1 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5 set in_cold_section_p to the partition containing loc_list->first.
6 When seeing loc_list->last_before_switch node, update secname and
7 perform range_across_switch second partition handling only after that.
10 * run-rtl-passes.c: Include debug.h.
11 (run_rtl_passes): Call debug_hooks->assembly_start.
12 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
16 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
17 OPTAB_DIRECT in calls to expand_simple_binop.
19 2017-11-20 David Malcolm <dmalcolm@redhat.com>
22 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
24 2017-11-20 David Malcolm <dmalcolm@redhat.com>
27 * spellcheck.h (best_match::blithely_get_best_candidate): New
30 2017-11-20 Jakub Jelinek <jakub@redhat.com>
32 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
33 with lower case letter.
35 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
37 * config/i386/i386.md (bswaphi2): New expander.
38 (*bswaphi2_movbe): New insn pattern.
39 (bswaphi -> rorhi pepehole2): New peephole pattern.
41 2017-11-20 Jeff Law <law@redhat.com>
43 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
44 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
45 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
46 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
47 gimple-ssa-evrp-analyze.h.
49 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
51 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
53 2017-11-20 Bin Cheng <bin.cheng@arm.com>
55 * tree-predcom.c: Add general comment on Store-Store chains.
56 (split_data_refs_to_components): Postpone clearing eliminate_store_p
58 (get_chain_last_ref_at): Rename into...
59 (get_chain_last_write_at): ...this.
60 (get_chain_last_write_before_load): New function.
61 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
62 CT_STORE_STORE when write reference is added.
63 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
64 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
65 (initialize_root_vars_store_elim_1): Ditto.
66 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
67 definition is created.
68 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
69 chain by replacing it with dominant stored value.
71 2017-11-20 Bin Cheng <bin.cheng@arm.com>
73 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
75 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
77 * vr-values.c (extract_range_from_binary_expr): Use a full range
80 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
82 * config/arm/arm.md (R4_REGNUM): Define constant.
83 (nonsecure_call_internal): Remove r4 clobber.
84 (nonsecure_call_value_internal): Likewise.
85 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
86 clobber and resequence match_operands.
87 (nonsecure_call_value_reg_thumb1_v5): Likewise.
88 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
89 (nonsecure_call_value_reg_thumb2): Likewise.
91 2017-11-20 Jakub Jelinek <jakub@redhat.com>
93 PR tree-optimization/78821
94 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
95 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
96 to the address of the base rather than the base itself.
97 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
98 (find_bswap_or_nop_finalize): New function.
99 (find_bswap_or_nop): Use it.
100 (bswap_replace): Return a tree rather than bool, change first
101 argument from gimple * to gimple_stmt_iterator, allow inserting
102 into an empty sequence, allow ins_stmt to be NULL - then emit
103 all stmts into gsi. Fix up MEM_REF address gimplification.
104 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
105 (struct store_immediate_info): Add N and INS_STMT non-static
107 (store_immediate_info::store_immediate_info): Initialize them
108 from newly added ctor args.
109 (merged_store_group::apply_stores): Formatting fixes. Sort by
111 (stmts_may_clobber_ref_p): For stores call also
112 refs_anti_dependent_p.
113 (gather_bswap_load_refs): New function.
114 (imm_store_chain_info::try_coalesce_bswap): New method.
115 (imm_store_chain_info::coalesce_immediate_stores): Use it.
116 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
117 (imm_store_chain_info::output_merged_store): Fail if number of
118 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
119 and NOP_EXPR rhs_code.
120 (pass_store_merging::process_store): Compute n and ins_stmt, if
121 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
122 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
124 (pass_store_merging::execute): Calculate dominators.
126 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
127 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
128 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
129 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
130 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
131 class pass_optimize_bswap, bswap_replace,
132 pass_optimize_bswap::execute): Moved to ...
133 * gimple-ssa-store-merging.c: ... this file.
134 Include optabs-tree.h.
135 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
136 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
137 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
138 anonymous namespace, remove static keywords.
139 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
140 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
142 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
145 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
147 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
149 * vec.h (debug_helper): New function.
150 (DEFINE_DEBUG_VEC): New macro.
151 * hash-set.h (debug_helper): New function.
152 (DEFINE_DEBUG_HASH_SET): New macro.
153 * cfg.c (debug_slim (edge)): New function.
154 Call DEFINE_DEBUG_VEC for edges.
155 Call DEFINE_DEBUG_HASH_SET for edges.
156 * cfghooks.c (debug_slim (basic_block)): New function.
157 Call DEFINE_DEBUG_VEC for basic blocks.
158 Call DEFINE_DEBUG_HASH_SET for basic blocks.
159 * print-tree.c (debug_slim): New function to handle trees.
160 Call DEFINE_DEBUG_VEC for trees.
161 Call DEFINE_DEBUG_HASH_SET for trees.
162 (debug (vec<tree, va_gc>) &): Remove.
163 (debug (<vec<tree, va_gc>) *): Remove.
164 * print-rtl.c (debug_slim): New function to handle const_rtx.
165 Call DEFINE_DEBUG_VEC for rtx_def.
166 Call DEFINE_DEBUG_VEC for rtx_insn.
167 Call DEFINE_DEBUG_HASH_SET for rtx_def.
168 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
169 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
170 (debug (vec<rtx_insn *> *ptr): Remove.
171 (debug_insn_vector): Remove.
172 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
174 2017-11-20 Tom de Vries <tom@codesourcery.com>
176 PR rtl-optimization/82020
177 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
178 IF_THEN_ELSE condition.
180 2017-11-19 Jeff Law <law@redhat.com>
182 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
183 of degenerates resulting from ignoring an edge.
185 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
188 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
190 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
193 * profile-count.c (profile_count::to_sreal_scale): Fix return value
194 for uninitialied counts.
196 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
199 * tree-inline.c (estimate_num_insns): Set to 1 at least.
201 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
204 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
206 2017-11-19 Tom de Vries <tom@codesourcery.com>
208 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
211 2017-11-19 Tom de Vries <tom@codesourcery.com>
213 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
216 2017-11-19 Tom de Vries <tom@codesourcery.com>
218 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
221 2017-11-19 Tom de Vries <tom@codesourcery.com>
223 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
225 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
227 2017-11-19 Tom de Vries <tom@codesourcery.com>
229 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
231 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
232 semicolon after MCORE_EXPORT_NAME call.
234 2017-11-19 Tom de Vries <tom@codesourcery.com>
237 * vmsdbgout.c (vmsdbgout_early_finish): New function.
238 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
240 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
242 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
243 (cgraph_edge::clone): Cleanup updating of profile.
244 * ipa-cp.c (update_profiling_info): Likewise.
245 * ipa-inline-transform.c (inline_transform): Likewise.
246 * ipa-inline.c (inline_small_functions): Add missing space to dump.
247 * ipa-split.c (execute_split_functions): Do not split when function
249 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
250 * profile-count.c (profile_count::dump): Add global0.
251 (profile_count::to_cgraph_frequency): Do not ICE when entry is
253 (profile_count::to_sreal_scale): Likewise.
254 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
255 (profile_count::combine_with_ipa_count): New function.
256 * profile-count.h (profile_guessed_global0adjusted): New.
257 (profile_count::adjusted_zero): New.
258 (profile_count::global0adjusted): New.
259 (profile_count::combine_with_ipa_count): New.
260 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
261 correct profile of return block of split functions.
262 (copy_cfg_body): Remove unused profile_count.
263 (copy_body): Likewise.
264 (expand_call_inline): Update.
265 (tree_function_versioning): Update.
267 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
269 * hash-set.h (hash_set::empty): New.
270 * tree-ssa-threadbackward.h: Delete.
271 * tree-ssa-threadbackward.c (class thread_jumps): New.
272 Move max_threaded_paths into class.
273 (fsm_find_thread_path): Remove arguments that are now in class.
274 (profitable_jump_thread_path): Rename to...
275 (thread_jumps::profitable_jump_thread_path): ...this.
276 (convert_and_register_jump_thread_path): Rename to...
277 (thread_jumps::convert_and_register_current_path): ...this.
278 (check_subpath_and_update_thread_path): Rename to...
279 (thread_jumps::check_subpath_and_update_thread_path): ...this.
280 (register_jump_thread_path_if_profitable): Rename to...
281 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
282 (handle_phi): Rename to...
283 (thread_jumps::handle_phi): ...this.
284 (handle_assignment): Rename to...
285 (thread_jumps::handle_assignment): ...this.
286 (fsm_find_control_statement_thread_paths): Rename to...
287 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
288 (find_jump_threads_backwards): Rename to...
289 (thread_jumps::find_jump_threads_backwards): ...this.
290 Initialize path local data.
291 (pass_thread_jumps::execute): Call find_jump_threads_backwards
292 from within thread_jumps class.
293 (pass_early_thread_jumps::execute): Same.
295 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
297 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
299 2017-11-17 Steve Ellcey <sellcey@cavium.com>
301 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
302 to canonical location.
304 2017-11-17 Steve Ellcey <sellcey@cavium.com>
307 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
309 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
311 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
313 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
314 rcrt1.o%s/grcrt1.o%s for -static-pie.
316 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
318 * i386.c (ix86_multiplication_cost, ix86_division_cost,
319 ix86_shift_rotate_cost): Break out from ...
320 (ix86_rtx_costs): ... here.
321 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
324 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
326 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
327 when entry block was promoted unlikely.
328 (estimate_bb_frequencies): Increase frequency scale.
329 * profile-count.h (profile_count): Export precision info.
331 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
333 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
334 disturbin profile of entry block.
336 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
338 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
341 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
343 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
344 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
346 2017-11-17 Jeff Law <law@redhat.com>
348 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
349 from evrp_dom_walker class. Various methods moved into new class.
350 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
351 (evrp_range_analyzer::enter): New method.
352 (evrp_range_analyzer::leave): New method.
353 (evrp_dom_walker): Remove delegators no longer needed by this class.
354 Replace vr_values data member with evrp_range_analyzer
356 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
357 method extracted from evrp_dom_walker::before_dom_children.
358 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
359 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
361 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
362 Add private copy constructor and move assignment operators.
363 Privatize methods and class data where trivially possible.
364 (evrp_dom_walker::cleanup): New function, extracted from
365 execute_early_vrp. Simplify access to class data.
367 * vr-values.h (get_output_for_vrp): Prototype.
368 * vr-values.c (get_output_for_vrp): New function extracted from
369 vrp_visit_assignment_or_call and extract_range_from_stmt.
370 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
372 2017-11-17 Luis Machado <luis.machado@linaro.org>
374 * config/aarch64/aarch64.c
375 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
376 (qdf24xx_tunings) <autoprefetcher_model>: Set to
377 tune_params::AUTOPREFETCHER_WEAK.
379 2017-11-17 Tamar Christina <tamar.christina@arm.com>
382 * config/arm/arm.c (arm_valid_target_attribute_rec):
383 Parse "arch=" and "+<ext>".
384 (arm_valid_target_attribute_tree): Re-init global options.
385 (arm_option_override): Make non-static.
386 (arm_options_perform_arch_sanity_checks): Make errors fatal.
387 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
388 (__ARM_FEATURE_CRC32): Support undef.
389 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
390 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
392 2017-11-17 David Malcolm <dmalcolm@redhat.com>
394 * gdbinit.in (break-on-diagnostic): New command.
396 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
398 * config/i386/i386.c (ix86_expand_epilogue): Change simple
399 return to indirect jump for EH return if control-flow
400 protection is enabled. Change explicit 'false' argument in
401 pro_epilogue_adjust_stack with a value of
403 * config/i386/i386.md (simple_return_indirect_internal):
404 Remove SImode restriction to support 64-bit.
406 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
408 * combine.c (added_notes_insn): New.
409 (try_combine): Handle added_notes_insn like added_links_insn.
410 Rewrite return value code.
411 (distribute_notes): Set added_notes_insn to the earliest insn we added
414 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
416 PR rtl-optimization/82621
417 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
418 dest of one of those SETs is unused.
420 2017-11-17 Richard Biener <rguenther@suse.de>
423 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
424 * tree-pretty-print.c (dump_generic_node): Handle
425 annot_expr_parallel_kind.
426 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
427 * gimplify.c (gimple_boolify): Likewise.
429 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
431 * config.gcc (extra_headers): Add cet.h for x86 targets.
432 * config/i386/cet.h: New file.
433 * doc/install.texi: Add --enable-cet/--disable-cet.
435 2017-11-17 Richard Biener <rguenther@suse.de>
437 PR tree-optimization/83017
438 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
439 (gen_parallel_loop): Properly count iterations.
440 (parallelize_loops): Handle loop->can_be_parallel independent
441 of flag_loop_parallelize_all. Make static profitability test match
443 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
444 * invoke.texi (parloops-min-per-thread): Document.
446 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
448 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
449 upstreaming review comments.
451 2017-11-17 Tamar Christina <tamar.christina@arm.com>
453 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
454 with fast unaligned access.
455 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
457 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
459 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
460 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
462 2017-11-17 Richard Biener <rguenther@suse.de>
464 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
465 folding of references.
467 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
470 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
471 of replacing call to strncmp with corresponding call to strcmp when
474 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
476 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
477 option prefer-avx256 for skylake-avx512 configuration.
478 * config/i386/i386.c (ix86_option_override_internal): Ditto.
479 (get_builtin_code_for_version): Ditto.
481 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
482 Monk Chiang <sh.chiang04@gmail.com>
484 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
485 (FIXED_REGISTERS): Reserve more register numbers.
486 (CALL_USED_REGISTERS): Likewise.
487 (REG_ALLOC_ORDER): Likewise.
488 (REG_CLASS_CONTENTS): Likewise.
489 (REGISTER_NAMES): Likewise.
491 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
492 Kito Cheng <kito.cheng@gmail.com>
494 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
496 * config/nds32/iterators.md: Add vector mode iterators and attributes.
498 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
500 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
501 parameter list for vec_splats.
503 2017-11-16 Joseph Myers <joseph@codesourcery.com>
505 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
507 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
509 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
512 * doc/install.texi: Document incompatibility of Perl >=5.6.26
513 with the required version of automake 1.11.6.
515 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
517 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
519 (DU_C2_3_power9): Correct reservation combinations.
520 (FP_DIV_power9, VEC_DIV_power9): New.
521 (power9-alu): Split out rotate/shift...
522 (power9-rot): ...to here, correct dispatch resource.
523 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
525 (power9-fp): Correct latency.
526 (power9-sdiv): Add div/sqrt resource.
527 (power9-ddiv): Correct latency, add div/sqrt resource.
528 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
529 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
531 (power9-qpdiv, power9-qpmul): Adjust resource usage.
533 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
535 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
536 switch statement mapping KF built-ins to TF built-ins if we don't
537 have the proper ISA 3.0 assembler support.
539 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
541 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
542 (lower_emutls_function_body): Do not compute it.
544 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
546 * ipa-split.c (split_bb_info): Turn time to sreal.
547 (split_point): Likewise.
548 (dump_split_point): Likewise.
549 (fine_split_points): Likewise.
550 (execute_split_functions): Only zero split_bbs; turn time to sreals.
552 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
554 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
557 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
559 * predict.c (combine_predictions_for_bb): Preserve zero predicted
561 (expensive_function_p): Remove useless assert.
562 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
564 2017-11-16 Martin Sebor <msebor@redhat.com>
566 PR tree-optimization/82588
567 PR tree-optimization/82583
568 * tree-vrp.c (check_array_ref): Handle flexible array members,
569 string literals, and inner indices.
570 (search_for_addr_array): Add detail to diagnostics.
572 2017-11-16 Nathan Sidwell <nathan@acm.org>
576 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
577 (SET_DECL_ASSEMBLER_NAME): Forward to
578 overwrite_decl_assembler_name.
579 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
580 (overwrite_decl_assembler_name): Declare.
581 * tree.c (overwrite_decl_assembler_name): New.
582 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
583 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
584 (LANG_HOOKS_INITIALIZER): Add it.
585 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
586 * langhooks.c (lhd_set_decl_assembler_name): Use
587 SET_DECL_ASSEMBLER_NAME.
588 (lhd_overwrite_decl_assembler_name): Default implementation.
590 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
591 Jackson Woodruff <jackson.woodruff@arm.com>
593 PR tree-optimization/71026
594 * match.pd: Canonicalize constant multiplies in division.
596 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
598 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
599 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
600 Remove OPT_fomit_frame_pointer entry.
601 * common/config/arc/arc-common.c: Likewise.
602 * common/config/arm/arm-common.c: Likewise.
603 * common/config/avr/avr-common.c: Likewise.
604 * common/config/c6x/c6x-common.c: Likewise.
605 * common/config/cr16/cr16-common.c: Likewise.
606 * common/config/cris/cris-common.c: Likewise.
607 * common/config/epiphany/epiphany-common.c: Likewise.
608 * common/config/fr30/fr30-common.c: Likewise.
609 * common/config/frv/frv-common.c: Likewise.
610 * common/config/ia64/ia64-common.c: Likewise.
611 * common/config/iq2000/iq2000-common.c: Likewise.
612 * common/config/lm32/lm32-common.c: Likewise.
613 * common/config/m32r/m32r-common.c: Likewise.
614 * common/config/mcore/mcore-common.c: Likewise.
615 * common/config/microblaze/microblaze-common.c: Likewise.
616 * common/config/mips/mips-common.c: Likewise.
617 * common/config/mmix/mmix-common.c: Likewise.
618 * common/config/mn10300/mn10300-common.c: Likewise.
619 * common/config/nios2/nios2-common.c: Likewise.
620 * common/config/pa/pa-common.c: Likewise.
621 * common/config/pdp11/pdp11-common.c: Likewise.
622 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
623 * common/config/riscv/riscv-common.c: Likewise.
624 * common/config/rs6000/rs6000-common.c: Likewise.
625 * common/config/rx/rx-common.c: Likewise.
626 * common/config/s390/s390-common.c: Likewise.
627 * common/config/sh/sh-common.c: Likewise.
628 * common/config/sparc/sparc-common.c: Likewise.
629 * common/config/tilegx/tilegx-common.c: Likewise.
630 * common/config/tilepro/tilepro-common.c: Likewise.
631 * common/config/v850/v850-common.c: Likewise.
632 * common/config/visium/visium-common.c: Likewise.
633 * common/config/xstormy16/xstormy16-common.c: Likewise.
634 * common/config/xtensa/xtensa-common.c: Likewise.
635 * invoke.texi (-fomit-frame-pointer): Update documentation.
637 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
639 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
641 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
643 * cfg.c (scale_bbs_frequencies_int,
644 cale_bbs_frequencies_gcov_type): Remove.
645 * cfg.h (scale_bbs_frequencies_int,
646 cale_bbs_frequencies_gcov_type): Remove.
648 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
650 * tree-ssa-loop-manip.c
651 (scale_dominated_blocks_in_loop): Update to profile counts.
652 (tree_transform_and_unroll_loop): Likewise.
654 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
656 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
657 scale_bbs_frequencies_int.
659 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
661 * final.c (compute_alignments): Use counts rather than frequencies.
663 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
665 * cfgloopanal.c: Include sreal.h
666 (average_num_loop_insns): Use counts and sreal for accounting.
668 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
670 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
673 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
674 Kito Cheng <kito.cheng@gmail.com>
676 * config/nds32/constraints.md: Provide more constraints.
677 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
678 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
679 support constraints usage.
681 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
683 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
685 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
686 Kito Cheng <kito.cheng@gmail.com>
688 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
689 * config/nds32/nds32.opt: Refine the layout.
690 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
691 TARGET_EXT_STRING): Support new options.
692 * config/nds32/nds32.h: Likewise.
693 * config/nds32/nds32.md: Likewise.
694 * config/nds32/nds32-predicates.c: Likewise.
695 * config/nds32/constraints.md: Likewise.
696 * common/config/nds32/nds32-common.c: Likewise.
698 2017-11-16 Julia Koval <julia.koval@intel.com>
701 * config/i386/gfniintrin.h: Add sse check.
702 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
704 2017-11-16 Julia Koval <julia.koval@intel.com>
706 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
707 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
708 (ix86_handle_option): Handle -mavx512vbmi2.
709 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
710 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
711 * config/i386/i386-c.c (__AVX512VBMI2__): New.
712 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
713 (ix86_valid_target_attribute_inner_p): Ditto.
714 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
715 * config/i386/i386.opt (mavx512vbmi2): New option.
716 * doc/invoke.texi: Add new option.
718 [2017-11-16 Julia Koval <julia.koval@intel.com>
720 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
721 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
722 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
723 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
724 _mm512_gf2p8mul_epi8): New intrinsics.
725 * config/i386/i386-builtin-types.def
726 (V64QI_FTYPE_V64QI_V64QI): New type.
727 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
728 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
729 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
730 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
731 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
732 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
734 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
736 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
737 explicitly as a stream of bytes.
739 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
741 * config/rs6000/altivec.h (vec_xst_be): New #define.
742 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
743 and externalize from *altivec_vperm_<mode>_internal.
744 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
746 (XL_BE_V8HI): Likewise.
747 (XL_BE_V4SI): Likewise.
748 (XL_BE_V4SI): Likewise.
749 (XL_BE_V2DI): Likewise.
750 (XL_BE_V4SF): Likewise.
751 (XL_BE_V2DF): Likewise.
752 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
753 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
754 all array entries with these keys: VSX_BUILTIN_VEC_XL,
755 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
756 VSX_BUILTIN_VEC_XST_BE.
757 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
758 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
760 (altivec_init_builtins): Replace conditional calls to def_builtin
761 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
762 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
763 with unconditional calls. Remove calls to def_builtin for
764 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
765 __builtin_vec_xst_be.
766 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
767 to define_expand, and add alternate RTL generation for P8.
768 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
770 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
771 add alternate RTL generation for P8.
772 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
773 vsx_ld_elemrev_v16qi.
774 (vsx_st_elemrev_v8hi): Convert define_insn
775 to define_expand, and add alternate RTL generation for P8.
776 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
778 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
779 add alternate RTL generation for P8.
780 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
781 vsx_st_elemrev_v16qi.
783 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
786 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
787 TARGET_AVX512ER check.
788 (ix86_option_override_internal): Set MASK_VZEROUPPER if
789 neither -mzeroupper nor -mno-zeroupper is used and
790 TARGET_EMIT_VZEROUPPER is set.
791 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
792 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
794 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
796 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
797 folding of vector compares.
798 (fold_build_vec_cmp): New helper function.
799 (fold_compare_helper): New helper function.
800 (builtin_function_type): Add compare builtins to the list of functions
801 having unsigned arguments. Cosmetic updates to comment indentation.
802 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
803 the not+eq combination.
805 2017-11-15 Bin Cheng <bin.cheng@arm.com>
807 PR tree-optimization/82726
808 PR tree-optimization/70754
809 * tree-predcom.c (order_drefs_by_pos): New function.
810 (combine_chains): Move code setting has_max_use_after to...
811 (try_combine_chains): ...here. New parameter. Sort combined chains
812 according to position information.
813 (tree_predictive_commoning_loop): Update call to above function.
814 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
816 2017-11-15 Bin Cheng <bin.cheng@arm.com>
818 PR tree-optimization/82726
820 2017-01-23 Bin Cheng <bin.cheng@arm.com>
822 PR tree-optimization/70754
823 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
824 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
825 combined stmt before it if not NULL.
826 (combine_chains): Process refs reversely and compute dominance point
830 2017-02-23 Bin Cheng <bin.cheng@arm.com>
832 PR tree-optimization/79663
833 * tree-predcom.c (combine_chains): Process refs in reverse order
834 only for ZERO length chains, and add explaining comment.
836 2017-11-15 Tamar Christina <tamar.christina@arm.com>
838 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
839 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
840 * doc/invoke.texi (ARM Options): Add armv8.3-a.
842 2017-11-15 Tamar Christina <tamar.christina@arm.com>
844 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
846 2017-11-15 Martin Liska <mliska@suse.cz>
848 * tree-cfg.c (pass_warn_function_return::execute):
849 Compare warn_return_type for greater than zero.
851 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
855 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
856 to return true on Xeon and not on Xeon Phi.
857 (ix86_check_avx256_register): Changed to ...
858 (ix86_check_avx_upper_register): ... this. Add extra check for
859 VALID_AVX512F_REG_OR_XI_MODE.
860 (ix86_avx_u128_mode_needed): Changed
861 ix86_check_avx256_register to ix86_check_avx_upper_register.
862 (ix86_check_avx256_stores): Changed to ...
863 (ix86_check_avx_upper_stores): ... this. Changed
864 ix86_check_avx256_register to ix86_check_avx_upper_register.
865 (ix86_avx_u128_mode_after): Changed
866 avx_reg256_found to avx_upper_reg_found. Changed
867 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
868 (ix86_avx_u128_mode_entry): Changed
869 ix86_check_avx256_register to ix86_check_avx_upper_register.
870 (ix86_avx_u128_mode_exit): Ditto.
871 * config/i386/i386.h: (host_detect_local_cpu): New define.
873 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
875 * config/arm/xgene1.md (xgene1): Split into automatons
876 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
877 (xgene1_f_load): Adjust reservations and/or types.
878 (xgene1_f_store): Likewise.
879 (xgene1_load_pair): Likewise.
880 (xgene1_store_pair): Likewise.
881 (xgene1_fp_load1): Likewise.
882 (xgene1_load1): Likewise.
883 (xgene1_store1): Likewise.
884 (xgene1_move): Likewise.
885 (xgene1_alu): Likewise.
886 (xgene1_simd): Likewise.
887 (xgene1_bfm): Likewise.
888 (xgene1_neon_load1): Likewise.
889 (xgene1_neon_store1): Likewise.
890 (xgene1_neon_logic): Likewise.
891 (xgene1_neon_st1): Likewise.
892 (xgene1_neon_ld1r): Likewise.
893 (xgene1_alu_cond): Added.
894 (xgene1_shift_reg): Likwise.
895 (xgene1_bfx): Likewise.
896 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
898 2017-11-15 Jakub Jelinek <jakub@redhat.com>
901 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
903 (can_widen_mult_without_libcall): New function.
904 (expand_mul_overflow): If only checking unsigned mul overflow,
905 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
906 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
907 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
908 (expand_DIVMOD): Formatting fix.
909 * expmed.h (expand_mult): Add NO_LIBCALL argument.
910 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
911 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
913 PR tree-optimization/82977
914 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
915 constructed temporary to strlen_to_stridx.put.
917 2017-11-15 Martin Liska <mliska@suse.cz>
919 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
920 * configure: Regenerate.
922 2017-11-15 Martin Liska <mliska@suse.cz>
925 * config/sh/sh-mem.cc: Use proper probability for
928 2017-11-14 Jeff Law <law@redhat.com>
930 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
931 the red zone for stack_clash_protection_final_dynamic_probe targets
932 when the total dynamic stack size is zero bytes.
934 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
935 blocks is post order.
937 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
939 * dumpfile.h (TDF_COMPARE_DEBUG): New.
940 * final.c (rest_of_clean_state): Set it for the
941 -fcompare-debug dump.
942 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
943 class when TDF_COMPARE_DEBUG is set.
945 * dwarf2out.c (gen_producer_string): Discard
948 2017-11-15 Joseph Myers <joseph@codesourcery.com>
951 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
952 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
953 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
954 (__TGMATH_REAL_2_3): Remove macros.
955 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
956 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
958 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
960 (remquo): Define using __TGMATH_REAL_3.
962 2017-11-14 Jeff Law <law@redhat.com>
964 * vr-values.c: New file with contents extracted from tree-vrp.c.
965 * Makefile.in (OBJS): Add vr-values.o
966 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
967 (set_value_range, set_and_canonicalize_value_range): Likewise.
968 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
969 (value_range_constant_singleton, symbolic_range_p): Likewise.
970 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
971 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
972 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
973 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
974 (find_case_label_range, find_case_label_index): Likewise.
975 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
976 (range_int_cst_singleton_p, value_inside_range): Likewise.
977 (get_single_symbol): Likewise.
978 (switch_update): Move structure definition here.
979 (to_remove_edges, to_update_switch_stmts): Provide externs.
980 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
981 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
982 (vrp_val_is_min, set_value_range): Likewise.
983 (set_and_canonicalize_value_range, copy_value_range): Likewise.
984 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
985 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
986 (range_is_nonnull, range_int_cst_p): Likewwise.
987 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
988 (get_single_symbol, operand_less_p): Likewise
989 (compare_values_warnv, compare_values): Likewise.
990 (value_inside_range, value_range_constant_singleton): Likewise.
991 (zero_nonzero_bitgs_from_vr): Likewise.
992 (extract_range_from_binary_expr_1): Likewise.
993 (overflow_comparison_p): Likewise.
994 (to_remove_edges, to_update_switch_stmts): Likewise.
995 (find_case_label-index, find_case_label_range): Likewise.
996 (switch_update, set_value_range_to_nonnegative): Remove.
997 (set_value_range_to_truthvalue): Likewise.
998 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
999 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
1000 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
1001 (find_case_label_ranges, test_for_singularity): Likewise.
1002 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
1003 (x_vr_values): Move to its remaining use site.
1005 2017-11-10 Jeff Law <law@redhat.com>
1007 * vr-values.h (VR_INITIALIZER): Move #define here.
1008 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
1009 * Makefile.in (OBJS): Add tree-evrp.o
1010 * tree-vrp.h (assert_info): Move structure definition here.
1011 (set_value_range_to_varying): Prototype.
1012 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1013 (infer_value_range, register_edge_assert_for): Likewise.
1014 (stmt_interesting_for_vrp): Likewise.
1015 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
1016 (set_value_range_to_varying): No longer static.
1017 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
1018 (infer_value_range, register_edge_assert_for): Likewise.
1020 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1022 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
1023 generate the XXBRD instruction.
1025 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
1026 (rs6000_builtin_type_compatible): Treat _Float128 and long double
1027 as being compatible if -mabi=ieeelongdouble.
1028 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
1029 to setup float128 built-ins with hardware support.
1030 (BU_FLOAT128_HW_2): Likewise.
1031 (BU_FLOAT128_HW_3): Likewise.
1032 (BU_FLOAT128_HW_VSX_1): Likewise.
1033 (BU_FLOAT128_HW_VSX_2): Likewise.
1034 (scalar_extract_expq): Change float128 built-in functions to
1035 accommodate having both KFmode and TFmode functions. Use the
1036 KFmode variant as the default.
1037 (scalar_extract_sigq): Likewise.
1038 (scalar_test_neg_qp): Likewise.
1039 (scalar_insert_exp_q): Likewise.
1040 (scalar_insert_exp_qp): Likewise.
1041 (scalar_test_data_class_qp): Likewise.
1042 (sqrtf128_round_to_odd): Delete processing the round to odd
1043 built-in functions as special built-in functions, and define them
1044 as float128 built-ins. Use the KFmode variant as the default.
1045 (truncf128_round_to_odd): Likewise.
1046 (addf128_round_to_odd): Likewise.
1047 (subf128_round_to_odd): Likewise.
1048 (mulf128_round_to_odd): Likewise.
1049 (divf128_round_to_odd): Likewise.
1050 (fmaf128_round_to_odd): Likewise.
1051 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1052 support for KFmode and TFmode xststdcqp calls.
1053 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1054 point, switch the built-in handlers for the get/set float128
1055 exponent, get float128 mantissa, float128 test built-ins, and the
1056 float128 round to odd built-in functions. Eliminate creating the
1057 float128 round to odd built-in functions as special built-ins.
1058 (rs6000_init_builtins): Eliminate special creation of the float128
1059 round to odd built-in functions.
1060 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1061 function insns to support both TFmode and KFmode variants.
1062 (xsxsigqp_<mode>): Likewise.
1063 (xsiexpqpf_<mode>): Likewise.
1064 (xsiexpqp_<mode>): Likewise.
1065 (xststdcqp_<mode>): Likewise.
1066 (xststdcnegqp_<mode>): Likewise.
1067 (xststdcqp_<mode>): Likewise.
1069 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1071 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1072 unused path_in_freq_ptr parameter.
1073 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1075 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1077 * ipa-inline.c (edge_badness): Dump sreal frequency.
1078 (compute_inlined_call_time): Match natural implementaiton ...
1079 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1080 forgotten division by CGRAPH_FREQ_BASE.
1082 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1084 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1085 Solaris 11. Update comment.
1086 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1088 * config/sol2.h (STARTFILE_SPEC): Likewise.
1089 * configure: Regenerate.
1091 2017-11-14 Carl Love <cel@us.ibm.com>
1093 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1094 le_ and be_ prefixes to swap* variables. Remove
1095 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1098 2017-11-14 Jason Merrill <jason@redhat.com>
1100 Support GTY((cache)) on hash_map.
1101 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1102 (ggc_cache_remove): Override it instead of ggc_mx.
1103 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1104 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1105 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1106 (simple_cache_map_traits): Override maybe_mx.
1107 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1108 (hash_map): Friend gt_cleare_cache.
1109 (gt_cleare_cache): New.
1110 * tree.h (tree_cache_traits): New hash_map traits class.
1111 (tree_cache_map): New typedef.
1113 2017-11-14 Richard Biener <rguenther@suse.de>
1115 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1116 paramter and handling.
1117 (cleanup_control_flow_bb): Likewise.
1118 (cleanup_control_flow_pre): New helper performing a DFS walk
1119 to call cleanup_control_flow_bb in PRE order.
1120 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1121 via cleanup_control_flow_pre.
1123 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1125 * config/aarch64/aarch64-simd.md
1126 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1127 (*aarch64_simd_bsl<mode>_alt): Likewise.
1128 (aarch64_simd_bsldi_internal): New.
1129 (aarch64_simd_bsldi_alt): Likewise.
1131 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1133 * tracer.c (better_p): Do not compare frequencies.
1134 * reg-stack.c (better_edge): Likewise.
1135 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1138 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1140 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1141 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1142 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1143 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1144 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1145 * predict.c (maybe_hot_frequency_p): Inline to ...
1146 (maybe_hot_count_p): ... here; rewrite to counts.
1147 (counts_to_freqs): Rename to ...
1148 (update_max_bb_count): ... this one.
1149 (expensive_function_p): Use counts.
1150 (estimate_bb_frequencies): Update.
1151 (rebuild_frequencies): Update.
1152 * predict.h (counts_to_freqs): Rename to ...
1153 (update_max_bb_count): ... this one.
1154 * profile.c (compute_branch_probabilities): Add debug info
1155 * tree-inline.c (expand_call_inline): Update debug info.
1156 (optimize_inline_calls): Use update_max_bb_count..
1157 (tree_function_versioning): Use update_max_bb_count..
1158 * value-prof.c (gimple_value_profile_transformations):
1159 Do not use update_max_bb_count.
1161 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1163 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1164 always use frequencies.
1166 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1168 * bb-reorder.c: Remove frequencies from comments.
1169 (better_edge_p): Use profile counts.
1170 (find_traces): Dump profile counts.
1171 (rotate_loop): Use profile counts.
1172 (find_traces_1_round): Likewise.
1173 (connect_better_edge_p): Use counts instead of probabilities for
1175 (copy_bb_p): Drop early check for non-0 frequency.
1176 (sanitize_hot_paths): Update comments.
1178 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1180 * ipa-split.c (struct split_point): Add count.
1181 (consider_split): Do not compute incoming frequency; compute incoming
1182 count and store it to split_point.
1183 (split_function): Set count of the call to split part correctly.
1185 2017-11-13 Carl Love <cel@us.ibm.com>
1187 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1189 2017-11-13 Tom Tromey <tom@tromey.com>
1191 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1193 2017-11-13 Carl Love <cel@us.ibm.com>
1195 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1196 Add support for builtins:
1197 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1198 vector {un,}signed {char,int,short},
1199 vector {un,}signed {char,int,short}) arguments.
1200 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1201 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1202 Add BU_P9V_AV_2 expansions for the builtins.
1203 * config/rs6000/altivec.h (vec_first_match_index,
1204 vec_first_mismatch_index, vec_first_match_or_eos_index,
1205 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1206 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1207 new extern declaration.
1208 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1209 * config/rs6000/vsx.md (first_match_index_<mode>,
1210 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1211 first_mismatch_or_eos_index_<mode>): Add define expand.
1212 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1213 * doc/extend.texi: Update the built-in documenation file for the new
1216 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1218 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1219 into the min/max operations for _Float<N> and _Float<N>X types.
1221 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1224 * dwarf2out.c (do_eh_frame): New static variable.
1225 (dwarf2out_begin_prologue): Set it.
1226 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1228 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1230 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1232 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1234 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1237 * debug.h (dwarf2out_do_eh_frame): Declare.
1238 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1239 (dwarf2out_do_frame): Use it.
1240 (dwarf2out_do_cfi_asm): Likewise.
1241 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1242 (dwarf2out_assembly_start): Likewise.
1243 (dwarf2out_begin_prologue): Fix comment.
1244 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1245 if the target needs either debug or unwind DWARF2 info.
1246 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1247 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1248 -ftrapping-math, -ftrapv and -fwrapv.
1250 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1252 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1253 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1254 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1255 (estimate_edge_size_and_time): Likewise.
1256 (ipa_merge_fn_summary_after_inlining): Likewise.
1257 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1258 (compute_uninlined_call_time): Use sreal_frequency.
1259 (compute_inlined_call_time): Likewise.
1260 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1261 * profile-count.c: Include sreal.h.
1262 (profile_count::to_sreal_scale): New.
1263 * profile-count.h: Forward declare sreal.
1264 (profile_count::to_sreal_scale): Declare.
1266 2017-11-13 Nathan Sidwell <nathan@acm.org>
1268 * diagnostic.c (maybe_line_and_column): New.
1269 (diagnostic_get_location_text): Use it.
1270 (diagnostic_report_current_module): Likewise.
1271 (test_diagnostic_get_location_text): Add tests.
1273 2017-11-13 Luis Machado <luis.machado@linaro.org>
1275 * doc/md.texi (Specifying processor pipeline description): Fix
1276 incorrect latency for the div instruction example.
1278 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1280 PR tree-optimization/78821
1281 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1282 that bit_not_p is the same.
1283 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1284 (split_group): Count precisely bit_not_p bits in each statement.
1285 (invert_op): New function.
1286 (imm_store_chain_info::output_merged_store): Use invert_op to
1287 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1288 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1290 2017-11-13 Martin Liska <mliska@suse.cz>
1292 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1293 (struct source_info): Likewise.
1294 (add_branch_counts): Likewise.
1295 (add_line_counts): Likewise.
1296 (function_summary): Likewise.
1297 (output_intermediate_line): Likewise.
1298 (generate_results): Likewise.
1300 2017-11-13 Martin Liska <mliska@suse.cz>
1302 * gcov.c (struct block_info): Remove typedef for block_t.
1303 (struct line_info): Likewise.
1304 (line_info::has_block): Likewise.
1305 (EXIT_BLOCK): Likewise.
1306 (unblock): Likewise.
1307 (circuit): Likewise.
1308 (get_cycles_count): Likewise.
1309 (process_file): Likewise.
1310 (read_graph_file): Likewise.
1311 (solve_flow_graph): Likewise.
1312 (find_exception_blocks): Likewise.
1313 (add_line_counts): Likewise.
1314 (accumulate_line_info): Likewise.
1315 (output_line_details): Likewise.
1317 2017-11-13 Martin Liska <mliska@suse.cz>
1319 * gcov.c (struct arc_info): Remove typedef for arc_t.
1320 (struct line_info): Likewise.
1321 (add_branch_counts): Likewise.
1322 (output_branch_count): Likewise.
1323 (function_info::~function_info): Likewise.
1324 (circuit): Likewise.
1325 (output_intermediate_line): Likewise.
1326 (read_graph_file): Likewise.
1327 (solve_flow_graph): Likewise.
1328 (find_exception_blocks): Likewise.
1329 (add_line_counts): Likewise.
1330 (accumulate_line_info): Likewise.
1331 (output_line_details): Likewise.
1332 (output_function_details): Likewise.
1334 2017-11-13 Martin Liska <mliska@suse.cz>
1336 * gcov.c (struct function_info): Remove typedef for function_t.
1337 (struct source_info): Likewise.
1338 (source_info::get_functions_at_location): Likewise.
1339 (solve_flow_graph): Likewise.
1340 (find_exception_blocks): Likewise.
1341 (add_line_counts): Likewise.
1342 (output_intermediate_file): Likewise.
1343 (process_file): Likewise.
1344 (generate_results): Likewise.
1345 (release_structures): Likewise.
1346 (read_graph_file): Likewise.
1347 (read_count_file): Likewise.
1348 (accumulate_line_counts): Likewise.
1349 (output_lines): Likewise.
1351 2017-11-13 Martin Liska <mliska@suse.cz>
1353 * gcov.c (function_info::function_info): Remove num_counts
1354 and add vector<gcov_type>.
1355 (function_info::~function_info): Use the vector.
1356 (process_file): Likewise.
1357 (read_graph_file): Likewise.
1358 (read_count_file): Likewise.
1359 (solve_flow_graph): Likewise.
1361 2017-11-13 Martin Liska <mliska@suse.cz>
1363 * gcov.c (function_info::is_artificial): New function.
1364 (process_file): Erase all artificial early.
1365 (generate_results): Skip as all artificial are already
1368 2017-11-13 Martin Liska <mliska@suse.cz>
1370 * gcov.c (read_graph_file): Store to global vector of functions.
1371 (read_count_file): Iterate the vector.
1372 (process_file): Likewise.
1373 (generate_results): Likewise.
1374 (release_structures): Likewise.
1376 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1378 PR tree-optimization/82954
1379 * gimple-ssa-store-merging.c
1380 (imm_store_chain_info::coalesce_immediate_stores): If
1381 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1383 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1385 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1386 Upddate call to ENDIAN_LANE_N.
1387 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1388 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1389 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1390 and use aarch64_endian_lane_rtx.
1391 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1393 2017-11-12 Tom de Vries <tom@codesourcery.com>
1395 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1397 2017-11-12 Tom de Vries <tom@codesourcery.com>
1399 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1401 2017-11-12 Tom de Vries <tom@codesourcery.com>
1403 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1405 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1406 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1407 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1409 2017-11-11 Martin Sebor <msebor@redhat.com>
1412 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1415 * prefic.c (translate_name): Replace strncpy with memcpy to
1416 avoid -Wstringop-truncation.
1418 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1420 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1422 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1424 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1426 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1428 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1431 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1433 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1434 enable generating XXBRH if the value is in a vector register.
1435 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1436 value is in a vector register.
1437 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1438 register to register bswap64's instead of doing the GPR sequence
1439 used on previous machines.
1440 (bswapdi2_xxbrd): New insn.
1441 (bswapdi2_reg): Disallow on ISA 3.0.
1442 (register to register bswap64 splitter): Do not split the insn on
1443 ISA 3.0 systems that use XXBRD.
1445 2017-11-10 Martin Sebor <msebor@redhat.com>
1448 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1450 (find_subframework_file): Same.
1452 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1454 * auto-profile.c (afdo_indirect_call): Drop frequency.
1455 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1456 (cgraph_node::create_edge): Drop frequency argument.
1457 (cgraph_node::create_indirect_edge): Drop frequency argument.
1458 (cgraph_edge::make_speculative): Drop frequency arguments.
1459 (cgraph_edge::resolve_speculation): Do not update frequencies
1460 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1461 (cgraph_node::dump): Check consistency in IPA mode.
1462 (cgraph_edge::maybe_hot_p): Use IPA counter.
1463 (cgraph_edge::verify_count_and_frequency): Rename to ...
1464 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1465 (cgraph_node::verify_node): Update.
1466 * cgraph.h (struct cgraph_edge): Drop frequency.
1467 (cgraph_edge::frequency): New function.
1468 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1470 (cgraph_edge::rebuild_edges): Likewise.
1471 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1472 (duplicate_thunk_for_node): Do not pass frequency.
1473 (cgraph_node::create_clone): Scale only counts.
1474 (cgraph_node::create_virtual_clone): Do not pass frequency.
1475 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1476 (cgraph_node::create_version_clone): Do not pass frequency.
1477 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1478 (cgraph_node::expand_thunk): Do not pass frequency.
1479 (cgraph_node::create_wrapper): Do not pass frequency.
1480 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1482 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1483 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1484 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1485 (gather_caller_stats): Use frequency function.
1486 (ipcp_cloning_candidate_p): Use frequency function.
1487 (ipcp_propagate_stage): Use frequency function.
1488 (get_info_about_necessary_edges): Use frequency function.
1489 (update_profiling_info): Update only IPA profile.
1490 (update_specialized_profile): Use frequency functoin.
1491 (perhaps_add_new_callers): Update only IPA profile.
1492 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1493 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1494 (dump_ipa_call_summary): Use frequency function.
1495 (estimate_edge_size_and_time): Use frequency function.
1496 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1497 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1498 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1499 (update_noncloned_counts): ... ths one; scale counts only.
1500 (clone_inlined_nodes): Do not scale frequency.
1501 (inline_call): Do not pass frequency.
1502 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1503 (compute_inlined_call_time): Use IPA profile.
1504 (want_inline_small_function_p): Use IPA profile.
1505 (want_inline_self_recursive_call_p): Use IPA profile.
1506 (edge_badness): Use IPA profile.
1507 (lookup_recursive_calls): Use IPA profile.
1508 (recursive_inlining): Do not pass frequency.
1509 (resolve_noninline_speculation): Do not update frequency.
1510 (inline_small_functions): Collect max of IPA profile.
1511 (dump_overall_stats): Dump IPA porfile.
1512 (dump_inline_stats): Dump IPA porfile.
1513 (ipa_inline): Collect IPA stats.
1514 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1515 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1516 (ipa_propagate_frequency): Use frequency function.
1517 (ipa_profile): Cleanup.
1518 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1519 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1520 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1521 (input_node): Do not stream frequency.
1522 (input_edge): Do not stream frequency.
1523 (merge_profile_summaries): Scale only IPA profiles.
1524 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1525 * predict.c (drop_profile): Do not recompute frequency.
1526 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1527 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1528 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1529 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1530 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1531 * tree-inline.c (copy_bb): Do not scale frequency.
1532 (expand_call_inline): Do not scale frequency.
1533 (tree_function_versioning): Do not scale frequency.
1534 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1536 2017-11-10 Julia Koval <julia.koval@intel.com>
1538 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1539 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1540 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1541 (_mm256_maskz_gf2p8affine_epi64_epi8)
1542 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1543 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1544 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1545 (__builtin_ia32_vgf2p8affineqb_v32qi)
1546 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1547 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1549 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1553 (arm_option_override): Refactor.
1554 (arm_option_reconfigure_globals): New.
1555 (arm_options_perform_arch_sanity_checks): New.
1556 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1558 (arm_options_perform_arch_sanity_checks): Likewise
1560 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1562 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1563 (power9-qpmul): New.
1564 * rs6000/rs6000.md ("type" attr): Add qmul.
1565 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1566 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1567 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1569 2017-11-10 Martin Sebor <msebor@redhat.com>
1572 * builtins.c (compute_objsize): Handle arrays that
1573 compute_builtin_object_size likes to fail for. Make extern.
1574 * builtins.h (compute_objsize): Declare.
1575 (check_strncpy_sizes): New function.
1576 (expand_builtin_strncpy): Call check_strncpy_sizes.
1577 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1578 -Wstringop-truncation.
1579 (gimple_fold_builtin_strncat): Same.
1580 * gimple.c (gimple_build_call_from_tree): Set call location.
1581 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1582 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1583 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1584 (handle_builtin_strlen): Use strlen_to_stridx.
1585 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1587 Use strlen_to_stridx.
1588 (pass_strlen::execute): Release strlen_to_stridx.
1589 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1590 (-Wstringop-truncation): Document new option.
1592 2017-11-10 Martin Liska <mliska@suse.cz>
1594 PR gcov-profile/82702
1595 * gcov.c (main): Handle intermediate files in a different
1597 (get_gcov_intermediate_filename): New function.
1598 (output_gcov_file): Remove support of intermediate files.
1599 (generate_results): Allocate intermediate file.
1600 (release_structures): Clean-up properly fn_end.
1601 (output_intermediate_file): Start iterating with line 1.
1603 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1605 PR tree-optimization/82929
1606 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1607 ops_swapped_p non-static data member.
1608 (store_immediate_info::store_immediate_info): Clear it.
1609 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1610 ops set ops_swapped_p.
1611 (count_multiple_uses): Handle ops_swapped_p.
1613 2017-11-10 Martin Liska <mliska@suse.cz>
1615 * coverage.c (coverage_init): Stream information about
1616 support of has_unexecuted_blocks.
1617 * doc/gcov.texi: Document that.
1618 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1619 * gcov.c (read_graph_file): Likewise.
1620 (output_line_beginning): Fix a small issue with
1623 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1625 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1626 reference of trivial component.
1628 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1631 * gimple-ssa-store-merging.c
1632 (pass_store_merging::terminate_all_aliasing_chains): For
1633 gimple_store_p stmts also call refs_output_dependent_p.
1635 PR rtl-optimization/82913
1636 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1639 2017-11-09 Jeff Law <law@redhat.com>
1641 * vr-values.h: New file with vr_values class.
1642 * tree-vrp.c: Include vr-values.h
1643 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1644 data objects into the vr_values class.
1645 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1646 (get_value_range): Make it a member function within vr_values class.
1647 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1648 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1649 (op_with_constant_singleton_value_range): Likewise.
1650 (extract_range_for_var_from_comparison_expr): Likewise.
1651 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1652 (extract_range_from_binary_expr): Likewise.
1653 (extract_range_from_unary_expr): Likewise.
1654 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1655 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1656 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1657 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1658 (compare_name_with_value, compare_names): Likewise.
1659 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1660 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1661 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1662 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1663 (extract_range_from_phi_node): Likewise.
1664 (simplify_truth_ops_using_ranges): Likewise.
1665 (simplify_div_or_mod_using_ranges): Likewise.
1666 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1667 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1668 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1669 (simplify_float_conversion_using_ranges): Likewise.
1670 (simplify_internal_call_using_ranges): Likewise.
1671 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1672 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1673 poitner via x_vr_values for calls into gimple folder.
1674 (vrp_initialize_lattice): Make this the vr_values ctor.
1675 (vrp_free_lattice): Make this the vr_values dtor.
1676 (set_vr_value): New function.
1677 (class vrp_prop): Add vr_values data member. Add various member
1678 functions as well as member functions that delegate to vr_values.
1679 (check_array_ref): Make a member function within vrp_prop class.
1680 (search_for_addr_array, vrp_initialize): Likewise.
1681 (vrp_finalize): Likewise. Revamp to avoid direct access to
1682 vr_value, values_propagated, etc.
1683 (check_array_bounds): Extract vrp_prop class instance pointer from
1684 walk info structure. Use it to call member functions.
1685 (check_all_array_refs): Make a member function within vrp_prop class.
1686 Smuggle class instance pointer via walk info structure.
1687 (x_vr_values): New local static.
1688 (vrp_valueize): Use x_vr_values to get class instance.
1689 (vr_valueize_1): Likewise.
1690 (class vrp_folder): Add vr_values data member. Add various member
1691 functions as well as member functions that delegate to vr_values.
1692 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1693 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1694 class instance from vr_values. Use it to call member functions.
1695 (vrp_dom_walker): Add vr_values data member.
1696 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1697 instance via x_vr_values.
1698 (identify_jump_threads): Accept vr_values as argument. Store
1699 it into the walker structure.
1700 (evrp_dom_walker): Add vr_values class data member. Add various
1702 (evrp_dom_walker::try_find_new_range): Use vr_values data
1703 member to access the memory allocator.
1704 (evrp_dom_walker::before_dom_children): Store vr_values class
1705 instance into the vrp_folder class.
1706 (evrp_dom_walker::push_value_range): Rework to avoid direct
1707 access to num_vr_values and vr_value.
1708 (evrp_dom_walker::pop_value_range): Likewise.
1709 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1710 Use vr_values to get to dump_all_value_ranges member function.
1711 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1712 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1713 Pass vr_values class instance down to identify_jump_threads.
1714 Remove call to vrp_free_lattice.
1715 (debug_all_value_ranges): Remove.
1717 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1718 (vrp_folder): Likewise.
1720 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1721 Get it from the existing bitmap instead.
1722 (vrp_intersect_ranges_1): Likewise.
1724 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1726 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1728 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1729 set bit_not_p to it.
1730 (imm_store_chain_info::coalesce_immediate_stores): Break group
1731 if bit_not_p is different.
1732 (count_multiple_uses, split_group,
1733 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1734 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1735 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1736 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1737 to store_immediate_info ctor.
1739 2017-11-09 Jim Wilson <jimw@sifive.com>
1741 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1742 (scan_prog_file): Likewise.
1744 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1746 * bb-reorder.c (max_entry_frequency): Remove.
1747 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1748 connect_traces, push_to_next_round_p): Remove prototypes.
1749 (find_traces_1_round): Use counts only.
1750 (push_to_next_round_p): Likewise.
1751 (find_traces): Likewise.
1752 (rotate_loop): Likewise.
1753 (find_traces_1_round): Likewise.
1754 (connect_traces): Likewise.
1755 (edge_order): Likewise.
1757 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1759 * config/arm/arm.c (output_return_instruction): Add comments to
1760 indicate requirement for cmse_nonsecure_entry return to account
1761 for the size of clearing instruction output here.
1762 (thumb_exit): Likewise.
1763 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1764 return in hardfloat mode.
1766 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1768 * config/rs6000/rs6000.c (machine_function): Add a bool,
1769 "toc_is_wrapped_separately".
1770 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1771 if it wasn't explicitly set or unset, we are optimizing for speed, and
1772 doing separate shrink-wrapping.
1773 (rs6000_get_separate_components): Enable the TOC component if
1774 saving the TOC register in the prologue.
1775 (rs6000_components_for_bb): Handle the TOC component.
1776 (rs6000_emit_prologue_components): Store the TOC register where needed.
1777 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1778 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1780 2017-11-09 Martin Jambor <mjambor@suse.cz>
1782 * ipa-param-manipulation.c: New file.
1783 * ipa-param-manipulation.h: Likewise.
1784 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1785 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1786 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1787 (ipa_parm_adjustment): Likewise.
1788 (ipa_parm_adjustment_vec): Likewise.
1789 (ipa_get_vector_of_formal_parms): Moved declaration to
1790 ipa-param-manipulation.h.
1791 (ipa_get_vector_of_formal_parm_types): Likewise.
1792 (ipa_modify_formal_parameters): Likewise.
1793 (ipa_modify_call_arguments): Likewise.
1794 (ipa_combine_adjustments): Likewise.
1795 (ipa_dump_param_adjustments): Likewise.
1796 (ipa_modify_expr): Likewise.
1797 (ipa_get_adjustment_candidate): Likewise.
1798 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1799 ipa-param-manipulation.c.
1800 (ipa_get_vector_of_formal_parm_types): Likewise.
1801 (ipa_modify_formal_parameters): Likewise.
1802 (ipa_modify_call_arguments): Likewise.
1803 (ipa_modify_expr): Likewise.
1804 (get_ssa_base_param): Likewise.
1805 (ipa_get_adjustment_candidate): Likewise.
1806 (index_in_adjustments_multiple_times_p): Likewise.
1807 (ipa_combine_adjustments): Likewise.
1808 (ipa_dump_param_adjustments): Likewise.
1809 * tree-sra.c: Also include ipa-param-manipulation.h
1810 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1813 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1814 Alan Hayward <alan.hayward@arm.com>
1815 David Sherwood <david.sherwood@arm.com>
1817 * doc/sourcebuild.texi (vect_masked_store): Document.
1819 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1820 Alan Hayward <alan.hayward@arm.com>
1821 David Sherwood <david.sherwood@arm.com>
1823 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1825 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1826 Alan Hayward <alan.hayward@arm.com>
1827 David Sherwood <david.sherwood@arm.com>
1829 * doc/sourcebuild.texi (vect_variable_length): Document.
1831 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1832 Alan Hayward <alan.hayward@arm.com>
1833 David Sherwood <david.sherwood@arm.com>
1835 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1837 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1838 Alan Hayward <alan.hayward@arm.com>
1839 David Sherwood <david.sherwood@arm.com>
1841 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1843 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1844 Alan Hayward <alan.hayward@arm.com>
1845 David Sherwood <david.sherwood@arm.com>
1847 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1848 previously undocumented selectors.
1849 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1851 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1853 * doc/rtl.texi (const_vector): Say that elements can be
1854 const_wide_ints too.
1855 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1856 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1857 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1858 * optabs.c (expand_vector_broadcast): Likewise.
1860 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1865 scaled index even if the unscaled address was invalid.
1866 Don't increase the complexity of using a scale in that case.
1868 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1869 Alan Hayward <alan.hayward@arm.com>
1870 David Sherwood <david.sherwood@arm.com>
1872 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1873 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1874 UNITS_PER_WORD bytes.
1875 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1876 based on REGMODE_NATURAL_SIZE of the inner mode.
1877 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1878 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1879 * expmed.c (lowpart_bit_field_p): Divide the value up into
1880 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1881 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1882 whether something is likely to occupy more than one register.
1884 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1887 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1888 profile_count::adjust_for_ipa_scaling.
1889 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1890 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1892 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1894 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1896 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1897 (split_group): Add total_orig and total_new arguments, estimate the
1898 number of statements related to the store group without store merging
1899 and with store merging.
1900 (imm_store_chain_info::output_merged_store): Adjust split_group
1901 callers, punt if estimated number of statements with store merging
1902 is not smaller than estimated number of statements without it.
1904 (handled_load): Remove has_single_use checks.
1905 (pass_store_merging::process_store): Likewise.
1907 2017-11-09 Richard Biener <rguenther@suse.de>
1909 PR tree-optimization/82902
1910 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1912 2017-11-09 Martin Liska <mliska@suse.cz>
1915 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1918 2017-11-09 Martin Liska <mliska@suse.cz>
1920 PR tree-optimization/82669
1921 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1923 2017-11-09 Martin Liska <mliska@suse.cz>
1925 PR gcov-profile/48463
1926 * coverage.c (coverage_begin_function): Output also end locus
1927 of a function and information whether the function is
1929 * gcov-dump.c (tag_function): Parse and print the information.
