_mrrc_qualifiers): ... this. New.
35158 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35160 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35161 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35162 (MRRCI, mrrc, MRRC): New.
35163 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35164 VUNSPEC_MRRC2): New.
35166 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35168 * config/arm/arm.md (<mcr>): New.
35170 * config/arm/arm.c (arm_coproc_builtin_available): Add
35171 support for mcr, mrc, mcr2 and mrc2.
35172 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35173 (arm_mcr_qualifiers): ... this. New.
35174 (MRC_QUALIFIERS): Define to ...
35175 (arm_mrc_qualifiers): ... this. New.
35176 (MCR_QUALIFIERS): Define to ...
35177 (arm_mcr_qualifiers): ... this. New.
35178 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35180 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35181 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35182 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35183 VUNSPEC_MRC2): New.
35185 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35187 * config/arm/arm.md (*ldc): New.
35191 * config/arm/arm.c (arm_coproc_builtin_available): Add
35192 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35193 (arm_coproc_ldc_stc_legitimate_address): New.
35194 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35195 'qualifier_const_pointer'.
35196 (LDC_QUALIFIERS): Define to...
35197 (arm_ldc_qualifiers): ... this. New.
35198 (STC_QUALIFIERS): Define to...
35199 (arm_stc_qualifiers): ... this. New.
35200 * config/arm/arm-protos.h
35201 (arm_coproc_ldc_stc_legitimate_address): New.
35202 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35203 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35204 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35205 stc2, stcl, stc2l): New.
35206 * config/arm/constraints.md (Uz): New.
35207 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35208 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35209 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35210 VUNSPEC_STC2L): New.
35212 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35214 * config/arm/arm.md (<cdp>): New.
35215 * config/arm/arm.c (neon_const_bounds): Rename this ...
35216 (arm_const_bounds): ... this.
35217 (arm_coproc_builtin_available): New.
35218 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35219 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35220 (CDP_QUALIFIERS): Define to...
35221 (arm_cdp_qualifiers): ... this. New.
35223 (arm_expand_builtin_args): Add case for 6 arguments.
35224 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35225 (arm_const_bounds): ... this.
35226 (arm_coproc_builtin_available): New.
35227 * config/arm/arm_acle.h (__arm_cdp): New.
35229 * config/arm/arm_acle_builtins.def (cdp): New.
35231 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35232 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35233 'arm_const_bounds'.
35234 * config/arm/types.md (coproc): New.
35235 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35236 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35237 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35238 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35240 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35242 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35243 (UBINOP_QUALIFIERS): New.
35245 (acle_builtin_data): New. Change comment.
35246 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35247 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35248 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35249 arm_acle_builtins.def.
35250 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35251 (arm_init_acle_builtins): New.
35252 (CRC32_BUILTIN): Remove.
35253 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35254 crc32cb, crc32ch and crc32cw.
35255 (arm_init_crc32_builtins): Remove.
35256 (arm_init_builtins): Use arm_init_acle_builtins rather
35257 than arm_init_crc32_builtins.
35258 (arm_expand_acle_builtin): New.
35259 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35260 * config/arm/arm_acle_builtins.def: New.
35262 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35264 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35265 (arm_builtin_datum): ... this.
35266 (arm_init_neon_builtin): Rename to ...
35267 (arm_init_builtin): ... this. Add a new parameters PREFIX
35268 and USE_SIG_IN_NAME.
35269 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35270 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35271 'arm_builtin_datum'.
35272 (arm_init_vfp_builtins): Likewise.
35273 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35274 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35275 (arm_expand_neon_args): Rename to ...
35276 (arm_expand_builtin_args): ... this. Rename builtin_arg
35277 enum values and differentiate between ARG_BUILTIN_MEMORY
35278 and ARG_BUILTIN_NEON_MEMORY.
35279 (arm_expand_neon_builtin_1): Rename to ...
35280 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35281 values, arm_expand_builtin_args and add bool parameter NEON.
35282 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35283 (arm_expand_vfp_builtin): Likewise.
35284 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35286 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35288 PR middle-end/77484
35289 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35290 * predict.c (tree_estimate_probability_bb): Reverse direction of
35291 polymorphic call predictor.
35293 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35295 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35296 (determine_pass_name_match): ...this new function and...
35297 (should_skip_pass_p): ...this new function.
35299 2017-01-06 Nathan Sidwell <nathan@acm.org>
35301 * ipa-visibility.c (function_and_variable_visibility): Reformat
35302 comments and long lines. Remove extrneous if.
35303 * symtab.c (symtab_node::make_decl_local): Fix code format.
35304 (symtab_node::set_section_for_node): Fix comment typo.