1930 * gcov.c (INCLUDE_MAP): Add include.
1931 (INCLUDE_SET): Likewise.
1932 (struct line_info): Move earlier in the source file because
1933 of vector<line_info> in function_info structure.
1934 (line_info::line_info): Likewise.
1935 (line_info::has_block): Likewise.
1936 (struct source_info): Add new member index.
1937 (source_info::get_functions_at_location): New function.
1938 (function_info::group_line_p): New function.
1939 (output_intermediate_line): New function.
1940 (output_intermediate_file): Use the mentioned function.
1941 (struct function_start): New.
1942 (struct function_start_pair_hash): Likewise.
1943 (process_file): Add code that identifies group functions.
1944 Assign lines either to global or function scope.
1945 (generate_results): Skip artificial functions.
1946 (find_source): Assign index for each source file.
1947 (read_graph_file): Read new flag artificial and end_line.
1948 (add_line_counts): Assign it either to global of function scope.
1949 (accumulate_line_counts): Isolate core of the function to
1950 accumulate_line_info and call it for both function and global
1952 (accumulate_line_info): New function.
1953 (output_line_beginning): Fix GNU coding style.
1954 (print_source_line): New function.
1955 (output_line_details): Likewise.
1956 (output_function_details): Likewise.
1957 (output_lines): Iterate both source (global) scope and function
1959 (struct function_line_start_cmp): New class.
1960 * doc/gcov.texi: Reflect changes in documentation.
1962 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1965 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1966 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1967 and similarly for not instead of neg.
1969 2017-11-08 Andi Kleen <ak@linux.intel.com>
1971 * config/i386/i386.opt: Add -mforce-indirect-call.
1972 * config/i386/predicates.md: Check for flag_force_indirect_call.
1973 * doc/invoke.texi: Document -mforce-indirect-call
1975 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1977 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1979 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1980 config/riscv/riscv.c (predict.h): New include.
1981 (riscv_slow_unaligned_access_p): No longer static.
1982 (riscv_block_move_straight): Add require.
1983 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1985 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1988 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1989 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1990 nonimmediate_operand predicate for operand 1 instead of
1993 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1995 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1997 * config/aarch64/constraints.md (Uml): New constraint.
1998 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
2001 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2003 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
2004 of two vec_duplicates into a vec_concat.
2006 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2008 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2009 Simplify vec_merge of vec_duplicate and vec_concat.
2010 * config/aarch64/constraints.md (Utq): New constraint.
2011 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
2014 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2016 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
2017 Simplify vec_merge of vec_duplicate and const_vector.
2018 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
2020 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
2021 mode iterator. Update predicate on operand 1 to
2022 handle non-const_vec constants. Delete constraints.
2023 (*aarch64_combinez_be<mode>): Likewise for operand 2.
2025 2017-11-08 Jakub Jelinek <jakub@redhat.com>
2027 PR tree-optimization/78821
2028 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
2030 (store_operand_info::store_operand_info): Initialize it to false.
2031 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
2032 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
2033 store in the group, and if chain_info is non-NULL, to ignore altogether
2035 (compatible_load_p): Fail if bit_not_p does not match.
2036 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2037 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2038 (pass_store_merging::process_store): Adjust
2039 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2040 call terminate_all_aliasing_chains newly when adding a store into
2041 a chain with non-NULL chain_info.
2043 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2045 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2047 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2049 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2051 (aarch64_layout_frame): Initialise emit_frame_chain.
2052 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2053 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2055 2017-11-08 Martin Liska <mliska@suse.cz>
2057 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2058 of gimple_call_internal_p.
2060 2017-11-07 Tom de Vries <tom@codesourcery.com>
2062 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2064 2017-11-07 Tom de Vries <tom@codesourcery.com>
2066 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2069 2017-11-08 Martin Liska <mliska@suse.cz>
2072 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2074 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2076 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2078 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2081 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2082 (*cmp<mode>_ccz_1): New insn with $k alternative.
2085 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2086 RTX_COMM_COMPARE as commutative as well.
2087 (ix86_binary_operator_ok): Formatting fix.
2088 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2089 *<code><mode>3<mask_name><round_saeonly_name>,
2090 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2091 *<s>mul<mode>3_highpart<mask_name>,
2092 *vec_widen_umult_even_v16si<mask_name>,
2093 *vec_widen_umult_even_v8si<mask_name>,
2094 *vec_widen_umult_even_v4si<mask_name>,
2095 *vec_widen_smult_even_v16si<mask_name>,
2096 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2097 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2098 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2099 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2100 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2101 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2102 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2103 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2104 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2105 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2106 ix86_binary_operator_ok. Formatting fixes.
2107 (*<plusminus_insn><mode>3<mask_name><round_name>,
2108 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2111 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2113 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2114 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2115 eq and ne if TARGET_ISEL.
2116 (cmp): New code_iterator.
2117 (UNS, UNSU_, UNSIK): New code_attrs.
2118 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2119 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2121 ("eq<mode>3"): ... this.
2122 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2124 ("ne<mode>3"): ... this.
2126 2017-11-07 Julia Koval <julia.koval@intel.com>
2129 * common/config/i386/i386-common.c
2130 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2131 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2132 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2133 * config/i386/i386.opt: Ditto.
2134 * config/i386/i386.c (ix86_target_string): Ditto.
2135 (ix86_option_override_internal): Ditto.
2136 (ix86_init_mpx_builtins): Move MPX to args2.
2137 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2138 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2139 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2140 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2141 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2142 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2143 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2145 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2148 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2149 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2150 (zero-extendsidi peephole2): Remove peephole.
2152 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2155 * stor-layout.c: Include attribs.h.
2156 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2157 explicit lookup of "aligned" attribute.
2159 2017-11-07 Andrew Waterman <andrew@sifive.com>
2161 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2162 (riscv_expand_block_move): Likewise.
2163 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2165 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2166 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2167 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2168 (riscv_adjust_block_mem): Likewise.
2169 (riscv_block_move_loop): Likewise.
2170 (riscv_expand_block_move): Likewise.
2171 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2173 2017-11-07 Michael Clark <michaeljclark@mac.com>
2175 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2176 (MUSL_DYNAMIC_LINKER): Likewise.
2178 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2180 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2181 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2184 2017-11-07 Richard Biener <rguenther@suse.de>
2186 * match.pd: Fix build.
2188 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2189 Jackson Woodruff <jackson.woodruff@arm.com>
2191 PR tree-optimization/71026
2192 * match.pd: Canonicalize negate in division.
2194 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2197 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2199 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2201 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2202 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2204 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2206 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2207 non-scalar integral types.
2208 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2209 (-(A-B), -(~A)): New transformations.
2211 2017-11-07 Tom de Vries <tom@codesourcery.com>
2213 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2214 semicolon after "do {} while (0)".
2215 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2216 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2217 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2218 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2219 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2221 2017-11-07 Tom de Vries <tom@codesourcery.com>
2223 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2224 after "do {} while (0)".
2225 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2226 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2227 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2228 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2229 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2231 2017-11-07 Tom de Vries <tom@codesourcery.com>
2234 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2236 (arm_rtx_costs_internal): Add missing semicolon after
2237 HANDLE_NARROW_SHIFT_ARITH call.
2239 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2241 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2242 disable isel if it was not set explicitly.
2244 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2246 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
2247 (add_type_attribute) likewise.
2249 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2251 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2252 of struct ix86_frame.
2253 (ix86_initial_elimination_offset): Likewise.
2254 (ix86_expand_split_stack_prologue): Likewise.
2256 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2258 * tree-vrp.h (enum value_range_type): Update stale comment.
2260 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayward <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2265 (aarch64_expand_vec_perm_const): Take the number of units too.
2266 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2267 (aarch64_expand_vec_perm_const): Likewise.
2268 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2269 (vec_perm<mode>): Update accordingly.
2271 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2276 Take the number of units too.
2277 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2278 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2279 but check for a vector mode before rather than after the call.
2280 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2281 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2282 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2283 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2284 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2285 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2286 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2287 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2288 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2289 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2290 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2291 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2292 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2293 (aarch64_sqdmull2_n<mode>): Update accordingly.
2295 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2296 Alan Hayward <alan.hayward@arm.com>
2297 David Sherwood <david.sherwood@arm.com>
2299 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2300 the number of units too.
2301 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2302 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2303 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2304 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2305 (vec_store_lanesxi<mode>): Update accordingly.
2307 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2308 Alan Hayward <alan.hayward@arm.com>
2309 David Sherwood <david.sherwood@arm.com>
2311 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2312 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2313 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2314 of units rather than the mode.
2315 * config/aarch64/iterators.md (nunits): New mode attribute.
2316 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2317 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2318 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2319 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2320 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2321 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2322 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2323 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2324 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2325 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2326 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2327 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2328 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2329 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2330 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2331 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2332 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2333 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2334 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2335 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2336 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2337 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2338 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2339 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2340 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2341 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2342 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2343 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2344 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2345 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2346 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2347 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2348 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2349 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2350 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2351 (aarch64_simd_vec_setv2di): Likewise.
2353 2017-11-06 Carl Love <cel@us.ibm.com>
2355 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2357 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2358 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2360 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2362 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2363 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2364 Add power 8 macro expansions.
2365 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2366 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2367 power 8 instructions. (VSX_XXBR): Add iterator.
2369 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2371 * config/arm/arm.md (predicable_short_it): Change default to "no",
2372 improve documentation, remove uses that are identical to the default.
2373 (enabled_for_depr_it): Rename to enabled_for_short_it.
2374 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2375 Remove default uses.
2376 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2377 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2378 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2379 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2381 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2384 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2385 float128 helper macros, which are no longer used after deleting
2386 the old 'q' built-in functions, and moving the round to odd
2387 built-in functions to being special built-in functions.
2388 (BU_FLOAT128_2): Likewise.
2389 (BU_FLOAT128_1_HW): Likewise.
2390 (BU_FLOAT128_2_HW): Likewise.
2391 (BU_FLOAT128_3_HW): Likewise.
2392 (FABSQ): Delete old 'q' built-in functions.
2393 (COPYSIGNQ): Likewise.
2394 (SQRTF128_ODD): Move round to odd built-in functions to be
2395 special built-in functions, so that we can handle
2396 -mabi=ieeelongdouble.
2397 (TRUNCF128_ODD): Likewise.
2398 (ADDF128_ODD): Likewise.
2399 (SUBF128_ODD): Likewise.
2400 (MULF128_ODD): Likewise.
2401 (DIVF128_ODD): Likewise.
2402 (FMAF128_ODD): Likewise.
2403 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2404 built-in names to 'f128'.
2405 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2406 old 'q' built-in functions, as the machine independent code for
2407 'f128' built-in functions handles this.
2408 (rs6000_expand_builtin): Add expansion for float128 round to odd
2409 functions, keying off on -mabi=ieeelongdouble of whether to use
2410 the KFmode or TFmode variant.
2411 (rs6000_init_builtins): Initialize the _Float128 round to odd
2413 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2414 _Float128 'q' built-in functions are now mapped into the new
2415 'f128' built-in functions.
2417 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2419 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2420 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2422 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2425 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2426 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2427 * toplev.c: Include "ipa-fnsummary.h".
2428 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2430 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2432 PR tree-optimization/82838
2433 * gimple-ssa-store-merging.c
2434 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2435 on a separate gimple_seq which is then appended to seq.
2437 2017-11-06 Jeff Law <law@redhat.com>
2440 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2441 (get_probe_interval): New functions.
2442 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2443 (ix86_adjust_stack_and_probe): Likewise.
2444 (output_adjust_stack_and_probe): Likewise.
2445 (ix86_emit_probe_stack_range): Likewise.
2446 (ix86_expand_prologue): Likewise.
2448 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2450 PR tree-optimization/82816
2451 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2452 if the modes of the two types are the same.
2453 (convert_plusminus_to_widen): Likewise.
2455 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2457 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2459 (usadv16qi): New define_expand.
2460 (usadv8hi): New define_expand.
2462 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2465 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2467 (inline_transform): Likewise.
2468 * predict.c (counts_to_freqs): Remove useless conditional.
2469 * profile-count.h (profile_count::apply_scale): Move sanity check.
2470 * tree-inline.c (copy_bb): Always scale.
2471 (copy_cfg_body): Likewise.
2473 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2476 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2478 (*cmp_ite0): Add enabled_for_depr_it attribute.
2479 (*cmp_ite1): Likewise.
2481 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2483 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2486 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2488 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2489 return the value by pointer.
2490 (extract_range_from_multiplicative_op_1): Update accordingly.
2491 Return as soon as an operation fails.
2493 2017-11-05 Tom de Vries <tom@codesourcery.com>
2496 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2497 (DEF_SANITIZER_BUILTIN): ... here.
2498 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2499 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2501 2017-11-05 Tom de Vries <tom@codesourcery.com>
2504 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2506 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2507 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2508 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2510 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2511 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2512 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2514 2017-11-05 Tom de Vries <tom@codesourcery.com>
2517 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2520 2017-11-04 Michael Clark <michaeljclark@mac.com>
2522 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2523 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2525 (*addsi3_extended): Likewise.
2526 (*addsi3_extended2): Likewise.
2527 (<optab>si3): Likewise.
2528 (<optab>di3): Likewise.
2529 (<optab><mode>3): Likewise.
2530 (<*optabe>si3_internal): Likewise.
2531 (zero_extendqi<SUPERQI:mode>2): Likewise.
2532 (*add<mode>hi3): Likewise.
2533 (*xor<mode>hi3): Likewise.
2534 (<optab>di3): Likewise.
2535 (*<optab>si3_extend): Likewise.
2536 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2537 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2538 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2540 2017-11-04 Andrew Waterman <andrew@sifive.com>
2542 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2543 TARGET_STRICT_ALIGN based upon -mtune argument.
2545 2017-11-04 Andrew Waterman <andrew@sifive.com>
2547 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2549 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2551 config/i386/i386.c (choose_basereg): Use optional scratch
2552 register and add assertion.
2553 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2554 needed, and don't allocate stack.
2555 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2556 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2557 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2559 2017-11-03 Jeff Law <law@redhat.com>
2561 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2562 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2563 to probe at the start of a noreturn function.
2565 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2567 PR tree-optimization/78821
2568 * gimple-ssa-store-merging.c: Update the file comment.
2569 (MAX_STORE_ALIAS_CHECKS): Define.
2570 (struct store_operand_info): New type.
2571 (store_operand_info::store_operand_info): New constructor.
2572 (struct store_immediate_info): Add rhs_code and ops data members.
2573 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2574 and op1r arguments to the ctor, initialize corresponding data members.
2575 (struct merged_store_group): Add load_align_base and load_align
2577 (merged_store_group::merged_store_group): Initialize them.
2578 (merged_store_group::do_merge): Update them.
2579 (merged_store_group::apply_stores): Pick the constant for
2580 encode_tree_to_bitpos from one of the two operands, or skip
2581 encode_tree_to_bitpos if neither operand is a constant.
2582 (class pass_store_merging): Add process_store method decl. Remove
2583 bool argument from terminate_all_aliasing_chains method decl.
2584 (pass_store_merging::terminate_all_aliasing_chains): Remove
2585 var_offset_p argument and corresponding handling.
2586 (stmts_may_clobber_ref_p): New function.
2587 (compatible_load_p): New function.
2588 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2589 if there is overlap and rhs_code is not INTEGER_CST. For
2590 non-overlapping stores terminate group if rhs is not mergeable.
2591 (get_alias_type_for_stmts): Change first argument from
2592 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2593 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2594 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2596 (get_location_for_stmts): Change first argument from
2597 auto_vec<gimple *> & to vec<gimple *> &.
2598 (struct split_store): Remove orig_stmts data member, add orig_stores.
2599 (split_store::split_store): Create orig_stores rather than orig_stmts.
2600 (find_constituent_stmts): Renamed to ...
2601 (find_constituent_stores): ... this. Change second argument from
2602 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2603 to info structures rather than the statements.
2604 (split_group): Rename ALLOW_UNALIGNED argument to
2605 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2606 it. Adjust find_constituent_stores caller.
2607 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2608 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2609 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2610 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2611 (mem_valid_for_store_merging): New function.
2612 (handled_load): New function.
2613 (pass_store_merging::process_store): New method.
2614 (pass_store_merging::execute): Use process_store method. Adjust
2615 terminate_all_aliasing_chains caller.
2617 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2619 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2620 Return true for more constants, symbols and label references.
2621 (aarch64_valid_floating_const): Remove unused function.
2623 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2626 * config/aarch64/aarch64.c (aarch64_layout_frame):
2627 Undo forcing of LR at bottom of frame.
2629 2017-11-03 Jeff Law <law@redhat.com>
2632 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2633 for int_registers_saved.
2635 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2636 extracted from tree-ssa-dom.c.
2637 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2638 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2639 (record_equivalences_from_incoming_edge): Add additional argument
2640 to single_pred_edge_ignoring_loop_edges call.
2641 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2642 (uncprop_dom_walker::before_dom_children): Add additional argument
2643 to single_pred_edge_ignoring_loop_edges call.
2644 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2645 single_pred_edge_ignoring_loop_edges rather than open coding.
2646 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2648 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2650 * match.pd (-(-A)): Rewrite.
2652 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2654 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2655 (rs6000_emit_int_cmove): New declaration.
2656 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2657 (rs6000_emit_sISEL): Delete.
2658 (rs6000_emit_int_cmove): Make non-static.
2659 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2660 instead of rs6000_emit_sISEL.
2662 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2664 * asan.c (create_cond_insert_point): Maintain profile.
2665 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2667 * basic-block.h (struct basic_block_def): Remove frequency.
2668 (EDGE_FREQUENCY): Use to_frequency
2669 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2671 (find_traces): Update to use to_frequency.
2672 (find_traces_1_round): Likewise; use only IPA counts.
2673 (bb_to_key): Likewise.
2674 (connect_traces): Use IPA counts only.
2675 (copy_bb_p): Update to use to_frequency.
2676 (fix_up_crossing_landing_pad): Likewise.
2677 (sanitize_hot_paths): Likewise.
2678 * bt-load.c (basic_block_freq): Likewise.
2679 * cfg.c (init_flow): Set count_max to uninitialized.
2680 (check_bb_profile): Remove frequencies; check counts.
2681 (dump_bb_info): Do not dump frequencies.
2682 (update_bb_profile_for_threading): Update counts only.
2683 (scale_bbs_frequencies_int): Likewise.
2684 (MAX_SAFE_MULTIPLIER): Remove.
2685 (scale_bbs_frequencies_gcov_type): Update counts only.
2686 (scale_bbs_frequencies_profile_count): Update counts only.
2687 (scale_bbs_frequencies): Update counts only.
2688 * cfg.h (struct control_flow_graph): Add count-max.
2689 (update_bb_profile_for_threading): Update prototype.
2690 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2691 (find_many_sub_basic_blocks): Likewise.
2692 * cfgcleanup.c (try_forward_edges): Likewise.
2693 (try_crossjump_to_edge): Likewise.
2694 * cfgexpand.c (expand_gimple_cond): Likewise.
2695 (expand_gimple_tailcall): Likewise.
2696 (construct_init_block): Likewise.
2697 (construct_exit_block): Likewise.
2698 * cfghooks.c (verify_flow_info): Check consistency of counts.
2699 (dump_bb_for_graph): Do not dump frequencies.
2700 (split_block_1): Do not update frequencies.
2701 (split_edge): Do not update frequencies.
2702 (make_forwarder_block): Do not update frequencies.
2703 (duplicate_block): Do not update frequencies.
2704 (account_profile_record): Do not update frequencies.
2705 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2706 for global heuristics.
2707 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2708 (expected_loop_iterations_unbounded): Use counts only.
2709 * cfgloopmanip.c (scale_loop_profile): Simplify.
2710 (create_empty_loop_on_edge): Simplify
2712 (duplicate_loop_to_header_edge): Simplify
2713 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2714 (update_br_prob_note): Take care of removing note when profile
2716 (relink_block_chain): Do not dump frequency.
2717 (rtl_account_profile_record): Use to_frequency.
2718 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2719 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2720 (cgraph_update_edges_for_call_stmt_node): Likewise.
2721 (cgraph_edge::verify_count_and_frequency): Update.
2722 (cgraph_node::verify_node): Temporarily disable frequency verification.
2723 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2724 to_cgraph_frequency.
2725 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2726 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2728 (cgraph_node::expand_thunk): Update profile.
2729 * except.c (dw2_build_landing_pads): Do not update frequency.
2730 * final.c (compute_alignments): Use to_frequency.
2731 (dump_basic_block_info): Do not dump frequency.
2732 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2733 (dump_gimple_bb_header): Do not dump frequency.
2734 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2736 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2737 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2738 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2739 (init_before_recovery): Do not update frequency.
2740 (sched_create_recovery_edges): Do not update frequency.
2741 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2742 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2743 (ipa_cp_c_finalize): Set max_count to uninitialized.
2744 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2745 (param_change_prob): Use counts.
2746 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2748 * ipa-split.c (consider_split): Use to_frequency.
2749 (split_function): Use to_frequency.
2750 * ira-build.c (loop_compare_func): Likewise.
2751 (mark_loops_for_removal): Likewise.
2752 (mark_all_loops_for_removal): Likewise.
2753 * loop-doloop.c (doloop_modify): Do not update frequency.
2754 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2756 * lto-streamer-in.c (input_function): Update count_max.
2757 * omp-expand.c (expand_omp_taskreg): Update count_max.
2758 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2759 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2760 (maybe_hot_count_p): Use ipa counts only.
2761 (maybe_hot_bb_p): Simplify.
2762 (maybe_hot_edge_p): Simplify.
2763 (probably_never_executed): Do not take frequency argument.
2764 (probably_never_executed_bb_p): Do not pass frequency.
2765 (probably_never_executed_edge_p): Likewise.
2766 (combine_predictions_for_bb): Check that profile is nonzero.
2767 (propagate_freq): Do not set frequency.
2768 (drop_profile): Simplify.
2769 (counts_to_freqs): Simplify.
2770 (expensive_function_p): Use to_frequency.
2771 (propagate_unlikely_bbs_forward): Simplify.
2772 (determine_unlikely_bbs): Simplify.
2773 (estimate_bb_frequencies): Add hack to silence graphite issues.
2774 (compute_function_frequency): Use ipa counts.
2775 (pass_profile::execute): Update.
2776 (rebuild_frequencies): Use counts only.
2777 (force_edge_cold): Use counts only.
2778 * profile-count.c (profile_count::dump): Dump new count types.
2779 (profile_count::differs_from_p): Check compatiblity.
2780 (profile_count::to_frequency): New function.
2781 (profile_count::to_cgraph_frequency): New function.
2782 * profile-count.h (struct function): Declare.
2783 (enum profile_quality): Add profile_guessed_local and
2784 profile_guessed_global0.
2785 (class profile_proability): Decrease number of bits to 29;
2786 update from_reg_br_prob_note and to_reg_br_prob_note.
2787 (class profile_count: Update comment; decrease number of bits
2788 to 61. Check compatibility.
2789 (profile_count::compatible_p): New private member function.
2790 (profile_count::ipa_p): New member function.
2791 (profile_count::operator<): Handle global zero correctly.
2792 (profile_count::operator>): Handle global zero correctly.
2793 (profile_count::operator<=): Handle global zero correctly.
2794 (profile_count::operator>=): Handle global zero correctly.
2795 (profile_count::nonzero_p): New member function.
2796 (profile_count::force_nonzero): New member function.
2797 (profile_count::max): New member function.
2798 (profile_count::apply_scale): Handle IPA scalling.
2799 (profile_count::guessed_local): New member function.
2800 (profile_count::global0): New member function.
2801 (profile_count::ipa): New member function.
2802 (profile_count::to_frequency): Declare.
2803 (profile_count::to_cgraph_frequency): Declare.
2804 * profile.c (OVERLAP_BASE): Delete.
2805 (compute_frequency_overlap): Delete.
2806 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2807 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2808 * sched-ebb.c (rank): Use counts only.
2809 * shrink-wrap.c (handle_simple_exit): Use counts only.
2810 (try_shrink_wrapping): Use counts only.
2811 (place_prologue_for_one_component): Use counts only.
2812 * tracer.c (find_best_predecessor): Use to_frequency.
2813 (find_trace): Use to_frequency.
2814 (tail_duplicate): Use to_frequency.
2815 * trans-mem.c (expand_transaction): Do not update frequency.
2816 * tree-call-cdce.c: Do not update frequency.
2817 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2818 (gimple_merge_blocks): Likewise.
2819 (gimple_split_edge): Likewise.
2820 (gimple_duplicate_sese_region): Likewise.
2821 (gimple_duplicate_sese_tail): Likewise.
2822 (move_sese_region_to_fn): Likewise.
2823 (gimple_account_profile_record): Likewise.
2824 (insert_cond_bb): Likewise.
2825 * tree-complex.c (expand_complex_div_wide): Likewise.
2826 * tree-eh.c (lower_resx): Update profile.
2827 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2829 (initialize_cfun): Do not initialize frequencies
2830 (freqs_to_counts): Delete.
2831 (copy_cfg_body): Ignore count parameter.
2832 (copy_body): Update.
2833 (expand_call_inline): Update count_max.
2834 (optimize_inline_calls): Update count_max.
2835 (tree_function_versioning): Update count_max.
2836 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2837 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2839 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2840 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2841 (try_peel_loop): Likewise.
2842 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2844 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2845 (tree_transform_and_unroll_loop): Do not use frequencies
2846 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2847 Use reliable prediction only.
2848 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2849 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2850 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2851 probability scaling.
2852 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2853 not update frequency
2854 (any_remaining_duplicated_blocks): Likewise.
2855 (update_profile): Likewise.
2856 (estimated_freqs_path): Delete.
2857 (freqs_to_counts_path): Delete.
2858 (clear_counts_path): Delete.
2859 (ssa_fix_duplicate_block_edges): Likewise.
2860 (duplicate_thread_path): Likewise.
2861 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2862 * tree-tailcall.c (decrease_profile): Do not update frequency.
2863 (eliminate_tail_call): Likewise.
2864 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2865 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2866 (optimize_mask_stores): Likewise.
2867 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2868 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2869 (ubsan_expand_ptr_ifn): Update profile.
2870 * value-prof.c (gimple_ic): Simplify.
2871 * value-prof.h (gimple_ic): Update prototype.
2872 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2873 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2875 (want_inline_self_recursive_call_p): Likewise.
2876 (resolve_noninline_speculation): Only cummulate defined counts.
2877 (inline_small_functions): Use nonzero_p.
2878 (ipa_inline): Do not access freed node.
2880 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2882 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2883 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2885 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2887 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2888 non-legitimate address.
2890 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2892 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2893 (*lt0_<mode>di, *lt0_<mode>si): New.
2895 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2897 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2898 TARGET_PAIRED_FLOAT.
2900 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2901 Jim Wilson <jim.wilson@linaro.org>
2903 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2904 * config/aarch64/aarch64-tune.md: Regenerated.
2905 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2906 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2908 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2910 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2911 (arc_expand_prologue): Restore blink for millicode.
2912 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2914 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2917 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2918 gen_vec_duplicate after forcing the scalar into a register.
2920 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2922 * combine (try_combine): Print the insns input to try_combine to the
2925 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2928 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2929 Remove second argument from aarch64_process_target_attr call.
2930 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2932 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2934 (aarch64_handle_attr_arch): Remove second argument.
2935 (aarch64_handle_attr_cpu): Ditto.
2936 (aarch64_handle_attr_tune): Ditto.
2937 (aarch64_handle_attr_isa_flags): Ditto.
2938 (aarch64_process_one_target_attr): Ditto.
2939 (aarch64_process_target_attr): Ditto.
2940 (aarch64_option_valid_attribute_p): Remove second argument.
2941 on aarch64_process_target_attr call.
2943 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2945 * diagnostic.c: Include "selftest-diagnostic.h".
2946 (selftest::assert_location_text): New function.
2947 (selftest::test_diagnostic_get_location_text): New function.
2948 (selftest::diagnostic_c_tests): Call it.
2950 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2952 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2953 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2954 (class selftest::test_diagnostic_context): Move to...
2955 * selftest-diagnostic.c: New file.
2956 * selftest-diagnostic.h: New file.
2958 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2960 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2961 offset range for FT32B.
2962 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2963 * config/ft32/ft32.md: Add TARGET_NOPM.
2964 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2965 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2967 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2969 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2971 2017-11-02 Jeff Law <law@redhat.com>
2973 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2974 virtual keyword on FINAL OVERRIDE members.
2976 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2977 virtuals together. Add virtual destructor.
2978 (substitute_and_fold_engine): Similarly.
2980 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2982 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2984 2017-11-02 Richard Biener <rguenther@suse.de>
2986 PR tree-optimization/82795
2987 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2989 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2991 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2993 (gcc_GAS_CHECK_FEATURE): Remove.
2994 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2995 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2996 * configure: Regenerate.
2998 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
3000 * config/arc/arc.c (hwloop_optimize): Account for empty
3003 2017-11-02 Richard Biener <rguenther@suse.de>
3006 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
3007 Truncate ARRAY_REF index and element size.
3009 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3011 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
3013 2017-11-01 Jeff Law <law@redhat.com>
3015 * tree-ssa-ccp.c (ccp_folder): New class derived from
3016 substitute_and_fold_engine.
3017 (ccp_folder::get_value): New member function.
3018 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
3019 (ccp_fold_stmt): Remove prototype.
3020 (ccp_finalize): Call substitute_and_fold from the ccp_class.
3021 * tree-ssa-copy.c (copy_folder): New class derived from
3022 substitute_and_fold_engine.
3023 (copy_folder::get_value): Renamed from get_value.
3024 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
3025 * tree-vrp.c (vrp_folder): New class derived from
3026 substitute_and_fold_engine.
3027 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
3028 (vrp_folder::get_value): New member function.
3029 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
3030 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
3031 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
3032 provide a class interface to folder/substitute routines.
3033 (ssa_prop_fold_stmt_fn): Remove typedef.
3034 (ssa_prop_get_value_fn): Likewise.
3035 (subsitute_and_fold): Remove prototype.
3036 (replace_uses_in): Likewise.
3037 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3038 Renamed from replace_uses_in. Call the virtual member function
3039 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3040 (substitute_and_fold_dom_walker): Remove initialization of
3041 data member entries for calbacks. Add substitute_and_fold_engine
3042 member and initialize it.
3043 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3044 member functions for get_value, replace_phi_args_in c
3045 replace_uses_in, and fold_stmt calls.
3046 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3047 substitute_and_fold. Remove assert. Update ctor call.
3049 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3050 (ssa_prop_visit_phi_fn): Likewise.
3051 (class ssa_propagation_engine): New class to provide an interface
3053 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3055 (ssa_prop_visit_phi): Likewise.
3056 (ssa_propagation_engine::simulate_stmt): Moved into class.
3057 Call visit_phi/visit_stmt from the class rather than via
3058 file scoped static variables.
3059 (ssa_propagation_engine::simulate_block): Moved into class.
3060 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3061 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3062 set file scoped statics for the visit_stmt/visit_phi callbacks.
3063 * tree-complex.c (complex_propagate): New class derived from
3064 ssa_propagation_engine.
3065 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3066 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3067 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3069 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3070 ssa_propagation_engine.
3071 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3072 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3073 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3074 * tree-ssa-copy.c (copy_prop): New class derived from
3075 ssa_propagation_engine.
3076 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3077 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3078 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3079 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3080 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3081 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3082 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3084 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3086 PR rtl-optimization/82778
3087 PR rtl-optimization/82597
3088 * compare-elim.c (struct comparison): Add in_a_setter field.
3089 (find_comparison_dom_walker::before_dom_children): Remove killed
3090 bitmap and df_simulate_find_defs call, instead walk the defs.
3091 Compute last_setter and initialize in_a_setter. Merge definitions
3092 with first initialization for a few variables.
3093 (try_validate_parallel): Use insn_invalid_p instead of
3094 recog_memoized. Return insn rather than just the pattern.
3095 (try_merge_compare): Fix up comment. Don't uselessly test if
3096 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3098 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3101 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3102 Alan Hayward <alan.hayward@arm.com>
3103 David Sherwood <david.sherwood@arm.com>
3105 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3106 aarch64_hard_regno_nregs to get the number of registers
3109 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3110 Alan Hayward <alan.hayward@arm.com>
3111 David Sherwood <david.sherwood@arm.com>
3113 * config/aarch64/constraints.md (Upl): Rename to...
3115 * config/aarch64/aarch64.md
3116 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3119 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3120 Alan Hayward <alan.hayward@arm.com>
3121 David Sherwood <david.sherwood@arm.com>
3123 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3124 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3132 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3133 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3135 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3136 of a single element.
3137 * config/aarch64/iterators.md: Add a comment above the permute
3138 unspecs to say that they are generated directly by
3139 aarch64_expand_vec_perm_const.
3140 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3142 2017-11-01 Nathan Sidwell <nathan@acm.org>
3144 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3146 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3148 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3149 and medany code models, and describe what they do.
3151 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3153 Revert accidental duplicate:
3155 * combine.c (can_change_dest_mode): Reject changes in
3156 REGMODE_NATURAL_SIZE.
3158 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3160 PR rtl-optimization/64682
3161 PR rtl-optimization/69567
3162 PR rtl-optimization/69737
3163 PR rtl-optimization/82683
3164 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3165 register mentioned in the note, drop the note, unless it came from I3,
3166 in which case it should go to I3 again.
3168 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3170 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3171 and return false if not.
3172 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3174 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3176 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3177 range is known to be empty.
3179 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3180 Alan Hayward <alan.hayward@arm.com>
3181 David Sherwood <david.sherwood@arm.com>
3183 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3184 and CONST_VECTOR_NUNITS instead of computing the number of units from
3185 the byte sizes of the vector and element.
3186 (simplify_binary_operation_1): Likewise.
3187 (simplify_const_binary_operation): Likewise.
3188 (simplify_ternary_operation): Likewise.
3190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3191 Alan Hayward <alan.hayward@arm.com>
3192 David Sherwood <david.sherwood@arm.com>
3194 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3195 (int_mem_offset): ...this new function.
3196 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3197 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3198 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3201 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * lower-subreg.c (interesting_mode_p): New function.
3206 (compute_costs, find_decomposable_subregs, decompose_register)
3207 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3208 (resolve_clobber, dump_choices): Use it.
3210 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3211 Alan Hayward <alan.hayward@arm.com>
3212 David Sherwood <david.sherwood@arm.com>
3214 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3216 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3217 Alan Hayward <alan.hayward@arm.com>
3218 David Sherwood <david.sherwood@arm.com>
3220 * alias.c (find_base_value, find_base_term): Only process integer
3221 truncations. Check the precision rather than the size.
3223 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3224 Alan Hayward <alan.hayward@arm.com>
3225 David Sherwood <david.sherwood@arm.com>
3227 * machmode.h (is_narrower_int_mode): New function
3228 * optabs.c (expand_float, expand_fix): Use it.
3229 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3231 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3232 Alan Hayward <alan.hayward@arm.com>
3233 David Sherwood <david.sherwood@arm.com>
3235 * rtl.h (narrower_subreg_mode): New function.
3236 * ira-color.c (update_costs_from_allocno): Use it.
3238 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3239 Alan Hayward <alan.hayward@arm.com>
3240 David Sherwood <david.sherwood@arm.com>
3242 * optabs-query.h (convert_optab_p): New function, split out from...
3243 (convert_optab_handler): ...here.
3244 (widening_optab_handler): Delete.
3245 (find_widening_optab_handler): Remove permit_non_widening parameter.
3246 (find_widening_optab_handler_and_mode): Likewise. Provide an
3247 override that operates on mode class wrappers.
3248 * optabs-query.c (widening_optab_handler): Delete.
3249 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3250 parameter. Assert that the two modes are the same class and that
3251 the "from" mode is narrower than the "to" mode. Use
3252 convert_optab_handler instead of widening_optab_handler.
3253 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3254 instead of widening_optab_handler.
3255 * expr.c (expand_expr_real_2): Update calls to
3256 find_widening_optab_handler.
3257 * optabs.c (expand_widen_pattern_expr): Likewise.
3258 (expand_binop_directly): Take the insn_code as a parameter.
3259 (expand_binop): Only call find_widening_optab_handler for
3260 conversion optabs; use optab_handler otherwise. Update calls
3261 to find_widening_optab_handler and expand_binop_directly.
3262 Use convert_optab_handler instead of widening_optab_handler.
3263 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3264 find_widening_optab_handler and use scalar_mode rather than
3266 (convert_plusminus_to_widen): Likewise.
3268 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3269 Alan Hayward <alan.hayward@arm.com>
3270 David Sherwood <david.sherwood@arm.com>
3272 * machmode.h (fixed_size_mode): New class.
3273 * rtl.h (get_pool_mode): Return fixed_size_mode.
3274 * gengtype.c (main): Add fixed_size_mode.
3275 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3276 (get_raw_arg_mode): Likewise.
3277 * doc/tm.texi: Regenerate.
3278 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3279 * targhooks.c (default_get_reg_raw_mode): Likewise.
3280 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3281 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3282 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3283 (msp430_get_raw_result_mode): Likewise.
3284 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3285 * dbxout.c (dbxout_parms): Require fixed-size modes.
3286 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3287 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3288 * omp-low.c (lower_oacc_reductions): Likewise.
3289 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3290 (simplify_subreg): Update accordingly.
3291 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3292 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3293 that aren't fixed-size.
3294 (get_pool_mode): Return a fixed_size_mode.
3295 (output_constant_pool_2): Take a fixed_size_mode.
3297 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3298 Alan Hayward <alan.hayward@arm.com>
3299 David Sherwood <david.sherwood@arm.com>
3301 * doc/rtl.texi (vec_series): Document.
3302 (const): Say that the operand can be a vec_series.
3303 * rtl.def (VEC_SERIES): New rtx code.
3304 * rtl.h (const_vec_series_p_1): Declare.
3305 (const_vec_series_p): New function.
3306 * emit-rtl.h (gen_const_vec_series): Declare.
3307 (gen_vec_series): Likewise.
3308 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3309 (gen_vec_series): Likewise.
3310 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3311 * simplify-rtx.c (simplify_unary_operation): Handle negations
3313 (simplify_binary_operation_series): New function.
3314 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3315 (test_vector_ops_series): New function.
3316 (test_vector_ops): Call it.
3317 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3318 gen_const_vec_series.
3319 (altivec_lvsr): Likewise.
3320 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3322 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3323 Alan Hayward <alan.hayward@arm.com>
3324 David Sherwood <david.sherwood@arm.com>
3326 * doc/rtl.texi (const): Update description of address constants.
3327 Say that vector constants are allowed too.
3328 * common.md (E, F): Use CONSTANT_P instead of checking for
3330 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3331 checking for CONST_VECTOR.
3332 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3334 * expr.c (expand_expr_real_2): Check for vector modes instead
3335 of checking for CONST_VECTOR.
3336 * rtl.h (const_vec_p): New function.
3337 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3338 (unwrap_const_vec_duplicate): Handle them here too.
3340 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3341 David Malcolm <dmalcolm@redhat.com>
3342 Alan Hayward <alan.hayward@arm.com>
3343 David Sherwood <david.sherwood@arm.com>
3345 * rtl.h (vec_duplicate_p): New function.
3346 * selftest-rtl.c (assert_rtx_eq_at): New function.
3347 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3348 (assert_rtx_eq_at): Declare.
3349 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3350 * selftest-run-tests.c (selftest::run_tests): Call it.
3351 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3352 (simplify_unary_operation_1): Recursively handle vector duplicates.
3353 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3355 (simplify_subreg): Handle subregs of vector duplicates.
3356 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3357 (selftest::simplify_rtx_c_tests): New functions.
3359 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3360 Alan Hayward <alan.hayward@arm.com>
3361 David Sherwood <david.sherwood@arm.com>
3363 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3364 (gen_vec_duplicate): Likewise.
3365 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3367 (gen_const_vector): ...here.
3368 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3369 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3370 whose elements are all equal.
3371 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3372 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3373 (simplify_relational_operation): Likewise.
3374 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3376 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3377 (aarch64_expand_vector_init): Likewise.
3378 * config/arm/arm.c (neon_vdup_constant): Likewise.
3379 (neon_expand_vector_init): Likewise.
3380 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3381 (arm_block_set_unaligned_vect): Likewise.
3382 (arm_block_set_aligned_vect): Likewise.
3383 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3384 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3385 (expand_vec_perm_even_odd_pack): Likewise.
3386 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3387 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3388 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3389 gen_const_vec_duplicate.
3390 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3391 * config/mips/mips.c (mips_gen_const_int_vector): Use
3392 gen_const_vec_duplicate.
3393 (mips_expand_vector_init): Use CONST0_RTX.
3394 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3395 (define_split): Use gen_const_vec_duplicate.
3396 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3397 (define_split): Use gen_const_vec_duplicate.
3398 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3399 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3400 * config/spu/spu.c (spu_const): Likewise.
3402 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3403 Alan Hayward <alan.hayward@arm.com>
3404 David Sherwood <david.sherwood@arm.com>
3406 * combine.c (can_change_dest_mode): Reject changes in
3407 REGMODE_NATURAL_SIZE.
3409 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3411 * configure.ac (--enable-libssp): New.
3412 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3413 trying to determine target-specific default. Adjust indentation.
3414 * configure: Regenerated.
3415 * doc/install.texi (Configuration): Expand --disable-libssp
3418 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3420 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3423 2017-10-31 Martin Jambor <mjambor@suse.cz>
3426 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3428 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3430 * auto-profile.c (autofdo_source_profile::read): Use
3431 UNKNOWN_LOCATION rather than 0.
3432 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3433 (warning_at): ...this overload.
3434 (warning_at_rich_loc_n): Rename to...
3435 (warning_n): ...this overload.
3436 (error_at_rich_loc): Rename to...
3437 (error_at): ...this overload.
3438 (pedwarn_at_rich_loc): Rename to...
3439 (pedwarn): ...this overload.
3440 (permerror_at_rich_loc): Rename to...
3441 (permerror): ...this overload.
3442 (inform_at_rich_loc): Rename to...
3443 (inform): ...this overload.
3444 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3445 (diagnostic_n_impl_richloc): Rename to...
3446 (diagnostic_n_impl): ...this rich_location *-based decl.
3447 (inform_at_rich_loc): Rename to...
3448 (inform): ...this, and add an assertion.
3449 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3450 (warning_at_rich_loc): Rename to...
3451 (warning_at): ...this, and add an assertion.
3452 (warning_at_rich_loc_n): Rename to...
3453 (warning_n): ...this, and add an assertion.
3454 (warning_n): Update location_t-based implementation for removal of
3455 location_t-based diagnostic_n_impl.
3456 (pedwarn_at_rich_loc): Rename to...
3457 (pedwarn): ...this, and add an assertion.
3458 (permerror_at_rich_loc): Rename to...
3459 (permerror): ...this, and add an assertion.
3460 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3461 (error_at_rich_loc): Rename to...
3462 (error_at): ...this, and add an assertion.
3463 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3464 (driver::do_spec_on_infiles): Likewise.
3465 * substring-locations.c (format_warning_va): Update for renaming
3466 of inform_at_rich_loc.
3468 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3470 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3471 _Float<N>X built-in functions so that the variant without the
3472 "__builtin_" prefix is only enabled for the GNU C and Objective C
3473 languages when they are in non-strict ANSI/ISO mode.
3474 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3475 * target.def (floatn_builtin_p): Add a target hook to control
3476 whether _Float<N> and _Float<N>X built-in functions without the
3477 "__builtin_" prefix are enabled, and return true for C and
3478 Objective C in the default hook. Include langhooks.h in
3480 * targhooks.h (default_floatn_builtin_p): Likewise.
3481 * targhooks.c (default_floatn_builtin_p): Likewise.
3482 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3483 floatn_builtin_p target hook.
3484 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3486 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3487 Eric Botcazou <ebotcazou@adacore.com>
3489 PR rtl-optimization/81803
3490 * lra-constraints.c (curr_insn_transform): Also reload the whole
3491 register for a strict subreg no wider than a word if this is for
3492 a WORD_REGISTER_OPERATIONS target.
3494 2017-10-31 Jason Merrill <jason@redhat.com>
3496 * gdbinit.in: Skip over inlines from timevar.h.
3498 2017-10-31 Martin Liska <mliska@suse.cz>
3500 * doc/gcov.texi: Document new option.
3501 * gcov.c (print_usage): Likewise print it.
3502 (process_args): Support the argument.
3503 (format_count): New function.
3504 (format_gcov): Use the function.
3506 2017-10-31 Martin Liska <mliska@suse.cz>
3508 * gcov.c (struct name_map): do not use typedef.
3509 Define operator== and operator<.
3510 (name_search): Remove.
3511 (name_sort): Remove.
3512 (main): Do not allocate names.
3513 (process_file): Add vertical space.
3514 (generate_results): Use std::find.
3515 (release_structures): Do not release memory.
3516 (find_source): Use std::find.
3518 2017-10-31 Martin Liska <mliska@suse.cz>
3520 * gcov.c (struct line_info): Remove it's typedef.
3521 (line_info::line_info): Add proper ctor.
3522 (line_info::has_block): Do not use a typedef.
3523 (struct source_info): Do not use typedef.
3524 (circuit): Likewise.
3525 (get_cycles_count): Likewise.
3526 (output_intermediate_file): Iterate via vector iterator.
3527 (add_line_counts): Use std::vector methods.
3528 (accumulate_line_counts): Likewise.
3529 (output_lines): Likewise.
3531 2017-10-31 Martin Liska <mliska@suse.cz>
3533 * gcov.c (struct source_info): Remove typedef.
3534 (source_info::source_info): Add proper ctor.
3535 (accumulate_line_counts): Use struct, not it's typedef.
3536 (output_gcov_file): Likewise.
3537 (output_lines): Likewise.
3538 (main): Do not allocate an array.
3539 (output_intermediate_file): Use size of vector container.
3540 (process_file): Resize the vector.
3541 (generate_results): Do not preallocate, use newly added vector
3543 (release_structures): Do not release sources.
3544 (find_source): Use vector methods.
3545 (add_line_counts): Do not use typedef.
3547 2017-10-31 Martin Liska <mliska@suse.cz>
3549 * doc/gcov.texi: Document that.
3550 * gcov.c (add_line_counts): Mark lines with a non-executed
3552 (output_line_beginning): Handle such lines.
3553 (output_lines): Pass new argument.
3554 (output_intermediate_file): Print it in intermediate format.
3556 2017-10-31 Martin Liska <mliska@suse.cz>
3558 * color-macros.h: New file.
3559 * diagnostic-color.c: Factor out color related to macros to
3561 * doc/gcov.texi: Document -k option.
3562 * gcov.c (INCLUDE_STRING): Include string.h.
3563 (print_usage): Add -k option.
3564 (process_args): Parse it.
3565 (pad_count_string): New function.
3566 (output_line_beginning): Likewise.
3567 (DEFAULT_LINE_START): New macro.
3568 (output_lines): Support color output.
3570 2017-10-31 Martin Liska <mliska@suse.cz>
3572 PR gcov-profile/82633
3573 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3574 their interaction with GCOV infrastructure.
3575 * configure.ac: Add -fkeep-{inline,static}-functions to
3577 * configure: Regenerate.
3579 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3582 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3584 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3587 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3588 into a register if it does not fit into an immediate offset field.
3590 2017-10-31 Olivier Hainque <hainque@adacore.com>
3592 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3594 2017-10-31 Julia Koval <julia.koval@intel.com>
3596 * config.gcc: Add gfniintrin.h.
3597 * config/i386/gfniintrin.h: New.
3598 * config/i386/i386-builtin-types.def
3599 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3600 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3601 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3602 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3603 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3604 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3605 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3606 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3607 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3608 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3609 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3610 * config/i386/immintrin.h: Include gfniintrin.h.
3611 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3613 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3615 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3617 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3619 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3620 (arm_ashldi3_1bit): Remove pattern.
3621 (ashrdi3): Remove shift by 1 expansion.
3622 (arm_ashrdi3_1bit): Remove pattern.
3623 (lshrdi3): Remove shift by 1 expansion.
3624 (arm_lshrdi3_1bit): Remove pattern.
3625 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3626 cost of ashldi3 by 1.
3627 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3628 (<shift>di3_neon): Likewise.
3630 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3632 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3633 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3634 and (*aarch64_simd_mov<VQ:mode>).
3635 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3636 pattern alternative.
3637 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3638 attributes to match pattern alternative.
3640 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3642 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3643 * config/rs6000/emmintrin.h: New file.
3644 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3646 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3648 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3649 * (movdi_vfp_cortexa8): Remove pattern.
3651 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3653 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3655 (Specific, alpha*-dec-osf5.1): Remove.
3656 (Specific, mips-sgi-irix5): Remove.
3657 (Specific, mips-sgi-irix6): Remove.
3659 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3662 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3663 arguments to clear_bit_region_be.
3665 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3667 * gimplify.c: Include memmodel.h.
3669 2017-10-30 Martin Jambor <mjambor@suse.cz>
3671 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3672 condition whether loop should be executed at all.
3674 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3677 gimple folding of vec_madd() intrinsics.
3678 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3679 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3680 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3682 2017-10-30 Richard Biener <rguenther@suse.de>
3684 PR tree-optimization/82762
3686 2017-10-23 Richard Biener <rguenther@suse.de>
3688 PR tree-optimization/82129
3690 2017-08-01 Richard Biener <rguenther@suse.de>
3692 PR tree-optimization/81181
3693 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3694 (compute_antic): ... end of iteration here.
3696 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3698 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3699 -std=iso9899:2017 and -std=gnu17.
3700 * doc/standards.texi (C Language): Document C17 support.
3701 * doc/cpp.texi (Overview): Mention -std=c17.
3702 (Standard Predefined Macros): Document C11 and C17 values of
3703 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3704 * doc/extend.texi (Inline): Do not list individual options for
3705 standards newer than C99.
3706 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3708 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3711 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3713 * asan.c (asan_finish_file): Align asan globals array by shadow
3716 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3719 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3720 (struct store_immediate_info): Add bitregion_start and bitregion_end
3722 (store_immediate_info::store_immediate_info): Add brs and bre
3723 arguments and initialize bitregion_{start,end} from those.
3724 (struct merged_store_group): Add bitregion_start, bitregion_end,
3725 align_base and mask fields. Drop unnecessary struct keyword from
3726 struct store_immediate_info. Add do_merge method.
3727 (clear_bit_region_be): Use memset instead of loop storing zeros.
3728 (merged_store_group::do_merge): New method.
3729 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3730 stores as long as the surrounding bitregions have no gaps.
3731 (merged_store_group::merge_overlapping): Use do_merge.
3732 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3733 is byte aligned, rather than requiring that start and width are
3734 byte aligned. Drop unnecessary struct keyword from
3735 struct store_immediate_info. Allocate and populate also mask array.
3736 Make start of the arrays relative to bitregion_start rather than
3737 start and size them according to bitregion_{end,start} difference.
3738 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3739 struct store_immediate_info.
3740 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3741 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3742 struct keyword from struct store_immediate_info.
3743 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3744 between stores as long as the surrounding bitregions have no gaps.
3746 (struct split_store): Add orig non-static data member.
3747 (split_store::split_store): Initialize orig to false.
3748 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3749 if there is exactly a single original stmt. Change stmts argument
3750 to pointer from reference, if NULL, don't push anything to it. Add
3751 first argument, use it to optimize skipping over orig stmts that
3752 are known to be before bitpos already. Simplify.
3753 (split_group): Return unsigned int count how many stores are or
3754 would be needed rather than a bool. Add allow_unaligned argument.
3755 Change split_stores argument from reference to pointer, if NULL,
3756 only do a dry run computing how many stores would be produced.
3757 Rewritten algorithm to use both alignment and misalign if
3758 !allow_unaligned and handle bitfield stores with gaps.
3759 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3760 from bitregion_start instead of start. Compute allow_unaligned
3761 here, if true, do 2 split_group dry runs to compute which one
3762 produces fewer stores and prefer aligned if equal. Punt if
3763 new count is bigger or equal than original before emitting any
3764 statements, rather than during that. Remove no longer needed
3765 new_ssa_names tracking. Replace num_stmts with
3766 split_stores.length (). Use 32-bit stack allocated entries
3767 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3768 if possible. Handle bitfields with gaps.
3769 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3770 Compute bitregion_{start,end} for the stores and construct
3771 store_immediate_info with that. Formatting fixes.
3773 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3776 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3777 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3779 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3781 * gimplify.c: Include tm_p.h.
3783 * common.opt (gcoff): Re-add as ignored option.
3784 (gcoff1, gcoff2, gcoff3): Likewise.
3786 * Makefile.in (OBJS): Delete sdbout.o.
3787 (GTFILES): Delete $(srcdir)/sdbout.c.
3788 * debug.h: Delete sdb_debug_hooks.
3789 * final.c: Delete sdbout.h include.
3790 (final_scan_insn): Delete SDB_DEBUG check.
3791 (rest_of_clean_state): Likewise.
3792 * output.h: Delete sdb_begin_function_line.
3795 * toplev.c: Delete sdbout.h include.
3796 (process_options): Delete SDB_DEBUG check.
3797 * tree-core.h (tree_type_common): Delete pointer field of
3799 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3800 TYPE_SYMTAB_POINTER.
3801 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3802 (TYPE_SYMTAB_IS_POINTER): Delete.
3803 (TYPE_SYMTAB_IS_DIE): Renumber.
3804 * xcoffout.c: Refer to former sdbout.c file.
3805 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3807 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3808 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3809 (Debugging Options): Delete -gcoff.
3810 (-gstabs): Delete SDB reference.
3812 (-gcoff@var{level}): Delete.
3813 * doc/passes.texi (Debugging information output): Delete SDB and
3814 sdbout.c references.
3815 * doc/tm.texi: Regenerate.
3816 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3817 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3819 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3820 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3821 (SDB_DEBUGGING_INFO): Delete.
3822 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3823 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3824 * target.def (output_source_filename): Delete COFF reference.
3826 * common.opt (gcoff): Delete.
3827 (gxcoff+): Update Negative chain.
3828 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3830 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3831 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3832 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3834 (expand_function_start): Change sdb reference to past tense.
3835 (expand_function_end): Change sdb reference to past tense.
3836 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3837 * opts.c (debug_type_names): Delete coff entry.
3838 (common_handle_option): Delete OPT_gcoff case.
3839 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3841 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3842 * config/cris/cris.h: Delete SDB reference in comment.
3843 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3844 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3845 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3846 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3848 (ix86_expand_prologue): Likewise.
3849 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3850 * config/ia64/ia64.h: Likewise.
3851 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3852 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3854 * config/mmix/mmix.h: Likewise.
3855 * config/nds32/nds32.c: Likewise.
3856 * config/stormy/storym16.h: Likewise.
3857 * config/visium/visium.h: Likewise.
3858 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3860 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3862 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3863 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3864 FRAME_POINTER_REGNUM point at high end of local var area.
3866 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3868 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3869 Move comment around. Do not reset best_edge for a copiable
3870 destination if the copy would cause a partition change.
3871 (better_edge_p): Remove redundant check.
3873 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3875 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3877 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3879 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3880 for math functions that have _Float<N> and _Float<N>X variants.
3881 (mathfn_built_in_2): Add support for math functions that have
3882 _Float<N> and _Float<N>X variants.
3883 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3884 (expand_builtin_mathfn_ternary): Add support for fma with
3885 _Float<N> and _Float<N>X variants.
3886 (expand_builtin): Likewise.
3887 (fold_builtin_3): Likewise.
3888 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3889 create math function _Float<N> and _Float<N>X variants as external
3891 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3892 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3893 the __builtin_ prefix and if not strict ansi, without the prefix.
3894 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3895 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3896 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3897 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3898 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3899 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3900 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3901 function signatures for fma _Float<N> and _Float<N>X variants.
3902 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3903 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3904 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3905 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3906 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3907 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3908 * gencfn-macros.c (print_case_cfn): Add support for math functions
3909 that have _Float<N> and _Float<N>X variants.
3910 (print_define_operator_list): Likewise.
3911 (fltfn_suffixes): Likewise.
3913 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3914 for math functions that have _Float<N> and _Float<N>X variants.
3915 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3916 and _Float<N>X variants.
3917 (COPYSIGN): Likewise.
3920 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3921 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3923 (integer_valued_read_call_p): Likewise.
3924 * fold-const-call.c (fold_const_call_ss): Likewise.
3925 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3926 _Float<N> and _Float<N>X variants.
3927 (fold_const_call_ssss): Add support for fma _Float<N> and
3928 _Float<N>X variants.
3929 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3930 support for copysign and fma _Float<N> and _Float<N>X variants.
3931 (backprop::process_builtin_call_use): Likewise.
3932 * tree-call-cdce.c (can_test_argument_range); Add support for
3933 sqrt _Float<N> and _Float<N>X variants.
3934 (edom_only_function): Likewise.
3935 (get_no_error_domain): Likewise.
3936 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3937 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3938 copysign _Float<N> and _Float<N>X variants.
3939 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3940 handled by machine independent code.
3941 (FMAF128): Likewise.
3942 * doc/cpp.texi (Common Predefined Macros): Document defining
3943 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3944 fma _Float<N> and _Float<N>X variants.
3946 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3949 * config/i386/i386-modes.def (CCFPU): Remove definition.
3950 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3951 (ix86_cc_modes_compatible): Ditto.
3952 (ix86_expand_carry_flag_compare): Ditto.
3953 (ix86_expand_int_movcc): Ditto.
3954 (ix86_expand_int_addcc): Ditto.
3955 (ix86_reverse_condition): Ditto.
3956 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3957 Return true/false for unordered/ordered fp comparisons.
3958 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3959 (ix86_prepare_fp_compare_args): Update for rename.
3960 (ix86_expand_fp_compare): Update for rename. Generate unordered
3961 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3962 (ix86_expand_sse_compare_and_jump): Ditto.
3963 * config/i386/predicates.md (fcmov_comparison_operator):
3964 Remove CCFPU mode handling.
3965 (ix86_comparison_operator): Ditto.
3966 (ix86_carry_flag_operator): Ditto.
3967 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3968 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3969 (*cmpu<mode>_cc_i387): Ditto.
3970 (FPCMP): Remove mode iterator.
3971 (unord): Remove mode attribute.
3972 (unord_subst): New define_subst transformation
3973 (unord): New define_subst attribute.
3975 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3976 (*cmpi<unord>xf_i387): Ditto.
3977 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3978 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3979 using unord_subst transformation.
3980 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3981 (round_saeonly): Also handle CCFP mode.
3982 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3983 Remove UNSPEC_SAHF unspec handling.
3985 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3987 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3989 2017-10-27 Jeff Law <law@redhat.com>
3991 * gimple-ssa-sprintf.c: Include domwalk.h.
3992 (class sprintf_dom_walker): New class, derived from dom_walker.
3993 (sprintf_dom_walker::before_dom_children): New function.
3994 (struct call_info): Moved into sprintf_dom_walker class
3995 (compute_formath_length, handle_gimple_call): Likewise.
3996 (sprintf_length::execute): Call the dominator walker rather
3997 than walking the statements.
3999 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
4000 gimple statement locations.
4001 (check_array_bounds): Corresponding changes. Get the statement's
4002 location directly from wi->stmt.
4004 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
4007 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
4009 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
4011 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
4012 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
4014 2017-10-27 Jakub Jelinek <jakub@redhat.com>
4017 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
4018 * config/i386/i386.c (maybe_get_pool_constant): Removed.
4019 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
4020 maybe_get_pool_constant.
4021 * config/i386/predicates.md (zero_extended_scalar_load_operand):
4024 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4026 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
4027 2.26 caveat. Update gas and gld versions.
4028 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
4031 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4033 * cgraph.h (set_malloc_flag): Declare.
4034 * cgraph.c (set_malloc_flag_1): New function.
4035 (set_malloc_flag): Likewise.
4036 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4037 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4039 (read_ipa_call_summary): Add support for reading is_return_callee.
4040 (write_ipa_call_summary): Stream is_return_callee.
4041 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4042 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4043 ipa-prop.h, ipa-fnsummary.h.
4044 (pure_const_names): Change to static.
4045 (malloc_state_e): Define.
4046 (malloc_state_names): Define.
4047 (funct_state_d): Add field malloc_state.
4048 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4049 (check_retval_uses): New function.
4050 (malloc_candidate_p): Likewise.
4051 (analyze_function): Add support for malloc attribute.
4052 (pure_const_write_summary): Stream malloc_state.
4053 (pure_const_read_summary): Add support for reading malloc_state.
4054 (dump_malloc_lattice): New function.
4055 (propagate_malloc): New function.
4056 (warn_function_malloc): New function.
4057 (ipa_pure_const::execute): Call propagate_malloc and
4058 ipa_free_fn_summary.
4059 (pass_local_pure_const::execute): Add support for malloc attribute.
4060 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4061 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4063 2017-10-27 Martin Liska <mliska@suse.cz>
4065 PR gcov-profile/82457
4066 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4067 for fork-like functions to be properly instrumented.
4069 2017-10-27 Richard Biener <rguenther@suse.de>
4072 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4073 info when we redirected EH.
4075 2017-10-26 Michael Collison <michael.collison@arm.com>
4077 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4079 (<optab>_trunchf<GPI:mode>2: New pattern.
4080 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4081 * config/aarch64/iterators.md (wv): New mode attribute.
4082 (vf, VF): New mode attributes.
4083 (vgp, VGP): New mode attributes.
4084 (s): Update attribute with SImode and DImode prefixes.
4086 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4088 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4089 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4090 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4091 (nios2_option_overide): Initialize it. Don't allow R0-relative
4092 addressing with PIC.
4093 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4094 (nios2_symbolic_constant_p): Likewise.
4095 (nios2_legitimate_address_p): Likewise.
4096 (nios2_r0rel_section_name_p): New.
4097 (nios2_symbol_ref_in_r0rel_data_p): New.
4098 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4099 (r0rel_constant_p): New.
4100 (nios2_print_operand_address): Handle r0rel_constant_p.
4101 (nios2_cdx_narrow_form_p): Likewise.
4102 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4103 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4104 (Nios II Options): Document -mr0rel-sec.
4106 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4108 * config/nios2/nios2.c: Include xregex.h.
4109 (nios2_gprel_sec_regex): New.
4110 (nios2_option_overide): Initialize it. Don't allow GP-relative
4111 addressing with PIC.
4112 (nios2_small_section_name_p): Check for regex match.
4113 * config/nios2/nios2.opt (mgprel-sec=): New option.
4114 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4115 (Nios II Options): Document -mgprel-sec.
4117 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4119 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4121 2017-10-26 Tom de Vries <tom@codesourcery.com>
4123 PR tree-optimization/82707
4124 * gimple.c (gimple_copy): Fix unsharing of
4125 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4127 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4129 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4130 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4131 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4132 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4133 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4134 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4135 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4136 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4137 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4138 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4139 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4140 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4141 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4142 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4143 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4144 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4145 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4147 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4149 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4151 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4152 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4153 warning to rs6000.c. Remove the Undocumented flag, since it has
4155 (-mabi=ibmlongdouble): Likewise.
4156 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4157 already set, set the default format for long double.
4158 (rs6000_debug_reg_global): Print whether long double is IBM or
4160 (rs6000_option_override_internal): Rework setting long double
4161 format. Only warn if the user is changing the long double default
4162 and they did not use -Wno-psabi.
4163 * doc/invoke.texi (PowerPC options): Update the documentation for
4164 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4166 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4167 Alan Hayward <alan.hayward@arm.com>
4168 David Sherwood <david.sherwood@arm.com>
4170 * rtl.h (wider_subreg_mode): New function.
4171 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4172 rather than an unsigned int *.
4173 * ira-color.c (regno_max_ref_width): Replace with...
4174 (regno_max_ref_mode): ...this new variable.
4175 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4176 Use wider_subreg_mode.
4177 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4178 rather than an unsigned int *.
4179 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4180 (process_alt_operands): Likewise.
4181 (invariant_p): Likewise.
4182 * lra-spills.c (assign_mem_slot): Likewise.
4183 (add_pseudo_to_slot): Likewise.
4184 * lra.c (collect_non_operand_hard_regs): Likewise.
4185 (add_regs_to_insn_regno_info): Likewise.
4186 * reload1.c (regno_max_ref_width): Replace with...
4187 (regno_max_ref_mode): ...this new variable.
4188 (reload): Update accordingly. Update call to
4189 ira_sort_regnos_for_alter_reg.
4190 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4191 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4192 (scan_paradoxical_subregs): Likewise.
4194 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4196 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4197 (aarch64_frame): Add emit_frame_chain boolean.
4198 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4199 Move eh_return case to aarch64_layout_frame.
4200 (aarch64_layout_frame): Initialize emit_frame_chain.
4201 (aarch64_expand_prologue): Use emit_frame_chain.
4203 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4205 * config/aarch64/aarch64.c (aarch64_layout_frame):
4206 Ensure LR is always stored at the bottom of the callee-saves.
4207 Remove rarely used frame layout which saves callee-saves at top of
4208 frame, so the store of LR can be used as a valid probe in all cases.
4210 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4212 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4213 Improve unaligned TImode/TFmode base/offset split.
4215 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4216 Alan Hayward <alan.hayward@arm.com>
4217 David Sherwood <david.sherwood@arm.com>
4219 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4220 * combine.c (find_single_use_1): Likewise.
4221 (expand_field_assignment): Likewise.
4222 (move_deaths): Likewise.
4223 * lra-constraints.c (simplify_operand_subreg): Likewise.
4224 (curr_insn_transform): Likewise.
4225 * lra.c (collect_non_operand_hard_regs): Likewise.
4226 (add_regs_to_insn_regno_info): Likewise.
4227 * rtlanal.c (reg_referenced_p): Likewise.
4228 (covers_regno_no_parallel_p): Likewise.
4230 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4232 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4233 Don't print any bits outside the precision of the value.
4234 * wide-int.cc (test_printing): Add some new tests.
4236 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4238 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4239 supports -xbrace_comment option.
4240 * configure: Regenerate.
4241 * config.in: Regenerate.
4242 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4243 (ASM_CPU_SPEC): Use it.
4245 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4247 * target.def (static_rtx_alignment): New hook.
4248 * targhooks.h (default_static_rtx_alignment): Declare.
4249 * targhooks.c (default_static_rtx_alignment): New function.
4250 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4251 * doc/tm.texi: Regenerate.
4252 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4253 instead of targetm.constant_alignment. Remove call to
4255 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4256 (cris_preferred_mininum_alignment): New function, split out from...
4257 (cris_constant_alignment): ...here.
4258 (cris_static_rtx_alignment): New function.
4259 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4261 (ix86_constant_alignment): ...here.
4262 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4263 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4264 (mmix_static_rtx_alignment): New function.
4265 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4266 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4268 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4271 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4272 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4274 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4276 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4277 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4278 mark as a sign-extending load.
4279 (local_pic_load_u): Define.
4281 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4284 * fold-const.c (operand_equal_for_comparison_p): Also return true
4285 if ARG0 is a simple variant of ARG1 with narrower precision.
4286 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4288 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4290 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4292 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4293 scatter_static, scatter_per_elt.
4294 * x86-tune-costs.h: Add new cost entries.
4296 2017-10-25 Richard Biener <rguenther@suse.de>
4298 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4299 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4300 class pass_fre): Move to ...
4301 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4302 class pass_fre): ... here and adjust for statistics.
4304 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4307 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4308 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4309 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4310 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4313 2017-10-25 Richard Biener <rguenther@suse.de>
4315 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4316 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4317 eliminate_push_avail, eliminate_insert): Move inside...
4318 (class eliminate_dom_walker): ... this class in preparation
4320 (fini_eliminate): Remove by merging with ...
4321 (eliminate): ... this function. Adjust for class changes.
4322 (pass_pre::execute): Remove fini_eliminate call.
4323 (pass_fre::execute): Likewise.
4325 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4328 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4329 (VPERMI2, VPERMI2I): New mode iterators.
4330 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4331 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4333 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4334 mode iterator. Remove 3 old define_insn patterns.
4335 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4336 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4337 VPERMI2 mode iterator, remove the other two expanders.
4338 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4339 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4340 instructions, remove the other two patterns.
4341 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4342 mode iterator, remove the other two patterns.
4343 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4344 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4345 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4346 and adjust argument order accordingly.
4347 (ix86_expand_vec_perm): Adjust caller.
4348 (expand_vec_perm_1): Likewise.
4349 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4350 (expand_vec_perm_vpermt2_vpshub2): ... this.
4351 (ix86_expand_vec_perm_const_1): Adjust caller.
4352 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4355 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4356 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4357 (vec_shl_<mode>): Remove unused expander.
4358 (avx512bw_<shift_insn><mode>3): New define_insn.
4359 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4360 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4362 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4365 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4368 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4370 PR rtl-optimization/82396
4371 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4372 (autopref_multipass_init): Simplify initialization.
4373 (autopref_rank_data): Simplify sort order.
4374 * gcc/sched-int.h (autopref_multipass_data_): Remove
4375 multi_mem_insn_p, min_offset and max_offset.
4377 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4380 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4381 Check special value of flag_omit_frame_pointer.
4382 (aarch64_can_eliminate): Likewise.
4383 (aarch64_override_options_after_change_1): Simplify handling of
4384 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4386 2017-10-24 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/82697
4389 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4390 zero for conditional load and unconditional store.
4392 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4394 * doc/install.texi: Document bootstrap-cet.
4396 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4399 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4400 ENDBR instruction at function entrance if function is only
4403 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4406 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4407 patterns to better describe from which operation the CF is computed.
4408 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4409 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4410 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4411 is 0, use _0 suffixed expanders instead of emitting a comparison
4414 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4416 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4417 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4419 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4422 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4423 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4424 * loop-iv.c (iv_get_reaching_def): Likewise.
4425 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4426 variable is promoted and the partition contains undefined values.
4428 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4430 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4432 (nios2_address_cost): Define.
4433 (nios2_legitimize_address): Recognize (exp + constant) directly.
4434 (TARGET_ADDRESS_COST): Define.
4436 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4438 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4439 (nios2_symbolic_memory_operand_p): Declare.
4440 (nios2_split_large_constant): Declare.
4441 (nios2_split_symbolic_memory_operand): Declare.
4442 * config/nios2/nios2.c: Adjust includes.
4443 (nios2_symbolic_constant_allowed): New.
4444 (nios2_symbolic_constant_p): New.
4445 (nios2_plus_symbolic_constant_p): New.
4446 (nios2_valid_addr_expr_p): Recognize addresses involving
4448 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4449 (nios2_symbolic_memory_operand_p): New.
4450 (nios2_large_constant_p): New.
4451 (nios2_split_large_constant): New.
4452 (nios2_split_plus_large_constant): New.
4453 (nios2_split_symbolic_memory_operand): New.
4454 (nios2_legitimize_address): Code refactoring. Handle addresses
4455 involving symbolic constants.
4456 (nios2_emit_move_sequence): Likewise.
4457 (nios2_print_operand): Improve error output.
4458 (nios2_print_operand_address): Handle LO_SUM.
4459 (nios2_cdx_narrow_form_p): Likewise.
4460 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4461 operands involving symbolic constants.
4462 (movhi_internal, movsi_internal): Likewise.
4463 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4464 (extendhisi2, extendqi<mode>2): Likewise.
4466 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4468 * tree-pass.h (PROP_rtl_split_insns): Define.
4469 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4471 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4473 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4475 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4478 * target.def (const_not_ok_for_debug_p): Default to
4479 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4480 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4481 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4482 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4483 which targetm.const_not_ok_for_debug_p returned true.
4484 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4486 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4488 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4489 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4490 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4492 (ix86_const_not_ok_for_debug_p): New function.
4493 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4494 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4496 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4499 * system.h: Conditionally include "unique-ptr.h" if
4500 INCLUDE_UNIQUE_PTR is defined.
4501 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4502 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4504 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4506 * config/rl78/rl78.md: New define_expand "subdi3".
4508 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4511 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4512 DF_REF_INSN if DF_REF_INSN_INFO is false.
4514 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4516 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4517 xmm_move instead of sse_move.
4518 (sse_store_index): New function.
4519 (ix86_register_move_cost): Be more sensible about mismatch stall;
4520 model AVX moves correctly; make difference between sse->integer and
4522 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4523 moves; make difference between SSE and AVX.
4524 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4525 and zmm_move. Increase size of sse load and store tables;
4526 add unaligned load and store tables; add ssemmx_to_integer.
4527 * x86-tune-costs.h: Update all entries according to real
4528 move latencies from Agner Fog's manual and chip documentation.
4530 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4533 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4534 * config/i386/constraints.md (Wf): New constraint.
4535 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4536 (cmp<dwi>_doubleword): Removed.
4537 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4538 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4539 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4540 expand with cmp<dwi>_doubleword. For LTU and GEU use
4541 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4543 * common.opt (gcolumn-info): Enable by default.
4544 * doc/invoke.texi (gcolumn-info): Document new default.
4546 2017-10-23 Richard Biener <rguenther@suse.de>
4548 PR tree-optimization/82672
4549 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4550 Fold the stmt if we propagated into it.
4552 2017-10-23 Richard Biener <rguenther@suse.de>
4554 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4555 (bitmap_remove_expr_from_set): ... this. All callers call this
4556 for non-constant values.
4557 (bitmap_set_subtract): Rename to...
4558 (bitmap_set_subtract_expressions): ... this. Adjust and
4560 (bitmap_set_contains_value): Remove superfluous check.
4561 (bitmap_set_replace_value): Inline into single caller ...
4562 (bitmap_value_replace_in_set): ... here and simplify.
4563 (dependent_clean): Merge into ...
4564 (clean): ... this using an overload. Adjust.
4565 (prune_clobbered_mems): Adjust.
4566 (compute_antic_aux): Likewise.
4567 (compute_partial_antic_aux): Likewise.
4569 2017-10-23 Richard Biener <rguenther@suse.de>
4571 PR tree-optimization/82129
4573 2017-08-01 Richard Biener <rguenther@suse.de>
4575 PR tree-optimization/81181
4576 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4577 (compute_antic): ... end of iteration here.
4579 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4581 * target.def (starting_frame_offset): New hook.
4582 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4583 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4584 * doc/tm.texi.in: Regenerate.
4585 * hooks.h (hook_hwi_void_0): Declare.
4586 * hooks.c (hook_hwi_void_0): New function.
4587 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4588 STARTING_FRAME_OFFSET.
4589 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4590 * reload1.c (reload): Likewise.
4591 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4592 instead of STARTING_FRAME_OFFSET.
4593 * function.c (try_fit_stack_local): Likewise.
4594 (assign_stack_local_1): Likewise
4595 (instantiate_virtual_regs): Likewise.
4596 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4597 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4598 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4599 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4600 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4601 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4602 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4603 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4604 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4605 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4606 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4607 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4608 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4609 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4610 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4611 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4612 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4613 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4614 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4615 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4616 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4617 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4618 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4619 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4620 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4621 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4622 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4623 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4624 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4625 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4626 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4627 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4628 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4629 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4630 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4631 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4632 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4633 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4634 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4635 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4636 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4637 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4638 return a HOST_WIDE_INT.
4639 (avr_builtin_setjmp_frame_value): Use it instead of
4640 STARTING_FRAME_OFFSET.
4641 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4642 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4643 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4645 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4646 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4647 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4648 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4649 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4650 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4651 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4652 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4653 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4654 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4655 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4656 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4658 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4659 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4660 * config/mips/mips.c (mips_compute_frame_info): Refer to
4661 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4662 (mips_starting_frame_offset): New function.
4663 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4664 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4665 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4666 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4667 and return a HOST_WIDE_INT.
4668 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4669 (mmix_initial_elimination_offset): Refer to
4670 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4671 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4672 * config/pa/pa.c (pa_starting_frame_offset): New function.
4673 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4674 (pa_expand_prologue): Likewise.
4675 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4676 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4677 !FRAME_GROWS_DOWNWARD handling to...
4678 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4679 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4680 !FRAME_GROWS_DOWNWARD handling to...
4681 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4682 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4683 !FRAME_GROWS_DOWNWARD handling to...
4684 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4685 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4687 (rs6000_starting_frame_offset): New function.
4688 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4689 !FRAME_GROWS_DOWNWARD handling to...
4690 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4691 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4692 !FRAME_GROWS_DOWNWARD handling to...
4693 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4694 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4695 !FRAME_GROWS_DOWNWARD handling to...
4696 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4697 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4698 (rs6000_starting_frame_offset): New function.
4699 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4700 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4701 * config/vax/vax.c (vax_starting_frame_offset): New function.
4702 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4703 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4704 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4705 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4706 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4707 * system.h (STARTING_FRAME_OFFSET): Poison.
4709 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4711 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4712 SCALAR_TYPE_MODE instead of TYPE_MODE.
4714 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4715 Alan Hayward <alan.hayward@arm.com>
4716 David Sherwood <david.sherwood@arm.com>
4718 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4720 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4721 Alan Hayward <alan.hayward@arm.com>
4722 David Sherwood <david.sherwood@arm.com>
4724 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4726 2017-10-23 Richard Biener <rguenther@suse.de>
4728 PR tree-optimization/82129
4729 * tree-ssa-pre.c (bitmap_set_and): Remove.
4730 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4731 canonicalizing expressions in the set to those with lowest
4732 ID rather than taking that from the first edge.
4734 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4736 * combine.c (rtx_equal_for_field_assignment_p): Use
4737 byte_lowpart_offset.
4739 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4740 Alan Hayward <alan.hayward@arm.com>
4741 David Sherwood <david.sherwood@arm.com>
4743 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4744 to a SUBREG_PROMOTED_VAR.
4746 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4747 Alan Hayward <alan.hayward@arm.com>
4748 David Sherwood <david.sherwood@arm.com>
4750 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4751 (expand_debug_source_expr): Likewise.
4752 * combine.c (combine_simplify_rtx): Likewise.
4753 * cse.c (fold_rtx): Likewise.
4754 * optabs.c (expand_float): Likewise.
4755 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4756 (simplify_binary_operation_1): Likewise.
4758 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4759 Alan Hayward <alan.hayward@arm.com>
4760 David Sherwood <david.sherwood@arm.com>
4762 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4763 (record_promoted_value): Likewise.
4764 * expr.c (expand_expr_real_2): Likewise.
4765 * ree.c (update_reg_equal_equiv_notes): Likewise.
4766 (combine_set_extension): Likewise.
4767 * rtlanal.c (low_bitmask_len): Likewise.
4768 * simplify-rtx.c (neg_const_int): Likewise.
4769 (simplify_binary_operation_1): Likewise.
4771 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4772 Alan Hayward <alan.hayward@arm.com>
4773 David Sherwood <david.sherwood@arm.com>
4775 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4776 * regcprop.c (maybe_mode_change): Likewise.
4777 * reload1.c (alter_reg): Likewise.
4779 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4781 * inchash.h (inchash::hash::add_wide_int): New function.
4782 * lto-streamer-out.c (hash_tree): Use it.
4784 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4786 * inchash.h (inchash::hash::add_wide_int): Rename to...
4787 (inchash::hash::add_hwi): ...this.
4788 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4789 (polymorphic_call_target_hasher::hash): Likewise.
4790 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4791 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4792 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4793 * lto-streamer-out.c (hash_tree): Likewise.
4794 * optc-save-gen.awk: Likewise.
4795 * tree.c (add_expr): Likewise.
4797 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4800 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4801 for ordered inequality comparisons even with TARGET_IEEE_FP.
4803 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4806 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4807 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4808 Expand with cmp<dwi>_doubleword.
4810 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4812 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4813 List CET intrinsics.
4814 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4815 specific to -fcf-protection option.
4817 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4819 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4820 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4821 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4822 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4823 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4824 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4825 (extra_objs): Add cet.o for Linux/x86 targets.
4826 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4827 * config/i386/cet.c: New file.
4828 * config/i386/cetintrin.h: Likewise.
4829 * config/i386/t-cet: Likewise.
4830 * config/i386/cpuid.h (bit_SHSTK): New.
4831 (bit_IBT): Likewise.
4832 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4833 pass IBT and SHSTK bits.
4834 * config/i386/i386-builtin-types.def
4835 (VOID_FTYPE_UNSIGNED_PVOID): New.
4836 (VOID_FTYPE_UINT64_PVOID): Likewise.
4837 * config/i386/i386-builtin.def: Add CET intrinsics.
4838 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4839 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4840 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4841 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4843 * config/i386/i386.c (rest_of_insert_endbranch): New.
4844 (pass_data_insert_endbranch): Likewise.
4845 (pass_insert_endbranch): Likewise.
4846 (make_pass_insert_endbranch): Likewise.
4847 (ix86_notrack_prefixed_insn_p): Likewise.
4848 (ix86_target_string): Add -mibt, -mshstk flags.
4849 (ix86_option_override_internal): Add flag_cf_protection
4851 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4852 (ix86_print_operand): Add 'notrack' prefix output.
4853 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4854 (ix86_expand_builtin): Expand CET intrinsics.
4855 (x86_output_mi_thunk): Add 'endbranch' instruction.
4856 * config/i386/i386.h (TARGET_IBT): New.
4857 (TARGET_IBT_P): Likewise.
4858 (TARGET_SHSTK): Likewise.
4859 (TARGET_SHSTK_P): Likewise.
4860 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4861 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4862 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4863 (builtin_setjmp_setup): New pattern.
4864 (builtin_longjmp): Likewise.
4865 (rdssp<mode>): Likewise.
4866 (incssp<mode>): Likewise.
4867 (saveprevssp): Likewise.
4868 (rstorssp): Likewise.
4869 (wrss<mode>): Likewise.
4870 (wruss<mode>): Likewise.
4871 (setssbsy): Likewise.
4872 (clrssbsy): Likewise.
4873 (nop_endbr): Likewise.
4874 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4876 * config/i386/immintrin.h: Include <cetintrin.h>.
4877 * config/i386/linux-common.h
4878 (file_end_indicate_exec_stack_and_cet): New prototype.
4879 (TARGET_ASM_FILE_END): New.
4881 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4883 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4884 latencies instead of having separate table; make difference between
4885 integer and float costs.
4886 * i386.h (processor_costs): Remove scalar_stmt_cost,
4887 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4888 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4890 * x86-tune-costs.h: Remove entries which has been removed in
4891 procesor_costs from all tables; make cond_taken_branch_cost
4892 and cond_not_taken_branch_cost COST_N_INSNS based.
4894 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4896 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4898 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4900 * config/i386/i386.md (isa): Remove fma_avx512f.
4901 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4902 <avx512>_fmadd_<mode>_mask3<round_name>,
4903 <avx512>_fmsub_<mode>_mask<round_name>,
4904 <avx512>_fmsub_<mode>_mask3<round_name>,
4905 <avx512>_fnmadd_<mode>_mask<round_name>,
4906 <avx512>_fnmadd_<mode>_mask3<round_name>,
4907 <avx512>_fnmsub_<mode>_mask<round_name>,
4908 <avx512>_fnmsub_<mode>_mask3<round_name>,
4909 <avx512>_fmaddsub_<mode>_mask<round_name>,
4910 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4911 <avx512>_fmsubadd_<mode>_mask<round_name>,
4912 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4913 (*vec_widen_umult_even_v16si<mask_name>,
4914 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4915 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4917 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4919 * extend.texi: Add 'nocf_check' documentation.
4920 * gimple.texi: Add second parameter to
4921 gimple_build_call_from_tree.
4922 * invoke.texi: Add -fcf-protection documentation.
4923 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4925 2017-10-20 Richard Biener <rguenther@suse.de>
4927 PR tree-optimization/82473
4928 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4929 the largest input type.
4931 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4933 * c-attribs.c (handle_nocf_check_attribute): New function.
4934 (c_common_attribute_table): Add 'nocf_check' handling.
4935 * gimple-parser.c: Add second argument NULL to
4936 gimple_build_call_from_tree.
4937 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4938 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4940 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4941 * common.opt: Add fcf-protection flag.
4942 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4943 * flag-types.h: Add enum cf_protection_level.
4944 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4945 Add 'nocf_check' attribute propagation to gimple call.
4946 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4947 (gimple_build_call_from_tree): Update prototype.
4948 (gimple_call_nocf_check_p): New function.
4949 (gimple_call_set_nocf_check): Likewise.
4950 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4951 * ipa-icf.c: Add nocf_check attribute in statement hash.
4952 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4953 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4954 * toplev.c (process_options): Add flag_cf_protection handling.
4956 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4958 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4960 2017-10-20 Richard Biener <rguenther@suse.de>
4962 PR tree-optimization/82603
4963 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4964 remove false predicated stores.
4966 2017-10-20 Richard Biener <rguenther@suse.de>
4968 * graphite-isl-ast-to-gimple.c
4969 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4970 Remove return value and simplify, dump copied stmt after lhs
4972 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4973 Reduce dump verbosity.
4974 (gsi_insert_earliest): Likewise.
4975 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4976 * graphite.c (print_global_statistics): Adjust dumping.
4977 (print_graphite_scop_statistics): Likewise.
4978 (print_graphite_statistics): Do not dump loops here.
4979 (graphite_transform_loops): But here.
4981 2017-10-20 Nicolas Roche <roche@adacore.com>
4983 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4984 * configure: Regenerate.
4986 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4989 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4990 functions when optimizing replace GIMPLE_RETURN stmts with
4991 calls to __builtin_unreachable ().
4994 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4995 for -fsanitize=thread link of executables.
4996 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4997 link of executables.
5000 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
5002 (<shift_insn><mode>3<mask_name>): Change the last of the 3
5003 define_insns for logical vector shifts to use VI248_AVX512BW
5004 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
5005 condition, useless isa and prefix attributes. Change the first
5006 2 of these define_insns to ...
5007 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
5008 define_insn for avx512vl.
5009 (<shift_insn><mode>3): ... and this, new define_insn without
5010 masking for non-avx512vl.
5013 * config/i386/sse.md (*andnot<mode>3,
5014 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
5015 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
5016 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
5017 not applied use empty suffix even for TARGET_AVX512VL.
5018 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
5019 is applied, supply evex,evex or evex,evex,evex instead of just
5022 2017-10-20 Julia Koval <julia.koval@intel.com>
5024 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
5025 (OPTION_MASK_ISA_GFNI_UNSET): New.
5026 (ix86_handle_option): Handle OPT_mgfni.
5027 * config/i386/cpuid.h (bit_GFNI): New.
5028 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
5029 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
5030 * config/i386/i386.c (ix86_target_string): Add -mgfni.
5031 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
5032 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
5033 * config/i386/i386.opt: Add mgfni.
5035 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5037 * config/msp430/msp430.c (msp430_option_override): Disable
5038 -fdelete-null-pointer-checks.
5039 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5041 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5043 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5044 of x87 and SSE instructions.
5046 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5048 * asan.c (create_cond_insert_point): Do not update edge count.
5049 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5050 (afdo_propagate_circuit): Likewise.
5051 (afdo_calculate_branch_prob): Likewise.
5052 (afdo_annotate_cfg): Likewise.
5053 * basic-block.h (struct edge_def): Remove count.
5054 (edge_def::count): New accessor.
5055 * bb-reorder.c (rotate_loop): Update.
5056 (find_traces_1_round): Update.
5057 (connect_traces): Update.
5058 (sanitize_hot_paths): Update.
5059 * cfg.c (unchecked_make_edge): Update.
5060 (make_single_succ_edge): Update.
5061 (check_bb_profile): Update.
5062 (dump_edge_info): Update.
5063 (update_bb_profile_for_threading): Update.
5064 (scale_bbs_frequencies_int): Update.
5065 (scale_bbs_frequencies_gcov_type): Update.
5066 (scale_bbs_frequencies_profile_count): Update.
5067 (scale_bbs_frequencies): Update.
5068 * cfganal.c (connect_infinite_loops_to_exit): Update.
5069 * cfgbuild.c (compute_outgoing_frequencies): Update.
5070 (find_many_sub_basic_blocks): Update.
5071 * cfgcleanup.c (try_forward_edges): Update.
5072 (try_crossjump_to_edge): Update
5073 * cfgexpand.c (expand_gimple_cond): Update
5074 (expand_gimple_tailcall): Update
5075 (construct_exit_block): Update
5076 * cfghooks.c (verify_flow_info): Update
5077 (redirect_edge_succ_nodup): Update
5078 (split_edge): Update
5079 (make_forwarder_block): Update
5080 (duplicate_block): Update
5081 (account_profile_record): Update
5082 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5083 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5084 * cfgloopmanip.c (scale_loop_profile): Update.
5086 (lv_adjust_loop_entry_edge): Update.
5087 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5088 (force_nonfallthru_and_redirect): Update.
5089 (purge_dead_edges): Update.
5090 (rtl_flow_call_edges_add): Update.
5091 * cgraphunit.c (init_lowered_empty_function): Update.
5092 (cgraph_node::expand_thunk): Update.
5093 * gimple-pretty-print.c (dump_probability): Update.
5094 (dump_edge_probability): Update.
5095 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5096 * haifa-sched.c (sched_create_recovery_edges): Update.
5097 * hsa-gen.c (convert_switch_statements): Update.
5098 * ifcvt.c (dead_or_predicable): Update.
5099 * ipa-inline-transform.c (inline_transform): Update.
5100 * ipa-split.c (split_function): Update.
5101 * ipa-utils.c (ipa_merge_profiles): Update.
5102 * loop-doloop.c (add_test): Update.
5103 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5104 * lto-streamer-in.c (input_cfg): Update.
5105 (input_function): Update.
5106 * lto-streamer-out.c (output_cfg): Update.
5107 * modulo-sched.c (sms_schedule): Update.
5108 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5109 * predict.c (maybe_hot_edge_p): Update.
5110 (unlikely_executed_edge_p): Update.
5111 (probably_never_executed_edge_p): Update.
5112 (dump_prediction): Update.
5113 (drop_profile): Update.
5114 (propagate_unlikely_bbs_forward): Update.
5115 (determine_unlikely_bbs): Update.
5116 (force_edge_cold): Update.
5117 * profile.c (compute_branch_probabilities): Update.
5118 * reg-stack.c (better_edge): Update.
5119 * shrink-wrap.c (handle_simple_exit): Update.
5120 * tracer.c (better_p): Update.
5121 * trans-mem.c (expand_transaction): Update.
5122 (split_bb_make_tm_edge): Update.
5123 * tree-call-cdce.c: Update.
5124 * tree-cfg.c (gimple_find_sub_bbs): Update.
5125 (gimple_split_edge): Update.
5126 (gimple_duplicate_sese_region): Update.
5127 (gimple_duplicate_sese_tail): Update.
5128 (gimple_flow_call_edges_add): Update.
5129 (insert_cond_bb): Update.
5130 (execute_fixup_cfg): Update.
5131 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5132 * tree-complex.c (expand_complex_div_wide): Update.
5133 * tree-eh.c (lower_resx): Update.
5134 (unsplit_eh): Update.
5135 (cleanup_empty_eh_move_lp): Update.
5136 * tree-inline.c (copy_edges_for_bb): Update.
5137 (freqs_to_counts): Update.
5138 (copy_cfg_body): Update.
5139 * tree-ssa-dce.c (remove_dead_stmt): Update.
5140 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5141 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5142 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5143 (unloop_loops): Update.
5144 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5145 * tree-ssa-loop-split.c (connect_loops): Update.
5146 (split_loop): Update.
5147 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5148 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5149 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5150 * tree-ssa-reassoc.c (branch_fixup): Update.
5151 * tree-ssa-tail-merge.c (replace_block_by): Update.
5152 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5153 (compute_path_counts): Update.
5154 (update_profile): Update.
5155 (recompute_probabilities): Update.
5156 (update_joiner_offpath_counts): Update.
5157 (estimated_freqs_path): Update.
5158 (freqs_to_counts_path): Update.
5159 (clear_counts_path): Update.
5160 (ssa_fix_duplicate_block_edges): Update.
5161 (duplicate_thread_path): Update.
5162 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5163 (case_bit_test_cmp): Update.
5164 (collect_switch_conv_info): Update.
5165 (gen_inbound_check): Update.
5166 (do_jump_if_equal): Update.
5167 (emit_cmp_and_jump_insns): Update.
5168 * tree-tailcall.c (decrease_profile): Update.
5169 (eliminate_tail_call): Update.
5170 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5171 (vect_do_peeling): Update.
5172 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5173 * ubsan.c (ubsan_expand_null_ifn): Update.
5174 (ubsan_expand_ptr_ifn): Update.
5175 * value-prof.c (gimple_divmod_fixed_value): Update.
5176 (gimple_mod_pow2): Update.
5177 (gimple_mod_subtract): Update.
5178 (gimple_ic): Update.
5179 (gimple_stringop_fixed_value): Update.
5181 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5184 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5186 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5188 PR rtl-optimization/82395
5189 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5190 based on non_spilled_static_chain_regno_p.
5192 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5194 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5195 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5196 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5198 2017-10-19 Martin Sebor <msebor@redhat.com>
5200 PR tree-optimization/82596
5201 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5203 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5205 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5206 (get_mem_refs_of_builtin_call): Likewise.
5207 * builtins.c (expand_builtin_apply): Adjust call to
5208 allocate_dynamic_stack_space.
5209 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5210 the third argument to allocate_dynamic_stack_space, otherwise -1.
5211 (expand_builtin): Deal with all alloca variants.
5212 (is_inexpensive_builtin): Likewise.
5213 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5214 * calls.c (special_function_p): Deal with all alloca variants.
5215 (initialize_argument_information): Adjust call to
5216 allocate_dynamic_stack_space.
5217 (expand_call): Likewise.
5218 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5219 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5220 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5221 use it for the stack usage computation.
5222 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5223 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5224 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5225 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5226 (in_loop_p): Remove first argument and useless check.
5227 (pass_walloca::execute): Remove useless test and adjust call to above.
5228 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5229 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5230 (gimplify_call_expr): Deal with all alloca variants.
5231 * hsa-gen.c (gen_hsa_alloca): Likewise.
5232 (gen_hsa_insns_for_call): Likewise.
5233 * ipa-pure-const.c (special_builtin_state): Likewise.
5234 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5235 * tree-object-size.c (alloc_object_size): Likewise.
5236 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5237 (call_may_clobber_ref_p_1): Likewise.
5238 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5239 (ccp_fold_stmt): Likewise.
5240 (optimize_stack_restore): Likewise.
5241 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5242 (mark_all_reaching_defs_necessary_1): Likewise.
5243 (propagate_necessity): Likewise.
5244 (eliminate_unnecessary_stmts): Likewise.
5245 * tree.c (build_common_builtin_nodes): Build
5246 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5247 (build_alloca_call_expr): New function.
5248 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5249 (CASE_BUILT_IN_ALLOCA): Likewise.
5250 (build_alloca_call_expr): Declare.
5251 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5253 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5256 * dwarf2out.c (new_die_raw): New static inline function.
5257 (new_die): Use it to create the DIE.
5258 (add_AT_external_die_ref): Likewise.
5259 (clone_die): Likewise.
5260 (clone_as_declaration): Likewise.
5261 (dwarf2out_vms_debug_main_pointer): Likewise.
5262 (base_type_die): Likewise. Remove early return for corner cases.
5263 Do not call add_pubtype on the DIE here.
5264 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5265 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5266 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5267 native order exists for base types, attach the DIE manually and call
5268 add_pubtype on it. Do not equate a reverse order DIE to the type.
5270 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5272 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5273 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5275 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5277 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5279 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5280 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5281 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5282 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5283 * opts.c (sanitizer_opts): Add builtin.
5284 * ubsan.c (instrument_builtin): New function.
5285 (pass_ubsan::execute): Call it.
5286 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5287 * doc/invoke.texi: Document -fsanitize=builtin.
5289 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5290 builtins, store max (log2 (align), 0) into uchar field instead of
5291 align into uptr field.
5292 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5293 store uchar 0 field instead of uptr 0 field.
5294 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5295 instead of passing one address of struct with 2 locations pass
5296 two addresses of structs with 1 location each.
5297 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5298 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5299 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5300 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5301 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5302 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5303 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5304 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5306 2017-10-19 Martin Liska <mliska@suse.cz>
5309 * file-find.c (remove_prefix): Remove.
5310 * file-find.h (remove_prefix): Likewise.
5311 * gcc-ar.c: Remove smartness of lookup.
5313 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5315 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5316 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5317 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5319 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5322 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5323 (setcc + and to xor + setcc): New peephole2.
5325 2017-10-19 Tom de Vries <tom@codesourcery.com>
5327 * doc/sourcebuild.texi (Test Directives, Variants of
5328 dg-require-support): Add dg-require-stack-size.
5330 2017-10-19 Martin Liska <mliska@suse.cz>
5333 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5334 that have a large alignment.
5335 (gimplify_target_expr): Likewise.
5337 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5339 PR rtl-optimization/82602
5340 * ira.c (rtx_moveable_p): Return false for volatile asm.
5342 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5345 * config/i386/i386-modes.def (CCGZ): New CC mode.
5346 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5347 * config/i386/predicates.md (ix86_comparison_operator):
5349 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5350 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5351 with double-word subtraction.
5352 (put_condition_code): Handle CCGZmode.
5354 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5356 * wide-int.cc (debug (const wide_int &)): New.
5357 (debug (const wide_int *)): New.
5358 (debug (const widest_int &)): New.
5359 (debug (const widest_int *)): New.
5361 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5364 * lra-constraints.c (curr_insn_transform): Use non-input operand
5365 instead of output one for matched reload.
5367 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5369 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5370 (tree-ssa-loop-ivopts.h): New header file.
5371 (struct builtin_info): New fields.
5372 (classify_builtin_1): Compute and record base and offset parts for
5373 memset builtin partition by calling strip_offset.
5374 (offset_cmp, fuse_memset_builtins): New functions.
5375 (finalize_partitions): Fuse adjacent memset partitions by calling
5377 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5378 Expose the interface.
5379 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5381 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5383 PR tree-optimization/82574
5384 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5385 that data reference must be executed exactly once per iteration
5386 against the outermost loop in nest.
5387 (classify_partition): Update call to above function.
5389 2017-10-18 Richard Biener <rguenther@suse.de>
5391 PR tree-optimization/82591
5392 * graphite.c (graphite_transform_loops): Move code gen message
5394 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5395 Here. Handle scop_to_isl_ast failing.
5396 (scop_to_isl_ast): Limit the number of ISL operations.
5398 2017-10-18 Richard Biener <rguenther@suse.de>
5400 * graphite-isl-ast-to-gimple.c
5401 (translate_isl_ast_to_gimple::set_rename): Simplify.
5402 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5403 (graphite_copy_stmts_from_block): ... here.
5404 (copy_bb_and_scalar_dependences): Simplify.
5405 (add_parameters_to_ivs_params): Canonicalize.
5406 (generate_entry_out_of_ssa_copies): Simplify.
5407 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5408 by passing in ISL dimension.
5409 (parameter_index_in_region_1): Rename to ...
5410 (parameter_index_in_region): ... this.
5411 (extract_affine): Adjust assert, pass down parameter index.
5412 (add_param_constraints): Use range-info when available.
5413 (build_scop_context): Adjust.
5414 * sese.c (new_sese_info): Adjust.
5415 (free_sese_info): Likewise.
5416 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5417 Remove unused typedefs.
5418 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5420 2017-10-18 Martin Liska <mliska@suse.cz>
5422 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5424 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5426 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5427 (s390_sched_init): Do not reset s390_sched_state if we entered the
5428 current basic block via a fallthru edge and all others are unlikely.
5430 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5432 * config/s390/s390.c (NUM_SIDES): New variable.
5433 (LONGRUNNING_THRESHOLD): New variable.
5434 (LATENCY_FACTOR): New variable.
5435 (s390_sched_score): Decrease score for long-running instructions on
5437 (s390_sched_variable_issue): Perform bookkeeping for long-running
5440 2017-10-18 Richard Biener <rguenther@suse.de>
5442 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5443 Simplify with removal of the parameter rename map.
5444 (set_rename): Likewise.
5445 (should_copy_to_new_region): Likewise.
5446 (graphite_copy_stmts_from_block): Likewise.
5447 (copy_bb_and_scalar_dependences): Remove initialization of
5448 unused copied_bb_map.
5450 (copy_internal_parameters): Likewise.
5451 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5452 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5453 Use INTEGRAL_TYPE_P.
5454 (parameter_index_in_region_1): Rename to ...
5455 (assign_parameter_index_in_region): ... this. Assert we have
5456 a parameter we handle.
5457 (scan_tree_for_params): Adjust.
5458 * sese.h (parameter_rename_map_t): Remove.
5459 (struct sese_info_t): Remove unused parameter_rename_map and
5460 copied_bb_map members.
5461 * sese.c (new_sese_info): Adjust.
5462 (free_sese_info): Likewise.
5464 2017-10-18 Martin Liska <mliska@suse.cz>
5467 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5468 on an abnormal edge.
5470 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5472 * doc/invoke.texi (ffunction-sections and fdata-sections):
5475 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5477 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5478 the use statement can throw internally.
5480 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5482 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5483 any RTX present on the RHS of a SET.
5484 * compare-elim.c (try_eliminate_compare): Restore comment.
5486 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5488 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5489 may be also called on tcc_type nodes.
5490 * langhooks.c (lhd_tree_size): Likewise.
5492 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5494 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5495 format_warning_at_substring.
5496 (maybe_warn): Convert source_range * param to a location_t. Pass
5497 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5498 (format_directive): Remove code to extract source_ranges and
5499 source_range * in favor of just a location_t.
5500 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5502 * substring-locations.c (format_warning_va): Convert
5503 source_range * param to a location_t.
5504 (format_warning_at_substring): Likewise.
5505 * substring-locations.h (format_warning_va): Likewise.
5506 (format_warning_at_substring): Likewise.
5508 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5510 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5512 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5513 and scatter/gather ops.
5515 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5516 vec_gather_load and vec_scatter_store.
5517 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5518 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5519 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5520 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5521 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5522 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5524 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5526 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5527 Detect FTST instruction and handle its register pops. Only pop
5528 second operand if can_pop_second_op is true.
5529 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5530 set can_pop_second_op to false in the compare_for_stack_reg call.
5532 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5533 output_fp_compare for stack register operands.
5534 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5535 instructions here. Do not emit stack register pops here. Assert
5536 that FCOMPP pops next to top stack register. Rewrite function.
5538 2017-10-17 Nathan Sidwell <nathan@acm.org>
5541 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5542 use DECL_ASSEMBLER_NAME_RAW.
5545 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5548 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5549 Wilco Dijkstra <wilco.dijkstra@arm.com>
5551 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5552 converstion to Pmode from the buf_addr.
5554 2017-10-17 Richard Biener <rguenther@suse.de>
5556 * graphite-dependences.c (scop_get_reads_and_writes): Change
5557 output parameters to references.
5559 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5561 PR 71026/tree-optimization
5562 * fold-const.c (distribute_real_division): Removed.
5563 (fold_binary_loc): Remove calls to distribute_real_divison.
5565 2017-10-17 Richard Biener <rguenther@suse.de>
5567 * graphite-scop-detection.c
5568 (scop_detection::stmt_has_simple_data_refs_p): Always use
5569 the full nest as region.
5570 (try_generate_gimple_bb): Likewise.
5571 * sese.c (scalar_evolution_in_region): Simplify now that
5572 SCEV can handle instantiation in regions.
5573 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5574 in the non-loop part of a function if requested.
5576 2017-10-17 Richard Biener <rguenther@suse.de>
5578 PR tree-optimization/82563
5579 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5581 (graphite_regenerate_ast_isl): Call it.
5582 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5584 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5586 PR tree-optimization/82549
5587 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5588 Formatting fixes. Instead of calling make_bit_field_ref with negative
5591 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5593 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5594 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5595 _mm_maskz_reduce_ss): New.
5596 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5597 __builtin_ia32_reducess_mask): Ditto..
5598 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5599 * config/i386/sse.md (reduces<mode>): Renamed to ...
5600 (reduces<mode><mask_scalar_name>): ... this.
5601 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5603 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5604 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5606 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5608 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5609 * selftest-run-tests.c (selftest::run_tests): Call
5610 selftest::unique_ptr_tests_cc_tests.
5611 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5612 * unique-ptr-tests.cc: New file.
5614 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5617 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5619 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5620 (make_hard_regno_born, make_hard_regno_dead): Update
5621 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5623 2017-10-16 Jeff Law <law@redhat.com>
5625 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5627 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5629 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5631 2017-10-16 Olivier Hainque <hainque@adacore.com>
5633 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5634 with_cpu if we were configured for an e500v2 target cpu name.
5636 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5638 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5639 * doc/invoke.texi: Document +nodsp as a valid extension for
5642 2017-10-16 Martin Liska <mliska@suse.cz>
5644 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5645 (test_set_range): Likewise.
5646 (test_range_functions): Rename to ...
5647 (test_bit_in_range): ... this.
5648 (sbitmap_c_tests): Add new test.
5650 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5652 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5654 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5655 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5657 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5659 * config/aarch64/aarch64-builtins.c
5660 (aarch64_types_quadopu_lane_qualifiers): New.
5661 (TYPES_QUADOPU_LANE): New.
5662 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5663 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5664 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5665 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5666 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5667 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5668 (Vdottype, DOTPROD): New.
5669 (sur): Add SDOT and UDOT.
5671 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5673 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5674 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5675 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5677 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5678 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5679 Enable TARGET_DOTPROD.
5680 (cortex-a75.cortex-a55): Likewise.
5681 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5683 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5685 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5686 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5688 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5690 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5691 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5692 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5693 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5694 * config/arm/types.md (neon_dot, neon_dot_q): New.
5695 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5697 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5699 * config/arm/arm.h (TARGET_DOTPROD): New.
5700 * config/arm/arm.c (arm_arch_dotprod): New.
5701 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5702 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5703 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5704 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5705 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5706 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5707 * doc/invoke.texi (armv8.2-a): Document dotprod
5709 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5711 * i386.c (ix86_vec_cost): New function.
5712 (ix86_rtx_costs): Handle vector operations better.
5713 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5714 * x86-tune-costs.h: Add new costs to all tables.
5716 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5718 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5720 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5721 divsd, sqrtss and sqrtsd
5722 * x86-tune-costs.h: Add new entries to all costs.
5723 (znver1_cost): Fix to match real instruction latencies.
5725 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5726 Michael Collison <michael.collison@arm.com>
5728 * compare-elim.c: Include emit-rtl.h.
5729 (can_merge_compare_into_arith): New function.
5730 (try_validate_parallel): Likewise.
5731 (try_merge_compare): Likewise.
5732 (try_eliminate_compare): Call the above when no previous clobber
5734 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5737 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5741 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5742 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5746 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5747 to be any operand_equal_p operands. For & (B - 1) require
5748 B to be power of 2. Recognize
5749 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5751 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5754 * optabs.c (expand_memory_blockage): Fix call of
5755 targetm.have_memory_blockage.
5757 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5760 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5761 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5762 objects to extra_objs instead of overwriting it.
5764 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5766 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5767 Use any_fp_register_operand as operand[3] predicate. Simplify
5768 equality test for operands[2] and operands[4] memory location.
5769 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5770 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5771 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5772 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5773 any_fp_register_operand as operand[1] predicate. Simplify
5774 equality test for operands[0] and operands[3] memory location.
5775 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5776 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5777 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5779 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5781 * target-insns.def: Add memory_blockage.
5782 * optabs.c (expand_memory_blockage): New function.
5783 (expand_asm_memory_barrier): Rename ...
5784 (expand_asm_memory_blockage): ... to this.
5785 (expand_mem_thread_fence): Call expand_memory_blockage
5786 instead of expand_asm_memory_barrier.
5787 (expand_mem_singnal_fence): Ditto.
5788 (expand_atomic_load): Ditto.
5789 (expand_atomic_store): Ditto.
5790 * doc/md.texi (Standard Pattern Names For Generation):
5791 Document memory_blockage instruction pattern.
5793 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5795 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5796 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5797 * config/rl78/rl78.md: New define_expand "adddi3".
5799 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5801 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5804 2017-10-13 Jeff Law <law@redhat.com>
5806 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5808 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5811 * internal-fn.c (expand_mul_overflow): If both operands have
5812 the same highpart of -1 or 0 and the topmost bit of lowpart
5813 is different, overflow is if res <= 0 rather than res < 0.
5815 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5817 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5818 TARGET_P9_VECTOR code for unaligned_load case.
5820 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5822 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5824 2017-10-13 Nathan Sidwell <nathan@acm.org>
5826 * tree-core.h (tree_contains_struct): Make bool.
5827 * tree.c (tree_contains_struct): Likewise.
5828 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5829 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5830 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5831 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5832 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5834 2017-10-13 Richard Biener <rguenther@suse.de>
5836 * graphite-isl-ast-to-gimple.c
5837 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5838 parameters and dominance check.
5839 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5840 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5841 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5842 Do not update SSA form here or do intermediate IL verification.
5843 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5844 (graphite_initialize): Remove check on the number of loops in
5845 the function and inline into graphite_transform_loops.
5846 (graphite_finalize): Inline into graphite_transform_loops.
5847 (graphite_transform_loops): Perform SSA update and IL verification
5849 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5851 2017-10-13 Richard Biener <rguenther@suse.de>
5853 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5854 graphite_expression_type_precision): Avoid global constructor
5856 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5857 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5858 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5859 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5860 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5861 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5863 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5866 * config/i386/i386.h (ix86_red_zone_size): New.
5867 * config/i386/i386.md (push peephole2s): Replace
5868 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5870 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5871 Alan Hayward <alan.hayward@arm.com>
5872 David Sherwood <david.sherwood@arm.com>
5874 * combine.c (can_change_dest_mode): Reject changes in
5875 REGMODE_NATURAL_SIZE.
5877 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5878 Alan Hayward <alan.hayward@arm.com>
5879 David Sherwood <david.sherwood@arm.com>
5881 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5882 (expand_debug_source_expr): Likewise.
5883 * combine.c (combine_simplify_rtx): Likewise.
5884 * cse.c (fold_rtx): Likewise.
5885 * fwprop.c (canonicalize_address): Likewise.
5886 * targhooks.c (default_shift_truncation_mask): Likewise.
5888 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5889 Alan Hayward <alan.hayward@arm.com>
5890 David Sherwood <david.sherwood@arm.com>
5892 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5893 (widened_mode): Likewise.
5894 (expand_unop): Likewise.
5895 * ree.c (transform_ifelse): Likewise.
5896 (merge_def_and_ext): Likewise.
5897 (combine_reaching_defs): Likewise.
5898 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5900 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5901 Alan Hayward <alan.hayward@arm.com>
5902 David Sherwood <david.sherwood@arm.com>
5904 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5905 * combine.c (gen_lowpart_for_combine): Likewise.
5906 * dwarf2out.c (rtl_for_decl_location): Likewise.
5907 * final.c (alter_subreg): Likewise.
5908 * rtlhooks.c (gen_lowpart_general): Likewise.
5909 (gen_lowpart_if_possible): Likewise.
5911 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5912 Alan Hayward <alan.hayward@arm.com>
5913 David Sherwood <david.sherwood@arm.com>
5915 * calls.c (expand_call): Use subreg_lowpart_offset.
5916 * cse.c (cse_insn): Likewise.
5917 * regcprop.c (copy_value): Likewise.
5918 (copyprop_hardreg_forward_1): Likewise.
5920 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5923 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5924 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5925 =Q constraints to +Q and into insn condition add check
5926 that operands[0] and operands[1] are equal.
5927 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5928 =Q constraints to +Q and into insn condition add check
5929 that operands[0] is equal to either operands[1] or operands[2].
5932 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5933 instead of handling MINUS_EXPR twice (once for each argument),
5934 canonicalize operand order and handle just once, use rtype where
5935 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5938 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5939 any values of __C while still being pattern recognizable as a simple
5942 2017-10-13 Richard Biener <rguenther@suse.de>
5944 PR tree-optimization/82451
5946 2017-10-02 Richard Biener <rguenther@suse.de>
5948 PR tree-optimization/82355
5949 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5950 a mapping for the enclosing loop but avoid generating one for
5952 (copy_bb_and_scalar_dependences): Remove premature codegen
5953 error on PHIs in blocks duplicated into multiple places.
5954 * graphite-scop-detection.c
5955 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5956 in the region use it as loop and nest to analyze the DR in.
5957 (try_generate_gimple_bb): Likewise.
5958 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5959 (add_loop_constraints): For blocks in a loop not in the region
5960 create a dimension with a single iteration.
5961 * sese.h (gbb_loop_at_index): Remove assert.
5963 * cfgloop.c (loop_preheader_edge): For the loop tree root
5964 return the single successor of the entry block.
5965 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5966 Reset the SCEV hashtable and niters.
5967 * graphite-scop-detection.c
5968 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5969 assert that we only have POLYNOMIAL_CHREC that vary in loops
5970 contained in the region.
5971 (scop_detection::graphite_can_represent_expr): Adjust.
5972 (scop_detection::stmt_has_simple_data_refs_p): For loops
5973 not in the region set loop to NULL. The nest is now the
5974 entry edge to the region.
5975 (try_generate_gimple_bb): Likewise.
5976 * sese.c (scalar_evolution_in_region): Adjust for
5977 instantiate_scev change.
5978 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5979 Make nest parameter the edge into the region.
5980 (create_data_ref): Likewise.
5981 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5982 entry edge into a region and adjust instantiate_scev calls.
5983 (create_data_ref): Likewise.
5984 (graphite_find_data_references_in_stmt): Likewise.
5985 (find_data_references_in_stmt): Pass the loop preheader edge
5986 from the nest argument.
5987 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5988 parameter the edge into the region.
5989 (instantiate_parameters): Use the loop preheader edge as entry.
5990 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5992 (get_instantiated_value_entry): Make instantiate_below parameter
5993 the edge into the region.
5994 (instantiate_scev_name): Likewise. Adjust dominance checks,
5995 when we cannot use loop-based instantiation instantiate by
5996 walking use-def chains.
5997 (instantiate_scev_poly): Adjust.
5998 (instantiate_scev_binary): Likewise.
5999 (instantiate_scev_convert): Likewise.
6000 (instantiate_scev_not): Likewise.
6001 (instantiate_array_ref): Remove.
6002 (instantiate_scev_3): Likewise.
6003 (instantiate_scev_2): Likewise.
6004 (instantiate_scev_1): Likewise.
6005 (instantiate_scev_r): Do not blindly handle N-operand trees.
6006 Do not instantiate array-refs. Handle all constants and invariants.
6007 (instantiate_scev): Make instantiate_below parameter
6008 the edge into the region.
6009 (resolve_mixers): Use the loop preheader edge for the region
6010 parameter to instantiate_scev_r.
6011 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
6013 2017-10-13 Richard Biener <rguenther@suse.de>
6015 PR tree-optimization/82525
6016 * graphite-isl-ast-to-gimple.c
6017 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
6019 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
6020 Fail code generation when we cannot represent the isl integer.
6021 (binary_op_to_tree): Elide modulo operations that are no-ops
6022 in the type we code generate. Remove now superfluous code
6025 2017-10-13 Richard Biener <rguenther@suse.de>
6027 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
6028 (scop_detection::harmful_loop_in_region): Remove premature
6029 IV type restriction.
6030 (scop_detection::graphite_can_represent_scev): We can handle
6031 pointer IVs just fine.
6033 2017-10-13 Alan Modra <amodra@gmail.com>
6035 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6036 "Clobbers and Scratch Registers". Add paragraph on
6037 alternative to clobbers for scratch registers and OpenBLAS
6040 2017-10-13 Alan Modra <amodra@gmail.com>
6042 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6043 example of a memory input for a string of known length. Move
6044 commentary out of table. Add a number of new examples
6045 covering array memory inputs.