35306 2017-01-06 Martin Liska <mliska@suse.cz>
35309 * lra-constraints.c: Rename invariant to lra_invariant.
35310 * predict.c (set_even_probabilities): Initialize e to NULL.
35312 2017-01-05 Martin Sebor <msebor@redhat.com>
35314 PR tree-optimization/78910
35315 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35316 (format_integer): Correct off-by-one error in the handling
35317 of precision with negative numbers in signed conversions..
35319 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35321 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35323 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35325 PR tree-optimization/71016
35326 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35327 factor_out_conditional_conversion. Formatting fix.
35328 (factor_out_conditional_conversion): Add cond_stmt argument.
35329 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35330 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35333 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35335 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35336 read-rtl-function.o, and selftest-rtl.o.
35337 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35338 (selftest::aarch64_test_loading_full_dump): New function.
35339 (selftest::aarch64_run_selftests): New function.
35340 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35341 selftest::aarch64_run_selftests.
35342 * config/i386/i386.c
35343 (selftest::ix86_test_loading_dump_fragment_1): New function.
35344 (selftest::ix86_test_loading_call_insn): New function.
35345 (selftest::ix86_test_loading_full_dump): New function.
35346 (selftest::ix86_test_loading_unspec): New function.
35347 (selftest::ix86_run_selftests): Call the new functions.
35348 * emit-rtl.c (maybe_set_max_label_num): New function.
35349 * emit-rtl.h (maybe_set_max_label_num): New decl.
35350 * function.c (instantiate_decls): Guard call to
35351 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35352 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35354 * gensupport.c (gen_reader::gen_reader): Pass "false"
35355 for new "compact" param of rtx_reader.
35356 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35357 rather than an empty string for NULL strings.
35358 * read-md.c: Potentially include config.h rather than bconfig.h.
35359 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35360 (have_error): New global, copied from errors.c.
35361 (md_reader::read_name): Rename to...
35362 (md_reader::read_name_1): ...this, adding "out_loc" param,
35363 and converting "missing name or number" to returning false, rather
35365 (md_reader::read_name): Reimplement in terms of read_name_1.
35366 (md_reader::read_name_or_nil): New function.
35367 (md_reader::read_string): Handle "(nil)" by returning NULL.
35368 (md_reader::md_reader): Add new param "compact".
35369 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35370 (md_reader::read_file): New method.
35371 * read-md.h (md_reader::md_reader): Add new param "compact".
35372 (md_reader::read_file): New method.
35373 (md_reader::is_compact): New accessor.
35374 (md_reader::read_name): Convert return type from void to file_location.
35375 (md_reader::read_name_or_nil): New decl.
35376 (md_reader::read_name_1): New decl.
35377 (md_reader::m_compact): New field.
35378 (noop_reader::noop_reader): Pass "false" for new "compact" param
35380 (rtx_reader::rtx_reader): Add new "compact" param.
35381 (rtx_reader::read_rtx_operand): Make virtual and convert return
35382 type from void to rtx.
35383 (rtx_reader::read_until): New decl.
35384 (rtx_reader::handle_any_trailing_information): New virtual function.
35385 (rtx_reader::postprocess): New virtual function.
35386 (rtx_reader::finalize_string): New virtual function.
35387 (rtx_reader::m_in_call_function_usage): New field.
35388 (rtx_reader::m_reuse_rtx_by_id): New field.
35389 * read-rtl-function.c: New file.
35390 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35391 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35392 (selftest::verify_three_block_rtl_cfg): New decl.
35393 * read-rtl-function.h: New file.
35394 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35395 For host, include function.h, memmodel.h, and emit-rtl.h.
35396 (one_time_initialization): New function.
35397 (struct compact_insn_name): New struct.
35398 (compact_insn_names): New array.
35399 (find_code): Handle insn codes in compact dumps.
35400 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35401 (bind_subst_iter_and_attr): Likewise.
35402 (add_condition_to_string): Likewise.
35403 (add_condition_to_rtx): Likewise.
35404 (apply_attribute_uses): Likewise.
35405 (add_current_iterators): Likewise.
35406 (apply_iterators): Likewise.
35407 (initialize_iterators): Guard usage of apply_subst_iterator with
35408 #ifdef GENERATOR_FILE.
35409 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35410 (md_reader::read_mapping): Likewise.
35411 (add_define_attr_for_define_subst): Likewise.
35412 (add_define_subst_attr): Likewise.
35413 (read_subst_mapping): Likewise.
35414 (check_code_iterator): Likewise.
35415 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35417 (one_time_initialization): New function.
35418 (rtx_reader::read_until): New method.
35419 (read_flags): New function.
35420 (parse_reg_note_name): New function.