6047 2017-10-12 Martin Liska <mliska@suse.cz>
6049 PR tree-optimization/82493
6050 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6051 (test_range_functions): New function.
6052 (sbitmap_c_tests): Likewise.
6053 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6054 * selftest.h (sbitmap_c_tests): New function.
6056 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6058 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6060 * config/rs6000/amo.h: Fix spacing issue.
6062 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6065 * config/i386/i386.md (*ashl<mode>3_mask_1,
6066 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6067 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6070 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6072 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6073 (profile_probability): Set max_probability
6074 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6076 * profile-count.c (profile_probability::differs_from_p): Do not
6077 rely on max_probaiblity == 10000
6079 2017-10-12 Jeff Law <law@redhat.com>
6081 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6084 2017-10-12 Martin Sebor <msebor@redhat.com>
6088 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6089 (handle_alias_pairs): Call it.
6090 * common.opt (-Wattribute-alias): New option.
6091 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6092 * doc/invoke.texi (-Wattribute-alias): Document.
6094 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6097 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6099 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6101 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6102 (make_hard_regno_born, make_hard_regno_dead): Update
6103 bb_killed_pseudos and bb_gen_pseudos.
6105 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6107 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6109 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6111 * config/alpha/alpha.c (alpha_split_conditional_move):
6112 Use std::swap instead of manually swapping.
6113 (alpha_stdarg_optimize_hook): Ditto.
6114 (alpha_canonicalize_comparison): Ditto.
6116 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6118 * tree-loop-distribution.c (struct builtin_info): New struct.
6119 (struct partition): Refactor fields into struct builtin_info.
6120 (partition_free): Free struct builtin_info.
6121 (build_size_arg_loc, build_addr_arg_loc): Delete.
6122 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6123 information from struct builtin_info.
6124 (find_single_drs): New function refactored from classify_partition.
6125 Also moved builtin validity checks to this function.
6126 (compute_access_range, alloc_builtin): New functions.
6127 (classify_builtin_st, classify_builtin_ldst): New functions.
6128 (classify_partition): Refactor code into functions find_single_drs,
6129 classify_builtin_st and classify_builtin_ldst.
6130 (distribute_loop): Don't do runtime alias check when distributing
6132 (find_seed_stmts_for_distribution): New function.
6133 (pass_loop_distribution::execute): Refactor code finding seed
6134 stmts into above function. Support distribution for the innermost
6135 two-level loop nest. Adjust dump information.
6137 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6139 * tree-loop-distribution.c: Adjust the general comment.
6140 (NUM_PARTITION_THRESHOLD): New macro.
6141 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6142 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6143 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6144 in call to build_partition_graph.
6145 (finalize_partitions): New parameter. Make loop distribution more
6146 conservative by fusing more partitions.
6147 (distribute_loop): Don't do runtime alias check in case of loop nest
6149 (find_seed_stmts_for_distribution): New function.
6150 (prepare_perfect_loop_nest): New function.
6151 (pass_loop_distribution::execute): Refactor code finding seed stmts
6152 and loop nest into above functions. Support loop nest distribution.
6153 Adjust dump information accordingly.
6155 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6157 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6158 and set PTYPE_SEQUENTIAL for merged partition.
6160 2017-10-12 Richard Biener <rguenther@suse.de>
6162 PR tree-optimization/69728
6164 2017-09-19 Richard Biener <rguenther@suse.de>
6166 PR tree-optimization/69728
6167 * graphite-sese-to-poly.c (schedule_error): New global.
6168 (add_loop_schedule): Handle empty domain by failing the
6170 (build_original_schedule): Handle schedule_error.
6172 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6173 domain by returning an unchanged schedule.
6175 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6177 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6178 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6180 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6182 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6185 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6188 * expr.c (store_field): Don't optimize away bitsize == 0 store
6189 from CALL_EXPR with addressable return type.
6191 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6193 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6194 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6195 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6196 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6197 TARGET_ISEL instead of TARGET_ISEL<sel>.
6199 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6201 * config/rs6000/rs6000.c
6202 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6204 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6206 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6208 (reg_or_cint_operand): Fix comment.
6209 (reg_or_zero_operand): New predicate.
6210 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6211 * config/rs6000/rs6000.c (output_isel): Delete.
6212 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6213 instead of reg_or_cint_operand. Output instruction directly (not via
6215 (isel_unsigned_<mode>): Ditto.
6216 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6217 gpc_reg_operand. Add an instruction alternative for this. Output
6218 instruction directly.
6219 (*isel_reversed_unsigned_<mode>): Ditto.
6221 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6223 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6224 (TARGET_CANONICALIZE_COMPARISON): Define.
6226 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6229 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6230 Check whether the dest is REG before adding REG_EQUIV note.
6232 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6235 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6237 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6238 (make_hard_regno_born, make_hard_regno_dead): Update
6239 bb_killed_pseudos and bb_gen_pseudos.
6241 2017-10-11 Nathan Sidwell <nathan@acm.org>
6243 * incpath.h (enum incpath_kind): Name enum, prefix values.
6244 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6245 * incpath.c (heads, tails): Use INC_MAX.
6246 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6247 (merge_include_chains, split_quote_chain,
6248 register_include_chains): Update incpath_kind names.
6249 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6250 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6252 (add_framework_path, darwin_register_objc_includes): Likewise.
6253 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6255 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6257 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6258 Do not use float_operator operator predicate.
6259 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6260 * config/i386/predicates.md (float_operator): Remove predicate.
6262 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6264 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6265 (*jccxf_i387): Ditto.
6266 (*jcc<mode>_i387): Ditto.
6267 (*jccu<mode>_i387): Ditto.
6268 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6269 (*jcc_*_i387 splitters): Remove.
6270 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6271 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6272 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6275 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6277 * profile-count.h (slow_safe_scale_64bit): New function.
6278 (safe_scale_64bit): New inline.
6279 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6280 * profile-count.c: Include wide-int.h
6281 (slow_safe_scale_64bit): New.
6283 2017-10-11 Nathan Sidwell <nathan@acm.org>
6285 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6286 HAS_DECL_ASSEMBLER_NAME_P.
6287 * gimple-expr.c (gimple_decl_printable_name: Check
6288 HAS_DECL_ASSEMBLER_NAME_P too.
6289 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6290 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6291 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6292 * passes.c (rest_of_decl_compilation): Check
6293 HAS_DECL_ASSEMBLER_NAME_P too.
6294 * recog.c (verify_changes): Likewise.
6295 * tree-pretty-print.c (dump_decl_name): Likewise.
6296 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6298 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6299 (SET_DECL_ASSEMBLER_NAME): Use it.
6300 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6301 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6302 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6304 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6306 * config.gcc (i386, x86_64): Add extra objects.
6307 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6308 (ix86_min_insn_size): Declare.
6309 (ix86_issue_rate): Declare.
6310 (ix86_adjust_cost): Declare.
6311 (ia32_multipass_dfa_lookahead): Declare.
6312 (ix86_macro_fusion_p): Declare.
6313 (ix86_macro_fusion_pair_p): Declare.
6314 (ix86_bd_has_dispatch): Declare.
6315 (ix86_bd_do_dispatch): Declare.
6316 (ix86_core2i7_init_hooks): Declare.
6317 (ix86_atom_sched_reorder): Declare.
6318 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6319 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6320 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6321 (rip_relative_addr_p): Rename to ...
6322 (ix86_rip_relative_addr_p): ... this one; export.
6323 (memory_address_length): Update.
6324 (ix86_issue_rate): Move to x86-tune-sched.c.
6325 (ix86_flags_dependent): Move to x86-tune-sched.c.
6326 (ix86_agi_dependent): Move to x86-tune-sched.c.
6327 (exact_dependency_1): Move to x86-tune-sched.c.
6328 (exact_store_load_dependency): Move to x86-tune-sched.c.
6329 (ix86_adjust_cost): Move to x86-tune-sched.c.
6330 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6331 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6332 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6333 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6334 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6335 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6336 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6337 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6338 (min_insn_size): Rename to ...
6339 (ix86_min_insn_size): ... this one; export.
6340 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6341 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6342 (core2i7_first_cycle_multipass_backtrack): Move to
6343 x86-tune-sched-core.c.
6344 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6345 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6346 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6347 (ix86_avoid_jump_mispredicts): Update.
6348 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6349 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6350 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6351 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6352 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6353 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6354 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6355 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6356 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6357 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6358 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6359 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6360 (BIG): Move to ix86-tune-sched-bd.c.
6361 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6362 (enum insn_path): Move to ix86-tune-sched-bd.c.
6363 (get_mem_group): Move to ix86-tune-sched-bd.c.
6364 (is_cmp): Move to ix86-tune-sched-bd.c.
6365 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6366 (is_branch): Move to ix86-tune-sched-bd.c.
6367 (is_prefetch): Move to ix86-tune-sched-bd.c.
6368 (init_window): Move to ix86-tune-sched-bd.c.
6369 (allocate_window): Move to ix86-tune-sched-bd.c.
6370 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6371 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6372 (process_end_window): Move to ix86-tune-sched-bd.c.
6373 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6374 (find_constant): Move to ix86-tune-sched-bd.c.
6375 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6376 (has_immediate): Move to ix86-tune-sched-bd.c.
6377 (get_insn_path): Move to ix86-tune-sched-bd.c.
6378 (get_insn_group): Move to ix86-tune-sched-bd.c.
6379 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6380 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6381 (add_insn_window): Move to ix86-tune-sched-bd.c.
6382 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6383 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6384 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6385 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6386 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6387 (do_dispatch): Move to ix86-tune-sched-bd.c.
6388 (has_dispatch): Move to ix86-tune-sched-bd.c.
6389 * config/i386/t-i386: Add new object files.
6390 * config/i386/x86-tune-costs.h: New file.
6391 * config/i386/x86-tune-sched-atom.c: New file.
6392 * config/i386/x86-tune-sched-bd.c: New file.
6393 * config/i386/x86-tune-sched-core.c: New file.
6394 * config/i386/x86-tune-sched.c: New file.
6396 2017-10-11 Liu Hao <lh_mouse@126.com>
6398 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6399 the generic version below instead.
6400 (should_colorize): Recognize Windows consoles as terminals
6402 * pretty-print.c [__MINGW32__] (write_all): New function.
6403 [__MINGW32__] (find_esc_head): Likewise.
6404 [__MINGW32__] (find_esc_terminator): Likewise.
6405 [__MINGW32__] (eat_esc_sequence): Likewise.
6406 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6408 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6411 2017-10-11 Richard Biener <rguenther@suse.de>
6413 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6414 Properly call analyze_scalar_evolution with the loop of the stmt.
6416 2017-10-11 Richard Biener <rguenther@suse.de>
6418 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6419 * tree-core.h (tree_base): Add chrec_var union member.
6420 * tree.h (CHREC_VAR): Remove.
6421 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6422 * tree-chrec.h (build_polynomial_chrec): Adjust.
6423 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6424 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6426 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6428 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6429 * match.pd: ... here.
6430 ((T) X == (T) Y): Relax condition.
6432 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6434 PR tree-optimization/82472
6435 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6437 (break_alias_scc_partitions): Update postorder number.
6439 2017-10-11 Martin Liska <mliska@suse.cz>
6442 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6444 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6446 2017-10-11 Martin Liska <mliska@suse.cz>
6448 * print-rtl.c (print_insn): Move declaration of idbuf
6449 to same scope as name.
6451 2017-10-11 Martin Liska <mliska@suse.cz>
6456 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6459 2017-10-11 Martin Liska <mliska@suse.cz>
6462 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6465 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6467 * config.gcc (default_gnu_indirect_function): Default to yes for
6468 arm*-*-linux* with glibc.
6470 2017-10-11 Richard Biener <rguenther@suse.de>
6472 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6473 default-defs and types we do not want to analyze.
6474 (interpret_loop_phi): Replace unreachable code with an assert.
6475 (compute_scalar_evolution_in_loop): Remove and inline ...
6476 (analyze_scalar_evolution_1): ... here, replacing condition with
6477 what makes the intent clearer. Remove handling of cases
6478 get_scalar_evolution now handles.
6480 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6482 PR rtl-optimization/81434
6483 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6484 comment for main loop. In sched_group_found if, also add checks for
6485 pass and min_cost_group.
6487 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6489 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6490 (rs6000_insn_cost): New function.
6491 * config/rs6000/rs6000.md (cost): New attribute.
6493 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6494 H.J. Lu <hongjiu.lu@intel.com>
6498 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6499 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6500 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6501 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6502 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6503 OPTION_MASK_ISA_MMX and some other bit set require both
6504 mmx and the ISAs without the mmx bit.
6505 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6506 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6507 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6508 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6509 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6510 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6511 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6512 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6513 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6514 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6515 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6516 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6517 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6518 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6519 Add OPTION_MASK_ISA_MMX.
6521 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6523 * config.gcc (armv7*-*-freebsd*): New target.
6524 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6526 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6528 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6529 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6532 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6534 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6535 a template parameter.
6536 (WIDE_INT_REF_FOR): Update accordingly.
6537 * tree.h (wi::int_traits <const_tree>): Delete.
6538 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6539 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6540 (wi::tree_to_wide_ref): New typedef.
6541 (wi::to_wide): New function.
6542 * calls.c (get_size_range): Use wi::to_wide when operating on
6544 * cgraph.c (cgraph_node::create_thunk): Likewise.
6545 * config/i386/i386.c (ix86_data_alignment): Likewise.
6546 (ix86_local_alignment): Likewise.
6547 * dbxout.c (stabstr_O): Likewise.
6548 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6549 * expr.c (const_vector_from_tree): Likewise.
6550 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6551 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6552 (fold_negate_expr_1, int_const_binop_1, const_binop)
6553 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6554 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6555 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6556 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6557 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6558 (fold_not_const, round_up_loc): Likewise.
6559 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6560 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6561 (alloca_call_type): Likewise.
6562 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6563 * godump.c (go_output_typedef): Likewise.
6564 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6565 * internal-fn.c (get_min_precision): Likewise.
6566 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6567 * ipa-polymorphic-call.c
6568 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6569 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6570 (ipa_modify_call_arguments): Likewise.
6571 * match.pd: Likewise.
6572 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6573 * print-tree.c (print_node_brief, print_node): Likewise.
6574 * stmt.c (expand_case): Likewise.
6575 * stor-layout.c (layout_type): Likewise.
6576 * tree-affine.c (tree_to_aff_combination): Likewise.
6577 * tree-cfg.c (group_case_labels_stmt): Likewise.
6578 * tree-data-ref.c (dr_analyze_indices): Likewise.
6579 (prune_runtime_alias_test_list): Likewise.
6580 * tree-dump.c (dequeue_and_dump): Likewise.
6581 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6582 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6583 * tree-pretty-print.c (dump_generic_node): Likewise.
6584 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6585 (simple_iv_with_niters): Likewise.
6586 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6587 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6588 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6589 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6590 (refine_value_range_using_guard, number_of_iterations_ne_max)
6591 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6592 (get_cst_init_from_scev, record_nonwrapping_iv)
6593 (scev_var_range_cant_overflow): Likewise.
6594 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6595 * tree-ssa-pre.c (compute_avail): Likewise.
6596 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6597 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6598 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6599 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6600 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6601 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6602 (dump_case_nodes, try_switch_expansion): Likewise.
6603 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6604 (vect_do_peeling): Likewise.
6605 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6606 * tree-vect-stmts.c (vectorizable_load): Likewise.
6607 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6608 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6609 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6610 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6611 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6612 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6613 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6614 (evrp_dom_walker::before_dom_children): Likewise.
6615 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6616 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6617 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6618 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6619 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6620 (get_range_pos_neg): Likewise.
6621 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6622 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6623 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6624 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6625 * config/avr/avr.c (avr_fold_builtin): Likewise.
6626 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6627 * config/msp430/msp430.c (msp430_attr): Likewise.
6628 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6629 * config/powerpcspe/powerpcspe-c.c
6630 (altivec_resolve_overloaded_builtin): Likewise.
6631 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6632 (rs6000_expand_ternop_builtin): Likewise.
6633 * config/rs6000/rs6000-c.c
6634 (altivec_resolve_overloaded_builtin): Likewise.
6635 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6636 (rs6000_expand_ternop_builtin): Likewise.
6637 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6639 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6641 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6642 when copying loop nest with only one inner loop.
6644 2017-10-10 Richard Biener <rguenther@suse.de>
6646 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6647 blocks if SCEV is active.
6648 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6650 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6651 (scev_initialize): Assert we are not yet initialized.
6653 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6655 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6656 inner loop's exit stmt by making it always exit the loop, otherwise
6657 we would generate an infinite empty loop.
6659 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6661 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6662 renaming variables in new preheader if it's deleted.
6664 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6666 * tree-loop-distribution.c (struct partition): Remove unused field
6667 loops of the structure.
6668 (partition_alloc, partition_free): Ditto.
6669 (build_rdg_partition_for_vertex): Ditto.
6671 2017-10-09 Jeff Law <law@redhat.com>
6673 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6674 return type to match prototype and documentation.
6676 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6678 * config/rs6000/rs6000.c (processor_costs): Move to ...
6679 * config/rs6000/rs6000.h: ... here.
6680 (rs6000_cost): Declare.
6682 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6684 * except.c (setjmp_fn): New global variable.
6685 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6686 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6687 if DONT_USE_BUILTIN_SETJMP is defined.
6689 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6691 * target.def (insn_cost): New hook.
6692 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6693 * doc/tm.texi: Regenerate.
6694 * rtlanal.c (insn_cost): Use the new hook.
6696 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6698 * combine.c (combine_validate_cost): Compute the new insn_cost,
6699 not just pattern_cost.
6700 (try_combine): Adjust comment.
6702 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6704 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6706 * combine.c (uid_insn_cost): Adjust comment.
6707 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6709 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6710 * dse.c (find_shift_sequence): Ditto.
6711 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6712 (bb_valid_for_noce_process_p): Use pattern_cost.
6713 * rtl.h (insn_rtx_cost): Delete.
6714 (pattern_cost): New prototype.
6715 (insn_cost): New prototype.
6716 * rtlanal.c (insn_rtx_cost): Rename to...
6717 (pattern_cost): ... this.
6720 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6722 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6723 (*jcc<mode>_0_r_i387): Ditto.
6724 (*jccxf_r_i387): Ditto.
6725 (*jcc<mode>_r_i387): Ditto.
6726 (*jccu<mode>_r_i387): Ditto.
6727 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6728 (*jcc): Rename from *jcc_1.
6730 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6732 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6733 deferred rescans after the lvx/stvx recombination pre-pass.
6735 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6737 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6738 memory operation instruction support.
6739 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6740 (rs6000-ibm-aix[789]*): Likewise.
6741 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6742 Document new functions.
6744 2017-10-09 Richard Biener <rguenther@suse.de>
6746 PR tree-optimization/82397
6747 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6748 equality only for semantically equal trees.
6750 2017-10-09 Richard Biener <rguenther@suse.de>
6752 PR tree-optimization/82449
6753 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6754 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6755 allow constant addresses.
6756 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6759 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6761 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6764 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6767 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6770 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6773 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6775 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6778 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6779 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6781 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6783 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6784 (WI_BINARY_PREDICATE_RESULT): Likewise.
6785 (wi::binary_traits::operator_result): New type.
6786 (wi::binary_traits::predicate_result): Likewise.
6787 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6788 (generic_wide_int::operator==, generic_wide_int::operator!=)
6789 (generic_wide_int::operator&, generic_wide_int::and_not)
6790 (generic_wide_int::operator|, generic_wide_int::or_not)
6791 (generic_wide_int::operator^, generic_wide_int::operator+
6792 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6794 (operator~, unary operator-, operator==, operator!=, operator&)
6795 (operator|, operator^, operator+, binary operator-, operator*): New
6797 * expr.c (get_inner_reference): Use wi::bit_and_not.
6798 * fold-const.c (fold_binary_loc): Likewise.
6799 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6800 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6801 (bit_value_binop): Likewise.
6802 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6803 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6804 (extract_range_from_binary_expr_1): Likewise.
6805 (masked_increment): Likewise.
6806 (simplify_bit_ops_using_ranges): Likewise.
6808 2017-10-09 Martin Jambor <mjambor@suse.cz>
6811 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6812 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6813 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6815 (hsa_fixup_mov_insn_type): New function.
6816 (hsa_op_with_type::get_in_type): Use it.
6817 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6818 immediates in an assert.
6819 (hsa_op_with_type::extend_int_to_32bit): New method.
6820 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6821 types. Convert to dest type if necessary.
6822 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6823 (reg_for_gimple_ssa): Pass false as min32int to
6824 hsa_type_for_scalar_tree_type.
6825 (gen_hsa_addr): Fixup type when creating addresable temporary.
6826 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6827 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6828 necessary. Call hsa_fixup_mov_insn_type.
6829 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6830 extend operands and convert to dest type if necessary.
6831 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6832 to dest type if necessary.
6833 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6834 if conversion nt necessary and size matches.
6835 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6836 to dest type if necessary.
6837 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6838 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6840 (gen_hsa_clrsb): Likewise.
6841 (gen_hsa_ffs): Likewise.
6842 (gen_hsa_divmod): Extend operands and convert to dest type if
6844 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6846 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6848 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6849 Remove empty default arguments. Use a brace block as output
6851 (conditional return): Ditto.
6853 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6854 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6856 (group_ending_nop): Ditto.
6857 (doloop_end): Ditto.
6858 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6859 (splitters for those): Ditto.
6861 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6863 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6864 a conditional jump (and the compare for it) so that pc_rtx is the
6866 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6867 for the deleted and renamed ctr<mode>_internal[234] patterns.
6868 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6869 Delete second conditional return pattern.
6870 (ctr<mode>_internal2): Delete this second bdnz pattern.
6871 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6872 (ctr<mode>_internal4): Delete this second bdz pattern.
6874 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6876 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6877 (always_initialized_rtx_for_ssa_name_p): New predicate.
6878 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6879 (finish_out_of_ssa): Free new field of SA.
6880 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6881 * tree-ssa-coalesce.c: Include tree-ssa.h.
6882 (get_parm_default_def_partitions): Remove extern keyword.
6883 (get_undefined_value_partitions): New function.
6884 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6885 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6887 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6889 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6891 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6893 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6895 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6896 for targets that preffer 128bit.
6898 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6900 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6902 2017-10-08 Olivier Hainque <hainque@adacore.com>
6904 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6905 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6907 (thumb_set_return_address): Likewise.
6909 2017-10-08 Olivier Hainque <hainque@adacore.com>
6911 * common/config/arm/arm-common.c (arm_except_unwind_info):
6912 Handle DWARF2_UNWIND_INFO.
6914 2017-10-07 Michael Collison <michael.collison@arm.com>
6916 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6919 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6921 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6922 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6923 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6924 defined, force the creation of a new block for a dispatch label.
6926 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6928 * invoke.texi (Wsuggest-attribute=cold): Document.
6929 * common.opt (Wsuggest-attribute=cold): New
6930 * ipa-pure-const.c (warn_function_cold): New function.
6931 * predict.c (compute_function_frequency): Use it.
6932 * predict.h (warn_function_cold): Declare.
6934 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6936 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6939 2017-10-06 Martin Liska <mliska@suse.cz>
6941 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6942 keyword for member functions.
6943 (struct sanopt_tree_couple): New struct.
6944 (struct sanopt_tree_couple_hash): New function.
6945 (struct sanopt_ctx): Add new hash_map.
6946 (has_dominating_ubsan_ptr_check): New function.
6947 (record_ubsan_ptr_check_stmt): Likewise.
6948 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6949 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6950 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6952 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6955 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6956 aarch64_simd_valid_immediate on CONST_VECTORs.
6957 (aarch64_reg_or_bic_imm): Likewise.
6959 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6961 PR rtl-optimization/82396
6962 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6964 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6966 * graphite-dependences.c (scop_get_reads): Move code to...
6967 (scop_get_must_writes): Move code to...
6968 (scop_get_may_writes): Move code to...
6969 (scop_get_reads_and_writes): ... here.
6970 (scop_get_dependences): Call scop_get_reads_and_writes.
6972 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6974 PR tree-optimization/82434
6975 * fold-const.h (can_native_encode_type_p,
6976 can_native_encode_string_p): Remove.
6977 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6978 don't encode anything, just return what would be otherwise returned.
6979 (native_encode_fixed, native_encode_complex, native_encode_vector):
6981 (native_encode_string): Likewise. Inline by hand
6982 can_native_encode_string_p.
6983 (can_native_encode_type_p): Remove.
6984 (can_native_encode_string_p): Remove.
6985 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6986 STRING_CSTs using can_native_encode_string_p, test all
6987 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6988 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6989 argument from native_encode_expr.
6990 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6991 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6994 2017-10-06 Richard Biener <rguenther@suse.de>
6996 PR tree-optimization/82397
6997 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6998 operand_equal_p but rely on data_ref_compare_tree for detecting
7000 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
7001 to match up with dr_group_sort_cmp.
7003 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7006 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
7008 * config/s390/s390-builtin-types.def: Regenerate.
7010 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7013 * config/s390/s390-builtin-types.def: Regenerate.
7014 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
7015 Change flag from B_VXE to B_VX.
7016 (s390_vec_min_dbl): Remove B_VXE flag.
7018 2017-10-06 Richard Biener <rguenther@suse.de>
7020 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
7021 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
7022 translate_isl_ast_to_gimple::is_valid_rename,
7023 translate_isl_ast_to_gimple::get_rename,
7024 translate_isl_ast_to_gimple::get_def_bb_for_const,
7025 translate_isl_ast_to_gimple::get_new_name,
7026 translate_isl_ast_to_gimple::collect_all_ssa_names,
7027 translate_isl_ast_to_gimple::copy_loop_phi_args,
7028 translate_isl_ast_to_gimple::collect_all_ssa_names,
7029 translate_isl_ast_to_gimple::copy_loop_phi_args,
7030 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
7031 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
7032 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
7033 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
7034 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
7035 translate_isl_ast_to_gimple::copy_cond_phi_args,
7036 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7037 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7038 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7039 translate_isl_ast_to_gimple::rename_uses,
7040 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7041 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7042 (set_rename_for_each_def): Likewise.
7043 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7044 here. Handle rewriting SCEV analyzable uses here.
7045 (copy_bb_and_scalar_dependences): Generate code for PHI
7047 (graphite_regenerate_ast_isl): Adjust.
7048 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7049 (add_write, add_read): New functions.
7050 (build_cross_bb_scalars_def): Use it and simplify.
7051 (build_cross_bb_scalars_use): Likewise.
7052 (graphite_find_cross_bb_scalar_vars): Inline into...
7053 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7054 simulating out-of-SSA. Compute liveout and add dependencies.
7055 (build_scops): Force an empty entry block.
7056 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7058 (sese_build_liveouts): Declare.
7059 (sese_trivially_empty_bb_p): Likewise.
7060 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7061 compute liveout and debug_liveout.
7062 (sese_bad_liveouts_use): Remove.
7063 (sese_reset_debug_liveouts_bb): Likewise.
7064 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7065 (sese_build_liveouts): Build liveout and debug_liveout and store
7067 (new_sese_info): Adjust.
7068 (free_sese_info): Likewise.
7069 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7070 do not build liveout here.
7071 (move_sese_in_condition): Adjust region entry.
7072 (scev_analyzable_p): Match up with chrec_apply requirements.
7073 (sese_trivially_empty_bb_p): New.
7074 * tree-into-ssa.c (get_reaching_def): Properly support generating
7075 default-defs for incremental rewrite of anonymous names.
7077 2017-10-06 Richard Biener <rguenther@suse.de>
7079 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7080 precision do not perform modulo reduction.
7082 2017-10-06 Richard Biener <rguenther@suse.de>
7084 PR tree-optimization/82436
7085 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7086 conservatively choose the vectorization factor when checking
7087 whether we can perform the required load permutation.
7088 (vect_transform_slp_perm_load): Assert when we may not fail.
7090 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7092 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7093 message for incompatible -msdata=* and -mcall-* options.
7095 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7097 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7098 rate for post-reload scheduling.
7100 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7102 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7104 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7106 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7107 to improve monte carlo in scimark.
7109 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7111 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7112 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7113 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7114 pentium4_cost, nocona_cost): Set reassociation width to 1.
7115 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7116 width to 2 for fp operations and 1 otherwise.
7117 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7119 (atom_cost): Set reassociation width to 2.
7120 (slm_cost, generic_cost): Set fp reassociation width
7121 to 2 and 1 otherwise.
7122 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7123 (core_cost): Set fp reassociation width to 4 and vector to 2.
7124 (ix86_reassociation_width): Rewrite using cost table; special case
7125 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7126 and TARGET_AVX128_OPTIMAL.
7127 * config/i386/i386.h (processor_costs): Add
7128 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7129 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7130 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7131 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7132 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7133 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7135 2017-10-05 Nathan Sidwell <nathan@acm.org>
7137 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7139 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7141 * config/arm/arm.c (arm_test_fpu_data): New.
7142 (arm_run_selftests): Call arm_test_fpu_data.
7144 2017-10-04 Nathan Sidwell <nathan@acm.org>
7146 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7148 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7150 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7152 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7153 check type for aarch64_simd_valid_immediate.
7154 (aarch64_output_simd_mov_immediate): Update prototype.
7155 (aarch64_simd_valid_immediate): Update prototype.
7156 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7157 support for ORR-immediate.
7158 (and<mode>3): modified pattern to add support for BIC-immediate.
7159 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7160 now checks for valid immediate for BIC and ORR based on new enum
7162 (aarch64_output_simd_mov_immediate): Function now used to output
7163 BIC/ORR imm as well based on new enum argument.
7164 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7166 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7167 (aarch64_reg_or_bic_imm): Likewise.
7169 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7171 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7172 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7174 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7178 PR rtl-optimization/82396
7179 * haifa-sched.c (autopref_multipass_init): Simplify
7181 (autopref_rank_data): Simplify sort order.
7182 * sched-int.h (autopref_multipass_data_): Remove
7183 multi_mem_insn_p, min_offset and max_offset.
7185 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7187 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7189 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7191 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7194 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7196 * doc/sourcebuild.texi: Document vect_long_mult.
7198 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7200 PR tree-optimization/82413
7201 * fold-const.c (build_range_check): Use widest_int when comparing
7202 the maximum ETYPE value with HIGH.
7204 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7206 PR rtl-optimization/82396
7207 * haifa-sched.c (autopref_multipass_init): Simplify
7209 (autopref_rank_data): Simplify sort order.
7210 * sched-int.h (autopref_multipass_data_): Remove
7211 multi_mem_insn_p, min_offset and max_offset.
7213 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7215 PR tree-optimization/82381
7216 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7217 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7220 PR tree-optimization/82374
7221 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7222 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7223 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7224 current_function_decl to the new decl.
7226 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7228 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7229 helper macro for IEEE float128 hardware built-in functions.
7230 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7232 (TRUNCF128_ODD): Likewise.
7233 (ADDF128_ODD): Likewise.
7234 (SUBF128_ODD): Likewise.
7235 (MULF128_ODD): Likewise.
7236 (DIVF128_ODD): Likewise.
7237 (FMAF128_ODD): Likewise.
7238 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7239 UNSPEC_TRUNC_ROUND_TO_ODD.
7240 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7241 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7242 floating point round to odd instructions.
7243 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7244 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7245 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7246 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7247 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7248 (trunc<mode>sf2_hw): Change the truncate with round to odd
7249 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7250 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7251 to odd hardware instructions.
7252 (sub<mode>3_odd): Likewise.
7253 (mul<mode>3_odd): Likewise.
7254 (div<mode>3_odd): Likewise.
7255 (sqrt<mode>2_odd): Likewise.
7256 (fma<mode>4_odd): Likewise.
7257 (fms<mode>4_odd): Likewise.
7258 (nfma<mode>4_odd): Likewise.
7259 (nfms<mode>4_odd): Likewise.
7260 (trunc<mode>df2_odd): Change the truncate with round to odd
7261 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7263 * doc/extend.texi (PowerPC built-in functions): Update documentation
7264 for existing IEEE float128-bit built-in functions. Add built-in
7265 functions that generate the IEEE 128-bit floating point round to
7268 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7270 PR rtl-optimization/77729
7271 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7272 to (X&(C1&~C2))|C2 transformations.
7274 2017-10-03 Martin Jambor <mjambor@suse.cz>
7276 PR tree-optimization/82363
7277 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7278 mismatch, mark lacc written regardless of racc.
7280 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7282 PR tree-optimization/82381
7283 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7284 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7285 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7286 fallthrough into reassoc_stmt_dominates_stmt_p.
7289 * combine.c (combine_instructions): Don't combine in unreachable
7292 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7296 function to not use the have_cpu variable. Do not set cpu_index,
7297 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7299 (rs6000_valid_attribute_p): Remove duplicate initializations of
7300 old_optimize and func_optimize.
7301 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7302 (rs6000_activate_target_options): Make global.
7303 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7306 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7308 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7309 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7310 Set *poffset to 0 and *psize and *pmax_size to -1 if
7311 *poffset + *psize overflows in HOST_WIDE_INT.
7313 PR tree-optimization/82387
7314 PR tree-optimization/82388
7315 PR tree-optimization/82389
7316 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7317 instead of live_bytes non-NULL.
7319 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7322 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7325 2017-10-02 Richard Biener <rguenther@suse.de>
7327 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7328 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7329 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7331 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7333 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7334 requested precision matches the type's.
7335 * calls.c (alloc_max_size): Calculate the new candidate size as
7336 a widest_int and use wi::to_widest when comparing it with the
7337 current candidate size.
7338 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7339 zero rather than integer_zero_node.
7340 * match.pd: Check for a no-op conversion before using wi::add
7341 rather than after. Use tree_to_uhwi when summing small shift
7342 counts into an unsigned int.
7344 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7345 Alan Hayward <alan.hayward@arm.com>
7346 David Sherwood <david.sherwood@arm.com>
7349 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7350 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7351 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7352 POINTER_AND_FP_REGS.
7354 2017-10-02 Richard Biener <rguenther@suse.de>
7356 PR tree-optimization/82355
7357 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7358 a mapping for the enclosing loop but avoid generating one for
7360 (copy_bb_and_scalar_dependences): Remove premature codegen
7361 error on PHIs in blocks duplicated into multiple places.
7362 * graphite-scop-detection.c
7363 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7364 in the region use it as loop and nest to analyze the DR in.
7365 (try_generate_gimple_bb): Likewise.
7366 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7367 (add_loop_constraints): For blocks in a loop not in the region
7368 create a dimension with a single iteration.
7369 * sese.h (gbb_loop_at_index): Remove assert.
7371 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7373 * omp-expand.c (adjust_context_scope): New function.
7374 (expand_parallel_call): Call adjust_context_scope.
7376 2017-10-01 Jeff Law <law@redhat.com>
7378 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7379 dom_opt_dom_walker class with direct access to private members.
7380 Add comments. Call test_for_singularity.
7381 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7382 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7383 m_dummy_cond anymore.
7384 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7386 (pass_dominator:execute): Build the dummy_cond here and pass it
7387 to the dom_opt_dom_walker ctor.
7388 (test_for_singularity): New function.
7390 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7391 Maya Rashish <coypu@sdf.org>
7393 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7394 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7395 (alpha*-*-netbsd*) Use nbsd_tm_file.
7396 (arm*-*-netbsdelf*) Likewise.
7397 (i[34567]86-*-netbsdelf*) Likewise.
7398 (x86_64-*-netbsd*) Likewise.
7399 (mips*-*-netbsd*) Likewise.
7400 (powerpc-*-netbsd*) Likewise.
7401 (sh*-*-netbsd*) Likewise.
7402 (sparc-*-netbsdelf*) Likewise.
7403 (sparc64-*-netbsd*) Likewise.
7404 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7406 (vax-*-netbsdelf*) Likewise.
7407 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7408 (UINT_FAST8_TYPE) Likewise.
7409 (INT_FAST16_TYPE) Check CHAR_FAST16.
7410 (UINT_FAST16_TYPE) Likewise.
7412 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7415 * config/i386/i386.md
7416 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7417 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7418 *divmodsi4_zext_2): New define_insn_and_split.
7419 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7420 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7421 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7422 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7423 New define_insn_and_split.
7424 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7425 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7426 operands[1] having DImode when mode is SImode.
7428 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7429 always SImode for DIV and MOD in REG_EQUAL notes.
7431 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7434 * match.pd: Fix handling of NaNs in pattern.
7436 2017-09-29 Jeff Law <law@redhat.com>
7438 * sbitmap.c (bitmap_bit_in_range_p): New function.
7439 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7440 * tree-ssa-dse.c (live_bytes_read): New function.
7441 (dse_classify_store): Ignore reads of dead bytes.
7443 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7444 typos and whitespace errors.
7445 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7446 * config/s390/s390.c (s390_emit_prologue): Likewise.
7448 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7451 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7452 with a symbol for LRA.
7454 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7456 PR rtl-optimization/82338
7457 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7459 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7461 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7462 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7463 (qsort_chk): Declare.
7464 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7465 (qsort_chk): New function.
7467 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7469 PR tree-optimization/82337
7470 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7471 phi definition if the PHI result appears in an abnormal PHI.
7472 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7473 basis appears in an abnormal PHI.
7475 2017-09-29 Richard Biener <rguenther@suse.de>
7477 * graphite-isl-ast-to-gimple.c
7478 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7479 (binary_op_to_tree): Use it.
7480 (get_rename_from_scev): Likewise.
7481 (copy_loop_phi_nodes): Likewise.
7482 (copy_bb_and_scalar_dependences): Likewise.
7483 (translate_pending_phi_nodes): Likewise.
7485 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7488 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7489 for movabsq $(i32 << shift), r64.
7491 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7493 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7494 index when encoding %esp as %rsp to avoid 0x67 prefix.
7496 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7498 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7499 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7501 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7503 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7504 Extensions with more than 16 double VFP registers.
7505 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7506 to_clear_mask and all code related to it. Replace the remaining
7507 entry by a sbitmap and adapt code accordingly.
7509 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7511 * brig-builtins.def: Change pure attributes to const.
7513 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7515 * config.gcc (default_gnu_indirect_function): Default to yes for
7516 sparc*-*-linux* with glibc.
7518 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7520 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7521 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7522 when creating .init_array and .fini_array sections with priority
7525 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7528 * config/aarch64/aarch64.c
7529 (aarch64_builtin_support_vector_misalignment): Always return false
7530 when misalignment is unknown.
7532 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7534 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7535 this function to return false if the definition used by the swap
7536 instruction is artificial, or if the memory address from which the
7537 constant value is loaded is not represented by a base address held
7538 in a register or if the base address register is a frame or stack
7539 pointer. Additionally, return false if the base address of the
7540 loaded constant is a SYMBOL_REF but is not considered to be a
7542 (replace_swapped_load_constant): New function.
7543 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7544 loaded constant vector with a load of a swapped constant vector.
7546 2017-09-27 Carl Love <cel@us.ibm.com>
7548 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7549 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7550 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7553 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7555 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7556 first, always call autopref_rank_data otherwise.
7558 2017-09-27 Richard Biener <rguenther@suse.de>
7560 * graphite-scop-detection.c (find_scop_parameters): Move
7561 loop bound handling ...
7562 (gather_bbs::before_dom_children): ... here, avoiding the need
7563 to build scop_info->loop_nest.
7564 (record_loop_in_sese): Remove.
7565 * sese.h (sese_info_t::loop_nest): Remove.
7566 * sese.c (new_sese_info): Do not allocate loop_nest.
7567 (free_sese_info): Do not free loop_nest.
7569 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7572 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7573 lhs from calls if the lhs has addressable type.
7575 2017-09-27 Richard Biener <rguenther@suse.de>
7577 * graphite.h (scop::max_alias_set): New member.
7578 * graphite-scop-detection.c: Remove references to non-existing
7579 --param in comments.
7580 (build_alias_sets): Record the maximum alias set used for drs.
7581 (build_scops): Support zero as unlimited for
7582 --param graphite-max-arrays-per-scop.
7583 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7585 (build_poly_sr_1): ... here. Compute alias set based on the
7586 maximum alias set used for drs rather than
7587 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7589 2017-09-27 Richard Biener <rguenther@suse.de>
7591 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7592 --param loop-block-tile-size=0 to disable tiling.
7594 2017-09-27 Richard Biener <rguenther@suse.de>
7596 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7597 (graphite-max-nb-scop-params): Document special value zero.
7598 * domwalk.h (dom_walker::STOP): New symbolical constant.
7599 (dom_walker::dom_walker): Add optional parameter for bb to
7601 (dom_walker::~dom_walker): Declare.
7602 (dom_walker::before_dom_children): Document STOP return value.
7603 (dom_walker::m_user_bb_to_rpo): New member.
7604 (dom_walker::m_bb_to_rpo): Likewise.
7605 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7606 mapping here if not provided by the user.
7607 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7608 provided by the user.
7609 (dom_walker::STOP): Define.
7610 (dom_walker::walk): Do not compute bb to RPO mapping here.
7611 Support STOP return value from before_dom_children to stop
7613 * graphite-optimize-isl.c (optimize_isl): If the schedule
7614 is the same still generate code if -fgraphite-identity
7615 or -floop-parallelize-all are given.
7616 * graphite-scop-detection.c: Include cfganal.h.
7617 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7619 (gather_bbs::before_dom_children): Return STOP for BBs
7621 (build_scops): Compute bb to RPO mapping and pass it to
7622 the domwalk. Treat --param graphite-max-nb-scop-params=0
7623 as not limiting the number of params.
7624 * graphite.c (graphite_initialize): Remove limit on the
7625 number of basic-blocks in a function.
7626 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7627 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7628 default value of 10.
7630 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7632 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7633 Adjust code to eliminate needing to do the shift right 32-bits
7634 operation after XSCVDPSPN.
7636 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7638 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7639 ((X / Y) != 0 -> X >= Y): Likewise.
7641 2017-09-26 Carl Love <cel@us.ibm.com>
7643 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7644 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7645 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7646 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7647 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7648 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7649 definitions and overloading.
7650 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7651 statement for P9V_BUILTIN_XST_LEN_R.
7652 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7653 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7654 define_expand and define_insn for the instructions and builtins.
7655 * doc/extend.texi: Update the built-in documentation file for the new
7657 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7658 define_insn for the instructions
7660 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7663 * gcc/config/netbsd-protos.h: New file.
7664 * gcc/config/netbsd.c: New file.
7665 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7666 * gcc/config/t-netbsd: New file.
7667 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7668 (tmake_file) Add t-netbsd.
7669 (extra_objs) Add netbsd.o.
7671 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7675 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7677 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7679 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7680 sign extension from a vector register to a GPR by doing a 32-bit
7681 direct move and then an EXTSW.
7682 (extendsi<mode>2 splitter): Likewise.
7683 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7684 right or vector extract after doing XSCVDPSPN. Use
7685 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7687 (movdi_from_sf_zero_ext): Likewise.
7688 (reload_gpr_from_vsxsf): Likewise.
7689 (p8_mfvsrd_4_disf): Delete, no longer used.
7690 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7691 then needing to move the SFmode to a GPR to use the XSCVDPSP
7692 instruction instead of FRSP and XSCVDPSPN.
7693 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7694 it is adjacent to the other XSCVSPDP insns.
7695 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7696 SFmode to be in traditional Altivec registers.
7697 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7698 (vsx_xscvspdpn): Likewise.
7699 (vsx_xscvspdpn_scalar): Likewise.
7701 2017-09-26 Martin Jambor <mjambor@suse.cz>
7703 * tree-sra.c (compare_access_positions): Put integral types first,
7704 stabilize sorting of integral types, remove conditions putting
7705 non-full-precision integers last.
7706 (sort_and_splice_var_accesses): Disable scalarization if a
7707 non-integert would be represented by a non-full-precision integer.
7709 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7711 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7712 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7713 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7714 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7715 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7716 conditionals inside the function instead of around it. Call
7717 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7718 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7720 2017-09-26 Richard Biener <rguenther@suse.de>
7722 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7724 (scop_detection::build_scop_breadth): ... this. Removed.
7725 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7726 (scop_detection::harmful_stmt_in_bb): Likewise.
7727 (scop_detection::graphite_can_represent_stmt): Likewise.
7728 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7729 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7730 (scop_detection::can_represent_loop_1): ... this. Removed.
7731 (scop_detection::harmful_loop_in_region): Simplify after inlining
7732 the above and remove more quadraticness.
7733 (build_scops): Adjust.
7734 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7737 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7740 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7741 REGNO (base) == SP_REG if base is a REG.
7744 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7745 if it is different SSA_NAME.
7746 (optimize_range_tests_cmp_bitwise): New function.
7747 (optimize_range_tests): Call it.
7749 2017-09-26 Richard Biener <rguenther@suse.de>
7751 PR tree-optimization/82321
7752 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7753 for the def being inside the loop.
7755 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7757 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7759 * config/s390/s390-builtins.def: Fix constraint on op4.
7761 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7763 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7764 independent expanders.
7765 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7766 ("vec_ordered", "vec_unordered"): New expanders.
7768 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7770 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7773 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7775 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7776 vec_unpacks_lo_v16qi.
7777 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7779 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7781 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7782 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7783 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7785 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7787 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7789 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7791 ("*vec_slb<mode>"): New insn pattern.
7792 ("vec_shr_<mode>"): New expander.
7793 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7794 and force the shift count operand to V16QImode.
7795 ("vec_srb<mode>"): Set shift count mode to V16QI.
7797 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7799 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7800 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7801 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7803 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7806 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7808 2017-09-26 Richard Biener <rguenther@suse.de>
7810 PR tree-optimization/82320
7811 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7814 2017-09-25 Jeff Law <law@redhat.com>
7816 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7817 prototype for new argument.
7818 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7819 mostly extracted from rs6000_emit_allocate_stack.
7820 (rs6000_emit_probe_stack_range_stack_clash): New function.
7821 (rs6000_emit_allocate_stack): Call
7822 rs6000_emit_probe_stack_range_stack_clash as needed.
7823 (rs6000_emit_probe_stack_range): Add additional argument
7824 to call to gen_probe_stack_range{si,di}.
7825 (output_probe_stack_range): New.
7826 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7827 (output_probe_stack_range_stack_clash): New.
7828 (rs6000_emit_prologue): Emit notes into dump file as requested.
7829 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7830 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7831 Add additional operand and pass it to output_probe_stack_range.
7833 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7835 PR tree-optimization/82163
7836 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7837 (checking_verify_loop_closed_ssa): New parameter.
7838 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7839 (check_loop_closed_ssa_stmt): Delete.
7840 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7841 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7842 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7845 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7847 * brig-builtins.def: Treat HSAIL barrier builtins as
7848 setjmp/longjump style functions.
7850 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7852 * target.def (constant_alignment): New hook.
7853 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7854 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7855 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7856 * doc/tm.texi: Regenerate.
7857 * targhooks.h (default_constant_alignment): Declare.
7858 (constant_alignment_word_strings): Likewise.
7859 * targhooks.c (default_constant_alignment): New function.
7860 (constant_alignment_word_strings): Likewise.
7861 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7862 instead of CONSTANT_ALIGNMENT.
7863 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7864 (force_const_mem): Likewise.
7865 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7866 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7867 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7868 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7869 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7871 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7872 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7873 constant_alignment_word_strings.
7874 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7875 (CONSTANT_ALIGNMENT): Likewise.
7876 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7877 (arm_constant_alignment): New function.
7878 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7879 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7880 constant_alignment_word_strings.
7881 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7882 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7883 constant_alignment_word_strings.
7884 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7885 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7886 (cris_constant_alignment): New function.
7887 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7888 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7889 (epiphany_constant_alignment): New function.
7890 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7891 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7892 constant_alignment_word_strings.
7893 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7894 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7895 constant_alignment_word_strings.
7896 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7897 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7898 constant_alignment_word_strings.
7899 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7900 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7901 * config/i386/i386.c (ix86_constant_alignment): Make static.
7902 Use the same interface as the target hook.
7903 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7904 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7905 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7906 constant_alignment_word_strings.
7907 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7908 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7909 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7910 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7911 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7912 constant_alignment_word_strings.
7913 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7914 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7915 constant_alignment_word_strings.
7916 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7917 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7918 constant_alignment_word_strings.
7919 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7920 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7922 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7923 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7924 * config/mips/mips.c (mips_constant_alignment): New function.
7925 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7926 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7927 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7928 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7929 (mmix_constant_alignment): Make static. Use the same interface
7931 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7932 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7933 constant_alignment_word_strings.
7934 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7935 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7936 constant_alignment_word_strings.
7937 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7938 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7939 constant_alignment_word_strings.
7940 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7941 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7942 (rs6000_constant_alignment): New function.
7943 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7944 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7945 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7946 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7947 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7948 (rs6000_constant_alignment): New function.
7949 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7950 * config/s390/s390.c (s390_constant_alignment): New function.
7951 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7952 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7953 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7954 constant_alignment_word_strings.
7955 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7956 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7957 (sparc_constant_alignment): New function.
7958 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7959 * config/spu/spu.c (spu_constant_alignment): New function.
7960 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7961 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7962 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7963 constant_alignment_word_strings.
7964 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7965 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7966 constant_alignment_word_strings.
7967 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7968 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7969 constant_alignment_word_strings.
7970 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7971 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7972 (visium_constant_alignment): New function.
7973 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7974 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7975 (xtensa_constant_alignment): New function.
7976 * system.h (CONSTANT_ALIGNMENT): Poison.
7978 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7981 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7982 (rs6000_builtin_valid_without_lhs): New helper function.
7983 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7984 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7986 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7988 * target.h (vec_perm_indices): Use unsigned short rather than
7990 (auto_vec_perm_indices): Likewise.
7991 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7992 Use unsigned int rather than unsigned char.
7993 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7995 2017-09-25 Richard Biener <rguenther@suse.de>
7997 * cfgloop.h (sort_sibling_loops): Declare.
7998 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7999 (sort_sibling_loops): New function sorting the sibling loop list
8001 * graphite.c (graphite_transform_loops): Sort sibling loops.
8003 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
8005 * target.def (vec_perm_const_ok): Change sel parameter to
8007 * optabs-query.c (can_vec_perm_p): Update accordingly.
8008 * doc/tm.texi: Regenerate.
8009 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
8010 auto_vec_perm_indices and remove separate nelt field.
8011 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
8012 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
8013 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
8014 (aarch64_expand_vec_perm_const): Update accordingly.
8015 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
8016 to vec_perm_indices.
8017 * config/arm/arm.c (expand_vec_perm_d): Change perm to
8018 auto_vec_perm_indices and remove separate nelt field.
8019 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
8020 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
8021 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
8023 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
8024 to vec_perm_indices.
8025 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
8026 sel to vec_perm_indices.
8027 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
8028 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
8029 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
8031 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
8034 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8037 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8038 on the FUNCTION_DECL function context if it has a DIE that is a
8041 2017-09-25 Richard Biener <rguenther@suse.de>
8043 PR tree-optimization/82285
8044 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8047 2017-09-25 Tom de Vries <tom@codesourcery.com>
8051 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8054 2017-09-25 Richard Biener <rguenther@suse.de>
8056 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8057 ISL errors other than isl_error_quota happen. Dump if the
8058 schedule is the same.
8059 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8060 errors instead of aborting inside ISL.
8062 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8065 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8066 of libgcc_eh for m64.
8067 * config/i386/darwin64.h: Likewise.
8069 2017-09-25 Richard Biener <rguenther@suse.de>
8072 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8073 attribute for incomplete types nor twice for complete ones.
8075 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8078 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8079 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8082 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8085 * config/i386/i386.opt (mprefer-avx256): Use
8086 ix86_target_flags variable.
8087 * config/i386/i386.c (ix86_target_string): Move
8088 -mprefer-avx256 to flag2_opts.
8090 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8093 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8094 and x != -1 | y != -1 into (x & y) != -1.
8096 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8098 * config.gcc: Add new case statement to set
8099 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8100 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8101 s390x-*-linux* case statements. Added aarch64 to the list of
8102 supported architectures.
8104 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8106 PR tree-optimization/82289
8107 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8108 STMT_VINFO_RELEVANT_P.
8110 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8111 Alan Hayward <alan.hayward@arm.com>
8112 David Sherwood <david.sherwood@arm.com>
8114 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8115 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8116 (extract_range_from_binary_expr_1): Don't call
8117 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8119 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8120 Alan Hayward <alan.hayward@arm.com>
8121 David Sherwood <david.sherwood@arm.com>
8123 * target.def (preferred_vector_alignment): New hook.
8124 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8126 * doc/tm.texi: Regenerate.
8127 * targhooks.h (default_preferred_vector_alignment): Declare.
8128 * targhooks.c (default_preferred_vector_alignment): New function.
8129 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8131 (DR_TARGET_ALIGNMENT): New macro.
8132 (aligned_access_p): Update commentary.
8133 (vect_known_alignment_in_bytes): New function.
8134 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8136 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8137 Calculate the misalignment based on the target alignment rather than
8139 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8140 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8141 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8142 the target alignment, rather than masking the element misalignment
8143 with the number of elements in a vector. Also use the target
8144 alignment when calculating the maximum number of peels.
8145 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8146 instead of TYPE_ALIGN_UNIT.
8147 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8148 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8149 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8150 (vect_create_data_ref_ptr): Likewise.
8151 (vect_setup_realignment): Realign by ANDing with
8152 -DR_TARGET_MISALIGNMENT.
8153 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8154 the number of peels based on DR_TARGET_ALIGNMENT.
8155 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8156 with the guaranteed alignment boundary when deciding whether
8158 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8159 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8160 (ensure_base_align): Remove stmt_info parameter. Get the
8161 target base alignment from DR_TARGET_ALIGNMENT.
8162 (vectorizable_store): Update call accordingly. Interpret
8163 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8165 (vectorizable_load): Likewise.
8167 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8168 Alan Hayward <alan.hayward@arm.com>
8169 David Sherwood <david.sherwood@arm.com>
8171 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8172 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8173 (vect_enhance_data_refs_alignment): Likewise.
8175 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8177 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8178 error. Only quit immediately if parsing is complete.
8179 (BEGIN): Initialize fatal_err and parse_done.
8180 (begin fpu, end fpu): Check number of arguments.
8181 (begin arch, end arch): Likewise.
8182 (begin cpu, end cpu): Likewise.
8183 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8184 (optalias): Likewise.
8186 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8188 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8189 * config/arm/arm-isa.h: Delete. Move definitions to ...
8190 * arm-cpus.in: ... here. Use new feature and fgroup values.
8191 * config/arm/arm.c (arm_option_override): Use lower case for feature
8193 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8194 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8195 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8197 (print_isa_bits_for): New function.
8198 (gen_isa): New function.
8199 (gen_comm_data): Use print_isa_bits_for.
8200 (define feature): New keyword.
8201 (define fgroup): New keyword.
8202 * config/arm/t-arm (TM_H): Remove.
8203 (GTM_H): Add arm-isa.h.
8204 (arm-isa.h): Add rule to generate file.
8205 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8206 case for feature bit names.
8208 2017-09-22 Richard Biener <rguenther@suse.de>
8210 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8212 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8213 print of no dependency loops ...
8214 * graphite.c (graphite_transform_loops): ... here.
8215 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8217 (same_close_phi_node, remove_duplicate_close_phi,
8218 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8219 (canonicalize_loop_closed_ssa): ... here and simplify.
8220 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8221 (optimize_isl): Use dump_printf_loc to tell when we stopped
8222 optimizing because of an ISL timeout.
8224 2017-09-22 Richard Biener <rguenther@suse.de>
8226 PR tree-optimization/82291
8227 * tree-if-conv.c (predicate_mem_writes): Make sure to
8228 remove writes in blocks predicated with false.
8230 2017-09-22 Richard Biener <rguenther@suse.de>
8232 * sese.c: Include cfganal.h.
8233 (if_region_set_false_region): Remove.
8234 (create_if_region_on_edge): Likewise.
8235 (move_sese_in_condition): Re-implement without destroying
8238 2017-09-22 Richard Biener <rguenther@suse.de>
8240 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8241 Verify both BBs contain loop PHI nodes before dispatching to
8243 (graphite_regenerate_ast_isl): Do not recompute dominators,
8244 do not verify three times. Restructure for clarity.
8245 * graphite-scop-detection.c (same_close_phi_node,
8246 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8247 defined_in_loop_p, canonicalize_loop_closed_ssa,
8248 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8249 checking and SSA rewrite, move to ...
8250 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8251 (graphite_initialize): Do not pass in ctx, do not reset the
8252 SCEV cache, compute only dominators.
8253 (graphite_transform_loops): Allocate ISL ctx after
8254 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8255 Maintain post-dominators only around build_scops.
8256 * sese.c (if_region_set_false_region): Make static. Free
8257 and recompute dominators.
8258 (move_sese_in_condition): Assert we don't get called with
8259 post-dominators computed.
8260 * sese.h (if_region_set_false_region): Remove.
8262 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8264 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8265 mode attribute for TARGET_AVX512VL.
8267 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8269 * config/i386/i386.opt (mprefer-avx256): New option.
8270 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8272 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8273 for TARGET_PREFER_AVX256.
8274 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8276 2017-09-21 Jeff Law <law@redhat.com>
8278 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8279 Fix dump output if the only stack space is for pushed registers.
8281 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8283 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8286 2017-09-21 Martin Sebor <msebor@redhat.com>
8289 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8290 code (in C++) or code that triggers warnings.
8292 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8294 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8296 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8298 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8299 * sched-rgn.c: Ditto.
8300 * sel-sched-ir.c: Ditto.
8302 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8304 * toplev.h (set_random_seed): Adjust return type.
8305 * toplev.c (init_local_tick): Move eager initialization of random_seed
8306 to get_random_seed. Adjust comment.
8307 (init_random_seed): Inline to get_random_seed, delete.
8308 (get_random_seed): Initialize random_seed lazily.
8309 (set_random_seed): Do not return previous value.
8310 (print_switch_value): Do not call get_random_seed.
8312 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8314 * cgraph.c (delete_function_version): New, broken out from...
8315 (cgraph_node::delete_function_version): ...here. Rename to
8316 cgraph_node::delete_function_version_by_decl. Update all uses.
8317 (cgraph_node::remove): Call delete_function_version.
8319 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8322 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8323 VAR_DECLs to which addressable non-volatile parameters are mapped
8324 and for id->retvar after the return value assignment. Clear
8325 id->retval and id->retbnd after inlining.
8327 2017-09-21 Richard Biener <rguenther@suse.de>
8329 PR tree-optimization/82276
8330 PR tree-optimization/82244
8331 * tree-vrp.c (build_assert_expr_for): Set
8332 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8334 (remove_range_assertions): Revert earlier change.
8336 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8339 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8341 2017-09-21 Richard Biener <rguenther@suse.de>
8343 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8344 Restore valid IL after code generation errors.
8345 * graphite.c (graphite_transform_loops): Diagnose code
8346 generation issues as MSG_MISSED_OPTIMIZATION and continue
8347 with processing SCOPs.
8349 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8350 Alan Hayward <alan.hayward@arm.com>
8351 David Sherwood <david.sherwood@arm.com>
8353 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8354 rather than choosing between plus_constant and gen_rtx_<CODE>.
8355 * expr.c (emit_push_insn): Likewise.
8356 (expand_expr_real_2): Likewise.
8358 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8359 Alan Hayward <alan.hayward@arm.com>
8360 David Sherwood <david.sherwood@arm.com>
8362 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8364 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8365 Alan Hayward <alan.hayward@arm.com>
8366 David Sherwood <david.sherwood@arm.com>
8368 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8369 calling tree_to_uhwi.
8371 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8372 Alan Hayward <alan.hayward@arm.com>
8373 David Sherwood <david.sherwood@arm.com>
8375 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8376 INTEGER_CST rather than a negative test for ADDR_EXPR.
8378 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8379 Alan Hayward <alan.hayward@arm.com>
8380 David Sherwood <david.sherwood@arm.com>
8382 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8383 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8385 2017-09-21 Richard Biener <rguenther@suse.de>
8387 PR tree-optimization/71351
8388 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8389 graphite_create_new_loop_guard): Remove, fold remaining parts
8391 (translate_isl_ast_node_for): ... here and simplify.
8393 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8396 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8397 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8398 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8399 alternative always use QI mode, for -Os imov (=R,R) alternative
8400 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8403 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8404 Jeff Law <law@redhat.com>
8406 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8407 (allocate_stack_space): New function, partially extracted from
8409 (s390_emit_prologue): Track offset to most recent stack probe.
8410 Code to allocate space moved into allocate_stack_space.
8411 Dump actions when no stack is allocated.
8412 (s390_prologue_plus_offset): New function.
8413 (s390_emit_stack_probe): Likewise.
8415 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8417 * common.opt (Wa, Wl, Wp, g, gz=): Add
8419 (gno-column-info): Remove.
8420 (gcolumn-info): Drop RejectNegative.
8422 (gno-record-gcc-switches): Remove.
8423 (grecord-gcc-switches): Drop RejectNegative.
8424 (gno-split-dwarf): Remove.
8425 (gsplit-dwarf): Drop RejectNegative.
8426 (gno-strict-dwarf): Remove.
8427 (gstrict-dwarf): Drop RejectNegative.
8428 * config/darwin.opt (gfull, gused): Add RejectNegative.
8429 * dwarf2out.c (gen_producer_string): Drop
8430 gno-record-gcc-switches handler.
8431 * optc-gen.awk: Add g to prefixes with negative forms.
8432 * opts-common.c (remapping_prefix_p): New.
8433 (find_opt): Check it.
8434 (generate_canonical_option): Test g prefix.
8435 (option_map): Add -gno- mapping.
8436 (add_misspelling_candidates): Check remapping_prefix_p.
8438 2017-09-20 Jeff Law <law@redhat.com>
8440 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8441 thinko in stack clash protection support.
8443 * explow.c (compute_stack_clash_protection_loop_data): Use
8444 CONST_INT_P instead of explicit test. Verify object is a
8445 CONST_INT_P before looking at INTVAL.
8446 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8447 instead of explicit test.
8449 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8452 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8453 address instead of to r1 and r11.
8455 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8457 * config.gcc: Support "knm".
8458 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8459 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8461 * config/i386/i386.c (m_KNM): Define.
8462 (processor_target_table): Add "knm".
8464 (ix86_option_override_internal): Add "knm".
8465 (ix86_issue_rate): Add PROCESSOR_KNM.
8466 (ix86_adjust_cost): Ditto.
8467 (ia32_multipass_dfa_lookahead): Ditto.
8468 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8469 (fold_builtin_cpu): Add M_INTEL_KNM.
8470 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8471 (processor_type): Add PROCESSOR_KNM.
8472 * config/i386/x86-tune.def: Add m_KNM.
8473 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8475 2017-09-20 Richard Biener <rguenther@suse.de>
8477 PR tree-optimization/80213
8478 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8479 are allowed in empty BBs as well.
8480 (canonicalize_loop_closed_ssa): Also look for other complex
8482 (scop_detection::get_sese): Include the loop-closed PHI block
8484 (scop_detection::merge_sese): Remove code adding extra blocks.
8485 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8486 (build_scops): Assert the final returned scop is invalid.
8488 2017-09-20 Richard Biener <rguenther@suse.de>
8490 PR tree-optimization/82264
8491 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8493 (vn_phi_lookup): Likewise.
8494 (vn_phi_insert): Likewise.
8496 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8498 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8499 that fits into uhwi or shwi, add DW_AT_const_value regardless
8500 of early_dwarf without going through RTL, using add_AT_unsigned
8503 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8504 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8505 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8506 (DEBUG_MACRO_SECTION): Likewise.
8507 (DEBUG_LTO_MACRO_SECTION): Likewise.
8508 (DEBUG_STR_DWO_SECTION): Likewise.
8509 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8510 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8511 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8512 (DEBUG_LTO_LINE_STR_SECTION): Define.
8513 (init_sections_and_labels): Initialize debug_line_str_section
8514 variable. Initialize debug_loc_section for -gdwarf-5 to
8515 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8517 2017-09-20 Richard Biener <rguenther@suse.de>
8519 * graphite-sese-to-poly.c (extract_affine): Properly handle
8520 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8522 2017-09-20 Richard Biener <rguenther@suse.de>
8524 PR tree-optimization/81373
8525 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8526 Force SESE live-out defs to be handled even if they are
8529 2017-09-19 Jeff Law <law@redhat.com>
8531 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8532 nothing for stack adjustments with REG_STACK_CHECK.
8533 * sched-deps.c (parse_add_or_inc): Reject insns with
8534 REG_STACK_CHECK from dependency breaking.
8535 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8536 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8537 * reg-notes.def (STACK_CHECK): New note.
8539 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8540 (ix86_expand_prologue): Dump stack clash info as needed.
8541 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8543 * function.c (dump_stack_clash_frame_info): New function.
8544 * function.h (dump_stack_clash_frame_info): Prototype.
8545 (enum stack_clash_probes): New enum.
8547 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8548 flag_stack_clash_protection.
8549 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8550 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8551 (arm_frame_pointer_required): Likewise.
8552 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8553 (ia64_expand_prologue): Likewise.
8554 * config/mips/mips.c (mips_expand_prologue): Likewise.
8555 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8556 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8557 (sparc_flat_expand_prologue): Likewise.
8558 * config/spu/spu.c (spu_expand_prologue): Likewise.
8560 * explow.c: Include "params.h".
8561 (anti_adjust_stack_and_probe_stack_clash): New function.
8562 (get_stack_check_protect): Likewise.
8563 (compute_stack_clash_protection_loop_data): Likewise.
8564 (emit_stack_clash_protection_loop_start): Likewise.
8565 (emit_stack_clash_protection_loop_end): Likewise.
8566 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8567 Use anti_adjust_stack_and_probe_stack_clash.
8568 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8569 (emit_stack_clash_protection_loop_start): Likewise.
8570 (emit_stack_clash_protection_loop_end): Likewise.
8571 * rtl.h (get_stack_check_protect): Prototype.
8572 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8573 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8574 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8576 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8578 * doc/tm.texi: Rebuilt.
8579 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8580 get_stack_check_protect.
8581 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8582 * config/arm/arm.c (arm_expand_prologue): Likewise.
8583 (arm_frame_pointer_required): Likewise.
8584 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8585 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8586 * config/mips/mips.c (mips_expand_prologue): Likewise.
8587 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8588 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8589 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8590 (sparc_flat_expand_prologue): Likewise.
8592 * common.opt (-fstack-clash-protection): New option.
8593 * flag-types.h (enum stack_check_type): Note difference between
8594 -fstack-check= and -fstack-clash-protection.
8595 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8596 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8597 * toplev.c (process_options): Issue warnings/errors for cases
8598 not handled with -fstack-clash-protection.
8599 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8600 (-fstack-check): Note additional problem with -fstack-check=generic.
8601 Note that -fstack-check is primarily for Ada and refer users
8602 to -fstack-clash-protection for stack-clash-protection.
8603 Document new params for stack clash protection.
8605 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8607 * config/i386/i386.c (ix86_split_long_move): Do not handle
8608 address used for LEA in a special way.
8610 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8612 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8614 2017-09-19 Martin Sebor <msebor@redhat.com>
8617 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8618 of incompatible types.
8620 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8622 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8623 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8624 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8625 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8627 2017-09-19 Richard Biener <rguenther@suse.de>
8629 PR tree-optimization/82244
8630 * tree-vrp.c (remove_range_assertions): Do not propagate
8631 a constant to abnormals but replace the assert with a copy.
8633 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8635 PR rtl-optimization/57878
8636 PR rtl-optimization/68988
8637 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8638 avoidance test involving non_reload_pseudos. Move frequency test
8639 below the general fragmentation avoidance test.
8641 2017-09-19 Richard Biener <rguenther@suse.de>
8643 PR tree-optimization/69728
8644 * graphite-sese-to-poly.c (schedule_error): New global.
8645 (add_loop_schedule): Handle empty domain by failing the
8647 (build_original_schedule): Handle schedule_error.
8649 2017-09-19 Richard Biener <rguenther@suse.de>
8651 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8652 Do not iterate to sibling loops but only to siblings of inner
8655 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8658 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8659 that the registers are different.
8661 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8663 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8664 to processor_model and "amdfam17h" to arch_names_table.
8665 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8667 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8670 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8672 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8673 Alan Hayward <alan.hayward@arm.com>
8674 David Sherwood <david.sherwood@arm.com>
8676 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8678 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8680 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8681 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8682 vect_schedule_slp_instance.
8683 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8684 Update call to vect_slp_analyze_node_operations. Simplify return
8686 (vect_slp_analyze_bb_1): Update call accordingly.
8687 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8688 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8689 (vect_schedule_slp): Update call accordingly.
8691 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8692 Alan Hayward <alan.hayward@arm.com>
8693 David Sherwood <david.sherwood@arm.com>
8695 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8696 with types that aren't in fact scalar.
8698 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8700 * tree-vect-slp.c (vect_record_max_nunits): New function,
8702 (vect_build_slp_tree_1): ...here.
8703 (vect_build_slp_tree_2): Call it for phis too.
8705 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8707 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8708 to vect_get_vec_def_for_operand when getting the mask operand.
8710 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8711 Alan Hayward <alan.hayward@arm.com>
8712 David Sherwood <david.sherwood@arm.com>
8714 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8717 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8718 Alan Hayward <alan.hayward@arm.com>
8719 David Sherwood <david.sherwood@arm.com>
8721 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8722 calculation for vector booleans.
8724 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8725 Alan Hayward <alan.hayward@arm.com>
8726 David Sherwood <david.sherwood@arm.com>
8728 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8730 (vect_transform_stmt): ...here.
8731 (vect_analyze_stmt): Use it instead of calling
8732 vectorizable_live_operation directly.
8734 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8736 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8737 non-SIMT targets in acc vector loops.
8739 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8741 * configure.ac: Add arc and check if assembler supports gdwarf2.
8742 * configure: Regenerate.
8744 2017-09-18 Richard Biener <rguenther@suse.de>
8746 PR tree-optimization/82220
8747 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8748 epilogue niters from the min_profitable_iters compute.
8750 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8753 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8754 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8755 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8757 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8760 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8761 switching to a new text section.
8763 2017-09-18 Richard Biener <rguenther@suse.de>
8765 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8767 (build_alias_set): Reject aliases with no access function.
8769 2017-09-18 Richard Biener <rguenther@suse.de>
8771 PR tree-optimization/79622
8772 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8774 (build_cross_bb_scalars_use): Likewise.
8776 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8778 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8780 2017-09-18 Alan Modra <amodra@gmail.com>
8783 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8784 stack_pointer_rtx for count 0. Update comments. Break up
8785 large rtl expression.
8787 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8789 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8790 Increase to 20 bytes.
8791 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8792 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8793 or avx version of the stub.
8795 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8798 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8799 compute the minimum stack alignment. Also update preferred stack
8800 boundary for leaf functions.
8802 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8804 PR tree-optimization/82228
8805 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8808 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8810 * common/config/nds32/nds32-common.c
8811 (nds32_option_optimization_table): Refine formatting.
8812 (nds32_option_optimization_table): Use -fsched-pressure and
8813 -fomit-frame-pointer for specific optimization level.
8815 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8817 * config/nds32/nds32.c: Refine formatting and comments.
8818 * config/nds32/nds32.h: Likewise.
8819 * config/nds32/nds32.md: Likewise.
8820 * config/nds32/nds32-cost.c: Likewise.
8821 * config/nds32/nds32-isr.c: Likewise.
8822 * config/nds32/nds32-md-auxiliary.c: Likewise.
8823 * config/nds32/nds32-multiple.md: Likewise.
8824 * config/nds32/nds32-predicates.c: Likewise.
8826 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8827 Jakub Jelinek <jakub@redhat.com>
8829 Add support for -std=c++2a.
8830 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8832 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8834 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8837 * doc/extend.texi (Common Function Attributes): Add
8838 references to ARM, AArch64, and S/390 specific attributes.
8839 (Function Specific Option Pragmas): Add AArch64 and S/390
8840 to list of back ends that support the target pragma.
8842 2017-09-15 Nathan Sidwell <nathan@acm.org>
8844 * doc/standards.texi: Fix C++17 description. Update URLs for
8847 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8849 * common.opt (Wcast-align=strict): New warning option.
8850 * doc/invoke.texi: Document -Wcast-align=strict.
8852 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8854 * cgraph.h (cgraph_thunk_info): Add comments.
8855 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8856 assert for VIRTUAL_* arguments stricter.
8858 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8860 PR tree-optimization/71026
8861 * match.pd: Move RDIV patterns from fold-const.c
8862 * fold-const.c (distribute_real_division): Removed.
8863 (fold_binary_loc): Remove calls to distribute_real_divison.
8865 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8867 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8868 c++1z and gnu++1z as deprecated. Change other references to
8869 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8870 Change -Wc++1z-compat to -Wc++17-compat.
8871 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8872 * dwarf2out.c (highest_c_language): Handle C++17.
8873 (gen_compile_unit_die): Likewise.
8875 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8877 PR rtl-optimization/82192
8878 * combine.c (make_extraction): Don't look through non-paradoxical
8879 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8882 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8883 Alan Hayward <alan.hayward@arm.com>
8884 David Sherwood <david.sherwood@arm.com>
8886 * target.def (function_arg_offset): New hook.
8887 * targhooks.h (default_function_arg_offset): Declare.
8888 * targhooks.c (default_function_arg_offset): New function.
8889 * function.c (locate_and_pad_parm): Use
8890 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8891 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8892 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8893 * doc/tm.texi: Regenerate.
8894 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8895 * config/spu/spu.c (spu_function_arg_offset): New function.
8896 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8897 * system.h (FUNCTION_ARG_OFFSET): Poison.
8899 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8900 Alan Hayard <alan.hayward@arm.com>
8901 David Sherwood <david.sherwood@arm.com>
8903 * target.def (truly_noop_truncation): New hook.
8904 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8905 than TRULY_NOOP_TRUNCATION.
8906 * hooks.h (hook_bool_uint_uint_true): Declare.
8907 * hooks.c (hook_bool_uint_uint_true): New function.
8908 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8909 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8910 * doc/tm.texi: Regenerate.
8911 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8912 rather than TRULY_NOOP_TRUNCATION in comments.
8913 (simplify_comparison): Likewise.
8914 (record_truncated_value): Likewise.
8915 * expmed.c (extract_bit_field_1): Likewise.
8916 (extract_split_bit_field): Likewise.
8917 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8918 instead of TRULY_NOOP_TRUNCATION.
8919 * function.c (assign_parm_setup_block): Likewise.
8920 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8921 * rtlhooks.c: Include target.h.
8922 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8923 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8924 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8925 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8926 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8927 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8928 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8929 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8930 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8931 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8932 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8933 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8934 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8935 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8936 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8937 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8938 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8939 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8940 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8941 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8942 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8943 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8944 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8945 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8946 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8947 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8948 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8949 rather than TRULY_NOOP_TRUNCATION in comments.
8950 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8951 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8952 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8953 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8954 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8955 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8956 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8957 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8958 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8959 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8960 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8961 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8962 rather than TRULY_NOOP_TRUNCATION in comments.
8963 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8964 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8965 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8966 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8967 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8968 TRULY_NOOP_TRUNCATION condition.
8969 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8970 (TRULY_NOOP_TRUNCATION): Delete.
8971 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8972 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8973 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8974 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8975 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8976 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8977 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8978 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8979 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8980 rather than TRULY_NOOP_TRUNCATION in comments.
8981 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8982 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8983 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8984 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8985 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8986 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8988 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8991 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8992 (*cmp_ior): Likewise.
8993 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8994 (*ior_scc_scc_cmp): Likewise.
8995 (*and_scc_scc): Likewise.
8996 (*and_scc_scc_cmp): Likewise.
8998 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8999 Alan Hayard <alan.hayward@arm.com>
9000 David Sherwood <david.sherwood@arm.com>
9002 * target.def (can_change_mode_class): New hook.
9003 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
9004 (hard_regno_nregs): Likewise.
9005 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
9006 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
9007 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
9008 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
9009 (LOAD_EXTEND_OP): Update accordingly.
9010 * doc/tm.texi: Regenerate.
9011 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
9012 CANNOT_CHANGE_MODE_CLASS.
9013 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
9014 (REG_CAN_CHANGE_MODE_P): ...this new macro.
9015 * combine.c (simplify_set): Update accordingly.
9016 * emit-rtl.c (validate_subreg): Likewise.
9017 * recog.c (general_operand): Likewise.
9018 * regcprop.c (mode_change_ok): Likewise.
9019 * reload1.c (choose_reload_regs): Likewise.
9020 (inherit_piecemeal_p): Likewise.
9021 * rtlanal.c (simplify_subreg_regno): Likewise.
9022 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
9023 instead of CANNOT_CHANGE_MODE_CLASS.
9024 (reload_cse_simplify_operands): Likewise.
9025 * reload.c (push_reload): Use targetm.can_change_mode_class
9026 instead of CANNOT_CHANGE_MODE_CLASS.
9027 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
9028 REG_CANNOT_CHANGE_MODE_P.
9029 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9030 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
9031 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9032 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9033 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9034 (arm_can_change_mode_class): New function.
9035 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9036 than CANNOT_CHANGE_MODE_CLASS in comments.
9037 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9038 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9039 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9040 (ix86_can_change_mode_class): ...this new function, inverting the
9041 sense of the return value.
9042 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9043 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9044 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9045 (ia64_can_change_mode_class): New function.
9046 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9047 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9048 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9049 (m32c_can_change_mode_class): ...this new function, inverting the
9050 sense of the return value.
9051 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9052 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9053 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9054 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9055 (mips_can_change_mode_class): ...this new function, inverting the
9056 sense of the return value.
9057 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9058 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9059 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9060 (msp430_can_change_mode_class): New function.
9061 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9062 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9063 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9064 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9065 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9066 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9067 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9068 (pa_cannot_change_mode_class): Replace with...
9069 (pa_can_change_mode_class): ...this new function, inverting the
9070 sense of the return value.
9071 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9072 than CANNOT_CHANGE_MODE_CLASS in comments.
9073 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9074 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9075 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9076 (pdp11_cannot_change_mode_class): Replace with...
9077 (pdp11_can_change_mode_class): ...this new function, inverting the
9078 sense of the return value.
9079 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9080 * config/powerpcspe/powerpcspe-protos.h
9081 (rs6000_cannot_change_mode_class_ptr): Delete.
9082 * config/powerpcspe/powerpcspe.c
9083 (rs6000_cannot_change_mode_class_ptr): Delete.
9084 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9085 (rs6000_option_override_internal): Assign to
9086 targetm.can_change_mode_class instead of
9087 rs6000_cannot_change_mode_class_ptr.
9088 (rs6000_cannot_change_mode_class): Replace with...
9089 (rs6000_can_change_mode_class): ...this new function, inverting the
9090 sense of the return value.
9091 (rs6000_debug_cannot_change_mode_class): Replace with...
9092 (rs6000_debug_can_change_mode_class): ...this new function.
9093 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9094 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9095 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9096 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9097 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9099 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9100 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9101 (rs6000_option_override_internal): Assign to
9102 targetm.can_change_mode_class instead of
9103 rs6000_cannot_change_mode_class_ptr.
9104 (rs6000_cannot_change_mode_class): Replace with...
9105 (rs6000_can_change_mode_class): ...this new function, inverting the
9106 sense of the return value.
9107 (rs6000_debug_cannot_change_mode_class): Replace with...
9108 (rs6000_debug_can_change_mode_class): ...this new function.
9109 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9110 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9111 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9112 (s390_can_change_mode_class): ...this new function, inverting the
9113 sense of the return value.
9114 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9115 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9116 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9117 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9118 (sh_cannot_change_mode_class): Replace with...
9119 (sh_can_change_mode_class): ...this new function, inverting the
9120 sense of the return value.
9121 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9122 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9123 (sparc_can_change_mode_class): New function.
9124 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9125 * config/spu/spu.c (spu_can_change_mode_class): New function.
9126 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9127 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9128 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9129 (visium_can_change_mode_class): New function.
9130 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9132 2017-09-15 Richard Biener <rguenther@suse.de>
9134 PR tree-optimization/82217
9135 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9136 but not undefined case.
9138 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9141 * postreload.c (reload_cse_simplify_operands): Skip
9142 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9144 2017-09-15 Richard Biener <rguenther@suse.de>
9146 PR tree-optimization/68823
9147 * graphite-scop-detection.c (build_alias_set): If we have a
9148 possible dependence check whether we can handle them by just
9149 looking at the DRs DR_ACCESS_FNs.
9150 (build_scops): If build_alias_set fails, fail the SCOP.
9152 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9154 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9155 to support float128 built-in functions that require the ISA 3.0
9157 (BU_FLOAT128_3_HW): Likewise.
9158 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9160 (FMAF128): Likewise.
9162 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9163 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9164 floating point instructions.
9165 (rs6000_invalid_builtin): Likewise.
9166 (rs6000_builtin_mask_names): Likewise.
9167 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9168 (RS6000_BTM_FLOAT128_HW): Likewise.
9169 (RS6000_BTM_COMMON): Likewise.
9170 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9172 * doc/extend.texi (RS/6000 built-in functions): Document the
9173 IEEE 128-bit floating point square root and fused multiply-add
9176 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9178 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9179 reg (r2) isn't in the set of registers defined in the prologue.
9181 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9182 Alan Hayward <alan.hayward@arm.com>
9183 David Sherwood <david.sherwood@arm.com>
9185 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9186 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9187 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9188 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9189 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9191 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9192 max_vectorization_factor.
9193 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9195 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9196 Alan Hayward <alan.hayward@arm.com>
9197 David Sherwood <david.sherwood@arm.com>
9199 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9200 (vect_worthwhile_without_simd_p): Declare.
9201 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9202 (vectorizable_reduction): Use it.
9203 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9204 (vectorizable_operation): Likewise.
9206 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9207 Alan Hayward <alan.hayward@arm.com>
9208 David Sherwood <david.sherwood@arm.com>
9210 * tree-vectorizer.h (vect_get_num_copies): New function.
9211 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9212 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9213 (vectorizable_induction): Likewise.
9214 (vectorizable_live_operation): Likewise.
9215 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9216 (vectorizable_bswap): Likewise.
9217 (vectorizable_call): Likewise.
9218 (vectorizable_conversion): Likewise.
9219 (vectorizable_assignment): Likewise.
9220 (vectorizable_shift): Likewise.
9221 (vectorizable_operation): Likewise.
9222 (vectorizable_store): Likewise.
9223 (vectorizable_load): Likewise.
9224 (vectorizable_condition): Likewise.
9225 (vectorizable_comparison): Likewise.
9226 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9228 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9229 Alan Hayward <alan.hayward@arm.com>
9230 David Sherwood <david.sherwood@arm.com>
9232 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9233 of vect_init_vector.
9235 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9236 Alan Hayward <alan.hayward@arm.com>
9237 David Sherwood <david.sherwood@arm.com>
9239 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9240 an inline wrapper that provides a location.
9241 (gimple_build_vector): Likewise.
9242 * gimple-fold.c (gimple_build_vector_from_val): New function.
9243 (gimple_build_vector): Likewise.
9244 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9245 functions to build the initial value. Always return a gimple value.
9246 (get_initial_defs_for_reduction): Likewise. Only compute
9248 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9249 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9250 (vectorizable_induction): Use gimple_build_vector rather than
9253 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9254 Alan Hayward <alan.hayward@arm.com>
9255 David Sherwood <david.sherwood@arm.com>
9257 * target.h (vec_perm_indices): New typedef.
9258 (auto_vec_perm_indices): Likewise.
9259 * optabs-query.h: Include target.h
9260 (can_vec_perm_p): Take a vec_perm_indices *.
9261 * optabs-query.c (can_vec_perm_p): Likewise.
9262 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9263 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9264 * tree-vect-generic.c (lower_vec_perm): Likewise.
9265 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9266 (vect_grouped_load_supported): Likewise.
9267 (vect_shift_permute_load_chain): Likewise.
9268 (vect_permute_store_chain): Use auto_vec_perm_indices.
9269 (vect_permute_load_chain): Likewise.
9270 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9271 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9272 Update uses of can_vec_perm_p.
9273 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9274 mode with a number of elements. Take a vec_perm_indices *.
9275 (vect_create_epilog_for_reduction): Update accordingly.
9276 Use auto_vec_perm_indices.
9277 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9278 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9279 (vect_transform_slp_perm_load): Likewise.
9280 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9281 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9282 (vect_gen_perm_mask_checked): Likewise.
9283 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9284 (vect_gen_perm_mask_checked): Likewise.
9285 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9286 (vectorizable_store): Likewise.
9287 (vectorizable_load): Likewise.
9288 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9289 (vectorizable_bswap): Likewise.
9291 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9292 Alan Hayward <alan.hayward@arm.com>
9293 David Sherwood <david.sherwood@arm.com>
9295 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9296 * tree.c (build_vector): Likewise.
9297 (build_vector_from_ctor): Update accordingly.
9298 (build_vector_from_val): Likewise.
9299 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9300 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9301 * tree-vect-generic.c (add_rshift): Likewise.
9302 (expand_vector_divmod): Likewise.
9303 (optimize_vector_constructor): Likewise.
9304 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9305 (vect_transform_slp_perm_load): Likewise.
9306 (vect_schedule_slp_instance): Likewise.
9307 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9308 (vectorizable_call): Likewise.
9309 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9310 * expmed.c (make_tree): Likewise.
9311 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9312 a vector passed to build_vector.
9313 (fold_convert_const): Likewise.
9314 (exact_inverse): Likewise.
9315 (fold_ternary_loc): Likewise.
9316 (fold_relational_const): Likewise.
9317 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9318 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9319 (const_unop): Likewise. Store the reduction accumulator in a
9320 variable rather than an array.
9321 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9322 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9323 the new vector, rather than constructing it after the input arrays.
9324 (native_interpret_vector): Use auto_vec<tree> when building
9325 a vector passed to build_vector. Add elements in order.
9326 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9327 auto_vec<tree> when building a vector passed to build_vector.
9328 (vect_create_epilog_for_reduction): Likewise.
9329 (vectorizable_induction): Likewise.
9330 (get_initial_def_for_reduction): Likewise. Fix indentation of
9332 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9334 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9335 passed to build_vector.
9337 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9338 Alan Hayward <alan.hayward@arm.com>
9339 David Sherwood <david.sherwood@arm.com>
9341 * tree-core.h (tree_base::u): Add an "nelts" field.
9342 (tree_vector): Use VECTOR_CST_NELTS as the length.
9343 * tree.c (tree_size): Likewise.
9344 (make_vector): Initialize VECTOR_CST_NELTS.
9345 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9346 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9347 TYPE_VECTOR_SUBPARTS.
9348 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9349 as the number of units, setting it from VECTOR_CST_NELTS.
9350 (const_vector_from_tree): Likewise.
9351 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9352 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9353 (fold_negate_expr_1): Likewise.
9354 (fold_convert_const): Likewise.
9355 (const_binop): Likewise. Differentiate the number of output and
9357 (const_unop): Likewise.
9358 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9359 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9360 in cases that did the opposite.
9362 2017-09-14 Richard Biener <rguenther@suse.de>
9364 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9367 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9369 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9371 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9374 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9375 if and where to split a bb, except for splitting before debug insn
9376 sequences followed by non-label real insn. Delete debug insns
9377 in between basic blocks.
9379 * combine.c (make_compound_operation_int): Formatting fixes.
9381 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9382 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9383 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9384 * config/sol2.h (LINK_EH_SPEC): Likewise.
9385 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9386 * config/s390/linux.h (LINK_SPEC): Likewise.
9387 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9388 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9389 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9390 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9391 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9392 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9393 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9394 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9396 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9398 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9400 (ENDFILE_LINUX_SPEC): Likewise.
9401 (LINK_EH_SPEC): Likewise.
9402 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9403 (LINK_OS_LINUX_SPEC32): Likewise.
9404 (LINK_OS_LINUX_SPEC64): Likewise.
9405 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9406 (LINK_OS_LINUX_SPEC): Likewise.
9408 2017-09-13 Martin Liska <mliska@suse.cz>
9411 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9412 CASE_HIGH is NULL_TREE.
9414 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9415 Alan Hayward <alan.hayward@arm.com>
9416 David Sherwood <david.sherwood@arm.com>
9418 * target.def (secondary_memory_needed): New hook.
9419 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9420 instead of SECONDARY_MEMORY_NEEDED.
9421 (secondary_memory_needed_mode): Likewise.
9422 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9423 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9424 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9425 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9426 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9427 * doc/tm.texi: Regenerate.
9428 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9429 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9430 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9431 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9432 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9433 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9434 mode argument first and change the reg_class arguments to reg_class_t.
9435 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9436 Make static. Update the call to inline_secondary_memory_needed.
9437 (ix86_register_move_cost): Update the call to
9438 inline_secondary_memory_needed.
9439 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9440 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9442 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9443 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9445 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9446 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9447 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9448 and match hook interface. Add comment from mips.h.
9449 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9450 * config/mmix/mmix.md (truncdfsf2): Refer to
9451 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9453 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9454 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9455 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9456 (pa_secondary_memory_needed): New function.
9457 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9458 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9459 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9460 (pdp11_secondary_memory_needed): Make static and match hook interface.
9461 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9462 * config/powerpcspe/powerpcspe-protos.h
9463 (rs6000_secondary_memory_needed_ptr): Delete.
9464 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9466 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9467 (rs6000_option_override_internal): Assign to
9468 targetm.secondary_memory_needed rather than
9469 rs6000_secondary_memory_needed_ptr.
9470 (rs6000_secondary_memory_needed): Match hook interface.
9471 (rs6000_debug_secondary_memory_needed): Likewise.
9472 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9473 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9474 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9475 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9476 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9477 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9479 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9480 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9481 (rs6000_option_override_internal): Assign to
9482 targetm.secondary_memory_needed rather than
9483 rs6000_secondary_memory_needed_ptr.
9484 (rs6000_secondary_memory_needed): Match hook interface.
9485 (rs6000_debug_secondary_memory_needed): Likewise.
9486 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9487 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9488 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9489 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9490 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9491 (sparc_secondary_memory_needed): New function.
9492 * lra-constraints.c (check_and_process_move): Refer to
9493 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9495 (curr_insn_transform): Likewise.
9496 (process_alt_operands): Use targetm.secondary_memory_needed
9497 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9498 (check_secondary_memory_needed_p): Likewise.
9499 (choose_split_class): Likewise.
9500 * reload.c: Unconditionally include code that was previously
9501 conditional on SECONDARY_MEMORY_NEEDED.
9502 (push_secondary_reload): Use targetm.secondary_memory_needed
9503 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9504 (push_reload): Likewise.
9505 * reload1.c: Unconditionally include code that was previously
9506 conditional on SECONDARY_MEMORY_NEEDED.
9507 (choose_reload_regs): Use targetm.secondary_memory_needed
9508 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9509 (gen_reload): Likewise.
9510 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9512 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9513 Alan Hayward <alan.hayward@arm.com>
9514 David Sherwood <david.sherwood@arm.com>
9516 * target.def (secondary_memory_needed_mode): New hook:
9517 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9518 * targhooks.h (default_secondary_memory_needed_mode): New function.
9519 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9520 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9521 * doc/tm.texi: Regenerate.
9522 * lra-constraints.c (check_and_process_move): Use
9523 targetm.secondary_memory_needed_mode instead of
9524 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9525 (curr_insn_transform): Likewise.
9526 * reload.c (get_secondary_mem): Likewise.
9527 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9528 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9530 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9531 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9532 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9533 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9534 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9536 * config/powerpcspe/powerpcspe-protos.h
9537 (rs6000_secondary_memory_needed_mode): Delete.
9538 * config/powerpcspe/powerpcspe.c
9539 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9540 (rs6000_secondary_memory_needed_mode): Make static.
9541 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9542 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9544 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9546 (rs6000_secondary_memory_needed_mode): Make static.
9547 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9548 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9549 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9550 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9551 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9553 (sparc_secondary_memory_needed_mode): New function.
9554 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9556 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9558 * config/aarch64/constraints.md (Umq): New constraint.
9559 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9561 (mov<mode>): Update condition.
9563 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9565 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9566 when bitposition is the same.
9568 2017-09-13 Richard Biener <rguenther@suse.de>
9570 * dwarf2out.c (output_die_symbol): Remove.
9571 (output_die): Do not output a DIEs symbol.
9573 2017-09-13 Richard Biener <rguenther@suse.de>
9576 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9577 default-def to avoid breaking iterator update with the weird
9578 interaction with cgraph_update_edges_for_call_stmt_node.
9580 2017-09-13 Richard Biener <rguenther@suse.de>
9582 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9583 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9584 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9585 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9587 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9589 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9590 Disable pc relative literal load irrespective of
9591 TARGET_FIX_ERR_A53_84341 for default.
9593 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9595 * config/sparc/sparc.c (output_return): Output the source location of
9596 the insn in the delay slot, if any.
9597 (output_sibcall): Likewise.
9599 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9602 * common.opt (-static-pie): New alias.
9603 (shared): Negate static-pie.
9604 (-no-pie): Update help text.
9606 (static-pie): New option.
9607 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9608 -static-pie support.
9609 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9610 (LINK_EH_SPEC): Likewise.
9611 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9612 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9613 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9614 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9615 (init_gcc_specs): Likewise.
9616 (init_spec): Likewise.
9617 (display_help): Update help message for -pie.
9618 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9621 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9623 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9624 (movdi_aarch64): Likewise.
9625 (movti_aarch64): Likewise.
9627 2017-09-12 Simon Wright <simon@pushface.org>
9630 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9631 calculation of the minor version, always output as 0.
9633 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9636 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9637 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9638 on it early, rather than manual conversion late. For
9639 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9640 instead of performing manual conversion.
9642 2017-09-12 Carl Love <cel@us.ibm.com>
9644 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9645 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9647 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9648 VMULOSW): Add definitions.
9649 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9650 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9651 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9652 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9653 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9655 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9657 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9659 (movti_aarch64): Likewise.
9660 (movdf_aarch64): Likewise.
9661 (movtf_aarch64): Likewise.
9662 (load_pairdi): Likewise.
9663 (store_pairdi): Likewise.
9664 (load_pairdf): Likewise.
9665 (store_pairdf): Likewise.
9666 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9667 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9668 (ldr_got_small_<mode>): Likewise.
9669 (ldr_got_small_28k_<mode>): Likewise.
9670 (ldr_got_tiny): Likewise.
9671 * config/aarch64/iterators.md (ldst_sz): New.
9672 (ldpstp_sz): Likewise.
9673 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9675 (thunderx_load): Split load_8 to load_16.
9676 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9678 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9679 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9680 (xgene1_store_pair): Split store_8 to store_16.
9681 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9682 (falkor_st_0_st_sd): Split store_8 to store_16.
9684 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9686 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9687 and store1/2/3/4 to store_4/8/12/16.
9688 * config/aarch64/aarch64.md: Update for rename.
9689 * config/arm/arm.md: Likewise.: Likewise.
9690 * config/arm/arm.c: Likewise.
9691 * config/arm/thumb1.md: Likewise.
9692 * config/arm/thumb2.md: Likewise.
9693 * config/arm/vfp.md: Likewise.
9694 * config/arm/arm-generic.md: Likewise.
9695 * config/arm/arm1020e.md: Likewise.
9696 * config/arm/arm1026ejs.md: Likewise.
9697 * config/arm/arm1136jfs.md: Likewise.
9698 * config/arm/arm926ejs.md: Likewise.
9699 * config/arm/cortex-a15.md: Likewise.
9700 * config/arm/cortex-a17.md: Likewise.
9701 * config/arm/cortex-a5.md: Likewise.
9702 * config/arm/cortex-a53.md: Likewise.
9703 * config/arm/cortex-a57.md: Likewise.
9704 * config/arm/cortex-a7.md: Likewise.
9705 * config/arm/cortex-a8.md: Likewise.
9706 * config/arm/cortex-a9.md: Likewise.
9707 * config/arm/cortex-m4.md: Likewise.
9708 * config/arm/cortex-m7.md: Likewise.
9709 * config/arm/cortex-r4.md: Likewise.
9710 * config/arm/exynos-m1.md: Likewise.
9711 * config/arm/fa526.md: Likewise.
9712 * config/arm/fa606te.md: Likewise.
9713 * config/arm/fa626te.md: Likewise.
9714 * config/arm/fa726te.md: Likewise.
9715 * config/arm/fmp626.md: Likewise.
9716 * config/arm/iwmmxt.md: Likewise.
9717 * config/arm/ldmstm.md: Likewise.
9718 * config/arm/marvell-pj4.md: Likewise.
9719 * config/arm/xgene1.md: Likewise.
9720 * config/aarch64/thunderx.md: Likewise.
9721 * config/aarch64/thunderx2t99.md: Likewise.
9722 * config/aarch64/falkor.md: Likewise.
9724 2017-09-12 Martin Liska <mliska@suse.cz>
9726 * attribs.c (private_lookup_attribute): New function.
9727 * attribs.h (private_lookup_attribute): Declared here.
9728 (lookup_attribute): Called from this place.
9730 2017-09-12 Richard Biener <rguenther@suse.de>
9732 PR tree-optimization/82157
9733 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9734 stmts with side-effects.
9736 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9737 Alan Hayward <alan.hayward@arm.com>
9738 David Sherwood <david.sherwood@arm.com>
9740 * target.def (hard_regno_nregs): New hook.
9741 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9742 * targhooks.h (default_hard_regno_nregs): Declare.
9743 * targhooks.c (default_hard_regno_nregs): New function.
9744 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9745 (TARGET_HARD_REGNO_NREGS): ...this hook.
9746 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9747 (CLASS_MAX_NREGS): Likewise.
9748 * doc/tm.texi: Regenerate.
9749 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9750 instead of HARD_REGNO_NREGS.
9751 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9752 HARD_REGNO_NREGS in the comment.
9753 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9754 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9755 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9756 Return an unsigned int.
9757 (TARGET_HARD_REGNO_NREGS): Redefine.
9758 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9759 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9760 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9761 (arc_hard_regno_nregs): New function.
9762 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9763 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9764 (arm_hard_regno_nregs): New function.
9765 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9766 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9767 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9768 (TARGET_HARD_REGNO_NREGS): Redefine.
9769 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9770 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9771 (HARD_REGNO_NREGS): Delete.
9772 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9773 (cr16_hard_regno_nregs): New function.
9774 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9775 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9776 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9777 (cris_hard_regno_nregs): New function.
9778 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9779 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9780 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9781 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9782 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9783 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9784 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9785 (frv_hard_regno_nregs): Make static. Take and return an
9787 (frv_class_max_nregs): Remove outdated copy of documentation.
9788 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9789 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9790 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9791 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9792 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9793 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9794 (TARGET_HARD_REGNO_NREGS): Redefine.
9795 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9796 (CLASS_MAX_NREGS): Update comment.
9797 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9798 (ia64_hard_regno_nregs): New function.
9799 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9800 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9801 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9802 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9803 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9805 (m32c_hard_regno_nregs): Likewise. Make static.
9806 (TARGET_HARD_REGNO_NREGS): Redefine.
9807 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9808 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9809 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9810 (m68k_hard_regno_nregs): New function.
9811 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9812 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9813 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9814 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9815 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9816 Take and return an unsigned int.
9817 (TARGET_HARD_REGNO_NREGS): Redefine.
9818 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9819 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9820 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9821 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9822 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9823 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9824 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9825 (msp430_hard_regno_nregs): Make static. Take and return an
9827 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9828 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9829 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9830 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9831 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9832 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9833 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9834 (TARGET_HARD_REGNO_NREGS): Redefine.
9835 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9836 (PA_HARD_REGNO_NREGS): ...this.
9837 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9838 (PA_HARD_REGNO_NREGS): ...this.
9839 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9840 (pa_hard_regno_nregs): New function.
9841 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9842 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9843 (pdp11_hard_regno_nregs): New function.
9844 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9845 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9846 (rs6000_hard_regno_nregs_hook): New function.
9847 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9848 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9849 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9850 Take and return an unsigned int. Move earlier in file.
9851 (TARGET_HARD_REGNO_NREGS): Redefine.
9852 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9853 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9854 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9855 (rl78_hard_regno_nregs): Make static. Take and return an
9857 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9858 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9859 (rs6000_hard_regno_nregs_hook): New function.
9860 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9861 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9862 (TARGET_HARD_REGNO_NREGS): Redefine.
9863 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9864 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9865 instead of HARD_REGNO_NREGS.
9866 (s390_hard_regno_nregs): New function.
9867 (s390_hard_regno_mode_ok): Add comment from s390.h.
9868 (TARGET_HARD_REGNO_NREGS): Redefine.
9869 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9870 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9871 (sh_hard_regno_nregs): New function.
9872 (sh_pass_in_reg_p): Use it.
9873 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9874 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9875 (sparc_hard_regno_nregs): New function.
9876 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9877 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9878 (spu_function_arg_advance): Use it, supplying a valid register number.
9879 (TARGET_HARD_REGNO_NREGS): Redefine.
9880 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9881 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9882 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9883 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9884 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9885 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9886 (CLASS_MAX_NREGS): Remove copy of old documentation.
9887 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9888 (visium_hard_regno_nregs): New function.
9889 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9890 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9891 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9892 xtensa_hard_regno_nregs): New function.
9893 * system.h (HARD_REGNO_NREGS): Poison.
9895 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9897 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9898 hard_regno_nregs instead of HARD_REGNO_NREGS.
9899 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9900 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9901 (c6x_expand_epilogue): Likewise.
9902 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9903 (frv_read_iacc_argument): Likewise.
9904 * config/sh/sh.c: Include regs.h.
9905 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9906 (regs_used): Likewise.
9907 (output_stack_adjust): Likewise.
9908 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9909 * expmed.c: Include regs.h.
9910 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9911 * ree.c: Include regs.h.
9912 (combine_reaching_defs): Use hard_regno_nregs instead of
9914 (add_removable_extension): Likewise.
9916 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9918 * regs.h (hard_regno_nregs): Turn into a function.
9919 (end_hard_regno): Update accordingly.
9920 * caller-save.c (setup_save_areas): Likewise.
9921 (save_call_clobbered_regs): Likewise.
9922 (replace_reg_with_saved_mem): Likewise.
9923 (insert_restore): Likewise.
9924 (insert_save): Likewise.
9925 * combine.c (can_change_dest_mode): Likewise.
9926 (move_deaths): Likewise.
9927 (distribute_notes): Likewise.
9928 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9929 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9930 (rs6000_split_multireg_move): Likewise.
9931 (rs6000_register_move_cost): Likewise.
9932 (rs6000_memory_move_cost): Likewise.
9933 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9934 (rs6000_split_multireg_move): Likewise.
9935 (rs6000_register_move_cost): Likewise.
9936 (rs6000_memory_move_cost): Likewise.
9937 * cselib.c (cselib_reset_table): Likewise.
9938 (cselib_lookup_1): Likewise.
9939 * emit-rtl.c (set_mode_and_regno): Likewise.
9940 * function.c (aggregate_value_p): Likewise.
9941 * ira-color.c (setup_profitable_hard_regs): Likewise.
9942 (check_hard_reg_p): Likewise.
9943 (calculate_saved_nregs): Likewise.
9944 (assign_hard_reg): Likewise.
9945 (improve_allocation): Likewise.
9946 (calculate_spill_cost): Likewise.
9947 * ira-emit.c (modify_move_list): Likewise.
9948 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9949 (ira_hard_reg_in_set_p): Likewise.
9950 * ira.c (setup_reg_mode_hard_regset): Likewise.
9951 (clarify_prohibited_class_mode_regs): Likewise.
9952 (check_allocation): Likewise.
9953 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9954 (lra_setup_reg_renumber): Likewise.
9955 (setup_try_hard_regno_pseudos): Likewise.
9956 (spill_for): Likewise.
9957 (assign_hard_regno): Likewise.
9958 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9959 * lra-constraints.c (in_class_p): Likewise.
9960 (lra_constraint_offset): Likewise.
9961 (simplify_operand_subreg): Likewise.
9962 (lra_constraints): Likewise.
9963 (split_reg): Likewise.
9964 (split_if_necessary): Likewise.
9965 (invariant_p): Likewise.
9966 (inherit_in_ebb): Likewise.
9967 * lra-lives.c (process_bb_lives): Likewise.
9968 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9969 (get_hard_regs): Likewise.
9970 (do_remat): Likewise.
9971 * lra-spills.c (assign_spill_hard_regs): Likewise.
9972 * mode-switching.c (create_pre_exit): Likewise.
9973 * postreload.c (reload_combine_recognize_pattern): Likewise.
9974 * recog.c (peep2_find_free_register): Likewise.
9975 * regcprop.c (kill_value_regno): Likewise.
9976 (set_value_regno): Likewise.
9977 (copy_value): Likewise.
9978 (maybe_mode_change): Likewise.
9979 (find_oldest_value_reg): Likewise.
9980 (copyprop_hardreg_forward_1): Likewise.
9981 * regrename.c (check_new_reg_p): Likewise.
9982 (regrename_do_replace): Likewise.
9983 * reload.c (push_reload): Likewise.
9984 (combine_reloads): Likewise.
9985 (find_dummy_reload): Likewise.
9986 (operands_match_p): Likewise.
9987 (find_reloads): Likewise.
9988 (find_equiv_reg): Likewise.
9989 (reload_adjust_reg_for_mode): Likewise.
9990 * reload1.c (count_pseudo): Likewise.
9991 (count_spilled_pseudo): Likewise.
9992 (find_reg): Likewise.
9993 (clear_reload_reg_in_use): Likewise.
9994 (free_for_value_p): Likewise.
9995 (allocate_reload_reg): Likewise.
9996 (choose_reload_regs): Likewise.
9997 (reload_adjust_reg_for_temp): Likewise.
9998 (emit_reload_insns): Likewise.
9999 (delete_output_reload): Likewise.
10000 * rtlanal.c (subreg_get_info): Likewise.
10001 * sched-deps.c (sched_analyze_reg): Likewise.
10002 * sel-sched.c (init_regs_for_mode): Likewise.
10003 (mark_unavailable_hard_regs): Likewise.
10004 (choose_best_reg_1): Likewise.
10005 (verify_target_availability): Likewise.
10006 * valtrack.c (dead_debug_insert_temp): Likewise.
10007 * var-tracking.c (track_loc_p): Likewise.
10008 (emit_note_insn_var_location): Likewise.
10009 * varasm.c (make_decl_rtl): Likewise.
10010 * reginfo.c (choose_hard_reg_mode): Likewise.
10011 (init_reg_modes_target): Refer directly to
10012 this_target_regs->x_hard_regno_nregs.
10014 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10016 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
10017 instead of hard_regno_nregs.
10019 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10021 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
10022 end_hard_regno instead of hard_regno_nregs.
10023 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
10024 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
10025 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
10026 * ira-color.c (improve_allocation): Likewise.
10027 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10028 * lra-lives.c (mark_regno_live): Likewise.
10029 (mark_regno_dead): Likewise.
10030 * lra-remat.c (operand_to_remat): Likewise.
10031 * lra.c (collect_non_operand_hard_regs): Likewise.
10032 * postreload.c (reload_combine_note_store): Likewise.
10033 (move2add_valid_value_p): Likewise.
10034 * reload.c (regno_clobbered_p): Likewise.
10036 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10038 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10040 * config/v850/v850.c (v850_reorg): Likewise.
10041 * reload.c (refers_to_regno_for_reload_p): Likewise.
10042 (find_equiv_reg): Likewise.
10043 * reload1.c (reload_reg_reaches_end_p): Likewise.
10045 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10047 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10049 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10050 * config/arm/arm.c (output_move_neon): Likewise.
10051 (arm_attr_length_move_neon): Likewise.
10052 (neon_split_vcombine): Likewise.
10053 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10054 (c6x_mark_reg_written): Likewise.
10055 (c6x_dwarf_register_span): Likewise.
10056 * config/i386/i386.c (ix86_save_reg): Likewise.
10057 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10058 (rws_access_reg): Likewise.
10059 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10060 * mode-switching.c (create_pre_exit): Likewise.
10061 * ree.c (combine_reaching_defs): Likewise.
10062 (add_removable_extension): Likewise.
10063 * regcprop.c (find_oldest_value_reg): Likewise.
10064 (copyprop_hardreg_forward_1): Likewise.
10065 * reload.c (reload_inner_reg_of_subreg): Likewise.
10066 (push_reload): Likewise.
10067 (combine_reloads): Likewise.
10068 (find_dummy_reload): Likewise.
10069 (reload_adjust_reg_for_mode): Likewise.
10070 * reload1.c (find_reload_regs): Likewise.
10071 (forget_old_reloads_1): Likewise.
10072 (reload_reg_free_for_value_p): Likewise.
10073 (reload_adjust_reg_for_temp): Likewise.
10074 (emit_reload_insns): Likewise.
10075 (delete_output_reload): Likewise.
10076 * sel-sched.c (choose_best_reg_1): Likewise.
10077 (choose_best_pseudo_reg): Likewise.
10079 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10080 Alan Hayward <alan.hayward@arm.com>
10081 David Sherwood <david.sherwood@arm.com>
10083 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10084 * target.def (slow_unaligned_access): New hook.
10085 * targhooks.h (default_slow_unaligned_access): Declare.
10086 * targhooks.c (default_slow_unaligned_access): New function.
10087 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10088 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10089 * doc/tm.texi: Regenerate.
10090 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10091 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10092 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10094 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10095 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10097 (rs6000_slow_unaligned_access): New function.
10098 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10099 (expand_block_compare): Likewise.
10100 (expand_strn_compare): Likewise.
10101 (rs6000_rtx_costs): Likewise.
10102 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10103 (riscv_slow_unaligned_access): Likewise.
10104 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10105 (riscv_slow_unaligned_access_p): ...this and make static.
10106 (riscv_option_override): Update accordingly.
10107 (riscv_slow_unaligned_access): New function.
10108 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10109 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10110 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10111 (rs6000_slow_unaligned_access): New function.
10112 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10113 (rs6000_rtx_costs): Likewise.
10114 * config/rs6000/rs6000-string.c (expand_block_compare)
10115 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10116 of SLOW_UNALIGNED_ACCESS.
10117 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10118 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10119 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10120 of SLOW_UNALIGNED_ACCESS.
10121 * expmed.c (simple_mem_bitfield_p): Likewise.
10122 * expr.c (alignment_for_piecewise_move): Likewise.
10123 (emit_group_load_1): Likewise.
10124 (emit_group_store): Likewise.
10125 (copy_blkmode_from_reg): Likewise.
10126 (emit_push_insn): Likewise.
10127 (expand_assignment): Likewise.
10128 (store_field): Likewise.
10129 (expand_expr_real_1): Likewise.
10130 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10131 * lra-constraints.c (simplify_operand_subreg): Likewise.
10132 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10133 * gimple-ssa-store-merging.c: Likewise in block comment at start
10135 * tree-ssa-strlen.c: Include target.h.
10136 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10137 of SLOW_UNALIGNED_ACCESS.
10138 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10140 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10142 PR rtl-optimization/82185
10143 * expmed.c (emit_store_flag_int): Only test tem if it has been
10146 2017-09-12 Richard Biener <rguenther@suse.de>
10148 PR middle-end/82149
10149 * match.pd ((FTYPE) N CMP CST): Fix typo.
10151 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10153 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10155 (mips_near_type_p): Add 'short_call' attribute as a synonym
10157 * doc/extend.texi (short_call): Document new function attribute.
10159 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10162 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10163 assertion check that in the condition.
10164 (get_atomic_generic_size): Likewise. Before testing if parameter
10165 has pointer type, if it has array type, call for C++
10166 default_conversion to perform array-to-pointer conversion.
10168 2017-09-12 Richard Biener <rguenther@suse.de>
10170 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10171 for operations we cannot scalarize.
10173 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10175 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10176 vectors heap vectors. Clean up comments.
10177 Make visited_bbs a reference.
10178 (profitable_jump_thread_path): Make GC
10179 vectors heap vectors. Clean up comments.
10181 (convert_and_register_jump_thread_path): Make GC vectors heap
10183 (check_subpath_and_update_thread_path): Same. Clean up comments.
10184 Make visited_bbs a reference.
10185 (handle_phi): Abstract common code to to
10186 register_jump_thread_path_if_profitable.
10187 Rename VAR_BB to DEF_BB.
10189 Make GC vectors heap vectors.
10190 Make visited_bbs a reference.
10191 (handle_assignment): Same.
10192 (register_jump_thread_path_if_profitable): New.
10193 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10195 Make GC vectors heap vectors. Clean up comments.
10196 Make visited_bbs a reference.
10197 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10198 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10201 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10204 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10205 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10207 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10209 Revert r251800 and r251799.
10211 2017-09-11 Martin Jambor <mjambor@suse.cz>
10214 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10215 arguments in advance.
10216 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10217 use it to find predecessor edges.
10218 (naive_outof_ssa): Collect vector of predecessors.
10220 2017-09-08 Jason Merrill <jason@redhat.com>
10222 PR c++/70029 - ICE with ref-qualifier and -flto
10223 * langhooks.h (struct lang_hooks_for_types): Add
10224 copy_lang_qualifiers.
10225 * attribs.c (build_type_attribute_qual_variant): Use it.
10226 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10228 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10229 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10231 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10234 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10235 (*mulsi3_sp64): New instruction.
10236 (mulsi3): New expander.
10238 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10240 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10242 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10244 * sancov.c: Include memmodel.h.
10246 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10249 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10252 2017-09-07 Carl Love <cel@us.ibm.com>
10254 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10255 the sldi instruction.
10257 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10259 * sancov.c: Include tm_p.h.
10261 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10264 * output.h (switch_to_other_text_partition): New declaration.
10265 * varasm.c (switch_to_other_text_partition): New function.
10266 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10267 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10268 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10269 to the other text partition before emitting LCL label and switch back
10270 after emitting the word after it.
10272 2017-09-07 Richard Biener <rguenther@suse.de>
10274 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10275 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10276 critical edges here, after loop init.
10277 (pass_data_pre): Remove PROP_no_crit_edges flags.
10278 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10279 for valueization of call args to avoid leaking VN_TOP.
10280 (visit_use): Assert we do not visit default defs.
10281 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10282 Use error_mark_node to more easily detect leaking VN_TOP.
10283 All default-defs are varying, not VN_TOP. Mark them visited.
10284 (run_scc_vn): Make code match comment.
10286 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10288 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10289 OPTION_MASK_FLOAT128_KEYWORD.
10290 (POWERPC_MASKS): Likewise.
10291 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10292 support for the -mfloat128-type option, and make -mfloat128
10293 default on PowerPC Linux systems. Define or undefine
10294 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10295 Define __float128 to be __ieee128 if IEEE 128-bit support is
10296 enabled, or undefine it.
10297 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10298 Delete defining __FLOAT128_TYPE__.
10299 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10300 -mfloat128-type option and make -mfloat128 default on PowerPC
10302 (TARGET_FLOAT128_TYPE): Likewise.
10303 (-mfloat128-type): Likewise.
10304 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10305 Delete the -mfloat128-type option and make -mfloat128 default on
10306 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10307 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10308 128-bit floating point is enabled. Change tests from using
10309 -mfloat128-type to -mfloat128.
10310 (rs6000_mangle_type): Use the correct mangling for the __float128
10311 type even if normal long double is restricted to 64-bits.
10312 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10314 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10315 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10316 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10318 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10319 documentation for -mfloat128.
10321 2017-09-06 Olivier Hainque <hainque@adacore.com>
10323 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10325 2017-09-06 Wish Wu <wishwu007@gmail.com>
10326 Jakub Jelinek <jakub@redhat.com>
10328 * asan.c (initialize_sanitizer_builtins): Add
10329 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10330 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10331 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10332 BT_FN_VOID_UINT64_PTR variables.
10333 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10334 (BT_FN_VOID_UINT16_UINT16): Likewise.
10335 (BT_FN_VOID_UINT32_UINT32): Likewise.