35421 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35422 Handle reuse_rtx ids.
35423 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35424 Add parsing support for RTL dumps, mirroring the special-cases in
35425 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35426 values, and calling handle_any_trailing_information.
35427 (rtx_reader::read_rtx_operand): Convert return type from void
35428 to rtx, returning return_rtx. Handle case 'e'. Call
35429 finalize_string on XSTR and XTMPL fields.
35430 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35431 "(nil)" values were omitted. Call the postprocess vfunc on the
35433 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35434 class ctor. Initialize m_in_call_function_usage. Call
35435 one_time_initialization.
35436 * rtl-tests.c (selftest::test_uncond_jump): Call
35437 set_new_first_and_last_insn.
35438 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35439 * selftest-rtl.c: New file.
35440 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35441 (selftest::get_insn_by_uid): New decl.
35442 * selftest-run-tests.c (selftest::run_tests): Call
35443 read_rtl_function_c_tests.
35444 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35445 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35448 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35450 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35451 operands in a special way. Assert that pos+len <= mode precision.
35453 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35455 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35456 3 argument Alias with unlimited for the negative form.
35457 (fno-vect-cost-model): Removed.
35459 2017-01-05 Martin Liska <mliska@suse.cz>
35461 * hsa-gen.c (gen_hsa_divmod): New function.
35462 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35464 2017-01-05 Martin Liska <mliska@suse.cz>
35467 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35470 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35472 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35473 small constant length operands.
35475 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35477 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35478 between loop iterations.
35480 2017-01-05 Martin Liska <mliska@suse.cz>
35483 * gimplify.c (gimplify_decl_expr): Compare to
35484 asan_poisoned_variables instread of checking flags.
35485 (gimplify_target_expr): Likewise.
35486 (gimplify_expr): Likewise.
35487 (gimplify_function_tree): Conditionally initialize
35488 asan_poisoned_variables.
35490 2017-01-04 Jeff Law <law@redhat.com>
35492 PR tree-optimizatin/78812
35493 * rtl.h (contains_mem_rtx_p): Prototype.
35494 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35495 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35496 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35497 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35498 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35500 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35502 * input.c (assert_char_at_range): Default-initialize actual_range.
35504 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35506 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35507 to match the caller.
35509 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35511 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35512 insns after final jump in test to emit dummy move.
35514 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35516 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35517 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35519 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35521 * multiple_target.c (create_dispatcher_calls): Init e_next.
35522 * tree-ssa-loop-split.c (split_loop): Init border.
35523 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35526 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35531 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35532 (altivec_register_operand): Do not return true if the operand
35533 contains a SUBREG mixing SImode and SFmode.
35534 (vsx_register_operand): Likewise.
35535 (vsx_reg_sfsubreg_ok): New predicate.
35536 (vfloat_operand): Do not return true if the operand contains a
35537 SUBREG mixing SImode and SFmode.
35538 (vint_operand): Likewise.
35539 (vlogical_operand): Likewise.
35540 (gpc_reg_operand): Likewise.
35541 (int_reg_operand): Likewise.
35542 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35543 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35544 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35546 (rs6000_emit_move_si_sf_subreg): New helper function.
35547 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35548 fixup SUBREGs involving SImode and SFmode.
35549 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35550 numbers for the new peephole2 optimization.
35551 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35552 the GLIBC math library that do AND/IOR/XOR operations on single
35553 precision floating point.
35554 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35555 target macros to say whether we need to avoid SUBREGs mixing
35557 (TARGET_ALLOW_SF_SUBREG): Likewise.
35558 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35559 (UNSPEC_SI_FROM_SF): Likewise.
35560 (iorxor): Change spacing.
35561 (and_ior_xor): New iterator for AND, IOR, and XOR.
35562 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35563 (movdi_from_sf_zero_ext): Likewise.
35564 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35565 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35566 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35567 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35568 (fms<mode>4): Likewise.
35569 (fnma<mode>4): Likewise.
35570 (fnms<mode>4): Likewise.
35571 (nfma<mode>4): Likewise.
35572 (nfms<mode>4): Likewise.
35574 2017-01-04 Marek Polacek <polacek@redhat.com>
35577 * doc/invoke.texi: Document -Wpointer-compare.
35579 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35581 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35584 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35585 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35588 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35591 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35592 documentation of the powerpc_popcntb_ok attribute.
35593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35594 code to issue warning messages if a requested CPU configuration is
35595 not supported by the binary (assembler and loader) toolchain.
35596 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35597 made to define a built-in function that has been disabled.
35598 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35599 made to define a built-in function that has been disabled.