10336 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10337 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10338 (BT_FN_VOID_UINT64_PTR): Likewise.
10339 * common.opt (flag_sanitize_coverage): New variable.
10340 (fsanitize-coverage=trace-pc): Remove.
10341 (fsanitize-coverage=): Add.
10342 * flag-types.h (enum sanitize_coverage_code): New enum.
10343 * fold-const.c (fold_range_test): Disable non-short-circuit
10344 optimization if flag_sanitize_coverage.
10345 (fold_truth_andor): Likewise.
10346 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10347 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10348 (coverage_sanitizer_opts): New array.
10349 (get_closest_sanitizer_option): Add OPTS argument, handle also
10350 OPT_fsanitize_coverage_.
10351 (parse_sanitizer_options): Adjusted to also handle
10352 OPT_fsanitize_coverage_.
10353 (common_handle_option): Add OPT_fsanitize_coverage_.
10354 * sancov.c (instrument_comparison, instrument_switch): New function.
10355 (sancov_pass): Add trace-cmp support.
10356 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10357 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10358 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10359 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10360 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10361 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10362 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10363 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10364 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10365 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10367 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10369 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10370 error. Only quit immediately if parsing is complete.
10371 (BEGIN): Initialize fatal_err and parse_done.
10372 (begin fpu, end fpu): Check number of arguments.
10373 (begin arch, end arch): Likewise.
10374 (begin cpu, end cpu): Likewise.
10375 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10376 (optalias): Likewise.
10378 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10380 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10381 * config/arm/arm-isa.h: Delete. Move definitions to ...
10382 * arm-cpus.in: ... here. Use new feature and fgroup values.
10383 * config/arm/arm.c (arm_option_override): Use lower case for feature
10385 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10386 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10387 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10389 (print_isa_bits_for): New function.
10390 (gen_isa): New function.
10391 (gen_comm_data): Use print_isa_bits_for.
10392 (define feature): New keyword.
10393 (define fgroup): New keyword.
10394 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10395 (arm-isa.h): Add rule to generate file.
10396 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10397 case for feature bit names.
10399 2017-09-06 Richard Biener <rguenther@suse.de>
10401 * tree-ssa-pre.c (NECESSARY): Remove.
10402 (create_expression_by_pieces): Do not touch pass-local flags.
10403 (insert_into_preds_of_block): Likewise.
10404 (do_pre_regular_insertion): Likewise.
10405 (eliminate_insert): Likewise.
10406 (eliminate_dom_walker::before_dom_children): Likewise.
10407 (fini_eliminate): Do not look at inserted_exprs.
10408 (mark_operand_necessary): Remove.
10409 (remove_dead_inserted_code): Replace with simple work-list
10410 algorithm based on inserted_exprs and SSA uses.
10411 (pass_pre::execute): Re-order fini_eliminate and
10412 remove_dead_inserted_code.
10414 2017-09-06 Olivier Hainque <hainque@adacore.com>
10416 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10417 for VxWorks 7. Adjust surrounding comments.
10419 2017-09-06 Richard Biener <rguenther@suse.de>
10421 * gimple-ssa-strength-reduction.c
10422 (find_candidates_dom_walker::before_dom_children): Also allow
10425 2017-09-06 Richard Biener <rguenther@suse.de>
10427 PR tree-optimization/82108
10428 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10429 for gap in the non-permutation SLP case.
10431 2017-09-06 Martin Jambor <mjambor@suse.cz>
10433 PR tree-optimization/82078
10434 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10435 add_access_to_work_queue...
10436 (build_accesses_from_assign): ...here.
10437 (propagate_all_subaccesses): Make sure racc is the group
10438 representative, if there is one.
10440 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10442 PR middle-end/82095
10443 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10446 2017-09-06 Richard Biener <rguenther@suse.de>
10448 * gimple-ssa-strength-reduction.c
10449 (find_candidates_dom_walker::before_doom_children): Use a
10450 type and not a mode check.
10452 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10455 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10457 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10459 2017-09-05 Jeff Law <law@redhat.com>
10461 PR tree-optimization/64910
10462 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10463 cases where we have 3 or more operands.
10465 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10467 PR middle-end/81768
10468 * omp-low.c (lower_omp_for): Recompute tree invariant if
10469 gimple_omp_for_initial/final is ADDR_EXPR.
10471 PR middle-end/81768
10472 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10473 into gimple val before gimplification fo the COND_EXPR.
10475 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10477 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10478 REGION_COPY argument.
10479 (thread_through_all_blocks): Remove unused argument to
10480 duplicate_thread_path.
10482 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10483 Alan Hayward <alan.hayward@arm.com>
10484 David Sherwood <david.sherwood@arm.com>
10486 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10487 Take a scalar_mode rather than a machine_mode.
10488 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10489 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10490 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10491 (aarch64_gen_adjusted_ldpstp): Likewise.
10492 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10494 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10495 Alan Hayward <alan.hayward@arm.com>
10496 David Sherwood <david.sherwood@arm.com>
10498 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10499 Take a scalar_int_mode instead of a machine_mode.
10500 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10501 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10502 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10503 (aarch64_simd_attr_length_rglist): Delete.
10504 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10505 a scalar_int_mode instead of a machine_mode.
10506 (aarch64_add_offset): Likewise.
10507 (aarch64_internal_mov_immediate): Likewise
10508 (aarch64_add_constant_internal): Likewise.
10509 (aarch64_add_constant): Likewise.
10510 (aarch64_movw_imm): Likewise.
10511 (aarch64_rtx_arith_op_extract_p): Likewise.
10512 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10513 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10514 Remove assert that the mode isn't a vector.
10515 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10516 (aarch64_expand_mov_immediate): Update calls after above changes.
10517 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10518 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10519 (aarch64_move_imm): Likewise.
10520 (aarch64_can_const_movi_rtx_p): Likewise.
10521 (aarch64_strip_extend): Likewise.
10522 (aarch64_extr_rtx_p): Likewise.
10523 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10524 a CONST_INT when the mode parameter is VOIDmode.
10525 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10527 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10529 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10530 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10531 (bitwise_type_for_mode): Update accordingly.
10533 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10535 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10536 * stor-layout.c (mode_for_size_tree): Likewise.
10537 (mode_for_array): Update accordingly.
10538 (layout_decl): Likewise.
10539 (compute_record_mode): Likewise. Only set the mode once.
10541 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10543 * target.def (get_mask_mode): Change return type to opt_mode.
10545 * doc/tm.texi: Regenerate.
10546 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10547 * targhooks.c (default_get_mask_mode): Likewise.
10548 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10549 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10550 targetm.get_mask_mode.
10551 * tree.c (build_truth_vector_type): Likewise.
10553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10555 * machmode.h (mode_for_vector): Return an opt_mode.
10556 * stor-layout.c (mode_for_vector): Likewise.
10557 (mode_for_int_vector): Update accordingly.
10558 (layout_type): Likewise.
10559 * config/i386/i386.c (emit_memmov): Likewise.
10560 (ix86_expand_set_or_movmem): Likewise.
10561 (ix86_expand_vector_init): Likewise.
10562 (ix86_get_mask_mode): Likewise.
10563 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10565 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10566 * expmed.c (extract_bit_field_1): Likewise.
10567 * expr.c (expand_expr_real_2): Likewise.
10568 * optabs-query.c (can_vec_perm_p): Likewise.
10569 (can_vec_mask_load_store_p): Likewise.
10570 * optabs.c (expand_vec_perm): Likewise.
10571 * targhooks.c (default_get_mask_mode): Likewise.
10572 * tree-vect-stmts.c (vectorizable_store): Likewise.
10573 (vectorizable_load): Likewise.
10574 (get_vectype_for_scalar_type_and_size): Likewise.
10576 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10578 * machmode.h (mode_for_int_vector): New function.
10579 * stor-layout.c (mode_for_int_vector): Likewise.
10580 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10581 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10582 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10583 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10584 (s390_expand_vcond): Likewise.
10586 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10588 * machmode.h (opt_machine_mode): New type.
10589 (opt_mode<T>): Allow construction from anything that can be
10591 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10592 (mode_for_size): Return an opt_machine_mode.
10593 * stor-layout.c (mode_for_size): Likewise.
10594 (mode_for_size_tree): Update call accordingly.
10595 (bitwise_mode_for_mode): Likewise.
10596 (make_fract_type): Likewise.
10597 (make_accum_type): Likewise.
10598 * caller-save.c (replace_reg_with_saved_mem): Update call
10600 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10601 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10602 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10603 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10604 * expmed.c (extract_bit_field_1): Likewise.
10605 * reload.c (get_secondary_mem): Likewise.
10606 * varasm.c (assemble_integer): Likewise.
10607 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10610 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10612 * machmode.h (decimal_float_mode_for_size): New function.
10613 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10614 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10615 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10616 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10617 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10618 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10620 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10622 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10623 (get_builtin_sync_mode): Likewise.
10624 (expand_ifn_atomic_compare_exchange): Likewise.
10625 (expand_builtin_atomic_clear): Likewise.
10626 (expand_builtin_atomic_test_and_set): Likewise.
10627 (fold_builtin_atomic_always_lock_free): Likewise.
10628 * calls.c (compute_argument_addresses): Likewise.
10629 (emit_library_call_value_1): Likewise.
10630 (store_one_arg): Likewise.
10631 * combine.c (combine_instructions): Likewise.
10632 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10633 * config/arm/arm.c (arm_function_value): Likewise.
10634 (aapcs_allocate_return_reg): Likewise.
10635 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10636 * config/i386/i386.c (construct_container): Likewise.
10637 (ix86_gimplify_va_arg): Likewise.
10638 (ix86_expand_sse_cmp): Likewise.
10639 (emit_memmov): Likewise.
10640 (emit_memset): Likewise.
10641 (expand_small_movmem_or_setmem): Likewise.
10642 (ix86_expand_pextr): Likewise.
10643 (ix86_expand_pinsr): Likewise.
10644 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10645 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10647 * config/mips/mips.c (mips_function_value_1) Likewise.
10648 (mips_block_move_straight): Likewise.
10649 (mips_expand_ins_as_unaligned_store): Likewise.
10650 * config/powerpcspe/powerpcspe.c
10651 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10652 (rs6000_darwin64_record_arg_flush): Likewise.
10653 * config/rs6000/rs6000.c
10654 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10655 (rs6000_darwin64_record_arg_flush): Likewise.
10656 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10657 (sparc_function_value_1): Likewise.
10658 * config/spu/spu.c (adjust_operand): Likewise.
10659 (spu_emit_branch_or_set): Likewise.
10660 (arith_immediate_p): Likewise.
10661 * emit-rtl.c (gen_lowpart_common): Likewise.
10662 * expr.c (expand_expr_real_1): Likewise.
10663 * function.c (assign_parm_setup_block): Likewise.
10664 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10665 * reload1.c (alter_reg): Likewise.
10666 * stor-layout.c (mode_for_vector): Likewise.
10667 (layout_type): Likewise.
10669 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10671 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10672 (spu_convert_move): Likewise.
10673 * lower-subreg.c (resolve_simple_move): Likewise.
10675 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10678 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10679 define_insn to a define_expand.
10680 (altivec_vsum2sws_direct): New define_insn.
10681 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10683 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10685 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10688 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10693 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10694 realign stack if stack alignment needed is less than incoming
10697 2017-09-05 Marek Polacek <polacek@redhat.com>
10700 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10703 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10705 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10707 2017-09-05 Richard Biener <rguenther@suse.de>
10709 PR tree-optimization/82084
10710 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10712 2017-09-05 Richard Biener <rguenther@suse.de>
10714 PR tree-optimization/82102
10715 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10717 2017-09-05 Martin Liska <mliska@suse.cz>
10719 PR tree-optimization/82032
10720 * tree-cfg.c (generate_range_test): New function.
10721 * tree-cfg.h (generate_range_test): Declared here.
10722 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10723 (cleanup_control_expr_graph): Use it.
10724 * tree-switch-conversion.c (try_switch_expansion): Remove
10726 (emit_case_nodes): Use generate_range_test.
10728 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10731 * config/i386/i386.md (*<btsc><mode>_mask): Add
10732 TARGET_USE_BT to insn constraint.
10733 (*btr<mode>_mask): Ditto.
10735 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10737 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10738 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10740 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10743 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10744 TARGET_NEON and TARGET_IWMMXT.
10745 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10746 TARGET_NEON and TARGET_IWMMXT.
10747 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10749 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10751 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10752 (ix86_rewrite_tls_address): Ditto.
10753 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10754 (ix86_rewrite_tls_address_1): Ditto.
10755 (ix86_rewrite_tls_address): Ditto.
10756 * config/i386/predicates.md (tls_address_pattern): New predicate.
10757 * config/i386/i386.md (TLS address splitter): New splitter.
10759 2017-09-04 Richard Biener <rguenther@suse.de>
10761 PR tree-optimization/82084
10762 * fold-const.h (can_native_encode_string_p): Declare.
10763 * fold-const.c (can_native_encode_string_p): Factor out from ...
10764 (native_encode_string): ... here.
10765 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10766 vectorizing stores from constants we later cannot handle.
10768 2017-09-04 Marek Polacek <polacek@redhat.com>
10771 * doc/invoke.texi: Update -Wtautological-compare documentation.
10773 2017-09-04 Jeff Law <law@redhat.com>
10775 PR tree-optimization/64910
10776 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10777 swap the first and last operand if the last is a constant.
10779 2017-09-04 Marek Polacek <polacek@redhat.com>
10782 * convert.c (do_narrow): When sanitizing signed integer overflows,
10783 bail out for signed types.
10784 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10786 2017-09-04 Richard Biener <rguenther@suse.de>
10788 PR tree-optimization/82060
10789 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10790 Move devirtualization after stmt folding and before EH/AB/noreturn
10791 cleanup to get the stmt refs canonicalized. Use a bool instead
10792 of gimple_modified_p since that doesn't work for NOPs. Schedule
10793 NOPs generated by folding for removal.
10795 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10796 Alan Hayward <alan.hayward@arm.com>
10797 David Sherwood <david.sherwood@arm.com>
10799 * coretypes.h (pad_direction): New enum.
10800 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10801 (FUNCTION_ARG_PADDING): Likewise.
10802 * target.def (function_arg_padding): New hook.
10803 * targhooks.h (default_function_arg_padding): Declare.
10804 * targhooks.c (default_function_arg_padding): New function.
10805 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10806 (TARGET_FUNCTION_ARG_PADDING): ...this.
10807 * doc/tm.texi: Regenerate.
10808 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10809 instead of direction.
10810 (compute_argument_addresses): Likewise.
10811 (load_register_parameters): Likewise.
10812 (emit_library_call_value_1): Likewise.
10813 (store_one_arg): Use targetm.calls.function_arg_padding instead
10814 of FUNCTION_ARG_PADDING.
10815 (must_pass_in_stack_var_size_or_pad): Likewise.
10816 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10817 (emit_group_store): Likewise.
10818 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10819 instead of FUNCTION_ARG_PADDING.
10820 (emit_push_insn): Likewise, and propagate enum change throughout
10822 * function.h (direction): Delete.
10823 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10825 * function.c (assign_parm_find_stack_rtl): Likewise.
10826 (assign_parm_setup_block_p): Likewise.
10827 (assign_parm_setup_block): Likewise.
10828 (gimplify_parameters): Likewise.
10829 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10830 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10832 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10833 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10834 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10835 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10836 (aarch64_function_arg_padding): ...this new function.
10837 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10838 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10839 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10840 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10841 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10842 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10843 (arm_pad_arg_upward): Replace with...
10844 (arm_function_arg_padding): ...this new function.
10845 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10847 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10848 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10849 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10850 (ia64_hpux_function_arg_padding): Replace with...
10851 (ia64_function_arg_padding): ...this new function. Use pad_direction
10852 instead of direction. Check for TARGET_HPUX.
10853 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10854 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10855 (iq2000_function_arg_padding): New function.
10856 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10857 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10858 (mips_function_arg_padding): ...this new function.
10859 (mips_pad_reg_upward): Update accordingly.
10860 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10861 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10862 targetm.calls.function_arg_padding.
10863 (FUNCTION_ARG_PADDING): Delete.
10864 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10865 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10866 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10867 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10868 (nios2_block_reg_padding): Return pad_direction instead of direction.
10869 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10870 instead of direction.
10871 (nios2_function_arg_padding): Likewise. Make static.
10872 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10873 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10874 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10875 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10876 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10877 (pa_function_arg_padding): Make static. Return pad_direction instead
10879 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10880 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10881 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10882 instead of direction. Use targetm.calls.function_arg_padding.
10883 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10884 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10885 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10886 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10887 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10889 (function_arg_padding): Rename to...
10890 (rs6000_function_arg_padding): ...this. Make static. Return
10891 pad_direction instead of direction.
10892 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10893 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10894 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10895 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10896 instead of direction. Use targetm.calls.function_arg_padding.
10897 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10898 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10899 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10900 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10901 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10902 (function_arg_padding): Rename to...
10903 (rs6000_function_arg_padding): ...this. Make static. Return
10904 pad_direction instead of direction.
10905 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10906 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10907 * config/s390/s390.c (s390_function_arg_padding): New function.
10908 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10909 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10910 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10911 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10912 (function_arg_padding): Rename to...
10913 (sparc_function_arg_padding): ...this. Make static. Return
10914 pad_direction instead of direction.
10915 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10916 * config/spu/spu.c (spu_function_arg_padding): New function.
10917 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10918 * system.h (FUNCTION_ARG_PADDING): Poison.
10920 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10921 Alan Hayward <alan.hayward@arm.com>
10922 David Sherwood <david.sherwood@arm.com>
10924 * target.def (modes_tieable_p): New hook.
10925 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10926 (TARGET_MODES_TIEABLE_P): ...this.
10927 * doc/tm.texi.in: Regenerate.
10928 * hooks.h (hook_bool_mode_mode_true): Declare.
10929 * hooks.c (hook_bool_mode_mode_true): New function.
10930 * combine.c (subst): Use targetm.modes_tieable_p instead of
10932 * dse.c (find_shift_sequence): Likewise.
10933 * expmed.c (extract_low_bits): Likewise.
10934 * lower-subreg.c: Include target.h.
10935 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10937 * rtlanal.c (rtx_cost): Likewise.
10938 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10939 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10940 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10941 (TARGET_MODES_TIEABLE_P): Redefine.
10942 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10943 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10944 (TARGET_MODES_TIEABLE_P): Redefine.
10945 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10946 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10947 (arc_modes_tieable_p): New function.
10948 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10949 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10950 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10951 (arm_modes_tieable_p): Make static.
10952 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10953 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10954 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10955 (TARGET_MODES_TIEABLE_P): Redefine.
10956 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10957 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10958 (TARGET_MODES_TIEABLE_P): Redefine.
10959 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10960 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10961 (cr16_modes_tieable_p): New function.
10962 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10963 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10964 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10965 (TRULY_NOOP_TRUNCATION): Update comment.
10966 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10967 (TRULY_NOOP_TRUNCATION): Update comment.
10968 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10969 (frv_modes_tieable_p): New function.
10970 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10971 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10972 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10973 (TARGET_MODES_TIEABLE_P): Redefine.
10974 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10975 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10976 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10977 (TARGET_MODES_TIEABLE_P): Redefine.
10978 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10979 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10980 (ia64_modes_tieable_p): New function.
10981 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10982 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10983 (iq2000_modes_tieable_p): New function.
10984 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10985 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10986 (lm32_modes_tieable_p): New function.
10987 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10988 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10989 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10990 (TARGET_MODES_TIEABLE_P): Redefine.
10991 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10992 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10993 (m32r_modes_tieable_p): New function.
10994 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10995 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10996 (m68k_modes_tieable_p): New function.
10997 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10998 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10999 (mcore_modes_tieable_p): New function.
11000 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
11001 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
11003 (TARGET_MODES_TIEABLE_P): Redefine.
11004 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
11005 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
11006 * config/mips/mips.c (mips_modes_tieable_p): Make static.
11007 (TARGET_MODES_TIEABLE_P): Redefine.
11008 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
11009 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
11010 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
11011 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
11012 (mn10300_modes_tieable_p): ...this and make static.
11013 (TARGET_MODES_TIEABLE_P): Redefine.
11014 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
11015 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
11016 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
11017 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
11018 (msp430_modes_tieable_p): Make static.
11019 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
11020 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
11021 (TARGET_MODES_TIEABLE_P): Redefine.
11022 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
11023 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
11024 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
11025 (TARGET_MODES_TIEABLE_P): Redefine.
11026 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
11027 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
11028 * config/pa/pa.c (pa_modes_tieable_p): Make static.
11029 (TARGET_MODES_TIEABLE_P): Redefine.
11030 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
11031 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
11032 (pdp11_modes_tieable_p): New function.
11033 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
11034 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
11035 (rs6000_modes_tieable_p): New function.
11036 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11037 * config/powerpcspe/powerpcspe.md: Update comment.
11038 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11039 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11040 (TARGET_MODES_TIEABLE_P): Redefine.
11041 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11042 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11043 (rl78_modes_tieable_p): New function.
11044 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11045 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11046 (rs6000_modes_tieable_p): New function.
11047 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11048 * config/rs6000/rs6000.md: Update comment.
11049 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11050 * config/rx/rx.c (rx_modes_tieable_p): New function.
11051 (TARGET_MODES_TIEABLE_P): Redefine.
11052 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11053 * config/s390/s390.c (s390_modes_tieable_p): New function.
11054 (TARGET_MODES_TIEABLE_P): Redefine.
11055 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11056 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11057 (sh_modes_tieable_p): New function.
11058 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11059 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11060 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11061 (sparc_modes_tieable_p): Make static.
11062 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11063 * config/spu/spu.c (spu_modes_tieable_p): New function.
11064 (TARGET_MODES_TIEABLE_P): Redefine.
11065 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11066 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11067 (TARGET_MODES_TIEABLE_P): Redefine.
11068 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11069 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11070 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11071 * config/v850/v850.c (v850_modes_tieable_p): New function.
11072 (TARGET_MODES_TIEABLE_P): Redefine.
11073 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11074 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11075 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11076 (visium_modes_tieable_p): New function.
11077 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11078 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11079 (xtensa_modes_tieable_p): New function.
11080 * system.h (MODES_TIEABLE_P): Poison.
11082 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11083 Alan Hayward <alan.hayward@arm.com>
11084 David Sherwood <david.sherwood@arm.com>
11086 * target.def (hard_regno_mode_ok): New hook.
11087 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11088 (TARGET_HARD_REGNO_MODE_OK): ...this.
11089 * doc/tm.texi.in: Regenerate.
11090 * hooks.h (hook_bool_uint_mode_true): Declare.
11091 * hooks.c (hook_bool_uint_mode_true): New function.
11092 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11093 HARD_REGNO_MODE_OK.
11094 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11095 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11096 instead of HARD_REGNO_MODE_OK.
11097 * caller-save.c: Include target.h.
11098 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11099 HARD_REGNO_MODE_OK.
11100 * combine.c (can_combine_p): Likewise.
11101 (combinable_i3pat): Likewise.
11102 (can_change_dest_mode): Likewise.
11103 * expr.c (init_expr_target): Likewise.
11104 (convert_move): Likewise.
11105 (convert_modes): Likewise.
11106 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11107 (setup_prohibited_mode_move_regs): Likewise.
11108 * ira.h (target_ira): Likewise.
11109 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11110 * lra-constraints.c (process_alt_operands): Likewise.
11111 (split_reg): Likewise.
11112 * recog.c (peep2_find_free_register): Likewise.
11113 * ree.c (combine_reaching_defs): Likewise.
11114 * regcprop.c (maybe_mode_change): Likewise.
11115 * reginfo.c (init_reg_sets_1): Likewise.
11116 (choose_hard_reg_mode): Likewise.
11117 (simplifiable_subregs): Likewise.
11118 * regrename.c (check_new_reg_p): Likewise.
11119 * reload.c (find_valid_class): Likewise.
11120 (find_valid_class_1): Likewise.
11121 (reload_inner_reg_of_subreg): Likewise.
11122 (push_reload): Likewise.
11123 (combine_reloads): Likewise.
11124 (find_dummy_reload): Likewise.
11125 (find_reloads): Likewise.
11126 * reload1.c (find_reg): Likewise.
11127 (set_reload_reg): Likewise.
11128 (allocate_reload_reg): Likewise.
11129 (choose_reload_regs): Likewise.
11130 (reload_adjust_reg_for_temp): Likewise.
11131 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11132 (simplify_subreg_regno): Likewise.
11133 * sel-sched.c (init_regs_for_mode): Likewise.
11134 * varasm.c (make_decl_rtl): Likewise.
11135 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11136 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11137 HARD_REGNO_MODE_OK.
11138 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11139 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11141 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11142 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11143 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11144 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11145 (arc_mode_class): Delete.
11146 (HARD_REGNO_MODE_OK): Delete.
11147 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11148 (arc_hard_regno_mode_ok): Rename old array to...
11149 (arc_hard_regno_mode_ok_modes): ...this.
11150 (arc_conditional_register_usage): Update accordingly.
11151 (arc_mode_class): Make static.
11152 (arc_hard_regno_mode_ok): New function.
11153 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11154 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11155 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11156 (arm_hard_regno_mode_ok): Make static.
11157 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11158 HARD_REGNO_MODE_OK.
11159 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11160 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11161 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11164 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11165 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11166 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11167 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11169 * config/bfin/predicates.md (valid_reg_operand): Use
11170 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11171 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11172 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11174 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11175 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11176 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11177 (cr16_hard_regno_mode_ok): Make static and return a bool.
11178 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11179 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11180 (cris_hard_regno_mode_ok): New function.
11181 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11182 (epiphany_mode_class): Delete.
11183 (HARD_REGNO_MODE_OK): Delete.
11184 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11185 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11186 (hard_regno_mode_ok): Rename to...
11187 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11188 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11189 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11190 HARD_REGNO_MODE_OK.
11191 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11192 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11193 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11194 (frv_hard_regno_mode_ok): Make static and return a bool.
11195 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11196 HARD_REGNO_MODE_OK.
11197 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11198 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11199 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11200 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11203 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11204 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11205 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11207 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11208 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11209 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11210 (ia64_hard_regno_mode_ok): New function.
11211 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11212 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11213 (iq2000_hard_regno_mode_ok): New function.
11214 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11215 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11216 (lm32_hard_regno_mode_ok): New function.
11217 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11218 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11219 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11220 instead of HARD_REGNO_MODE_OK.
11221 (m32c_hard_regno_ok): Rename to...
11222 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11223 (m32c_cannot_change_mode_class): Update accordingly.
11224 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11225 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11226 (m32r_mode_class): Delete.
11227 (HARD_REGNO_MODE_OK): Delete.
11228 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11229 (m32r_hard_regno_mode_ok): Rename to...
11230 (m32r_hard_regno_modes): ...this.
11231 (m32r_mode_class): Make static.
11232 (m32r_hard_regno_mode_ok): New function.
11233 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11234 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11235 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11236 (m68k_hard_regno_mode_ok): Make static.
11237 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11238 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11239 (mcore_hard_regno_mode_ok): New function.
11240 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11241 (HARD_REGNO_MODE_OK): Delete.
11242 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11244 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11245 (microblaze_hard_regno_mode_ok): New function.
11246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11247 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11248 (mips_hard_regno_mode_ok): Delete.
11249 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11250 (mips_hard_regno_mode_ok_p): ...this and make static.
11251 (mips_hard_regno_mode_ok_p): Rename to...
11252 (mips_hard_regno_mode_ok_uncached): ...this.
11253 (mips_hard_regno_mode_ok): New function.
11254 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11255 of HARD_REGNO_MODE_OK.
11256 (mips_option_override): Update after above name changes.
11257 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11258 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11259 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11260 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11261 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11262 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11263 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11264 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11265 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11266 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11267 (msp430_hard_regno_mode_ok): Make static and return a bool.
11268 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11269 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11270 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11273 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11274 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11275 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11276 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11277 (PA_HARD_REGNO_MODE_OK): ...this
11278 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11279 (PA_HARD_REGNO_MODE_OK): ...this.
11280 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11281 (pa_hard_regno_mode_ok): New function.
11282 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11283 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11284 (pdp11_hard_regno_mode_ok): New function.
11285 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11286 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11288 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11290 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11291 (rs6000_hard_regno_mode_ok): Rename to...
11292 (rs6000_hard_regno_mode_ok_uncached): ...this.
11293 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11294 (rs6000_hard_regno_mode_ok): New function.
11295 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11296 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11297 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11298 (riscv_hard_regno_mode_ok): ...this and make static.
11299 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11300 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11301 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11302 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11303 (rl78_hard_regno_mode_ok): Make static and return bool.
11304 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11305 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11307 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11308 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11309 (rs6000_hard_regno_mode_ok): Rename to...
11310 (rs6000_hard_regno_mode_ok_uncached): ...this.
11311 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11312 (rs6000_hard_regno_mode_ok): New function.
11313 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11314 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11315 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11316 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11317 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11318 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11319 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11320 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11321 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11322 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11323 (sh_hard_regno_mode_ok): Make static.
11324 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11325 instead of HARD_REGNO_MODE_OK.
11326 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11327 (sparc_mode_class): Delete.
11328 (HARD_REGNO_MODE_OK): Delete.
11329 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11330 (hard_regno_mode_classes): Make static.
11331 (sparc_mode_class): Likewise.
11332 (sparc_hard_regno_mode_ok): New function.
11333 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11334 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11335 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11337 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11338 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11339 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11340 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11341 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11343 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11344 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11345 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11346 (visium_hard_regno_mode_ok): New function.
11347 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11348 instead of HARD_REGNO_MODE_OK.
11349 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11350 (HARD_REGNO_MODE_OK): Delete.
11351 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11352 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11353 (xtensa_option_override): Update accordingly.
11354 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11355 (xtensa_hard_regno_mode_ok): New function.
11356 * system.h (HARD_REGNO_MODE_OK): Poison.
11358 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11359 Alan Hayward <alan.hayward@arm.com>
11360 David Sherwood <david.sherwood@arm.com>
11362 * target.def (hard_regno_call_part_clobbered): New hook.
11363 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11364 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11365 * doc/tm.texi: Regenerate.
11366 * hooks.h (hook_bool_uint_mode_false): Declare.
11367 * hooks.c (hook_bool_uint_mode_false): New function.
11368 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11369 * cselib.c (cselib_process_insn): Use
11370 targetm.hard_regno_call_part_clobbered instead of
11371 HARD_REGNO_CALL_PART_CLOBBERED.
11372 * ira-conflicts.c (ira_build_conflicts): Likewise.
11373 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11374 * lra-constraints.c (need_for_call_save_p): Likewise.
11375 * lra-lives.c: Include target.h.
11376 (check_pseudos_live_through_calls): Use
11377 targetm.hard_regno_call_part_clobbered instead of
11378 HARD_REGNO_CALL_PART_CLOBBERED.
11379 * regcprop.c: Include target.h.
11380 (copyprop_hardreg_forward_1): Use
11381 targetm.hard_regno_call_part_clobbered instead of
11382 HARD_REGNO_CALL_PART_CLOBBERED.
11383 * reginfo.c (choose_hard_reg_mode): Likewise.
11384 * regrename.c (check_new_reg_p): Likewise.
11385 * reload.c (find_equiv_reg): Likewise.
11386 * reload1.c (emit_reload_insns): Likewise.
11387 * sched-deps.c (deps_analyze_insn): Likewise.
11388 * sel-sched.c (init_regs_for_mode): Likewise.
11389 (mark_unavailable_hard_regs): Likewise.
11390 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11391 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11392 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11394 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11395 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11396 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11398 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11400 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11401 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11402 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11404 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11405 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11406 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11408 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11409 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11411 * config/powerpcspe/powerpcspe.c
11412 (rs6000_hard_regno_call_part_clobbered): New function.
11413 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11414 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11415 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11417 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11418 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11419 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11421 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11422 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11423 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11425 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11426 Alan Hayward <alan.hayward@arm.com>
11427 David Sherwood <david.sherwood@arm.com>
11429 * rtl.h (subreg_memory_offset): Declare.
11430 * emit-rtl.c (subreg_memory_offset): New function.
11431 * expmed.c (store_bit_field_1): Use it.
11432 * expr.c (undefined_operand_subword_p): Likewise.
11433 * simplify-rtx.c (simplify_subreg): Likewise.
11435 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11437 PR rtl-optimization/57448
11440 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11441 using atomic_load pattern.
11442 (expand_atomic_store): Likewise.
11444 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11447 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11448 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11449 handling. Use replace_call_with_value with NULL instead of
11450 gsi_replace, unlink_stmt_vdef and release_defs.
11452 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11455 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11457 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11460 * rtl.h (emit_library_call_value_1): Declare.
11461 (emit_library_call): Replace declaration with a series of overloads.
11462 Remove the parameter count argument.
11463 (emit_library_call_value): Likewise.
11464 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11465 with an "rtx_mode_t *".
11466 (emit_library_call_value): Delete.
11467 (emit_library_call): Likewise.
11468 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11469 (asan_emit_allocas_unpoison): Likewise.
11470 * builtins.c (expand_builtin_powi): Likewise.
11471 (expand_asan_emit_allocas_unpoison): Likewise.
11472 * cfgexpand.c (expand_main_function): Likewise.
11473 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11474 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11475 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11476 * config/arm/arm.c (arm_trampoline_init): Likewise.
11477 (arm_call_tls_get_addr): Likewise.
11478 (arm_expand_divmod_libfunc): Likewise.
11479 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11480 (smulsi3_highpart): Likewise.
11481 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11482 (c6x_expand_compare): Likewise.
11483 (c6x_expand_movmem): Likewise.
11484 * config/frv/frv.c (frv_trampoline_init): Likewise.
11485 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11486 (ix86_expand_divmod_libfunc): Likewise.
11487 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11488 (ia64_expand_compare): Likewise.
11489 (ia64_profile_hook): Likewise.
11490 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11491 (nonlocal_goto): Likewise.
11492 (restore_stack_nonlocal): Likewise.
11493 * config/m32r/m32r.c (block_move_call): Likewise.
11494 (m32r_trampoline_init): Likewise.
11495 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11496 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11497 (m68k_call_m68k_read_tp): Likewise.
11498 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11499 (microblaze_expand_divide): Likewise.
11500 * config/mips/mips.h (mips_args): Likewise.
11501 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11502 (MIPS_ICACHE_SYNC): Likewise.
11503 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11504 (nios2_trampoline_init): Likewise.
11505 * config/pa/pa.c (hppa_tls_call): Likewise.
11506 (pa_trampoline_init): Likewise.
11507 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11508 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11509 (expand_strn_compare): Likewise.
11510 (rs6000_generate_compare): Likewise.
11511 (rs6000_expand_float128_convert): Likewise.
11512 (output_profile_hook): Likewise.
11513 (rs6000_trampoline_init): Likewise.
11514 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11515 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11516 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11517 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11518 (rs6000_generate_compare): Likewise.
11519 (rs6000_expand_float128_convert): Likewise.
11520 (output_profile_hook): Likewise.
11521 (rs6000_trampoline_init): Likewise.
11522 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11523 * config/sh/sh.c (sh_trampoline_init): Likewise.
11524 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11525 (sparc_emit_float_lib_cmp): Likewise.
11526 (sparc32_initialize_trampoline): Likewise.
11527 (sparc64_initialize_trampoline): Likewise.
11528 (sparc_profile_hook): Likewise.
11529 * config/spu/spu.c (ea_load_store): Likewise.
11530 * config/spu/spu.md (floatunssidf2): Likewise.
11531 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11532 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11533 * config/visium/visium.c (expand_block_move_4): Likewise.
11534 (expand_block_move_2): Likewise.
11535 (expand_block_move_1): Likewise.
11536 (expand_block_set_4): Likewise.
11537 (expand_block_set_2): Likewise.
11538 (expand_block_set_1): Likewise.
11539 (visium_trampoline_init): Likewise.
11540 (visium_profile_hook): Likewise.
11541 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11542 (xtensa_setup_frame_addresses): Likewise.
11543 (xtensa_trampoline_init): Likewise.
11544 * except.c (sjlj_emit_function_enter): Likewise.
11545 (sjlj_emit_function_exit): Likewise.
11546 * explow.c (allocate_dynamic_stack_space): Likewise.
11547 (probe_stack_range): Likewise.
11548 * expr.c (convert_mode_scalar): Likewise.
11549 * optabs.c (expand_binop): Likewise.
11550 (expand_twoval_binop_libfunc): Likewise.
11551 (expand_unop): Likewise.
11552 (prepare_cmp_insn): Likewise.
11553 (prepare_float_lib_cmp): Likewise.
11554 (expand_float): Likewise.
11555 (expand_fix): Likewise.
11556 (expand_fixed_convert): Likewise.
11557 (maybe_emit_sync_lock_test_and_set): Likewise.
11558 (expand_atomic_compare_and_swap): Likewise.
11559 (expand_mem_thread_fence): Likewise.
11560 (expand_atomic_fetch_op): Likewise.
11562 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11564 * doc/generic.texi (OpenACC): Adjust URL.
11565 * doc/invoke.texi (C Dialect Options): Ditto.
11567 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11569 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11570 predicate for operand 1. Add (m,<S>) constraint.
11571 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11572 Prevent memory operand 1 with register operand 2.
11574 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11576 PR rtl-optimization/82024
11577 * combine.c (try_combine): If the combination result is a PARALLEL,
11578 and we only need to retain the SET in there that would be placed
11579 at I2, check that we can place that at I3 instead, before doing so.
11581 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11584 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11586 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11587 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11590 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11591 Jeff Law <law@redhat.com>
11593 * varasm.c (bss_initializer_p): Do not put constants into .bss
11594 (categorize_decl_for_section): Handle bss_initializer_p returning
11595 false when DECL_INITIAL is NULL.
11597 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11600 * config/s390/s390.c (s390_can_inline_p): New function.
11602 2017-09-01 Jeff Law <law@redhat.com>
11604 PR tree-optimization/82052
11605 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11606 Always initialize the returned slot after a hash table miss
11607 when INSERT is true.
11609 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11611 * config/s390/s390.md (mem_signal_fence): Remove.
11612 * doc/md.texi (mem_signal_fence): Remove.
11613 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11615 * target-insns.def (mem_signal_fence): Remove.
11617 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11620 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11623 * asan.c (create_odr_indicator): Strip name encoding from assembler
11624 name before appending it after __odr_asan_.
11626 2017-09-01 Martin Liska <mliska@suse.cz>
11628 PR tree-optimization/82059
11629 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11630 frequency only when an edge is redirected.
11632 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11634 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11635 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11636 (arc_conditional_register_usage): Remove ARC600 lp_count
11638 (arc_file_start): Emit Tag_ARC_CPU_variation.
11639 (arc_can_use_doloop_p): New conditions to use ZOLs.
11640 (hwloop_fail): New function.
11641 (hwloop_optimize): Likewise.
11642 (hwloop_pattern_reg): Likewise.
11643 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11644 (arc_reorg_loops): New function, calls reorg_loops.
11645 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11646 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11648 (arc_loop_hazard): Remove function, functionality moved into
11650 (arc_hazard): Remove arc_loop_hazard call.
11651 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11652 into hwloop_optimize.
11653 (arc_label_align): Remove ZOL handling.
11654 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11655 * config/arc/arc.md (doloop_begin): Remove pattern.
11656 (doloop_begin_i): Likewise.
11657 (doloop_end_i): Likewise.
11658 (doloop_fallback): Likewise.
11659 (doloop_fallback_m): Likewise.
11660 (doloop_end): Reimplement expand.
11661 (arc_lp): New pattern for LP instruction.
11662 (loop_end): New pattern.
11663 (loop_fail): Likewise.
11664 (decrement_and_branch_until_zero): Likewise.
11665 * config/arc/arc.opt (mlpc-width): New option.
11666 * doc/invoke.texi (mlpc-width): Document option.
11668 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11670 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11671 (arc_ccfsm_advance): Fix checking for delay slots.
11672 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11674 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11676 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11678 (movhi_insn): Update store instruction constraint which are saving
11679 6-bit short immediates.
11680 (movsi_insn): Consider also short scaled load operations.
11681 (zero_extendhisi2_i): Use Usd constraint instead of T.
11682 (extendhisi2_i): Add q constraint.
11683 (arc_clzsi2): Add type and length attributes.
11684 (arc_ctzsi2): Likewise.
11685 * config/arc/constraints.md (Usc): Update constraint, the
11686 assembler can parse two relocations for a single instruction.
11688 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11690 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11691 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11693 2017-08-31 Olivier Hainque <hainque@adacore.com>
11695 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11696 match as powerpc-wrs-vxworks*.
11698 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11700 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11701 register constraint for by-element operand.
11702 (aarch64_mls_elt_merge<mode>): Likewise.
11704 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11706 * config/arc/arc.c (arc_can_follow_jump): Check for short
11709 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11711 * config.gcc: Use g.opt for arc.
11712 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11713 functionality moved to ...
11714 (legitimate_scaled_address_p): New function, ...here.
11715 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11716 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11717 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11719 (arc_override_options): Handle G option.
11720 (arc_output_pic_addr_const): Correct function definition.
11721 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11722 (arc_decl_anon_ns_mem_p): Delete.
11723 (arc_in_small_data_p): Overhaul this function to take into
11724 consideration the value given via G option.
11725 (arc_rewrite_small_data_1): Renamed and corrected old
11726 arc_rewrite_small_data function.
11727 (arc_rewrite_small_data): New function.
11728 (small_data_pattern): Don't use pic_offset_table_rtx.
11729 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11730 * config/arc/simdext.md (movmisalignv2hi): Use
11731 prepare_move_operands function.
11733 (movmisalign*): Likewise.
11734 * doc/invoke.texi (ARC options): Document -G option.
11736 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11738 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11740 * config/arc/arc.c (arc_print_operand): Output scalled address for
11741 sdata whenever is possible.
11742 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11743 load/stores are available.
11744 (compact_sda_memory_operand): Check for the alignment required by
11745 code density instructions.
11746 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11748 * config/arc/constraints.md (Usd): Update constraint.
11749 (Us0): New constraint.
11750 (Usc): Update constraint.
11752 2017-08-31 Richard Biener <rguenther@suse.de>
11754 PR middle-end/82054
11755 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11756 function only once.
11758 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11760 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11761 Resize type_signature.
11763 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11764 Alan Hayward <alan.hayward@arm.com>
11765 David Sherwood <david.sherwood@arm.com>
11767 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11768 subregs whose inner modes can be stored in GPRs.
11769 (aarch64_classify_index): Likewise.
11771 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11772 Alan Hayward <alan.hayward@arm.com>
11773 David Sherwood <david.sherwood@arm.com>
11775 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11776 (V_INT_EQUIV): ...this.
11777 (v_cmp_result): Rename to...
11778 (v_int_equiv): ...this.
11779 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11780 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11781 (copysign<mode>3): Likewise.
11782 (aarch64_simd_bsl<mode>_internal): Likewise.
11783 (aarch64_simd_bsl<mode>): Likewise.
11784 (vec_cmp<mode><mode>): Likewise.
11785 (vcond<mode><mode>): Likewise.
11786 (vcond<v_cmp_mixed><mode>): Likewise.
11787 (vcondu<mode><v_cmp_mixed>): Likewise.
11788 (aarch64_cm<optab><mode>): Likewise.
11789 (aarch64_cmtst<mode>): Likewise.
11790 (aarch64_fac<optab><mode>): Likewise.
11791 (vec_perm_const<mode>): Likewise.
11792 (vcond_mask_<mode><v_cmp_result>): Rename to...
11793 (vcond_mask_<mode><v_int_equiv>): ...this.
11794 (vec_cmp<mode><v_cmp_result>): Rename to...
11795 (vec_cmp<mode><v_int_equiv>): ...this.
11797 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11798 Alan Hayward <alan.hayward@arm.com>
11799 David Sherwood <david.sherwood@arm.com>
11801 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11803 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11804 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11805 (UNSPEC_LD4_DREG): New unspecs.
11806 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11807 (aarch64_ld2<mode>_dreg_be): Replace with...
11808 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11810 (aarch64_ld3<mode>_dreg_le)
11811 (aarch64_ld3<mode>_dreg_be): Replace with...
11812 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11814 (aarch64_ld4<mode>_dreg_le)
11815 (aarch64_ld4<mode>_dreg_be): Replace with...
11816 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11819 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11821 PR tree-optimization/81987
11822 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11823 insert an initializer in a location not dominated by the stride
11826 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11828 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11829 on the entire header of the finally block in the fallthru case.
11831 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11833 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11835 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11837 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11838 rs6000_emit_move_from_cr and call renamed function.
11839 (rs6000_emit_prologue): Call renamed functions.
11840 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11841 movesi_from_cr, remove volatile CRs.
11843 2017-08-30 Jon Beniston <jon@beniston.com>
11844 Richard Biener <rguenther@suse.de>
11846 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11847 of VECTOR_MODE_P check.
11848 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11849 element vector types.
11851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11853 * df.h (df_read_modify_subreg_p): Remove in favor of...
11854 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11855 const_rtx instead of an rtx.
11856 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11857 * df-problems.c (df_word_lr_mark_ref): Likewise.
11858 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11859 (mark_pseudo_reg_dead): Likewise.
11860 (mark_ref_dead): Likewise.
11861 * reginfo.c (init_subregs_of_mode): Likewise.
11862 * sched-deps.c (sched_analyze_1): Likewise.
11863 * df-scan.c (df_def_record_1): Likewise.
11864 (df_uses_record): Likewise.
11865 (df_read_modify_subreg_p): Remove in favor of...
11866 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11867 const_rtx instead of an rtx.
11869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11870 Alan Hayward <alan.hayward@arm.com>
11871 David Sherwood <david.sherwood@arm.com>
11873 * rtl.h (partial_subreg_p): New function.
11874 * caller-save.c (save_call_clobbered_regs): Use it.
11875 * calls.c (expand_call): Likewise.
11876 * combine.c (combinable_i3pat): Likewise.
11877 (simplify_set): Likewise.
11878 (make_extraction): Likewise.
11879 (make_compound_operation_int): Likewise.
11880 (gen_lowpart_or_truncate): Likewise.
11881 (force_to_mode): Likewise.
11882 (make_field_assignment): Likewise.
11883 (reg_truncated_to_mode): Likewise.
11884 (record_truncated_value): Likewise.
11885 (move_deaths): Likewise.
11886 * cse.c (record_jump_cond): Likewise.
11887 (cse_insn): Likewise.
11888 * cselib.c (cselib_lookup_1): Likewise.
11889 * expmed.c (extract_bit_field_using_extv): Likewise.
11890 * function.c (assign_parm_setup_reg): Likewise.
11891 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11892 * ira-build.c (create_insn_allocnos): Likewise.
11893 * lra-coalesce.c (merge_pseudos): Likewise.
11894 * lra-constraints.c (match_reload): Likewise.
11895 (simplify_operand_subreg): Likewise.
11896 (curr_insn_transform): Likewise.
11897 * lra-lives.c (process_bb_lives): Likewise.
11898 * lra.c (new_insn_reg): Likewise.
11899 (lra_substitute_pseudo): Likewise.
11900 * regcprop.c (mode_change_ok): Likewise.
11901 (maybe_mode_change): Likewise.
11902 (copyprop_hardreg_forward_1): Likewise.
11903 * reload.c (push_reload): Likewise.
11904 (find_reloads): Likewise.
11905 (find_reloads_subreg_address): Likewise.
11906 * reload1.c (alter_reg): Likewise.
11907 (eliminate_regs_1): Likewise.
11908 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11910 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11912 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11913 back to if statements, including unpack.
11915 2017-08-30 Martin Liska <mliska@suse.cz>
11917 PR inline-asm/82001
11918 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11920 (func_checker::compare_asm_inputs_outputs): ... this function.
11921 (func_checker::compare_gimple_asm): Use the function to compare
11922 also ASM constrains.
11923 * ipa-icf-gimple.h: Rename the function.
11925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11926 Alan Hayward <alan.hayward@arm.com>
11927 David Sherwood <david.sherwood@arm.com>
11929 * coretypes.h (complex_mode): New type.
11930 * gdbhooks.py (build_pretty_printer): Handle it.
11931 * machmode.h (complex_mode): New class.
11932 (complex_mode::includes_p): New function.
11933 (is_complex_int_mode): Likewise.
11934 (is_complex_float_mode): Likewise.
11935 * genmodes.c (get_mode_class): Handle complex mode classes.
11936 * function.c (expand_function_end): Use is_complex_int_mode.
11938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11939 Alan Hayward <alan.hayward@arm.com>
11940 David Sherwood <david.sherwood@arm.com>
11942 * coretypes.h (scalar_mode_pod): New typedef.
11943 * gdbhooks.py (build_pretty_printer): Handle it.
11944 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11945 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11946 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11947 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11948 as_a <scalar_mode>.
11950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11951 Alan Hayward <alan.hayward@arm.com>
11952 David Sherwood <david.sherwood@arm.com>
11954 * machmode.h (mode_for_vector): Take a scalar_mode instead
11956 * stor-layout.c (mode_for_vector): Likewise.
11957 * explow.c (promote_mode): Use as_a <scalar_mode>.
11958 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11961 Alan Hayward <alan.hayward@arm.com>
11962 David Sherwood <david.sherwood@arm.com>
11964 * target.def (preferred_simd_mode): Take a scalar_mode
11965 instead of a machine_mode.
11966 * targhooks.h (default_preferred_simd_mode): Likewise.
11967 * targhooks.c (default_preferred_simd_mode): Likewise.
11968 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11969 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11970 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11971 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11972 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11973 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11974 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11975 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11977 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11978 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11979 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11980 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11981 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11982 * doc/tm.texi: Regenerate.
11983 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11987 Alan Hayward <alan.hayward@arm.com>
11988 David Sherwood <david.sherwood@arm.com>
11990 * target.def (scalar_mode_supported_p): Take a scalar_mode
11991 instead of a machine_mode.
11992 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11993 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11994 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11995 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11996 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11997 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11998 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11999 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
12000 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
12001 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
12002 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
12003 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
12004 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
12005 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
12007 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
12008 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
12009 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12010 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12011 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
12013 * doc/tm.texi: Regenerate.
12015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12016 Alan Hayward <alan.hayward@arm.com>
12017 David Sherwood <david.sherwood@arm.com>
12019 * coretypes.h (opt_scalar_mode): New typedef.
12020 * gdbhooks.py (build_pretty_printers): Handle it.
12021 * machmode.h (mode_iterator::get_2xwider): Add overload for
12023 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
12025 * expr.c (convert_mode_scalar): Likewise.
12026 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12027 * optabs.c (expand_float): Likewise.
12028 (expand_fix): Likewise.
12029 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12032 Alan Hayward <alan.hayward@arm.com>
12033 David Sherwood <david.sherwood@arm.com>
12035 * optabs.c (expand_float): Explicitly check for scalars before
12036 using a branching expansion.
12037 (expand_fix): Likewise.
12039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12040 Alan Hayward <alan.hayward@arm.com>
12041 David Sherwood <david.sherwood@arm.com>
12043 * expr.c (convert_mode): Split scalar handling out into...
12044 (convert_mode_scalar): ...this new function. Treat the modes
12047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12048 Alan Hayward <alan.hayward@arm.com>
12049 David Sherwood <david.sherwood@arm.com>
12051 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12053 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12056 Alan Hayward <alan.hayward@arm.com>
12057 David Sherwood <david.sherwood@arm.com>
12059 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12060 rather than a machine_mode.
12061 (fixed_from_string): Likewise.
12062 (fixed_convert): Likewise.
12063 (fixed_convert_from_int): Likewise.
12064 (fixed_convert_from_real): Likewise.
12065 (real_convert_from_fixed): Likewise.
12066 * fixed-value.c (fixed_from_double_int): Likewise.
12067 (fixed_from_string): Likewise.
12068 (fixed_convert): Likewise.
12069 (fixed_convert_from_int): Likewise.
12070 (fixed_convert_from_real): Likewise.
12071 (real_convert_from_fixed): Likewise.
12072 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12075 Alan Hayward <alan.hayward@arm.com>
12076 David Sherwood <david.sherwood@arm.com>
12078 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12079 of separate mode class checks. Do not allow vector modes here.
12080 (immed_wide_int_const): Use as_a <scalar_mode>.
12081 * explow.c (trunc_int_for_mode): Likewise.
12082 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12083 (wi::shwi): Likewise.
12084 (wi::min_value): Likewise.
12085 (wi::max_value): Likewise.
12086 * dwarf2out.c (loc_descriptor): Likewise.
12087 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12088 for CONST_WIDE_INT.
12090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12091 Alan Hayward <alan.hayward@arm.com>
12092 David Sherwood <david.sherwood@arm.com>
12094 * tree.h (SCALAR_TYPE_MODE): New macro.
12095 * expr.c (expand_expr_addr_expr_1): Use it.
12096 (expand_expr_real_2): Likewise.
12097 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12098 (fold_convert_const_fixed_from_int): Likewise.
12099 (fold_convert_const_fixed_from_real): Likewise.
12100 (native_encode_fixed): Likewise
12101 (native_encode_complex): Likewise
12102 (native_encode_vector): Likewise.
12103 (native_interpret_fixed): Likewise.
12104 (native_interpret_real): Likewise.
12105 (native_interpret_complex): Likewise.
12106 (native_interpret_vector): Likewise.
12107 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12108 (simd_clone_adjust_argument_types): Likewise.
12109 (simd_clone_init_simd_arrays): Likewise.
12110 (simd_clone_adjust): Likewise.
12111 * stor-layout.c (layout_type): Likewise.
12112 * tree.c (build_minus_one_cst): Likewise.
12113 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12114 * tree-inline.c (estimate_move_cost): Likewise.
12115 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12116 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12117 (vectorizable_reduction): Likewise.
12118 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12119 (vect_recog_mixed_size_cond_pattern): Likewise.
12120 (check_bool_pattern): Likewise.
12121 (adjust_bool_pattern): Likewise.
12122 (search_type_for_mask_1): Likewise.
12123 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12124 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12125 (vectorizable_load): Likewise.
12126 (vectorizable_store): Likewise.
12127 * ubsan.c (ubsan_encode_value): Likewise.
12128 * varasm.c (output_constant): Likewise.
12130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12131 Alan Hayward <alan.hayward@arm.com>
12132 David Sherwood <david.sherwood@arm.com>
12134 * coretypes.h (scalar_mode): New class.
12135 * machmode.h (scalar_mode): Likewise.
12136 (scalar_mode::includes_p): New function.
12137 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12138 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12139 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12140 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12141 * expmed.c (store_bit_field_1): Likewise.
12142 (extract_bit_field_1): Likewise.
12143 * expr.c (write_complex_part): Likewise.
12144 (read_complex_part): Likewise.
12145 (emit_move_complex_push): Likewise.
12146 (expand_expr_real_2): Likewise.
12147 * function.c (assign_parm_setup_reg): Likewise.
12148 (assign_parms_unsplit_complex): Likewise.
12149 * optabs.c (expand_binop): Likewise.
12150 * rtlanal.c (subreg_get_info): Likewise.
12151 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12152 * varasm.c (output_constant_pool_2): Likewise.
12154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12155 Alan Hayward <alan.hayward@arm.com>
12156 David Sherwood <david.sherwood@arm.com>
12158 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12160 (expmed_mult_highpart_optab): Likewise.
12161 (expmed_mult_highpart): Likewise.
12163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12164 Alan Hayward <alan.hayward@arm.com>
12165 David Sherwood <david.sherwood@arm.com>
12167 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12168 instead of a machine_mode.
12169 (builtin_memset_read_str): Likewise.
12170 * builtins.c (c_readstr): Likewise.
12171 (builtin_memcpy_read_str): Likewise.
12172 (builtin_strncpy_read_str): Likewise.
12173 (builtin_memset_read_str): Likewise.
12174 (builtin_memset_gen_str): Likewise.
12175 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12176 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12177 instead of a machine_mode.
12178 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12180 (make_extraction): Likewise.
12181 (try_widen_shift_mode): Take and return scalar_int_modes instead
12183 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12184 a scalar_int_mode instead of a machine_mode.
12185 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12186 (avr_addr_space_pointer_mode): Likewise.
12187 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12188 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12189 (msp430_unwind_word_mode): Likewise.
12190 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12191 (spu_addr_space_pointer_mode): Likewise.
12192 (spu_addr_space_address_mode): Likewise.
12193 (spu_libgcc_cmp_return_mode): Likewise.
12194 (spu_libgcc_shift_count_mode): Likewise.
12195 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12196 (rl78_addr_space_pointer_mode): Likewise.
12197 (fl78_unwind_word_mode): Likewise.
12198 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12200 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12201 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12202 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12203 (mips_valid_pointer_mode): Likewise.
12204 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12205 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12206 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12208 (ft32_addr_space_address_mode): Likewise.
12209 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12210 scalar_int_mode instead of a machine_mode.
12211 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12213 (m32c_addr_space_address_mode): Likewise.
12214 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12215 (rs6000_eh_return_filter_mode): Likewise.
12216 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12217 (rs6000_eh_return_filter_mode): Likewise.
12218 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12219 (s390_libgcc_shift_count_mode): Likewise.
12220 (s390_unwind_word_mode): Likewise.
12221 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12223 * target.def (mode_rep_extended): Likewise.
12224 (valid_pointer_mode): Likewise.
12225 (addr_space.valid_pointer_mode): Likewise.
12226 (eh_return_filter_mode): Return a scalar_int_mode rather than
12228 (libgcc_cmp_return_mode): Likewise.
12229 (libgcc_shift_count_mode): Likewise.
12230 (unwind_word_mode): Likewise.
12231 (addr_space.pointer_mode): Likewise.
12232 (addr_space.address_mode): Likewise.
12233 * doc/tm.texi: Regenerate.
12234 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12236 (do_jump): Use scalar_int_mode for local variables.
12237 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12238 rather than a machine_mode.
12239 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12240 (scompare_loc_descriptor_wide): Likewise.
12241 (scompare_loc_descriptor_narrow): Likewise.
12242 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12244 * except.c (sjlj_emit_dispatch_table): Likewise.
12245 (expand_builtin_eh_copy_values): Likewise.
12246 * explow.c (convert_memory_address_addr_space_1): Likewise.
12247 Take a scalar_int_mode rather than a machine_mode.
12248 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12249 than a machine_mode.
12250 (memory_address_addr_space): Use scalar_int_mode for local variables.
12251 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12252 rather than a machine_mode.
12253 * expmed.c (mask_rtx): Likewise.
12254 (init_expmed_one_conv): Likewise.
12255 (expand_mult_highpart_adjust): Likewise.
12256 (extract_high_half): Likewise.
12257 (expmed_mult_highpart_optab): Likewise.
12258 (expmed_mult_highpart): Likewise.
12259 (expand_smod_pow2): Likewise.
12260 (expand_sdiv_pow2): Likewise.
12261 (emit_store_flag_int): Likewise.
12262 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12264 (extract_low_bits): Likewise.
12265 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12267 * expr.c (pieces_addr::adjust): Likewise.
12268 (can_store_by_pieces): Likewise.
12269 (store_by_pieces): Likewise.
12270 (clear_by_pieces_1): Likewise.
12271 (expand_expr_addr_expr_1): Likewise.
12272 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12273 (expand_expr_real_1): Likewise.
12274 (try_casesi): Likewise.
12275 * final.c (shorten_branches): Likewise.
12276 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12277 type of "mode" to machine_mode.
12278 * internal-fn.c (expand_arith_overflow_result_store): Take a
12279 scalar_int_mode rather than a machine_mode.
12280 (expand_mul_overflow): Use scalar_int_mode for local variables.
12281 * loop-doloop.c (doloop_modify): Likewise.
12282 (doloop_optimize): Likewise.
12283 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12284 than a machine_mode.
12285 (expand_doubleword_shift_condmove): Likewise.
12286 (expand_doubleword_shift): Likewise.
12287 (expand_doubleword_clz): Likewise.
12288 (expand_doubleword_popcount): Likewise.
12289 (expand_doubleword_parity): Likewise.
12290 (expand_absneg_bit): Use scalar_int_mode for local variables.
12291 (prepare_float_lib_cmp): Likewise.
12292 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12293 rather than a machine_mode.
12294 (convert_memory_address_addr_space): Likewise.
12295 (get_mode_bounds): Likewise.
12296 (get_address_mode): Return a scalar_int_mode rather than a
12298 * rtlanal.c (get_address_mode): Likewise.
12299 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12300 than a machine_mode.
12301 * targhooks.c (default_mode_rep_extended): Likewise.
12302 (default_valid_pointer_mode): Likewise.
12303 (default_addr_space_valid_pointer_mode): Likewise.
12304 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12305 than a machine_mode.
12306 (default_libgcc_cmp_return_mode): Likewise.
12307 (default_libgcc_shift_count_mode): Likewise.
12308 (default_unwind_word_mode): Likewise.
12309 (default_addr_space_pointer_mode): Likewise.
12310 (default_addr_space_address_mode): Likewise.
12311 * targhooks.h (default_eh_return_filter_mode): Likewise.
12312 (default_libgcc_cmp_return_mode): Likewise.
12313 (default_libgcc_shift_count_mode): Likewise.
12314 (default_unwind_word_mode): Likewise.
12315 (default_addr_space_pointer_mode): Likewise.
12316 (default_addr_space_address_mode): Likewise.
12317 (default_mode_rep_extended): Take a scalar_int_mode rather than
12319 (default_valid_pointer_mode): Likewise.
12320 (default_addr_space_valid_pointer_mode): Likewise.
12321 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12323 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12324 rather than a machine_mode.
12325 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12326 for local variables.
12327 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12328 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12329 than a machine_mode.
12331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12332 Alan Hayward <alan.hayward@arm.com>
12333 David Sherwood <david.sherwood@arm.com>
12335 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12336 the mode argument to scalar_int_mode.
12337 (do_jump_by_parts_zero_rtx): Likewise.
12338 (do_jump_by_parts_equality_rtx): Likewise.
12339 (do_jump_by_parts_greater): Take a mode argument.
12340 (do_jump_by_parts_equality): Likewise.
12341 (do_jump_1): Update calls accordingly.
12343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12344 Alan Hayward <alan.hayward@arm.com>
12345 David Sherwood <david.sherwood@arm.com>
12347 * is-a.h (safe_dyn_cast): New function.
12348 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12349 (jump_table_for_label): Likewise.
12350 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12351 instead of an rtx_insn *.
12352 (shorten_branches): Use dyn_cast instead of LABEL_P and
12353 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12354 rtx_jump_table_data::get_data_mode.
12355 (final_scan_insn): Likewise.
12357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12358 Alan Hayward <alan.hayward@arm.com>
12359 David Sherwood <david.sherwood@arm.com>
12361 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12362 trying to combine a full-register integer set with a subreg
12365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12366 Alan Hayward <alan.hayward@arm.com>
12367 David Sherwood <david.sherwood@arm.com>
12369 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12370 that is always either address_mode or pointer_mode.
12372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12373 Alan Hayward <alan.hayward@arm.com>
12374 David Sherwood <david.sherwood@arm.com>
12376 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12377 when the two are known to be equal.
12379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12380 Alan Hayward <alan.hayward@arm.com>
12381 David Sherwood <david.sherwood@arm.com>
12383 * simplify-rtx.c (simplify_const_unary_operation): Use
12384 is_a <scalar_int_mode> instead of checking for a nonzero
12385 precision. Forcibly convert op_mode to a scalar_int_mode
12386 in that case. More clearly differentiate the operand and
12387 result modes and use the former when deciding what the value
12388 of a count-bits operation should be. Use is_int_mode instead
12389 of checking for a MODE_INT. Remove redundant check for whether
12390 this mode has a zero precision.
12392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12393 Alan Hayward <alan.hayward@arm.com>
12394 David Sherwood <david.sherwood@arm.com>
12396 * optabs.c (widen_leading): Change the type of the mode argument
12397 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12398 (widen_bswap): Likewise.
12399 (expand_parity): Likewise.
12400 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12401 (expand_ffs): Likewise.
12402 (epand_unop): Check for scalar integer modes before calling the
12405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12406 Alan Hayward <alan.hayward@arm.com>
12407 David Sherwood <david.sherwood@arm.com>
12409 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12411 (expand_expr_real_1): Update call accordingly.
12413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12414 Alan Hayward <alan.hayward@arm.com>
12415 David Sherwood <david.sherwood@arm.com>
12417 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12418 value_mode arguments. Use scalar_int_mode internally.
12419 (store_bit_field_1): Rename the new integer mode from imode
12420 to op0_mode and use it instead of GET_MODE (op0). Update calls
12421 to store_split_bit_field, store_bit_field_using_insv and
12422 store_fixed_bit_field.
12423 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12424 Use scalar_int_mode internally. Use a bit count rather than a mode
12425 when calculating the largest bit size for get_best_mode.
12426 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12427 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12428 Remove assertion that OP0 has a scalar integer mode.
12429 (store_split_bit_field): Add op0_mode and value_mode arguments.
12430 Update calls to extract_fixed_bit_field.
12431 (extract_bit_field_using_extv): Add an op0_mode argument.
12432 Use scalar_int_mode internally.
12433 (extract_bit_field_1): Rename the new integer mode from imode to
12434 op0_mode and use it instead of GET_MODE (op0). Update calls to
12435 extract_split_bit_field, extract_bit_field_using_extv and
12436 extract_fixed_bit_field.
12437 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12438 to extract_split_bit_field and extract_fixed_bit_field_1.
12439 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12440 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12441 on the target mode.
12442 (extract_split_bit_field): Add an op0_mode argument. Update call
12443 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12450 * explow.c (hard_function_value): Likewise.
12451 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12452 convert_to_mode call outside the loop.
12453 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12454 for the mode iterator. Require the mode specified by max_pieces
12456 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12458 (copy_blkmode_to_reg): Likewise.
12459 (set_storage_via_setmem): Likewise.
12460 * optabs.c (prepare_cmp_insn): Likewise.
12461 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12462 * stor-layout.c (finish_bitfield_representative): Likewise.
12464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12465 Alan Hayward <alan.hayward@arm.com>
12466 David Sherwood <david.sherwood@arm.com>
12468 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12469 * expr.c (convert_move): Use them.
12470 (convert_modes): Likewise.
12471 (store_expr_with_bounds): Likewise.
12473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12474 Alan Hayward <alan.hayward@arm.com>
12475 David Sherwood <david.sherwood@arm.com>
12477 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12478 parameter for the mode of "x". Remove the "known_x", "known_mode"
12479 and "known_ret" arguments. Change the type of the mode argument
12480 to scalar_int_mode.
12481 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12482 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12483 (reg_num_sign_bit_copies_for_combine): Likewise.
12484 * rtlanal.c (nonzero_bits1): Likewise.
12485 (num_sign_bit_copies1): Likewise.
12486 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12487 (reg_num_sign_bit_copies_general): Likewise.
12488 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12489 (reg_nonzero_bits_general): Likewise.
12491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12492 Alan Hayward <alan.hayward@arm.com>
12493 David Sherwood <david.sherwood@arm.com>
12495 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12496 than in subroutines. Return 1 for non-integer modes.
12497 (cached_num_sign_bit_copies): Change the type of the mode parameter
12498 to scalar_int_mode.
12499 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12500 classes. Handle CONST_INT_P first and then check whether X also
12501 has a scalar integer mode. Check the same thing for inner registers
12502 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12509 in subroutines. Return the mode mask for non-integer modes.
12510 (cached_nonzero_bits): Change the type of the mode parameter
12511 to scalar_int_mode.
12512 (nonzero_bits1): Likewise. Remove early exit for other mode
12513 classes. Handle CONST_INT_P first and then check whether X
12514 also has a scalar integer mode.
12516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12517 Alan Hayward <alan.hayward@arm.com>
12518 David Sherwood <david.sherwood@arm.com>
12520 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12521 Return a scalar_int_mode and assert that the size is greater than
12523 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12524 (op_by_pieces_d::op_by_pieces_d): Likewise.
12525 (op_by_pieces_d::run): Likewise.
12526 (can_store_by_pieces): Likewise.
12528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12529 Alan Hayward <alan.hayward@arm.com>
12530 David Sherwood <david.sherwood@arm.com>
12532 * combine.c (extract_left_shift): Add a mode argument and update
12534 (make_compound_operation_int): Change the type of the mode parameter
12535 to scalar_int_mode and update the call to extract_left_shift.
12537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12538 Alan Hayward <alan.hayward@arm.com>
12539 David Sherwood <david.sherwood@arm.com>
12541 * combine.c (simplify_and_const_int): Change the type of the mode
12542 parameter to scalar_int_mode.
12543 (simplify_and_const_int_1): Likewise. Update recursive call.
12545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12546 Alan Hayward <alan.hayward@arm.com>
12547 David Sherwood <david.sherwood@arm.com>
12549 * combine.c (simplify_compare_const): Check that the mode is a
12550 scalar_int_mode (rather than VOIDmode) before testing its
12552 (simplify_comparison): Move COMPARISON_P handling out of the
12553 loop and restrict the latter part of the loop to scalar_int_modes.
12554 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12555 and when considering SUBREG_REGs. Use is_int_mode instead of
12556 checking GET_MODE_CLASS against MODE_INT.
12558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12559 Alan Hayward <alan.hayward@arm.com>
12560 David Sherwood <david.sherwood@arm.com>
12562 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12563 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12564 shift_unit_mode and for modes involved in scalar shifts.
12566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12567 Alan Hayward <alan.hayward@arm.com>
12568 David Sherwood <david.sherwood@arm.com>
12570 * combine.c (force_int_to_mode): New function, split out from...
12571 (force_to_mode): ...here. Keep xmode up-to-date and use it
12572 instead of GET_MODE (x).
12574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12575 Alan Hayward <alan.hayward@arm.com>
12576 David Sherwood <david.sherwood@arm.com>
12578 * optabs-query.h (extraction_insn::struct_mode): Change type to
12579 opt_scalar_int_mode and update comment.
12580 (extraction_insn::field_mode): Change type to scalar_int_mode.
12581 (extraction_insn::pos_mode): Likewise.
12582 * combine.c (make_extraction): Update accordingly.
12583 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12584 (get_optab_extraction_insn): Likewise.
12585 * recog.c (simplify_while_replacing): Likewise.
12586 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12587 parameter to opt_scalar_int_mode.
12589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12590 Alan Hayward <alan.hayward@arm.com>
12591 David Sherwood <david.sherwood@arm.com>
12593 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12594 to a scalar_int_mode instead of a machine_mode.
12595 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12596 (get_best_mode): Return a boolean and use a pointer argument to store
12597 the selected mode. Replace the limit mode parameter with a bit limit.
12598 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12599 for the values returned by bit_field_mode_iterator::next_mode.
12600 (store_bit_field): Update call to get_best_mode.
12601 (store_fixed_bit_field): Likewise.
12602 (extract_fixed_bit_field): Likewise.
12603 * expr.c (optimize_bitfield_assignment_op): Likewise.
12604 * fold-const.c (optimize_bit_field_compare): Likewise.
12605 (fold_truth_andor_1): Likewise.
12606 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12607 Update for new type of m_mode.
12608 (get_best_mode): As above.
12610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12611 Alan Hayward <alan.hayward@arm.com>
12612 David Sherwood <david.sherwood@arm.com>
12614 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12615 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12616 (store_bit_field): Check is_a <scalar_int_mode> before calling
12617 strict_volatile_bitfield_p.
12618 (extract_bit_field): Likewise.
12620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12621 Alan Hayward <alan.hayward@arm.com>
12622 David Sherwood <david.sherwood@arm.com>
12624 * target.def (cstore_mode): Return a scalar_int_mode.
12625 * doc/tm.texi: Regenerate.
12626 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12627 * targhooks.h (default_cstore_mode): Likewise.
12628 * targhooks.c (default_cstore_mode): Likewise, using a forced
12630 * expmed.c (emit_cstore): Expect the target of the cstore to be
12633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12634 Alan Hayward <alan.hayward@arm.com>
12635 David Sherwood <david.sherwood@arm.com>
12637 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12639 (niter_desc): Likewise mode.
12640 (iv_analyze): Add a mode parameter.
12642 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12643 and change its type to scalar_int_mode.
12644 * loop-iv.c: Update commentary at head of file.
12645 (iv_constant): Pass the mode paraeter before the rtx it describes
12646 and change its type to scalar_int_mode. Remove VOIDmode handling.
12647 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12648 (iv_extend): Likewise.
12649 (shorten_into_mode): Likewise.
12650 (iv_add): Use scalar_int_mode.
12651 (iv_mult): Likewise.
12652 (iv_shift): Likewise.
12653 (canonicalize_iv_subregs): Likewise.
12654 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12655 it describes and change its mode to scalar_int_mode. Also change
12656 the type of the returned inner_mode to scalar_int_mode.
12657 (get_biv_step): Likewise, turning outer_mode from a pointer
12658 into a direct parameter. Update call to get_biv_step_1.
12659 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12660 iv_constant and get_biv_step.
12661 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12662 and change its type to scalar_int_mode. Don't initialise iv->mode
12663 to VOIDmode and remove later checks for its still being VOIDmode.
12664 Update calls to iv_analyze_op and iv_analyze_expr. Check
12665 is_a <scalar_int_mode> when changing the mode under consideration.
12666 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12667 Update call to iv_analyze_expr.
12668 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12669 inner register is not also a scalar_int_mode. Update call to
12671 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12672 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12673 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12674 separate mode class checks. Update calls to iv_analyze. Remove
12675 fix-up of VOIDmodes after iv_analyze_biv.
12676 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12677 don't have a scalar_int_mode. Update call to biv_p.
12679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12680 Alan Hayward <alan.hayward@arm.com>
12681 David Sherwood <david.sherwood@arm.com>
12683 * cfgexpand.c (convert_debug_memory_address): Use
12684 as_a <scalar_int_mode>.
12685 * combine.c (expand_compound_operation): Likewise.
12686 (make_extraction): Likewise.
12687 (change_zero_ext): Likewise.
12688 (simplify_comparison): Likewise.
12689 * cse.c (cse_insn): Likewise.
12690 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12691 (mem_loc_descriptor): Likewise.
12692 (loc_descriptor): Likewise.
12693 * expmed.c (init_expmed_one_mode): Likewise.
12694 (synth_mult): Likewise.
12695 (emit_store_flag_1): Likewise.
12696 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12697 of a comparison with size.
12698 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12699 (reduce_to_bit_field_precision): Likewise.
12700 * function.c (expand_function_end): Likewise.
12701 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12702 * loop-doloop.c (doloop_modify): Likewise.
12703 * optabs.c (expand_binop): Likewise.
12704 (expand_unop): Likewise.
12705 (expand_copysign_absneg): Likewise.
12706 (prepare_cmp_insn): Likewise.
12707 (maybe_legitimize_operand): Likewise.
12708 * recog.c (const_scalar_int_operand): Likewise.
12709 * rtlanal.c (get_address_mode): Likewise.
12710 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12711 (simplify_cond_clz_ctz): Likewise.
12712 * tree-nested.c (get_nl_goto_field): Likewise.
12713 * tree.c (build_vector_type_for_mode): Likewise.
12714 * var-tracking.c (use_narrower_mode): Likewise.
12716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12717 Alan Hayward <alan.hayward@arm.com>
12718 David Sherwood <david.sherwood@arm.com>
12720 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12721 * builtins.c (expand_builtin_signbit): Use it.
12722 * cfgexpand.c (expand_debug_expr): Likewise.
12723 * dojump.c (do_jump): Likewise.
12724 (do_compare_and_jump): Likewise.
12725 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12726 * expmed.c (make_tree): Likewise.
12727 * expr.c (expand_expr_real_2): Likewise.
12728 (expand_expr_real_1): Likewise.
12729 (try_casesi): Likewise.
12730 * fold-const-call.c (fold_const_call_ss): Likewise.
12731 * fold-const.c (unextend): Likewise.
12732 (extract_muldiv_1): Likewise.
12733 (fold_single_bit_test): Likewise.
12734 (native_encode_int): Likewise.
12735 (native_encode_string): Likewise.
12736 (native_interpret_int): Likewise.
12737 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12738 * internal-fn.c (expand_addsub_overflow): Likewise.
12739 (expand_neg_overflow): Likewise.
12740 (expand_mul_overflow): Likewise.
12741 (expand_arith_overflow): Likewise.
12742 * match.pd: Likewise.
12743 * stor-layout.c (layout_type): Likewise.
12744 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12745 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12746 * tree-ssanames.c (get_range_info): Likewise.
12747 * tree-switch-conversion.c (array_value_type) Likewise.
12748 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12749 (vect_recog_divmod_pattern): Likewise.
12750 (vect_recog_mixed_size_cond_pattern): Likewise.
12751 * tree-vrp.c (extract_range_basic): Likewise.
12752 (simplify_float_conversion_using_ranges): Likewise.
12753 * tree.c (int_fits_type_p): Likewise.
12754 * ubsan.c (instrument_bool_enum_load): Likewise.
12755 * varasm.c (mergeable_string_section): Likewise.
12756 (narrowing_initializer_constant_valid_p): Likewise.
12757 (output_constant): Likewise.
12759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12760 Alan Hayward <alan.hayward@arm.com>
12761 David Sherwood <david.sherwood@arm.com>
12763 * machmode.h (NARROWEST_INT_MODE): New macro.
12764 * expr.c (alignment_for_piecewise_move): Use it instead of
12765 GET_CLASS_NARROWEST_MODE (MODE_INT).
12766 (push_block): Likewise.
12767 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12769 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12772 Alan Hayward <alan.hayward@arm.com>
12773 David Sherwood <david.sherwood@arm.com>
12775 * postreload.c (move2add_valid_value_p): Change the type of the
12776 mode parameter to scalar_int_mode.
12777 (move2add_use_add2_insn): Add a mode parameter and use it instead
12779 (move2add_use_add3_insn): Likewise.
12780 (reload_cse_move2add): Update accordingly.
12782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12783 Alan Hayward <alan.hayward@arm.com>
12784 David Sherwood <david.sherwood@arm.com>
12786 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12788 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12789 * optabs.c (expand_unop): Likewise.
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12796 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12797 (popcount_loc_descriptor): Likewise.
12798 (bswap_loc_descriptor): Likewise.
12799 (rotate_loc_descriptor): Likewise.
12800 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12801 calling the functions above.
12803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12804 Alan Hayward <alan.hayward@arm.com>
12805 David Sherwood <david.sherwood@arm.com>
12807 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12809 (try_combine): Likewise.
12810 (simplify_if_then_else): Likewise.
12811 * cse.c (cse_insn): Likewise.
12812 * dwarf2out.c (mem_loc_descriptor): Likewise.
12813 * emit-rtl.c (gen_lowpart_common): Likewise.
12814 * simplify-rtx.c (simplify_truncation): Likewise.
12815 (simplify_binary_operation_1): Likewise.
12816 (simplify_const_relational_operation): Likewise.
12817 (simplify_ternary_operation): Likewise.
12818 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12821 Alan Hayward <alan.hayward@arm.com>
12822 David Sherwood <david.sherwood@arm.com>
12824 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12825 * reload.c (push_reload): Likewise.
12826 (find_reloads): Likewise.
12828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12829 Alan Hayward <alan.hayward@arm.com>
12830 David Sherwood <david.sherwood@arm.com>
12832 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12833 (make_compound_operation_int): Likewise.
12834 (change_zero_ext): Likewise.
12835 * expr.c (convert_move): Likewise.
12836 (convert_modes): Likewise.
12837 * fwprop.c (forward_propagate_subreg): Likewise.
12838 * loop-iv.c (get_biv_step_1): Likewise.
12839 * optabs.c (widen_operand): Likewise.
12840 * postreload.c (move2add_valid_value_p): Likewise.
12841 * recog.c (simplify_while_replacing): Likewise.
12842 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12843 (simplify_binary_operation_1): Likewise. Remove redundant
12844 mode equality check.
12846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12847 Alan Hayward <alan.hayward@arm.com>
12848 David Sherwood <david.sherwood@arm.com>
12850 * combine.c (combine_simplify_rtx): Add checks for
12851 is_a <scalar_int_mode>.
12852 (simplify_if_then_else): Likewise.
12853 (make_field_assignment): Likewise.
12854 (simplify_comparison): Likewise.
12855 * ifcvt.c (noce_try_bitop): Likewise.
12856 * loop-invariant.c (canonicalize_address_mult): Likewise.
12857 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12860 Alan Hayward <alan.hayward@arm.com>
12861 David Sherwood <david.sherwood@arm.com>
12863 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12864 is_a <scalar_int_mode> instead of != BLKmode.
12866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12867 Alan Hayward <alan.hayward@arm.com>
12868 David Sherwood <david.sherwood@arm.com>
12870 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12871 instead of != VOIDmode.
12872 * combine.c (if_then_else_cond): Likewise.
12873 (change_zero_ext): Likewise.
12874 * dwarf2out.c (mem_loc_descriptor): Likewise.
12875 (loc_descriptor): Likewise.
12876 * rtlanal.c (canonicalize_condition): Likewise.
12877 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12880 Alan Hayward <alan.hayward@arm.com>
12881 David Sherwood <david.sherwood@arm.com>
12883 * simplify-rtx.c (simplify_binary_operation_1): Use
12884 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12887 Alan Hayward <alan.hayward@arm.com>
12888 David Sherwood <david.sherwood@arm.com>
12890 * wide-int.h (int_traits<unsigned char>) New class.
12891 (int_traits<unsigned short>) Likewise.
12892 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12893 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12895 * combine.c (set_nonzero_bits_and_sign_copies): Use
12896 is_a <scalar_int_mode>.
12897 (find_split_point): Likewise.
12898 (combine_simplify_rtx): Likewise.
12899 (simplify_logical): Likewise.
12900 (expand_compound_operation): Likewise.
12901 (expand_field_assignment): Likewise.
12902 (make_compound_operation): Likewise.
12903 (extended_count): Likewise.
12904 (change_zero_ext): Likewise.
12905 (simplify_comparison): Likewise.
12906 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12907 (ucompare_loc_descriptor): Likewise.
12908 (minmax_loc_descriptor): Likewise.
12909 (mem_loc_descriptor): Likewise.
12910 (loc_descriptor): Likewise.
12911 * expmed.c (init_expmed_one_mode): Likewise.
12912 * lra-constraints.c (lra_constraint_offset): Likewise.
12913 * optabs.c (prepare_libcall_arg): Likewise.
12914 * postreload.c (move2add_note_store): Likewise.
12915 * reload.c (operands_match_p): Likewise.
12916 * rtl.h (load_extend_op): Likewise.
12917 * rtlhooks.c (gen_lowpart_general): Likewise.
12918 * simplify-rtx.c (simplify_truncation): Likewise.
12919 (simplify_unary_operation_1): Likewise.
12920 (simplify_binary_operation_1): Likewise.
12921 (simplify_const_binary_operation): Likewise.
12922 (simplify_const_relational_operation): Likewise.
12923 (simplify_subreg): Likewise.
12924 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12925 * var-tracking.c (adjust_mems): Likewise.
12926 (prepare_call_arguments): Likewise.
12928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12929 Alan Hayward <alan.hayward@arm.com>
12930 David Sherwood <david.sherwood@arm.com>
12932 * machmode.h (is_int_mode): New fuction.
12933 * combine.c (find_split_point): Use it.
12934 (combine_simplify_rtx): Likewise.
12935 (simplify_if_then_else): Likewise.
12936 (simplify_set): Likewise.
12937 (simplify_shift_const_1): Likewise.
12938 (simplify_comparison): Likewise.
12939 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12940 * cse.c (notreg_cost): Likewise.
12941 (cse_insn): Likewise.
12942 * cselib.c (cselib_lookup_1): Likewise.
12943 * dojump.c (do_jump_1): Likewise.
12944 (do_compare_rtx_and_jump): Likewise.
12945 * dse.c (get_call_args): Likewise.
12946 * dwarf2out.c (rtl_for_decl_init): Likewise.
12947 (native_encode_initializer): Likewise.
12948 * expmed.c (emit_store_flag_1): Likewise.
12949 (emit_store_flag): Likewise.
12950 * expr.c (convert_modes): Likewise.
12951 (store_field): Likewise.
12952 (expand_expr_real_1): Likewise.
12953 * fold-const.c (fold_read_from_constant_string): Likewise.
12954 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12955 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12956 * optabs.c (expand_binop): Likewise.
12957 (expand_unop): Likewise.
12958 (expand_abs_nojump): Likewise.
12959 (expand_one_cmpl_abs_nojump): Likewise.
12960 * simplify-rtx.c (mode_signbit_p): Likewise.
12961 (val_signbit_p): Likewise.
12962 (val_signbit_known_set_p): Likewise.
12963 (val_signbit_known_clear_p): Likewise.
12964 (simplify_relational_operation_1): Likewise.
12965 * tree.c (vector_type_mode): Likewise.
12967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12968 Alan Hayward <alan.hayward@arm.com>
12969 David Sherwood <david.sherwood@arm.com>
12971 * machmode.h (smallest_mode_for_size): Fix formatting.
12972 (smallest_int_mode_for_size): New function.
12973 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12974 instead of smallest_mode_for_size.
12975 * combine.c (make_extraction): Likewise.
12976 * config/arc/arc.c (arc_expand_movmem): Likewise.
12977 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12978 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12979 * config/s390/s390.c (s390_expand_insv): Likewise.
12980 * config/sparc/sparc.c (assign_int_registers): Likewise.
12981 * config/spu/spu.c (spu_function_value): Likewise.
12982 (spu_function_arg): Likewise.
12983 * coverage.c (get_gcov_type): Likewise.
12984 (get_gcov_unsigned_t): Likewise.
12985 * dse.c (find_shift_sequence): Likewise.
12986 * expmed.c (store_bit_field_1): Likewise.
12987 * expr.c (convert_move): Likewise.
12988 (store_field): Likewise.
12989 * internal-fn.c (expand_arith_overflow): Likewise.
12990 * optabs-query.c (get_best_extraction_insn): Likewise.
12991 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12992 * stor-layout.c (layout_type): Likewise.
12993 (initialize_sizetypes): Likewise.
12994 * targhooks.c (default_get_mask_mode): Likewise.
12995 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12998 Alan Hayward <alan.hayward@arm.com>
12999 David Sherwood <david.sherwood@arm.com>
13001 * machmode.h (opt_mode::else_blk): New function.
13002 (int_mode_for_mode): Declare.
13003 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
13004 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
13006 * cfgexpand.c (expand_debug_expr): Likewise.
13007 * combine.c (gen_lowpart_or_truncate): Likewise.
13008 (gen_lowpart_for_combine): Likewise.
13009 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
13010 * config/avr/avr.c (avr_to_int_mode): Likewise.
13011 (avr_out_plus_1): Likewise.
13012 (avr_out_plus): Likewise.
13013 (avr_out_round): Likewise.
13014 * config/i386/i386.c (ix86_split_to_parts): Likewise.
13015 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13016 (s390_expand_vcond): Likewise.
13017 * config/spu/spu.c (spu_split_immediate): Likewise.
13018 (spu_expand_mov): Likewise.
13019 * dse.c (get_stored_val): Likewise.
13020 * expmed.c (store_bit_field_1): Likewise.
13021 (convert_extracted_bit_field): Use int_mode_for_mode instead of
13023 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
13024 (extract_low_bits): Likewise.
13025 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
13026 handling rather than repeating the check.
13027 (emit_group_store): Likewise.
13028 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
13029 * optabs.c (expand_absneg_bit): Likewise.
13030 (expand_copysign_absneg): Likewise.
13031 (expand_copysign_bit): Likewise.
13032 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
13033 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
13034 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
13035 * var-tracking.c (prepare_call_arguments): Likewise.
13036 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13037 int_mode_for_mode instead of mode_for_size.
13038 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13041 Alan Hayward <alan.hayward@arm.com>
13042 David Sherwood <david.sherwood@arm.com>
13044 * machmode.h (int_mode_for_size): New function.
13045 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13046 instead of mode_for_size.
13047 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13049 * combine.c (expand_field_assignment): Use int_mode_for_size
13050 instead of mode_for_size.
13051 (make_extraction): Likewise.
13052 (simplify_shift_const_1): Likewise.
13053 (simplify_comparison): Likewise.
13054 * dojump.c (do_jump): Likewise.
13055 * dwarf2out.c (mem_loc_descriptor): Likewise.
13056 * emit-rtl.c (init_derived_machine_modes): Likewise.
13057 * expmed.c (flip_storage_order): Likewise.
13058 (convert_extracted_bit_field): Likewise.
13059 * expr.c (copy_blkmode_from_reg): Likewise.
13060 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13061 * internal-fn.c (expand_mul_overflow): Likewise.
13062 * lower-subreg.c (simple_move): Likewise.
13063 * optabs-libfuncs.c (init_optabs): Likewise.
13064 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13065 * tree.c (vector_type_mode): Likewise.
13066 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13067 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13068 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13069 * tree-vect-stmts.c (vectorizable_load): Likewise.
13070 (vectorizable_store): Likewise.
13072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13073 Alan Hayward <alan.hayward@arm.com>
13074 David Sherwood <david.sherwood@arm.com>
13076 * coretypes.h (pod_mode): New type.
13077 (scalar_int_mode_pod): New typedef.
13078 * machmode.h (pod_mode): New class.
13079 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13080 * genmodes.c (emit_mode_int_n): Update accordingly.
13081 * lower-subreg.h (target_lower_subreg): Change type to
13082 scalar_int_mode_pod.
13083 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13084 scalar_int_mode_pod.
13086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13087 Alan Hayward <alan.hayward@arm.com>
13088 David Sherwood <david.sherwood@arm.com>
13090 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13091 machine_mode to scalar_int_mode.
13092 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13093 (rs6000_option_override_internal): Remove cast to int.
13094 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13095 machine_mode to scalar_int_mode.
13096 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13097 (rs6000_option_override_internal): Remove cast to int.
13098 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13099 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13101 * config/s390/s390.c (s390_expand_builtin): Likewise.
13102 * coretypes.h (scalar_int_mode): New type.
13103 (opt_scalar_int_mode): New typedef.
13104 * machmode.h (scalar_int_mode): New class.
13105 (scalar_int_mode::includes_p): New function.
13106 (byte_mode): Change type to scalar_int_mode.
13107 (word_mode): Likewise.
13108 (ptr_mode): Likewise.
13109 * emit-rtl.c (byte_mode): Likewise.
13110 (word_mode): Likewise.
13111 (ptr_mode): Likewise.
13112 (init_derived_machine_modes): Update accordingly.
13113 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13114 and MODE_PARTIAL_INT.
13115 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13116 opt_scalar_int_mode.
13118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13119 Alan Hayward <alan.hayward@arm.com>
13120 David Sherwood <david.sherwood@arm.com>
13122 * target.def (libgcc_floating_mode_supported_p): Take a
13124 * doc/tm.texi: Regenerate.
13125 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13127 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13128 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13132 Alan Hayward <alan.hayward@arm.com>
13133 David Sherwood <david.sherwood@arm.com>
13135 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13136 * doc/tm.texi: Regenerate.
13137 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13138 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13139 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13140 * targhooks.h (default_floatn_mode): Likewise.
13141 * targhooks.c (default_floatn_mode): Likewise.
13142 * tree.c (build_common_tree_nodes): Update accordingly.
13144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13145 Alan Hayward <alan.hayward@arm.com>
13146 David Sherwood <david.sherwood@arm.com>
13148 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13149 (mode_iterator::iterate_p): Likewise.
13150 (mode_iterator::get_wider): Likewise.
13151 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13154 Alan Hayward <alan.hayward@arm.com>
13155 David Sherwood <david.sherwood@arm.com>
13157 * coretypes.h (opt_scalar_float_mode): New typedef.
13158 * machmode.h (float_mode_for_size): New function.
13159 * emit-rtl.c (double_mode): Delete.
13160 (init_emit_once): Use float_mode_for_size.
13161 * stor-layout.c (layout_type): Likewise.
13162 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13165 Alan Hayward <alan.hayward@arm.com>
13166 David Sherwood <david.sherwood@arm.com>
13168 * output.h (assemble_real): Take a scalar_float_mode.
13169 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13170 * config/arm/arm.md (consttable_4): Likewise.
13171 (consttable_8): Likewise.
13172 (consttable_16): Likewise.
13173 * config/mips/mips.md (consttable_float): Likewise.
13174 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13175 * varasm.c (assemble_real): Take a scalar_float_mode.
13176 (output_constant_pool_2): Update accordingly.
13177 (output_constant): Likewise.
13179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13180 Alan Hayward <alan.hayward@arm.com>
13181 David Sherwood <david.sherwood@arm.com>
13183 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13184 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13185 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13186 (native_encode_real): Likewise.
13187 (native_interpret_real): Likewise.
13188 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13189 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13192 Alan Hayward <alan.hayward@arm.com>
13193 David Sherwood <david.sherwood@arm.com>
13195 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13196 <scalar_float_mode>. Simplify.
13197 (gen_extend_conv_libfunc): Likewise.
13199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13200 Alan Hayward <alan.hayward@arm.com>
13201 David Sherwood <david.sherwood@arm.com>
13203 * coretypes.h (scalar_float_mode): New type.
13204 * machmode.h (mode_traits::from_int): Use machine_mode if
13205 USE_ENUM_MODES is defined.
13206 (is_a): New function.
13208 (dyn_cast): Likewise.
13209 (scalar_float_mode): New class.
13210 (scalar_float_mode::includes_p): New function.
13211 (is_float_mode): Likewise.
13212 * gdbhooks.py (MachineModePrinter): New class.
13213 (build_pretty_printer): Use it for scalar_float_mode.
13214 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13215 (format_helper::format_helper): Turn into a template.
13216 * genmodes.c (get_mode_class): New function.
13217 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13218 or machine_mode if none.
13219 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13220 as_a <scalar_float_mode>.
13221 * dwarf2out.c (mem_loc_descriptor): Likewise.
13222 (insert_float): Likewise.
13223 (add_const_value_attribute): Likewise.
13224 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13225 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13226 (expand_unop): Update accordingly.
13227 (expand_abs_nojump): Likewise.
13228 (expand_copysign_absneg): Take a scalar_float_mode.
13229 (expand_copysign_bit): Likewise.
13230 (expand_copysign): Update accordingly.
13232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13233 Alan Hayward <alan.hayward@arm.com>
13234 David Sherwood <david.sherwood@arm.com>
13236 * coretypes.h (opt_mode): New class.
13237 * machmode.h (opt_mode): Likewise.
13238 (opt_mode::else_void): New function.
13239 (opt_mode::require): Likewise.
13240 (opt_mode::exists): Likewise.
13241 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13242 (GET_MODE_2XWIDER_MODE): Likewise.
13243 (mode_iterator::get_wider): Update accordingly.
13244 (mode_iterator::get_2xwider): Likewise.
13245 (mode_iterator::get_known_wider): Likewise, turning into a template.
13246 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13247 forcing a wider mode to exist.
13248 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13249 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13250 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13251 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13252 * lower-subreg.c (init_lower_subreg): Likewise.
13253 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13254 on the final iteration.
13255 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13256 a wider mode exists before asking for a move pattern.
13257 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13258 forcing a wider mode to exist.
13259 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13260 returning false if no such mode exists.
13261 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13262 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13263 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13264 Avoid checking for a MODE_INT if we already know the mode is not a
13266 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13267 forcing a wider mode to exist.
13268 (expmed_mult_highpart_optab): Likewise.
13269 (expmed_mult_highpart): Likewise.
13270 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13272 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13273 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13274 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13275 * internal-fn.c (expand_mul_overflow): Update use of
13276 GET_MODE_2XWIDER_MODE.
13277 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13278 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13279 GET_MODE_WIDER_MODE.
13280 (convert_plusminus_to_widen): Likewise.
13281 * tree-switch-conversion.c (array_value_type): Likewise.
13282 * var-tracking.c (emit_note_insn_var_location): Likewise.
13283 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13284 Return false inside rather than outside the loop if no wider mode
13286 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13287 and GET_MODE_2XWIDER_MODE
13288 (can_compare_p): Use else_void.
13289 * gdbhooks.py (OptMachineModePrinter): New class.
13290 (build_pretty_printer): Use it for opt_mode.
13292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13293 Alan Hayward <alan.hayward@arm.com>
13294 David Sherwood <david.sherwood@arm.com>
13296 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13297 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13300 Alan Hayward <alan.hayward@arm.com>
13301 David Sherwood <david.sherwood@arm.com>
13303 * machmode.h (mode_traits): New structure.
13304 (get_narrowest_mode): New function.
13305 (mode_iterator::start): Likewise.
13306 (mode_iterator::iterate_p): Likewise.
13307 (mode_iterator::get_wider): Likewise.
13308 (mode_iterator::get_known_wider): Likewise.
13309 (mode_iterator::get_2xwider): Likewise.
13310 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13311 (FOR_EACH_MODE): Likewise.
13312 (FOR_EACH_MODE_FROM): Likewise.
13313 (FOR_EACH_MODE_UNTIL): Likewise.
13314 (FOR_EACH_WIDER_MODE): Likewise.
13315 (FOR_EACH_2XWIDER_MODE): Likewise.
13316 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13317 * combine.c (simplify_comparison): Likewise
13318 * config/i386/i386.c (type_natural_mode): Likewise.
13319 * cse.c (cse_insn): Likewise.
13320 * dse.c (find_shift_sequence): Likewise.
13321 * emit-rtl.c (init_derived_machine_modes): Likewise.
13322 (init_emit_once): Likewise.
13323 * explow.c (hard_function_value): Likewise.
13324 * expmed.c (extract_fixed_bit_field_1): Likewise.
13325 (extract_bit_field_1): Likewise.
13326 (expand_divmod): Likewise.
13327 (emit_store_flag_1): Likewise.
13328 * expr.c (init_expr_target): Likewise.
13329 (convert_move): Likewise.
13330 (alignment_for_piecewise_move): Likewise.
13331 (widest_int_mode_for_size): Likewise.
13332 (emit_block_move_via_movmem): Likewise.
13333 (copy_blkmode_to_reg): Likewise.
13334 (set_storage_via_setmem): Likewise.
13335 (compress_float_constant): Likewise.
13336 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13337 * optabs-query.c (get_best_extraction_insn): Likewise.
13338 * optabs.c (expand_binop): Likewise.
13339 (expand_twoval_unop): Likewise.
13340 (expand_twoval_binop): Likewise.
13341 (widen_leading): Likewise.
13342 (widen_bswap): Likewise.
13343 (expand_parity): Likewise.
13344 (expand_unop): Likewise.
13345 (prepare_cmp_insn): Likewise.
13346 (prepare_float_lib_cmp): Likewise.
13347 (expand_float): Likewise.
13348 (expand_fix): Likewise.
13349 (expand_sfix_optab): Likewise.
13350 * postreload.c (move2add_use_add2_insn): Likewise.
13351 * reg-stack.c (reg_to_stack): Likewise.
13352 * reginfo.c (choose_hard_reg_mode): Likewise.
13353 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13354 * stor-layout.c (mode_for_size): Likewise.
13355 (smallest_mode_for_size): Likewise.
13356 (mode_for_vector): Likewise.
13357 (finish_bitfield_representative): Likewise.
13358 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13359 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13360 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13361 * var-tracking.c (prepare_call_arguments): Likewise.
13363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13364 Alan Hayward <alan.hayward@arm.com>
13365 David Sherwood <david.sherwood@arm.com>
13367 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13368 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13369 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13370 * machmode.h (mode_size): Move earlier in file.
13371 (mode_precision): Likewise.
13372 (mode_inner): Likewise.
13373 (mode_nunits): Likewise.
13374 (mode_unit_size): Likewise.
13375 (unit_unit_precision): Likewise.
13376 (mode_wider): Likewise.
13377 (mode_2xwider): Likewise.
13378 (machine_mode): New class.
13379 (mode_to_bytes): New function.
13380 (mode_to_bits): Likewise.
13381 (mode_to_precision): Likewise.
13382 (mode_to_inner): Likewise.
13383 (mode_to_unit_size): Likewise.
13384 (mode_to_unit_precision): Likewise.
13385 (mode_to_nunits): Likewise.
13386 (GET_MODE_SIZE): Use mode_to_bytes.
13387 (GET_MODE_BITSIZE): Use mode_to_bits.
13388 (GET_MODE_PRECISION): Use mode_to_precision.
13389 (GET_MODE_INNER): Use mode_to_inner.
13390 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13391 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13392 (GET_MODE_NUNITS): Use mode_to_nunits.
13393 * system.h (ALWAYS_INLINE): New macro.
13394 * config/powerpcspe/powerpcspe-c.c
13395 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13396 int for arg1_mode and arg2_mode.
13398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13399 Alan Hayward <alan.hayward@arm.com>
13400 David Sherwood <david.sherwood@arm.com>
13402 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13403 Prefix mode names with E_ in case statements.
13404 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13405 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13406 (aarch64_split_simd_move): Likewise.
13407 (aarch64_gen_storewb_pair): Likewise.
13408 (aarch64_gen_loadwb_pair): Likewise.
13409 (aarch64_gen_store_pair): Likewise.
13410 (aarch64_gen_load_pair): Likewise.
13411 (aarch64_get_condition_code_1): Likewise.
13412 (aarch64_constant_pool_reload_icode): Likewise.
13413 (get_rsqrte_type): Likewise.
13414 (get_rsqrts_type): Likewise.
13415 (get_recpe_type): Likewise.
13416 (get_recps_type): Likewise.
13417 (aarch64_gimplify_va_arg_expr): Likewise.
13418 (aarch64_simd_container_mode): Likewise.
13419 (aarch64_emit_load_exclusive): Likewise.
13420 (aarch64_emit_store_exclusive): Likewise.
13421 (aarch64_expand_compare_and_swap): Likewise.
13422 (aarch64_gen_atomic_cas): Likewise.
13423 (aarch64_emit_bic): Likewise.
13424 (aarch64_emit_atomic_swap): Likewise.
13425 (aarch64_emit_atomic_load_op): Likewise.
13426 (aarch64_evpc_trn): Likewise.
13427 (aarch64_evpc_uzp): Likewise.
13428 (aarch64_evpc_zip): Likewise.
13429 (aarch64_evpc_ext): Likewise.
13430 (aarch64_evpc_rev): Likewise.
13431 (aarch64_evpc_dup): Likewise.
13432 (aarch64_gen_ccmp_first): Likewise.
13433 (aarch64_gen_ccmp_next): Likewise.
13434 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13435 (alpha_emit_xfloating_libcall): Likewise.
13436 (emit_insxl): Likewise.
13437 (alpha_arg_type): Likewise.
13438 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13439 (arc_preferred_simd_mode): Likewise.
13440 (arc_secondary_reload): Likewise.
13441 (get_arc_condition_code): Likewise.
13442 (arc_print_operand): Likewise.
13443 (arc_legitimate_constant_p): Likewise.
13444 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13445 * config/arc/arc.md (casesi_load): Likewise.
13446 (casesi_compact_jump): Likewise.
13447 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13448 (cc_use_register): Likewise.
13449 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13450 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13451 (arm_init_iwmmxt_builtins): Likewise.
13452 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13453 (neon_expand_vector_init): Likewise.
13454 (arm_attr_length_move_neon): Likewise.
13455 (maybe_get_arm_condition_code): Likewise.
13456 (arm_emit_vector_const): Likewise.
13457 (arm_preferred_simd_mode): Likewise.
13458 (arm_output_iwmmxt_tinsr): Likewise.
13459 (thumb1_output_casesi): Likewise.
13460 (thumb2_output_casesi): Likewise.
13461 (arm_emit_load_exclusive): Likewise.
13462 (arm_emit_store_exclusive): Likewise.
13463 (arm_expand_compare_and_swap): Likewise.
13464 (arm_evpc_neon_vuzp): Likewise.
13465 (arm_evpc_neon_vzip): Likewise.
13466 (arm_evpc_neon_vrev): Likewise.
13467 (arm_evpc_neon_vtrn): Likewise.
13468 (arm_evpc_neon_vext): Likewise.
13469 (arm_validize_comparison): Likewise.
13470 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13471 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13472 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13473 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13474 (c6x_preferred_simd_mode): Likewise.
13475 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13476 (epiphany_rtx_costs): Likewise.
13477 * config/epiphany/predicates.md (proper_comparison_operator):
13479 * config/frv/frv.c (condexec_memory_operand): Likewise.
13480 (frv_emit_move): Likewise.
13481 (output_move_single): Likewise.
13482 (output_condmove_single): Likewise.
13483 (frv_hard_regno_mode_ok): Likewise.
13484 (frv_matching_accg_mode): Likewise.
13485 * config/h8300/h8300.c (split_adds_subs): Likewise.
13486 (h8300_rtx_costs): Likewise.
13487 (h8300_print_operand): Likewise.
13488 (compute_mov_length): Likewise.
13489 (output_logical_op): Likewise.
13490 (compute_logical_op_length): Likewise.
13491 (compute_logical_op_cc): Likewise.
13492 (h8300_shift_needs_scratch_p): Likewise.
13493 (output_a_shift): Likewise.
13494 (compute_a_shift_length): Likewise.
13495 (compute_a_shift_cc): Likewise.
13496 (expand_a_rotate): Likewise.
13497 (output_a_rotate): Likewise.
13498 * config/i386/i386.c (classify_argument): Likewise.
13499 (function_arg_advance_32): Likewise.
13500 (function_arg_32): Likewise.
13501 (function_arg_64): Likewise.
13502 (function_value_64): Likewise.
13503 (ix86_gimplify_va_arg): Likewise.
13504 (ix86_legitimate_constant_p): Likewise.
13505 (put_condition_code): Likewise.
13506 (split_double_mode): Likewise.
13507 (ix86_avx256_split_vector_move_misalign): Likewise.
13508 (ix86_expand_vector_logical_operator): Likewise.
13509 (ix86_split_idivmod): Likewise.
13510 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13511 (ix86_build_const_vector): Likewise.
13512 (ix86_build_signbit_mask): Likewise.
13513 (ix86_match_ccmode): Likewise.
13514 (ix86_cc_modes_compatible): Likewise.
13515 (ix86_expand_branch): Likewise.
13516 (ix86_expand_sse_cmp): Likewise.
13517 (ix86_expand_sse_movcc): Likewise.
13518 (ix86_expand_int_sse_cmp): Likewise.
13519 (ix86_expand_vec_perm_vpermi2): Likewise.
13520 (ix86_expand_vec_perm): Likewise.
13521 (ix86_expand_sse_unpack): Likewise.
13522 (ix86_expand_int_addcc): Likewise.
13523 (ix86_split_to_parts): Likewise.
13524 (ix86_vectorize_builtin_gather): Likewise.
13525 (ix86_vectorize_builtin_scatter): Likewise.
13526 (avx_vpermilp_parallel): Likewise.
13527 (inline_memory_move_cost): Likewise.
13528 (ix86_tieable_integer_mode_p): Likewise.
13529 (x86_maybe_negate_const_int): Likewise.
13530 (ix86_expand_vector_init_duplicate): Likewise.
13531 (ix86_expand_vector_init_one_nonzero): Likewise.
13532 (ix86_expand_vector_init_one_var): Likewise.
13533 (ix86_expand_vector_init_concat): Likewise.
13534 (ix86_expand_vector_init_interleave): Likewise.
13535 (ix86_expand_vector_init_general): Likewise.
13536 (ix86_expand_vector_set): Likewise.
13537 (ix86_expand_vector_extract): Likewise.
13538 (emit_reduc_half): Likewise.
13539 (ix86_emit_i387_round): Likewise.
13540 (ix86_mangle_type): Likewise.
13541 (ix86_expand_round_sse4): Likewise.
13542 (expand_vec_perm_blend): Likewise.
13543 (canonicalize_vector_int_perm): Likewise.
13544 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13545 (expand_vec_perm_1): Likewise.
13546 (expand_vec_perm_interleave3): Likewise.
13547 (expand_vec_perm_even_odd_pack): Likewise.
13548 (expand_vec_perm_even_odd_1): Likewise.
13549 (expand_vec_perm_broadcast_1): Likewise.
13550 (ix86_vectorize_vec_perm_const_ok): Likewise.
13551 (ix86_expand_vecop_qihi): Likewise.
13552 (ix86_expand_mul_widen_hilo): Likewise.
13553 (ix86_expand_sse2_abs): Likewise.
13554 (ix86_expand_pextr): Likewise.
13555 (ix86_expand_pinsr): Likewise.
13556 (ix86_preferred_simd_mode): Likewise.
13557 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13558 * config/i386/sse.md (*andnot<mode>3): Likewise.
13559 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13560 (*<code><mode>3): Likewise.
13561 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13562 (ia64_expand_atomic_op): Likewise.
13563 (ia64_arg_type): Likewise.
13564 (ia64_mode_to_int): Likewise.
13565 (ia64_scalar_mode_supported_p): Likewise.
13566 (ia64_vector_mode_supported_p): Likewise.
13567 (expand_vec_perm_broadcast): Likewise.
13568 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13569 (iq2000_function_arg_advance): Likewise.
13570 (iq2000_function_arg): Likewise.
13571 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13572 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13573 (m68k_libcall_value): Likewise.
13574 (m68k_function_value): Likewise.
13575 (sched_attr_op_type): Likewise.
13576 * config/mcore/mcore.c (mcore_output_move): Likewise.
13577 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13579 (microblaze_function_arg): Likewise.
13580 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13581 (mips_print_operand): Likewise.
13582 (mips_mode_ok_for_mov_fmt_p): Likewise.
13583 (mips_vector_mode_supported_p): Likewise.
13584 (mips_preferred_simd_mode): Likewise.
13585 (mips_expand_vpc_loongson_even_odd): Likewise.
13586 (mips_expand_vec_unpack): Likewise.
13587 (mips_expand_vi_broadcast): Likewise.
13588 (mips_expand_vector_init): Likewise.
13589 (mips_expand_vec_reduc): Likewise.
13590 (mips_expand_msa_cmp): Likewise.
13591 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13592 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13593 (cc_flags_for_mode): Likewise.
13594 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13595 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13596 (nds32_output_casesi_pc_relative): Likewise.
13597 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13598 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13599 (nvptx_gen_unpack): Likewise.
13600 (nvptx_gen_pack): Likewise.
13601 (nvptx_gen_shuffle): Likewise.
13602 (nvptx_gen_wcast): Likewise.
13603 (nvptx_preferred_simd_mode): Likewise.
13604 * config/pa/pa.c (pa_secondary_reload): Likewise.
13605 * config/pa/predicates.md (base14_operand): Likewise.
13606 * config/powerpcspe/powerpcspe-c.c
13607 (altivec_resolve_overloaded_builtin): Likewise.
13608 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13610 (rs6000_preferred_simd_mode): Likewise.
13611 (output_vec_const_move): Likewise.
13612 (rs6000_expand_vector_extract): Likewise.
13613 (rs6000_split_vec_extract_var): Likewise.
13614 (reg_offset_addressing_ok_p): Likewise.
13615 (rs6000_legitimate_offset_address_p): Likewise.
13616 (rs6000_legitimize_address): Likewise.
13617 (rs6000_emit_set_const): Likewise.
13618 (rs6000_const_vec): Likewise.
13619 (rs6000_emit_move): Likewise.
13620 (spe_build_register_parallel): Likewise.
13621 (rs6000_darwin64_record_arg_recurse): Likewise.
13622 (swap_selector_for_mode): Likewise.
13623 (spe_init_builtins): Likewise.
13624 (paired_init_builtins): Likewise.
13625 (altivec_init_builtins): Likewise.
13626 (do_load_for_compare): Likewise.
13627 (rs6000_generate_compare): Likewise.
13628 (rs6000_expand_float128_convert): Likewise.
13629 (emit_load_locked): Likewise.
13630 (emit_store_conditional): Likewise.
13631 (rs6000_output_function_epilogue): Likewise.
13632 (rs6000_handle_altivec_attribute): Likewise.
13633 (rs6000_function_value): Likewise.
13634 (emit_fusion_gpr_load): Likewise.
13635 (emit_fusion_p9_load): Likewise.
13636 (emit_fusion_p9_store): Likewise.
13637 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13638 (fusion_gpr_mem_load): Likewise.
13639 (fusion_addis_mem_combo_load): Likewise.
13640 (fusion_addis_mem_combo_store): Likewise.
13641 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13642 (fusion_gpr_mem_load): Likewise.
13643 (fusion_addis_mem_combo_load): Likewise.
13644 (fusion_addis_mem_combo_store): Likewise.
13645 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13647 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13648 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13649 (rs6000_preferred_simd_mode): Likewise.
13650 (output_vec_const_move): Likewise.
13651 (rs6000_expand_vector_extract): Likewise.
13652 (rs6000_split_vec_extract_var): Likewise.
13653 (reg_offset_addressing_ok_p): Likewise.
13654 (rs6000_legitimate_offset_address_p): Likewise.
13655 (rs6000_legitimize_address): Likewise.
13656 (rs6000_emit_set_const): Likewise.
13657 (rs6000_const_vec): Likewise.
13658 (rs6000_emit_move): Likewise.
13659 (rs6000_darwin64_record_arg_recurse): Likewise.
13660 (swap_selector_for_mode): Likewise.
13661 (paired_init_builtins): Likewise.
13662 (altivec_init_builtins): Likewise.
13663 (rs6000_expand_float128_convert): Likewise.
13664 (emit_load_locked): Likewise.
13665 (emit_store_conditional): Likewise.
13666 (rs6000_output_function_epilogue): Likewise.
13667 (rs6000_handle_altivec_attribute): Likewise.
13668 (rs6000_function_value): Likewise.
13669 (emit_fusion_gpr_load): Likewise.
13670 (emit_fusion_p9_load): Likewise.
13671 (emit_fusion_p9_store): Likewise.
13672 * config/rx/rx.c (rx_gen_move_template): Likewise.
13673 (flags_from_mode): Likewise.
13674 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13675 (s390_slb_comparison): Likewise.
13676 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13677 (s390_vector_mode_supported_p): Likewise.
13678 (s390_cc_modes_compatible): Likewise.
13679 (s390_match_ccmode_set): Likewise.
13680 (s390_canonicalize_comparison): Likewise.
13681 (s390_emit_compare_and_swap): Likewise.
13682 (s390_branch_condition_mask): Likewise.
13683 (s390_rtx_costs): Likewise.
13684 (s390_secondary_reload): Likewise.
13685 (__SECONDARY_RELOAD_CASE): Likewise.
13686 (s390_expand_cs): Likewise.
13687 (s390_preferred_simd_mode): Likewise.
13688 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13689 * config/sh/sh.c (sh_print_operand): Likewise.
13690 (dump_table): Likewise.
13691 (sh_secondary_reload): Likewise.
13692 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13693 * config/sh/sh.md (casesi_worker_1): Likewise.
13694 (casesi_worker_2): Likewise.
13695 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13696 (fcc_comparison_operator): Likewise.
13697 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13698 (emit_soft_tfmode_cvt): Likewise.
13699 (sparc_preferred_simd_mode): Likewise.
13700 (output_cbranch): Likewise.
13701 (sparc_print_operand): Likewise.
13702 (sparc_expand_vec_perm_bmask): Likewise.
13703 (vector_init_bshuffle): Likewise.
13704 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13705 (spu_vector_mode_supported_p): Likewise.
13706 (spu_expand_insv): Likewise.
13707 (spu_emit_branch_or_set): Likewise.
13708 (spu_handle_vector_attribute): Likewise.
13709 (spu_builtin_splats): Likewise.
13710 (spu_builtin_extract): Likewise.
13711 (spu_builtin_promote): Likewise.
13712 (spu_expand_sign_extend): Likewise.
13713 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13714 (tilegx_simd_int): Likewise.
13715 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13716 (tilepro_simd_int): Likewise.
13717 * config/v850/v850.c (const_double_split): Likewise.
13718 (v850_print_operand): Likewise.