35600 (altivec_init_builtins): Add comment explaining why definition
35601 of the DST built-in functions is not preceded by an assertion
35602 check. Add assertions to prevent ICE if attempts are made to
35603 define an altivec predicate or an abs* built-in function that has
35605 (htm_init_builtins): Add comment explaining why definition of the
35606 htm built-in functions is not preceded by an assertion check.
35608 2017-01-04 Jeff Law <law@redhat.com>
35610 PR tree-optimizatin/67955
35611 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35612 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35613 the points-to solution does not include pt_null. Use DECL_PT_UID
35616 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35618 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35619 Use gen_int_mode instead of gen_lopwart for const_int operands.
35621 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35623 PR tree-optimization/71563
35624 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35625 out of range value - has low bits known to be zero.
35627 2017-01-04 Alan Modra <amodra@gmail.com>
35629 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35630 * configure: Regenerate.
35631 * config.in: Regenerate.
35633 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35636 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35637 a substring of the message, but strcmp with the whole message. Ifdef
35638 ENABLE_NLS, translate the message first using dgettext.
35640 2017-01-03 Jeff Law <law@redhat.com>
35642 PR tree-optimizatin/78856
35643 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35644 (mark_threaded_blocks): Remove code to truncate thread paths that
35645 cross multiple loop headers. Instead invalidate the cached loop
35646 iteration information and handle case of a thread path walking
35647 into an irreducible region.
35649 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35652 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35653 assertions. Add support for doing the signbit if the IEEE 128-bit
35654 floating point value is in a GPR.
35655 * config/rs6000/rs6000.md (Fsignbit): Delete.
35656 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35657 Update the length attribute if the value is in a GPR.
35658 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35659 the sign or zero extension instruction, since the value is always 0/1.
35660 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35663 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35664 extracting SImode to a GPR register so that we can generate a
35665 store, limit the vector to be in a traditional Altivec register
35666 for the vextuwrx instruction.
35668 2017-01-03 Ian Lance Taylor <iant@google.com>
35670 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35672 2017-01-03 Martin Sebor <msebor@redhat.com>
35674 PR tree-optimization/78696
35675 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35676 precision. Use MPFR for %f for greater fidelity. Correct handling
35678 (pass_sprintf_length::compute_format_length): Set width and precision
35679 specified by asrerisk to void_node for vararg functions.
35680 (try_substitute_return_value): Adjust dump output.
35682 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35684 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35686 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35688 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35689 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35690 -mlra/-mno-lra was passed to the compiler.
35692 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35694 PR rtl-optimization/65618
35695 * emit-rtl.c (try_split): Move initialization of "before" and
35696 "after" to just before the call to emit_insn_after_setloc.
35698 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35700 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35702 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35704 * dwarf2out.c (gen_enumeration_type_die): When
35705 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35707 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35709 PR tree-optimization/78965
35710 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35711 Change first argument from const call_info & to call_info &. For %n
35712 set info.nowrite to false.
35714 PR middle-end/78901
35715 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35716 possibly throwing calls.
35718 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35719 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35720 and fns handling, rather than in a separate case SSA_NAME.
35722 2017-01-02 Jeff Law <law@redhat.com>
35724 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35725 fixes for first_period and second_period variables.
35727 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35730 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35731 (*insvqi_1): New insn pattern.
35732 (*insvqi_1_mem_rex64): Ditto.
35733 (*insvqi_2): Ditto.
35734 (*insvqi_3): Rename from *insvqi.
35736 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35738 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35740 * doc/cfg.texi (Edges): Remove reference to Java.
35741 (Maintaining the CFG): Ditto.
35743 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35745 PR middle-end/77674
35746 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35747 transparent aliases.
35749 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35751 PR middle-end/77484
35752 * predict.def (PRED_CALL): Update hitrate.
35753 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35754 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35755 into direct/indirect/polymorphic variants.
35757 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35759 Update copyright years.
35761 * gcc.c (process_command): Update copyright notice dates.
35762 * gcov-dump.c (print_version): Ditto.
35763 * gcov.c (print_version): Ditto.
35764 * gcov-tool.c (print_version): Ditto.
35765 * gengtype.c (create_file): Ditto.
35766 * doc/cpp.texi: Bump @copying's copyright year.
35767 * doc/cppinternals.texi: Ditto.
35768 * doc/gcc.texi: Ditto.
35769 * doc/gccint.texi: Ditto.
35770 * doc/gcov.texi: Ditto.
35771 * doc/install.texi: Ditto.
35772 * doc/invoke.texi: Ditto.
35774 Copyright (C) 2017 Free Software Foundation, Inc.
35776 Copying and distribution of this file, with or without modification,
35777 are permitted in any medium without royalty provided the copyright
35778 notice and this notice are preserved.