13719 (ep_memory_offset): Likewise.
13720 * config/vax/vax.c (vax_rtx_costs): Likewise.
13721 (vax_output_int_move): Likewise.
13722 (vax_output_int_add): Likewise.
13723 (vax_output_int_subtract): Likewise.
13724 * config/visium/predicates.md (visium_branch_operator): Likewise.
13725 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13726 (visium_print_operand_address): Likewise.
13727 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13728 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13729 (xtensa_expand_conditional_branch): Likewise.
13730 (xtensa_copy_incoming_a7): Likewise.
13731 (xtensa_output_literal): Likewise.
13732 * dfp.c (decimal_real_maxval): Likewise.
13733 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13736 Alan Hayward <alan.hayward@arm.com>
13737 David Sherwood <david.sherwood@arm.com>
13739 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13740 (mode_nunits_inline): Likewise.
13741 (mode_inner_inline): Likewise.
13742 (mode_unit_size_inline): Likewise.
13743 (mode_unit_precision_inline): Likewise.
13744 (emit_insn_modes_h): Likewise. Also emit a #define of the
13746 (emit_mode_wider): Add an E_ prefix to mode names.
13747 (emit_mode_complex): Likewise.
13748 (emit_mode_inner): Likewise.
13749 (emit_mode_adjustments): Likewise.
13750 (emit_mode_int_n): Likewise.
13751 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13752 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13753 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13754 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13755 (CRC32_BUILTIN, ENTRY): Likewise.
13756 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13757 (aarch64_pop_regs): Likewise.
13758 (aarch64_process_components): Likewise.
13759 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13760 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13761 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13762 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13763 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13764 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13766 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13767 (pcmpistr): Likewise.
13768 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13769 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13770 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13772 * config/rl78/rl78.c (mduc_regs): Likewise.
13773 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13774 (htm_expand_builtin): Likewise.
13775 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13776 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13777 * config/xtensa/xtensa.c (print_operand): Likewise.
13778 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13779 (NUM_MODE_VECTOR_INT): Likewise.
13780 * genoutput.c (null_operand): Likewise.
13781 (output_operand_data): Likewise.
13782 * genrecog.c (print_parameter_value): Likewise.
13783 * lra.c (debug_operand_data): Likewise.
13785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13787 * dwarf2out.c (scompare_loc_descriptor_wide)
13788 (scompare_loc_descriptor_narrow): New functions, split out from...
13789 (scompare_loc_descriptor): ...here.
13790 * expmed.c (emit_store_flag_int): New function, split out from...
13791 (emit_store_flag): ...here.
13793 2017-08-30 Richard Biener <rguenther@suse.de>
13795 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13796 (dwarf2out_early_finish): Move setting of AT_pubnames from
13797 early debug output to early finish.
13799 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13801 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13802 and -mdata-region to the assembler.
13804 2017-08-30 Richard Biener <rguenther@suse.de>
13806 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13808 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13809 (dwarf2out_early_global_decl): Only generate a DIE for the
13810 abstract origin if it doesn't already exist or is a declaration DIE.
13811 (resolve_addr): Do not add the linkage name twice when
13812 generating a stub DIE for the DW_TAG_GNU_call_site target.
13814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13816 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13817 Use machine_mode rather than int for arg1_mode.
13819 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13822 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13823 that the second argument of the built-in functions to unpack
13824 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13825 switch statement instead a lot of if statements.
13826 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13827 Allow 64-bit values to be in Altivec registers as well as
13828 traditional floating point registers.
13829 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13831 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13833 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13834 MAX_REGS_PER_ADDRESS == 1.
13836 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13838 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13839 * config/i386/i386.c (ix86_option_override_internal): Simplify
13840 setting of opts->x_flag_entry.
13842 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13843 Jakub Jelinek <jakub@redhat.com>
13844 Richard Biener <rguenther@suse.de>
13846 PR tree-optimization/81503
13847 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13848 folded constant fits in the target type; reorder tests for clarity.
13850 2017-08-29 Martin Liska <mliska@suse.cz>
13852 * passes.def: Include pass_lower_switch.
13853 * stmt.c (dump_case_nodes): Remove and move to
13854 tree-switch-conversion.
13855 (case_values_threshold): Likewise.
13856 (expand_switch_as_decision_tree_p): Likewise.
13857 (emit_case_decision_tree): Likewise.
13858 (expand_case): Likewise.
13859 (balance_case_nodes): Likewise.
13860 (node_has_low_bound): Likewise.
13861 (node_has_high_bound): Likewise.
13862 (node_is_bounded): Likewise.
13863 (emit_case_nodes): Likewise.
13864 (struct simple_case_node): New struct.
13865 (add_case_node): Remove.
13866 (emit_case_dispatch_table): Use vector instead of case_list.
13867 (reset_out_edges_aux): Remove.
13868 (compute_cases_per_edge): Likewise.
13869 (expand_case): Build list of simple_case_node.
13870 (expand_sjlj_dispatch_table): Use it.
13871 * tree-switch-conversion.c (struct case_node): Moved from
13872 stmt.c and adjusted.
13873 (emit_case_nodes): Likewise.
13874 (node_has_low_bound): Likewise.
13875 (node_has_high_bound): Likewise.
13876 (node_is_bounded): Likewise.
13877 (case_values_threshold): Likewise.
13878 (reset_out_edges_aux): Likewise.
13879 (compute_cases_per_edge): Likewise.
13880 (add_case_node): Likewise.
13881 (dump_case_nodes): Likewise.
13882 (balance_case_nodes): Likewise.
13883 (expand_switch_as_decision_tree_p): Likewise.
13884 (emit_jump): Likewise.
13885 (emit_case_decision_tree): Likewise.
13886 (try_switch_expansion): Likewise.
13887 (do_jump_if_equal): Likewise.
13888 (emit_cmp_and_jump_insns): Likewise.
13889 (fix_phi_operands_for_edge): New function.
13890 (record_phi_operand_mapping): Likewise.
13891 (class pass_lower_switch): New pass.
13892 (pass_lower_switch::execute): New function.
13893 (make_pass_lower_switch): Likewise.
13894 (conditional_probability):
13895 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13896 * tree-pass.h: Add make_pass_lower_switch.
13898 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13901 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13904 2017-08-29 Richard Biener <rguenther@suse.de>
13906 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13907 we do not add a DW_AT_inline attribute twice.
13908 (gen_subprogram_die): Remove code setting DW_AT_inline on
13909 DECL_ABSTRACT_P nodes.
13911 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13913 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13914 calls to internal functions.
13915 (gimplify_modify_expr): Likewise.
13916 * tree-call-cdce.c (use_internal_fn): Likewise.
13917 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13918 (convert_to_divmod): Set the nothrow flag.
13919 * tree-if-conv.c (predicate_mem_writes): Likewise.
13920 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13921 (vectorizable_call): Likewise.
13922 (vectorizable_store): Likewise.
13923 (vectorizable_load): Likewise.
13924 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13925 (vect_recog_mask_conversion_pattern): Likewise.
13927 2017-08-29 Martin Liska <mliska@suse.cz>
13930 * gcc.c (driver_handle_option): Add new argument.
13931 * opts-common.c (handle_option): Pass
13932 target_option_override_hook.
13933 * opts-global.c (lang_handle_option): Add new option.
13934 (set_default_handlers): Add new argument.
13935 (decode_options): Likewise.
13936 * opts.c (target_handle_option): Likewise.
13937 (common_handle_option): Call target_option_override_hook.
13938 * opts.h (struct cl_option_handler_func): Add hook for
13939 target option override.
13940 (struct cl_option_handlers): Likewise.
13941 (set_default_handlers): Add new argument.
13942 (decode_options): Likewise.
13943 (common_handle_option): Likewise.
13944 (target_handle_option): Likewise.
13945 * toplev.c (toplev::main): Pass targetm.target_option.override
13948 2017-08-29 Richard Biener <rguenther@suse.de>
13949 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13951 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13952 life to the active subtree.
13954 2017-08-28 Jeff Law <law@redhat.com>
13956 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13957 derive_equivalences.
13958 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13959 Code moved into....
13960 (edge_info::derive_equivalences): New private member function
13962 * tree-ssa-dom.c (class edge_info): Changed from a struct
13963 to a class. Add ctor/dtor, methods and data members.
13964 (edge_info::edge_info): Renamed from allocate_edge_info.
13965 Initialize additional members.
13966 (edge_info::~edge_info): New.
13967 (free_dom_edge_info): Delete the edge info.
13968 (record_edge_info): Use new class & associated member functions.
13969 Tighten forms for testing for edge equivalences.
13970 (record_temporary_equivalences): Iterate over the simple
13971 equivalences rather than assuming there's only one per edge.
13972 (cprop_into_successor_phis): Iterate over the simple
13973 equivalences rather than assuming there's only one per edge.
13974 (optimize_stmt): Use operand_equal_p rather than pointer
13975 equality for mini-DSE code.
13977 2017-08-28 Nathan Sidwell <nathan@acm.org>
13979 * gcc.c (execute): Fold SIGPIPE handling into switch
13980 statement. Adjust internal error message.
13982 2017-08-28 Richard Biener <rguenther@suse.de>
13985 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13986 Do nothing for removed DIEs.
13988 2017-08-28 Richard Biener <rguenther@suse.de>
13990 PR tree-optimization/81977
13991 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13994 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13997 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13998 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13999 using targetm.gen_mem_thread_fence.
14001 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
14004 * config/i386/i386.md (*<btsc><mode>): Change operand 2
14005 predicate to register_operand. Reorder operands.
14006 (*btr<mode>): Ditto.
14007 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
14008 (*btr<mode>_mask): Ditto.
14010 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
14012 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
14013 * config/rs6000/xmmintrin.h: New file.
14014 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
14016 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14019 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
14020 parameter and_insn and return it.
14021 (recombine_lvx_pattern): Insert a copy to ensure availability of
14022 the base register of the copied masking operation at the point of
14023 the instruction replacement.
14024 (recombine_stvx_pattern): Likewise.
14026 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
14028 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
14029 undocumented switches.
14030 (-mpower9-dform-vector): Likewise.
14031 (-mpower9-dform): Likewise.
14032 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14033 comments to delete references to -mpower9-dform* switches.
14034 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
14035 Delete reference to -mpower9-dform* switches, test for
14036 -mpower9-vector instead.
14037 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14038 (OTHER_P9_VECTOR_MASKS): Likewise.
14039 (POWERPC_MASKS): Likewise.
14040 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14041 tests against -mpower9-dform* to -mpower9-vector. Delete code
14042 that checked for -mpower9-dform* consistancy with other options.
14043 Add test for -mpower9-misc to enable other power9 switches.
14044 (rs6000_init_hard_regno_mode_ok): Likewise.
14045 (rs6000_option_override_internal): Likewise.
14046 (rs6000_emit_prologue): Likewise.
14047 (rs6000_emit_epilogue): Likewise.
14048 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14049 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14050 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14052 (emit_fusion_p9_store): Likewise.
14053 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14054 resetting these macros if the assembler does not support ISA 3.0
14056 (TARGET_P9_DFORM_VECTOR): Likewise.
14057 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14058 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14060 2017-08-25 Alan Modra <amodra@gmail.com>
14062 PR rtl-optimization/81747
14063 * cse.c (cse_extended_basic_block): Don't attempt to record
14064 equivalences for degenerate conditional branches that branch
14065 to their fall-through.
14067 2017-08-24 Martin Sebor <msebor@redhat.com>
14069 PR middle-end/81908
14070 * gimple-fold.c (size_must_be_zero_p): New function.
14071 (gimple_fold_builtin_memory_op): Call it.
14073 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14075 * config/rs6000/mm_malloc.h: New file.
14077 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14079 PR tree-optimization/81913
14080 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14081 analysis when either IVs in condition can wrap.
14083 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14085 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14086 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14088 2017-08-24 Richard Biener <rguenther@suse.de>
14091 * targhooks.c (default_target_can_inline_p): Properly
14092 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14093 is present and always compare.
14094 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14095 infer -mfpmath=sse from TARGET_SSE_P.
14096 (ix86_can_inline_p): Properly use target_option_default_node when
14097 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14099 2017-08-24 Richard Biener <rguenther@suse.de>
14102 * dwarf2out.c (output_die): Handle flag_generate_offload like
14104 (output_comp_unit): Likewise.
14105 (gen_array_type_die): Likewise.
14106 (dwarf2out_early_finish): Likewise.
14107 (note_variable_value_in_expr): Likewise.
14108 (dwarf2out_finish): Likewise. Adjust assert.
14109 * cgraphunit.c (symbol_table::compile): Move setting of
14110 flag_generate_offload earlier ...
14111 (symbol_table::finalize_compilation_unit): ... here, before
14112 early debug finalization.
14114 2017-08-24 Richard Biener <rguenther@suse.de>
14116 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14117 and ipa-fnsummary.h.
14118 (ix86_can_inline_p): When ix86_fpmath flags do not match
14119 check whether the callee uses FP math at all.
14121 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14123 PR middle-end/81931
14124 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14125 instead of TYPE_PRECISION.
14127 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14128 Alan Hayward <alan.hayward@arm.com>
14129 David Sherwood <david.sherwood@arm.com>
14131 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14133 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14135 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14137 (ix86_frame::stack_realign_allocate): New field.
14138 (struct machine_frame_state): Modify comments.
14139 (machine_frame_state::sp_realigned_fp_end): New field.
14140 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14141 layout calculation.
14142 (sp_valid_at): Add assertion to assure no attempt to access invalid
14143 offset of a realigned stack.
14144 (fp_valid_at): Likewise.
14145 (choose_baseaddr): Modify comments.
14146 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14147 ix86_expand_prologue.
14148 (ix86_expand_prologue): Modify stack realignment and allocation.
14149 (ix86_expand_epilogue): Modify comments.
14150 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14151 avx2_runtime, avx512f, and avx512f_runtime.
14153 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14155 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14156 (mstackrealign): Do not init to -1.
14157 * config/i386/i386.c (ix86_option_override_internal):
14158 Check opts_set, not opts when setting default value of
14159 opts->x_ix86_force_align_arg_pointer.
14161 2017-08-23 Richard Biener <rguenther@suse.de>
14163 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14164 lang_hooks.decl_printable_name.
14165 * print-rtl-function.c (print_rtx_function): Likewise.
14166 * tree-pretty-print.c (dump_function_header): Likewise.
14168 2017-08-23 Richard Biener <rguenther@suse.de>
14171 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14172 -g0 at compile-time.
14174 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14176 PR middle-end/19706
14177 * doc/sourcebuild.texi (Other hardware attributes):
14180 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14182 PR middle-end/19706
14183 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14184 Fix single-use check.
14186 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14188 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14190 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14192 * doc/install.texi: Modify to add more details on running selected
14195 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14197 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14198 is combined with -mabi=ms.
14199 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14202 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14204 PR tree-optimization/81488
14205 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14206 and cached_basis fields.
14207 (MAX_SPREAD): New constant.
14208 (alloc_cand_and_find_basis): Initialize new fields.
14209 (clear_visited): New function.
14210 (create_phi_basis_1): Rename from create_phi_basis, set visited
14211 and cached_basis fields.
14212 (create_phi_basis): New wrapper function.
14213 (phi_add_costs_1): Rename from phi_add_costs, add spread
14214 parameter, set visited field, short-circuit when limits reached.
14215 (phi_add_costs): New wrapper function.
14216 (record_phi_increments_1): Rename from record_phi_increments, set
14218 (record_phi_increments): New wrapper function.
14219 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14220 (phi_incr_cost): New wrapper function.
14221 (all_phi_incrs_profitable_1): Rename from
14222 all_phi_incrs_profitable, set visited field.
14223 (all_phi_incrs_profitable): New wrapper function.
14225 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14226 Alan Hayward <alan.hayward@arm.com>
14227 David Sherwood <david.sherwood@arm.com>
14229 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14230 that takes the outer and inner modes.
14231 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14232 comparison as the canonical test for a paradoxical subreg.
14233 * combine.c (simplify_set): Use paradoxical_subreg_p.
14234 (make_extraction): Likewise.
14235 (force_to_mode): Likewise.
14236 (rtx_equal_for_field_assignment_p): Likewise.
14237 (gen_lowpart_for_combine): Likewise.
14238 (simplify_comparison): Likewise.
14239 * cse.c (equiv_constant): Likewise.
14240 * expmed.c (store_bit_field_1): Likewise.
14241 * final.c (alter_subreg): Likewise.
14242 * fwprop.c (propagate_rtx): Likewise.
14243 (forward_propagate_subreg): Likewise.
14244 * ira-conflicts.c (ira_build_conflicts): Likewise.
14245 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14246 * lra-constraints.c (curr_insn_transform): Likewise.
14247 (split_reg): Likewise.
14248 * lra-eliminations.c (move_plus_up): Likewise.
14249 (lra_eliminate_regs_1): Likewise.
14250 * recog.c (general_operand): Likewise.
14251 * ree.c (combine_reaching_defs): Likewise.
14252 * reload.c (push_reload): Likewise.
14253 (find_reloads): Likewise.
14254 * reload1.c (elimination_effects): Likewise.
14255 (compute_reload_subreg_offset): Likewise.
14256 (choose_reload_regs): Likewise.
14257 * rtlanal.c (subreg_lsb_1): Likewise.
14258 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14259 (simplify_subreg): Likewise.
14260 * var-tracking.c (track_loc_p): Likewise.
14261 * emit-rtl.c (byte_lowpart_offset): Likewise.
14262 (paradoxical_subreg_p): Delete out-of-line definition.
14264 2017-08-22 Jeff Law <law@redhat.com>
14266 PR tree-optimization/81741
14267 PR tree-optimization/71947
14268 * tree-ssa-dom.c: Include tree-inline.h.
14269 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14270 equivalences if one is more expensive to compute than the other.
14271 * tree-ssa-scopedtables.h (class const_or_copies): Make
14272 record_const_or_copy_raw method private.
14273 (class avail_exprs_stack): New method simplify_binary_operation.
14274 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14275 avail_exprs_stack::simplify_binary_operation as needed.
14276 (avail_exprs_stack::simplify_binary_operation): New function.
14278 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14280 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14281 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14282 (DOT_SYMBOLS): Likewise.
14283 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14284 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14285 (RS6000_ABI_NAME): Likewise.
14286 (TARGET_CMODEL): Likewise.
14287 (TOC_SECTION_ASM_OP): Likewise.
14288 (SET_CMODEL): New macro.
14289 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14291 2017-08-22 Richard Biener <rguenther@suse.de>
14293 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14294 to allow for free-lang-data replacements similar to verify_type_variant.
14296 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14297 Alan Hayward <alan.hayward@arm.com>
14298 David Sherwood <david.sherwood@arm.com>
14300 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14301 VOIDmode for the LABEL_REF.
14303 2017-08-22 Richard Biener <rguenther@suse.de>
14305 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14307 2017-08-22 Richard Biener <rguenther@suse.de>
14309 * common.opt (feliminate-dwarf2-dups): Ignore.
14310 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14311 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14312 same_die_p_wrap, compute_section_prefix,
14313 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14314 (comdat_symbol_id, comdat_symbol_number): Likewise.
14315 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14317 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14318 (output_die): Mark unreachable path unreachable.
14319 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14320 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14321 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14322 (dwarf2out_early_finish): Likewise.
14324 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14326 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14328 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14331 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14332 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14333 (avr_attribute_table) <io, io_low, address>: Initialize
14334 .decl_required with true.
14336 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14338 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14339 undocumented debugging options.
14340 (-mvsx-scalar-double): Likewise.
14341 (-mallow-df-permute): Likewise.
14342 (-mvectorize-builtins): Likewise.
14343 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14344 (rs6000_builtin_vectorized_function): Likewise.
14345 (rs6000_builtin_md_vectorized_function): Likewise.
14346 (rs6000_opt_vars): Likewise.
14348 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14351 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14352 (*btrq_imm): Rename from *btrq.
14353 (*btcq_imm): Rename from *btcq.
14354 (btsc): New code attribute.
14355 (*<btsc><mode>): New insn pattern.
14356 (*btr<mode>): Ditto.
14357 (*<btsc><mode>_mask): New insn_and_split pattern.
14358 (*btr<mode>_mask): Ditto.
14360 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14361 Alan Hayward <alan.hayward@arm.com>
14362 David Sherwood <david.sherwood@arm.com>
14364 * function.c (pad_below): Simplify padding calculation.
14366 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14367 Alan Hayward <alan.hayward@arm.com>
14368 David Sherwood <david.sherwood@arm.com>
14370 * target.def (function_prologue): Remove frame size argument.
14371 (function_epilogue): Likewise.
14372 * doc/tm.texi: Regenerate.
14373 * final.c (final_start_function): Update call to function_prologue.
14374 (final_end_function): Update call to function_epilogue.
14375 (default_function_pro_epilogue): Remove frame size argument.
14376 * output.h (default_function_pro_epilogue): Likewise.
14377 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14378 (arm_output_function_prologue): Likewise.
14379 * config/frv/frv.c (frv_function_prologue): Likewise.
14380 (frv_function_epilogue): Likewise.
14381 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14382 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14383 (ia64_output_function_epilogue): Likewise.
14384 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14385 (m32r_output_function_epilogue): Likewise.
14386 * config/microblaze/microblaze.c (microblaze_function_prologue)
14387 (microblaze_function_epilogue): Likewise.
14388 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14389 (mips_output_function_epilogue): Likewise.
14390 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14391 (mmix_target_asm_function_epilogue): Likewise.
14392 * config/msp430/msp430.c (msp430_start_function): Likewise.
14393 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14394 (nds32_asm_function_epilogue): Likewise.
14395 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14396 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14397 (pa_output_function_epilogue): Likewise.
14398 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14399 (rs6000_output_function_epilogue): Likewise.
14400 * config/rl78/rl78.c (rl78_start_function): Likewise.
14401 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14402 (rs6000_output_function_epilogue): Likewise.
14403 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14404 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14405 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14406 (sparc_asm_function_epilogue): Likewise.
14408 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14410 * tree.h (type_has_mode_precision_p): New function.
14411 * convert.c (convert_to_integer_1): Use it.
14412 * expr.c (expand_expr_real_2): Likewise.
14413 (expand_expr_real_1): Likewise.
14414 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14415 * match.pd: Likewise.
14416 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14417 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14418 * tree-tailcall.c (process_assignment): Likewise.
14419 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14420 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14421 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14422 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14423 (vectorizable_assignment): Likewise.
14424 (vectorizable_shift): Likewise.
14425 (vectorizable_operation): Likewise.
14426 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14428 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14430 * match.pd: Add pow (C, x) simplification.
14432 2017-08-21 Richard Biener <rguenther@suse.de>
14434 PR tree-optimization/81900
14435 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14436 for blocks with abnormal predecessors.
14437 (compute_antic): Do not set visited flag prematurely.
14439 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14442 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14444 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14446 * stor-layout.h (vector_type_mode): Move to...
14447 * tree.h (vector_type_mode): ...here.
14448 * stor-layout.c (vector_type_mode): Move to...
14449 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14451 2017-08-21 Richard Biener <rguenther@suse.de>
14453 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14454 register_external_die hooks.
14455 (debug_false_tree_charstarstar_uhwistar): Declare.
14456 (debug_nothing_tree_charstar_uhwi): Likewise.
14457 * debug.c (do_nothing_debug_hooks): Adjust.
14458 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14459 (debug_nothing_tree_charstar_uhwi): Likewise.
14460 * dbxout.c (dbx_debug_hooks): Adjust.
14461 (xcoff_debug_hooks): Likewise.
14462 * sdbout.c (sdb_debug_hooks): Likewise.
14463 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14464 * dwarf2out.c (macinfo_label_base): New global.
14465 (dwarf2out_register_external_die): New function for the
14466 register_external_die hook.
14467 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14468 (dwarf2_debug_hooks): Use them.
14469 (dwarf2_lineno_debug_hooks): Adjust.
14470 (struct die_struct): Add with_offset flag.
14471 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14472 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14473 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14474 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14475 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14476 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14477 defining section names for the early LTO debug variants.
14478 (reset_indirect_string): New helper.
14479 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14480 (print_dw_val): Add support for offsetted symbol references.
14481 (get_ultimate_context): Split out from is_cxx.
14482 (is_cxx): Use get_ultimate_context.
14483 (is_fortran): Add decl overload.
14484 (compute_comp_unit_symbol): Split out worker from
14485 compute_section_prefix.
14486 (compute_section_prefix): Call compute_comp_unit_symbol and
14487 set comdat_type_p here.
14488 (output_die): Skip DIE symbol output for the LTO added one.
14489 Handle DIE symbol references with offset.
14490 (output_comp_unit): Guard section name mangling properly.
14491 For LTO debug sections emit a symbol at the section beginning
14492 which we use to refer to its DIEs.
14493 (add_abstract_origin_attribute): For DIEs registered via
14494 dwarf2out_register_external_die directly refer to the early
14495 DIE rather than indirectly through the shadow one we created.
14496 Remove obsolete call to dwarf2out_abstract_function for
14497 non-function/block origins.
14498 (gen_array_type_die): When generating early LTO debug do
14499 not emit DW_AT_string_length.
14500 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14501 late when in LTO. As suggested place a gcc_unreachable for
14502 the DECL_ABSTRACT_P case.
14503 (gen_subprogram_die): Avoid another specification DIE
14504 for early built declarations/definitions for the late LTO case.
14505 (gen_variable_die): Add type references for late duplicated VLA dies
14507 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14508 we have the abstract instance already.
14509 (process_scope_var): Adjust decl DIE contexts in LTO which
14510 first puts them in limbo.
14511 (gen_decl_die): Do not generate type DIEs late apart from
14512 types for VLAs or for decls we do not yet have a DIE. Do not
14513 call dwarf2out_abstract_function late.
14514 (dwarf2out_early_global_decl): Make sure to create DIEs
14515 for abstract instances of a decl first.
14516 (dwarf2out_late_global_decl): Adjust comment.
14517 (output_macinfo_op): With multiple macro sections use
14518 macinfo_label_base to distinguish labels.
14519 (output_macinfo): Likewise. Update macinfo_label_base.
14520 Pass in the line info label.
14521 (note_variable_value_in_expr): When generating LTO resolve
14522 all variable values here by generating DIEs as needed.
14523 (init_sections_and_labels): Add early LTO debug flag parameter
14524 and generate different sections and names if set. Add generation
14525 counter for the labels so we can have multiple of them.
14526 (reset_dies): Helper to allow DIEs to be output multiple times.
14527 (dwarf2out_finish): When outputting DIEs to the fat part of an
14528 LTO object first reset DIEs.
14529 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14530 (modified_type_die): Check for decl_ultimate_origin being self
14532 (gen_type_die_with_usage): Likewise.
14533 (gen_typedef_die): Allow decl_ultimate_origin being self.
14534 (set_decl_abstract_flags): Remove.
14535 (set_block_abstract_flags): Likewise.
14536 (dwarf2out_abstract_function): Treat the early generated DIEs
14537 as the abstract copy and only add DW_AT_inline and
14538 DW_AT_artificial here and call set_decl_origin_self.
14539 If the DIE has an abstract origin don't do anything.
14540 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14541 if we have none yet (Go fails to build one, PR78628).
14542 (variably_modified_type_p): Prevent endless recursion for Ada
14543 cyclic pointer types.
14544 * lto-streamer-in.c: Include debug.h.
14545 (dref_queue): New global.
14546 (lto_read_tree_1): Stream in DIE references.
14547 (lto_input_tree): Register DIE references.
14548 (input_function): Stream DECL_DEBUG_ARGS.
14549 * lto-streamer-out.c: Include debug.h.
14550 (lto_write_tree_1): Output DIE references.
14551 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14552 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14553 (output_function): Stream DECL_DEBUG_ARGS.
14554 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14555 Stream DECL_ABSTRACT_ORIGIN.
14556 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14557 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14558 DECL_CONTEXT for file-scope decls.
14559 * lto-streamer.h (struct dref_entry): Declare.
14560 (dref_queue): Likewise.
14561 * cfgexpand.c (pass_expand::execute): Do not call the
14562 outlining_inline_function hook here.
14563 * lto-wrapper.c (debug_obj): New global.
14564 (tool_cleanup): Unlink it if required.
14565 (debug_objcopy): New function.
14566 (run_gcc): Handle early debug sections in the IL files by
14567 extracting them to separate files, partially linkin them and
14568 feeding the result back as result to the linker.
14569 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14570 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14571 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14572 sections into a separate segment.
14573 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14575 (darwin_asm_dwarf_section): Likewise.
14576 (darwin_asm_output_dwarf_offset): Likewise.
14577 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14579 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14580 Alan Hayward <alan.hayward@arm.com>
14581 David Sherwood <david.sherwood@arm.com>
14583 * read-md.h (md_reader::record_potential_iterator_use): Replace
14584 pointer argument with an rtx and an index.
14585 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14586 (apply_mode_iterator): Likewise.
14587 (apply_code_iterator): Likewise.
14588 (apply_int_iterator): Likewise.
14589 (apply_subst_iterator): Likewise.
14590 (record_iterator_use): Likewise.
14591 (record_attribute_use): Likewise.
14592 (md_reader::record_potential_iterator_use): Likewise. Update calls
14593 to record_iterator_use and apply_iterator.
14594 (iterator_use): Replace ptr with x and index.
14595 (attribute_use): Likewise.
14596 (apply_attribute_uses): Update calls to apply_iterator.
14597 (apply_iterators): Likewise. Update initialization of iterator_use.
14598 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14599 and record_potential_iterator_use.
14600 (rtx_reader::read_rtx_operand): Likewise.
14602 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14603 Alan Hayward <alan.hayward@arm.com>
14604 David Sherwood <david.sherwood@arm.com>
14606 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14609 2017-08-21 Richard Biener <rguenther@suse.de>
14611 PR middle-end/81884
14612 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14613 at struct end conservatively when comparing common bases.
14615 2017-08-21 Richard Biener <rguenther@suse.de>
14617 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14618 (mem_ref_in_stmt): Remove.
14619 (determine_max_movement): Use ref index to get at the reference.
14620 (invariantness_dom_walker::before_dom_children): Deal with
14621 lim data already initialized.
14622 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14624 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14626 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14627 (TARGET_ISA_ROUND): Ditto.
14628 (TARGET_ROUND): Ditto.
14629 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14630 * config/i386/i386.md: Ditto.
14631 * config/i386/sse.md: Ditto.
14632 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14633 with OPTION_MASK_ISA_SSE4_1.
14635 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14638 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14639 __builtin_ia32_lzcnt_u16.
14641 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14644 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14645 (rs6000_set_current_function): Rewrite function to use it.
14647 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14650 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14651 and TYPE_WARN_IF_NOT_ALIGN.
14652 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14653 (handle_warn_if_not_align): New.
14654 (place_union_field): Call handle_warn_if_not_align.
14655 (place_field): Call handle_warn_if_not_align.
14656 Copy TYPE_WARN_IF_NOT_ALIGN.
14657 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14658 (layout_type): Likewise.
14659 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14661 (tree_decl_common): Add warn_if_not_align.
14662 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14663 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14664 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14665 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14666 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14667 * doc/extend.texi: Document warn_if_not_aligned attribute.
14668 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14670 2017-08-17 Martin Liska <mliska@suse.cz>
14673 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14674 (get_data_dependence): Use it as pointer type.
14675 (distribute_loop): Likewise.
14677 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14679 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14680 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14681 (p8_vmrgow_<mode>_direct): New define_insn.
14682 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14683 handle endianness for vmrgew and vmrgow permute patterns.
14685 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14687 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14688 * config/rs6000/rs6000-cpus.def: Remove comment.
14689 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14690 (POWERPC_MASKS): Likewise.
14691 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14692 use of TARGET_VSX_TIMODE.
14693 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14694 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14695 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14696 (rs6000_option_override_internal): Remove dead code.
14697 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14698 (rs6000_legitimize_reload_address): Likewise.
14699 (rs6000_legitimate_address_p): Likewise.
14700 (rs6000_opt_masks): Delete "vsx-timode".
14701 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14702 from function comment.
14703 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14704 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14705 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14707 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14708 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14709 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14710 (VSX_TI): Likewise.
14712 (define_peephole2): Likewise.
14714 2017-08-17 Martin Sebor <msebor@redhat.com>
14717 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14718 past the end of an array.
14719 (test_pp_format): Add test cases.
14721 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14723 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14724 missing ECF_NOTHROW flags.
14726 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14729 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14730 operands residing in integer registers.
14731 (*vsx_le_perm_load_<mode>): Likewise.
14732 (*vsx_le_perm_store_<mode>): Likewise.
14733 (define_peephole2): Add peepholes to optimize the above.
14735 2017-08-17 Marek Polacek <polacek@redhat.com>
14737 PR middle-end/81814
14738 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14739 to mimic what shorten_compare did. Change the return type to bool.
14740 (fold_cond_expr_with_comparison): Update call to
14741 operand_equal_for_comparison_p.
14742 (fold_ternary_loc): Likewise.
14744 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14746 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14748 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14750 2017-08-17 Richard Biener <rguenther@suse.de>
14752 * tree-ssa-structalias.c (solve_graph): When propagating
14753 to successors update the graphs succ edges and avoid duplicate work.
14755 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14758 * config/i386/i386.c (ix86_option_override_internal): Save target
14759 specific options after ix86_stack_protector_guard_reg was changed.
14761 2017-08-17 Richard Biener <rguenther@suse.de>
14763 PR tree-optimization/81827
14764 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14765 (new_var_info): Initialize it conservatively.
14766 (get_call_vi): Mark register vars.
14767 (new_scalar_tmp_constraint_exp): Likewise.
14768 (handle_rhs_call): Likewise.
14769 (handle_const_call): Likewise.
14770 (create_function_info_for): Likewise.
14771 (solve_constraints): Sort varinfos to separate register from
14772 non-register vars to pack points-to solution bitmaps during
14775 2017-08-17 Marek Polacek <polacek@redhat.com>
14777 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14779 2017-08-17 Richard Biener <rguenther@suse.de>
14781 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14782 to true when overflow is undefined and we saturated the result.
14784 2017-08-17 Alan Modra <amodra@gmail.com>
14787 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14788 Don't use store multiple if only one reg needs saving.
14789 (interesting_frame_related_regno): New function.
14790 (rs6000_frame_related): Don't emit frame info for regs that
14792 (rs6000_emit_epilogue): Likewise.
14794 2017-08-16 Nathan Sidwell <nathan@acm.org>
14796 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14797 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14798 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14799 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14800 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14801 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14802 (verify_type): Adjust for TYPE_BINFO move.
14803 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14804 process TYPE_BINFO directly.
14805 (hash_tree): Likewise.
14806 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14808 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14811 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14813 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14815 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14818 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14819 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14820 Add ix86_binary_operator_ok to insn constraint.
14821 (*iordi_1_bts): Ditto.
14822 (*xordi_1_btc): Ditto.
14823 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14824 Update corresponding peephole2 pattern.
14828 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14830 PR tree-optimization/81832
14831 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14832 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14834 2017-08-16 Marek Polacek <polacek@redhat.com>
14837 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14838 regarding size_zero_node.
14840 2017-08-16 Martin Liska <mliska@suse.cz>
14843 * config.gcc: Respect previously set extra_objs in case
14846 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14848 PR tree-optimization/81835
14849 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14850 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14851 not depend on the phi.
14853 2017-08-16 Alan Modra <amodra@gmail.com>
14855 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14858 2017-08-16 Alan Modra <amodra@gmail.com>
14860 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14861 (save_reg_p): ..into this. Update all callers.
14862 (first_reg_to_save): Simplify.
14864 2017-08-16 Alan Modra <amodra@gmail.com>
14866 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14869 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14873 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14874 constant count if that count is less than 32.
14876 2017-08-15 Nathan Sidwell <nathan@acm.org>
14878 * gcc.c (execute): Emit friendlier message if inferior is killed
14879 by an external cause.
14881 2017-08-15 Richard Biener <rguenther@suse.de>
14883 PR tree-optimization/81790
14884 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14885 CONSTRUCTORs from simplifying and VN.
14887 2017-08-14 Martin Sebor <msebor@redhat.com>
14889 * builtin-attrs.def: Add comments.
14891 2017-08-14 Martin Sebor <msebor@redhat.com>
14894 * doc/extend.texi (attribute nonstring): Document new attribute.
14896 2017-08-14 Martin Sebor <msebor@redhat.com>
14899 * tree-diagnostic.c (default_tree_printer): Handle %G.
14900 * gimple-pretty-print.h (percent_G_format): Declare new function.
14901 * gimple-pretty-print.c (percent_G_format): Define.
14902 * tree-pretty-print.c (percent_K_format): Add argument.
14904 2017-08-14 Martin Sebor <msebor@redhat.com>
14906 PR translation/79998
14907 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14908 Remove a stray space.
14910 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14913 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14914 (*iordi_1_bts): Ditto.
14915 (*xordi_1_btc): Ditto.
14917 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14920 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14923 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14924 quoted strings, and make more translator-friendly.
14925 (darwin_rs6000_override_options): Likewise.
14926 (rs6000_option_override_internal): Likewise.
14927 (rs6000_return_in_memory): Fix overlong line.
14928 (init_cmulative_args): Use quoted strings, and make more
14929 translator-friendly.
14930 (rs6000_pass_by_reference): Fix overlong line.
14931 (def_builtin): Use quoted strings.
14932 (altivec_expand_predicate_builtin): Use quoted strings, and make
14933 more translator-friendly.
14934 (htm_expand_builtin): Use quoted strings.
14935 (cpu_expand_builtin): Use quoted strings, and make more
14936 translator-friendly.
14937 (altivec_expand_builtin): Likewise.
14938 (paired_expand_predicate_builtin): Likewise.
14939 (rs6000_invalid_builtin): Likewise.
14940 (builtin_function_type): Use quoted strings.
14941 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14942 more translator-friendly.
14943 (rs6000_trampoline_init): Likewise.
14944 (rs6000_handle_altivec_attribute): Likewise.
14945 (rs6000_inner_target_options): Use quoted strings.
14946 (rs6000_disable_incompatible_switches): Likewise.
14947 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14948 strings, and make more translator-friendly.
14949 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14951 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14953 PR tree-optimization/81799
14954 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14955 cond_expr to simple gimple operand.
14957 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14959 PR middle-end/46932
14960 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14962 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14966 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14967 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14968 TARGET_GASISR_PROLOGUES.
14969 * config/avr/avr.c (avr_option_override): Same.
14970 (avr_pass_pre_proep::execute): Same.
14972 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14975 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14976 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14978 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14980 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14981 define_insn_and_split. Split to a memory load from 0 in
14982 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14983 using PTR mode iterator.
14984 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14985 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14986 (*add_tp_<mode>): Redefine as define_insn_and_split.
14987 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14988 address space. Merge with *add_tp_x32 using PTR mode iterator.
14989 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14990 Split to an add with a memory load from 0 in
14991 DEFAULT_TLS_SEG_REG address space.
14993 2017-08-12 Andrew Pinski <apinski@cavium.com>
14995 * config/aarch64/aarch64-option-extensions.def (rdma):
14996 Fix feature string to what Linux prints out in /proc/cpuinfo.
14998 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
15001 * dwarf2out.c (modified_type_die): For C typedef types that have
15002 an ultimate origin, process the ultimate origin instead of the
15004 (gen_typedef_die): Assert that input DECLs have no ultimate
15006 (gen_type_die_with_usage): For typedef variants that have an
15007 ultimate origin, just call gen_decl_die on the original DECL.
15008 (process_scope_var): Avoid creating DIEs for local typedefs and
15009 concrete static variables.
15011 2017-08-12 Alan Modra <amodra@gmail.com>
15015 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
15016 match gnu-user.h startfile.
15017 (ENDFILE_LINUX_SPEC): Similarly.
15019 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
15022 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
15024 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
15026 2017-08-11 Tamar Christina <tamar.christina@arm.com>
15027 * config/aarch64/aarch64.md (mov<mode>): Change.
15028 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15029 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
15030 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
15032 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
15034 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
15035 for storage order barriers.
15037 2017-08-11 Martin Liska <mliska@suse.cz>
15040 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15041 variables of void type.
15043 2017-08-11 Martin Liska <mliska@suse.cz>
15045 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15046 TARGET_SUPPORTS_ALIASES.
15047 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15048 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15049 (optimize_weakref): Likewise.
15050 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15051 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15052 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15054 2017-08-11 Martin Liska <mliska@suse.cz>
15057 * config/i386/i386.c (make_resolver_func): Do complete
15058 refactoring of the function.
15060 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15063 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15064 * config/i386/i386.c (ix86_stack_protect_guard): Use
15065 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15066 * doc/invoke.texi (x86 Options): Document
15067 -mstack-protector-guard-symbol= option.
15069 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15071 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15072 * config/i386/i386.c (ix86_split_stack_guard): New function.
15073 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15074 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15075 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15076 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15077 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15078 (split_stack_space_check): Call ix86_split_stack_guard.
15080 2017-08-10 Martin Sebor <msebor@redhat.com>
15082 * print-tree.c (print_node): Print location using the established
15084 Replace spaces with colons.
15085 (debug_raw, debug): Ditto.
15087 2017-08-10 Martin Sebor <msebor@redhat.com>
15090 * pretty-print.c (pp_format): Correct the handling of %s precision.
15092 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15095 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15097 (ix86_finalize_stack_frame_flags): This. Also clear
15098 frame_pointer_needed if -fno-omit-frame-pointer is used without
15100 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15101 with ix86_finalize_stack_frame_flags.
15102 (ix86_expand_epilogue): Likewise.
15103 (ix86_expand_split_stack_prologue): Likewise.
15104 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15106 2017-08-10 Martin Liska <mliska@suse.cz>
15109 * c-attribs.c (handle_target_attribute):
15110 Report warning for an empty string argument of target attribute.
15112 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15115 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15117 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15118 or DECL_NONLOCAL labels.
15119 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15120 or DECL_NONLOCAL labels here.
15122 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15124 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15125 to indicate when early gimple folding has been disabled.
15126 (rs6000_gimple_fold_builtin): Add debug content.
15127 (rs6000_invalid_builtin): Fix whitespace.
15128 (rs6000_expand_builtin): Fix whitespace.
15129 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15131 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15134 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15135 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15137 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15139 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15140 (qdf24xx): Likewise.
15141 * config/aarch64/aarch64.md: Include falkor.md.
15142 * config/aarch64/falkor.md: New.
15144 2017-08-09 Marek Polacek <polacek@redhat.com>
15147 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15148 * diagnostic.c (emit_diagnostic): Add a comment.
15149 (emit_diagnostic_valist): New function.
15151 2017-08-09 Marek Polacek <polacek@redhat.com>
15154 * input.c (make_location): New overload.
15155 * input.h (make_location): Declare.
15157 2017-08-08 Alan Modra <amodra@gmail.com>
15158 H.J. Lu <hongjiu.lu@intel.com>
15161 * gcc.c (NO_PIE_SPEC): Delete.
15162 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15164 (LINK_PIE_SPEC): ..to here.
15165 (LINK_COMMAND_SPEC): Support -no-pie.
15166 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15167 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15169 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15170 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15171 (ENDFILE_CRTEND_SPEC): Similarly.
15173 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15176 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15177 (mstack-protector-guard-offset=): Ditto.
15178 * config/i386/i386.c (ix86_option_override): Handle
15179 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15181 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15182 ix86_stack_protect_guard_offset variables.
15183 (TARGET_STACK_PROTECT_GUARD): Always define.
15184 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15185 and -mstack-protector-guard-offset= options.
15187 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15189 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15190 boundary case for the last candidate.
15192 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15194 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15195 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15197 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15199 PR middle-end/19706
15200 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15201 * config/aarch64/aarch64-builtins.c
15202 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15203 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15204 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15206 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15207 Andrew Pinski <pinskia@gmail.com>
15209 PR middle-end/19706
15210 * internal-fn.def (XORSIGN): New.
15211 * optabs.def (xorsign_optab): New.
15212 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15213 (convert_expand_mult_copysign): New.
15214 (pass_optimize_widening_mul::execute): Call
15215 convert_expand_mult_copysign.
15217 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15219 PR tree-optimization/81354
15220 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15221 Insert on edges rather than explicitly creating landing pads.
15222 (analyze_candidates_and_replace): Commit edge inserts.
15224 2017-08-08 Richard Biener <rguenther@suse.de>
15226 PR middle-end/81719
15227 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15228 (expand_simple_operations): Also look through ADDR_EXPRs with
15229 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15231 2017-08-08 Richard Biener <rguenther@suse.de>
15233 PR tree-optimization/81723
15234 * tree-vect-slp.c (struct bst_traits): New hash traits.
15235 (bst_fail): New global.
15236 (vect_build_slp_tree_2): New worker, split out from ...
15237 (vect_build_slp_tree): ... this now wrapping it with using
15238 bst_fail set to cache SLP tree build fails. Properly handle
15240 (vect_analyze_slp_instance): Allocate and free bst_fail.
15242 2017-08-08 Martin Liska <mliska@suse.cz>
15245 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15246 LABEL_DECLs that can be from a different function.
15248 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15250 PR tree-optimization/81744
15251 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15252 loop's number of iterations.
15254 2017-08-08 Martin Liska <mliska@suse.cz>
15256 * asan.c: Include header files.
15257 * attribs.c (build_decl_attribute_variant): New function moved
15259 (build_type_attribute_qual_variant): Likewise.
15260 (cmp_attrib_identifiers): Likewise.
15261 (simple_cst_list_equal): Likewise.
15262 (omp_declare_simd_clauses_equal): Likewise.
15263 (attribute_value_equal): Likewise.
15264 (comp_type_attributes): Likewise.
15265 (build_type_attribute_variant): Likewise.
15266 (lookup_ident_attribute): Likewise.
15267 (remove_attribute): Likewise.
15268 (merge_attributes): Likewise.
15269 (merge_type_attributes): Likewise.
15270 (merge_decl_attributes): Likewise.
15271 (merge_dllimport_decl_attributes): Likewise.
15272 (handle_dll_attribute): Likewise.
15273 (attribute_list_equal): Likewise.
15274 (attribute_list_contained): Likewise.
15275 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15276 (lookup_attribute_by_prefix): Likewise.
15277 * bb-reorder.c: Include header files.
15278 * builtins.c: Likewise.
15279 * calls.c: Likewise.
15280 * cfgexpand.c: Likewise.
15281 * cgraph.c: Likewise.
15282 * cgraphunit.c: Likewise.
15283 * convert.c: Likewise.
15284 * dwarf2out.c: Likewise.
15285 * final.c: Likewise.
15286 * fold-const.c: Likewise.
15287 * function.c: Likewise.
15288 * gimple-expr.c: Likewise.
15289 * gimple-fold.c: Likewise.
15290 * gimple-pretty-print.c: Likewise.
15291 * gimple.c: Likewise.
15292 * gimplify.c: Likewise.
15293 * hsa-common.c: Likewise.
15294 * hsa-gen.c: Likewise.
15295 * internal-fn.c: Likewise.
15296 * ipa-chkp.c: Likewise.
15297 * ipa-cp.c: Likewise.
15298 * ipa-devirt.c: Likewise.
15299 * ipa-fnsummary.c: Likewise.
15300 * ipa-inline.c: Likewise.
15301 * ipa-visibility.c: Likewise.
15303 * lto-cgraph.c: Likewise.
15304 * omp-expand.c: Likewise.
15305 * omp-general.c: Likewise.
15306 * omp-low.c: Likewise.
15307 * omp-offload.c: Likewise.
15308 * omp-simd-clone.c: Likewise.
15309 * opts-global.c: Likewise.
15310 * passes.c: Likewise.
15311 * predict.c: Likewise.
15312 * sancov.c: Likewise.
15313 * sanopt.c: Likewise.
15314 * symtab.c: Likewise.
15315 * toplev.c: Likewise.
15316 * trans-mem.c: Likewise.
15317 * tree-chkp.c: Likewise.
15318 * tree-eh.c: Likewise.
15319 * tree-into-ssa.c: Likewise.
15320 * tree-object-size.c: Likewise.
15321 * tree-parloops.c: Likewise.
15322 * tree-profile.c: Likewise.
15323 * tree-ssa-ccp.c: Likewise.
15324 * tree-ssa-live.c: Likewise.
15325 * tree-ssa-loop.c: Likewise.
15326 * tree-ssa-sccvn.c: Likewise.
15327 * tree-ssa-structalias.c: Likewise.
15328 * tree-ssa.c: Likewise.
15329 * tree-streamer-in.c: Likewise.
15330 * tree-vectorizer.c: Likewise.
15331 * tree-vrp.c: Likewise.
15332 * tsan.c: Likewise.
15333 * ubsan.c: Likewise.
15334 * varasm.c: Likewise.
15335 * varpool.c: Likewise.
15336 * tree.c: Remove functions moved to attribs.[ch].
15337 * tree.h: Likewise.
15338 * config/aarch64/aarch64.c: Add attrs.h header file.
15339 * config/alpha/alpha.c: Likewise.
15340 * config/arc/arc.c: Likewise.
15341 * config/arm/arm.c: Likewise.
15342 * config/avr/avr.c: Likewise.
15343 * config/bfin/bfin.c: Likewise.
15344 * config/c6x/c6x.c: Likewise.
15345 * config/cr16/cr16.c: Likewise.
15346 * config/cris/cris.c: Likewise.
15347 * config/darwin.c: Likewise.
15348 * config/epiphany/epiphany.c: Likewise.
15349 * config/fr30/fr30.c: Likewise.
15350 * config/frv/frv.c: Likewise.
15351 * config/ft32/ft32.c: Likewise.
15352 * config/h8300/h8300.c: Likewise.
15353 * config/i386/winnt.c: Likewise.
15354 * config/ia64/ia64.c: Likewise.
15355 * config/iq2000/iq2000.c: Likewise.
15356 * config/lm32/lm32.c: Likewise.
15357 * config/m32c/m32c.c: Likewise.
15358 * config/m32r/m32r.c: Likewise.
15359 * config/m68k/m68k.c: Likewise.
15360 * config/mcore/mcore.c: Likewise.
15361 * config/microblaze/microblaze.c: Likewise.
15362 * config/mips/mips.c: Likewise.
15363 * config/mmix/mmix.c: Likewise.
15364 * config/mn10300/mn10300.c: Likewise.
15365 * config/moxie/moxie.c: Likewise.
15366 * config/msp430/msp430.c: Likewise.
15367 * config/nds32/nds32-isr.c: Likewise.
15368 * config/nds32/nds32.c: Likewise.
15369 * config/nios2/nios2.c: Likewise.
15370 * config/nvptx/nvptx.c: Likewise.
15371 * config/pa/pa.c: Likewise.
15372 * config/pdp11/pdp11.c: Likewise.
15373 * config/powerpcspe/powerpcspe.c: Likewise.
15374 * config/riscv/riscv.c: Likewise.
15375 * config/rl78/rl78.c: Likewise.
15376 * config/rx/rx.c: Likewise.
15377 * config/s390/s390.c: Likewise.
15378 * config/sh/sh.c: Likewise.
15379 * config/sol2.c: Likewise.
15380 * config/sparc/sparc.c: Likewise.
15381 * config/spu/spu.c: Likewise.
15382 * config/stormy16/stormy16.c: Likewise.
15383 * config/tilegx/tilegx.c: Likewise.
15384 * config/tilepro/tilepro.c: Likewise.
15385 * config/v850/v850.c: Likewise.
15386 * config/vax/vax.c: Likewise.
15387 * config/visium/visium.c: Likewise.
15388 * config/xtensa/xtensa.c: Likewise.
15390 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15393 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15394 constraints since the -mupper-regs-* switches have been
15396 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15397 into a vector from a double word element that was extracted from
15398 another vector, and eliminate extra XXPERMDI instructions.
15399 (vsx_concat_<mode>_2): Likewise.
15400 (vsx_concat_<mode>_3): Likewise.
15401 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15402 concat to allow optimizing inserts from previous extracts.
15404 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15406 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15407 memory reference to a SSP offset in TLS address space.
15408 (ix86_print_operand) <case '@'>: Remove.
15409 (ix86_print_operand_punct_valid_p): Remove '@' code.
15410 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15411 UNSPEC_SP_TLS_TEST.
15412 (stack_tls_protect_set_<mode>): Remove.
15413 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15414 (stack_tls_protect_test_<mode>): Remove.
15415 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15417 2017-08-07 Olivier Hainque <hainque@adacore.com>
15420 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15422 2017-08-07 Douglas Rupp <rupp@adacore.com>
15424 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15425 variable was referenced as multidir in command.
15427 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15430 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15433 2017-08-07 Martin Liska <mliska@suse.cz>
15435 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15436 * config/rl78/rl78.c: Add include of attribs.h.
15437 * config/sh/sh.c: Likewise.
15438 * config/v850/v850.c: Likewise.
15440 2017-08-07 Tom de Vries <tom@codesourcery.com>
15442 PR middle-end/78266
15443 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15445 2017-08-07 Martin Liska <mliska@suse.cz>
15447 * config/mips/mips.c: Include attribs.h.
15449 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15452 * doc/invoke.texi: Document change in behvaior for -Ofast for
15455 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15457 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15459 (aarch64_pop_regs): Likewise.
15460 (aarch64_gen_load_pair): Likewise.
15461 (aarch64_save_callee_saves): Likewise.
15462 (aarch64_restore_callee_saves): Likewise.
15464 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15466 * config/i386/i386.c: Revert the last change.
15468 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15471 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15473 (ix86_finalize_stack_frame_flags): This. Also clear
15474 frame_pointer_needed if -fno-omit-frame-pointer is used without
15476 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15477 with ix86_finalize_stack_frame_flags.
15478 (ix86_expand_epilogue): Likewise.
15479 (ix86_expand_split_stack_prologue): Likewise.
15481 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15484 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15485 to P_AES for Westmere.
15487 2017-08-07 Jonathan Yong <10walls@gmail.com>
15489 * config/i386/mingw.opt (fset-stack-executable): Removed.
15490 * config/i386/cygming.opt (fset-stack-executable): Moved
15492 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15494 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15496 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15498 2017-08-07 Marek Polacek <polacek@redhat.com>
15500 PR middle-end/81737
15501 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15503 2017-08-07 Martin Liska <mliska@suse.cz>
15505 * attribs.h (canonicalize_attr_name): New function.
15506 (cmp_attribs): Move from c-format.c and adjusted.
15507 (is_attribute_p): Moved from tree.h.
15508 * tree-inline.c: Add new includes.
15509 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15510 (private_is_attribute_p): Remove.
15511 (private_lookup_attribute): Likewise.
15512 (private_lookup_attribute_by_prefix): Simplify.
15513 (remove_attribute): Use is_attribute_p.
15514 * tree.h: Remove removed declarations.
15516 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15518 PR middle-end/81698
15519 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15520 instead of computing it in the function. Formatting fix.
15521 (expand_case): Don't rely on default_edge being the first edge,
15522 clear it if removing it, pass default_edge to
15523 emit_case_dispatch_table.
15524 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15527 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15529 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15530 insn in the function, emit NOP after the insn.
15532 2017-08-06 Tom de Vries <tom@codesourcery.com>
15534 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15537 2017-08-06 Tom de Vries <tom@codesourcery.com>
15539 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15542 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15544 PR tree-optimization/57371
15545 * match.pd: New pattern.
15547 2017-08-04 Marek Polacek <polacek@redhat.com>
15549 PR middle-end/81695
15550 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15551 perform the computation in offset_int.
15553 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15555 PR tree-optimization/81136
15556 * tree-vectorizer.h: Include tree-hash-traits.h.
15557 (vec_base_alignments): New typedef.
15558 (vec_info): Add a base_alignments field.
15559 (vect_record_base_alignments): Declare.
15560 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15562 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15563 (create_data_ref): Add an is_conditional_in_stmt argument.
15564 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15565 the is_conditional_in_stmt field.
15566 (data_ref_loc): Add an is_conditional_in_stmt field.
15567 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15568 (find_data_references_in_stmt): Update call to create_data_ref.
15569 (graphite_find_data_references_in_stmt): Likewise.
15570 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15571 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15572 (vect_record_base_alignment): New function.
15573 (vect_record_base_alignments): Likewise.
15574 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15575 for nested statements even if we fail to compute a misalignment.
15576 Use pooled base alignments for unconditional references.
15577 (vect_find_same_alignment_drs): Compare base addresses instead
15579 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15580 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15582 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15584 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15585 Add an explicit name for the enum. Use auto_vec for slp_instances
15586 and grouped_stores.
15587 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15589 (_bb_vec_info): Add a constructor and destructor.
15590 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15591 (destroy_loop_vec_info): Delete.
15592 (vect_destroy_datarefs): Likewise.
15593 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15594 (vec_info::vec_info): New function.
15595 (vec_info::~vec_info): Likewise.
15596 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15597 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15598 destroy_loop_vec_info.
15599 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15600 (_loop_vec_info::_loop_vec_info): ...this.
15601 (destroy_loop_vec_info): Replace with...
15602 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15603 the stmt_vec_infos. Leave handling of vec_info information to its
15604 destructor. Remove explicit vector releases.
15605 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15606 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15607 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15608 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15609 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15610 (destroy_bb_vec_info): Replace with...
15611 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15612 information to its destructor.
15613 (vect_slp_analyze_bb_1): Use new and delete instead of
15614 new_bb_vec_info and destroy_bb_vec_info.
15615 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15618 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15620 * tree-data-ref.h (subscript): Add access_fn field.
15621 (data_dependence_relation): Add could_be_independent_p.
15622 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15623 (same_access_functions): Move to tree-data-ref.c.
15624 * tree-data-ref.c (ref_contains_union_access_p): New function.
15625 (access_fn_component_p): Likewise.
15626 (access_fn_components_comparable_p): Likewise.
15627 (dr_analyze_indices): Add a reference to access_fn_component_p.
15628 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15630 (constant_access_functions): Likewise.
15631 (add_other_self_distances): Likewise.
15632 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15633 (initialize_data_dependence_relation): Use XCNEW and remove
15634 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15635 of access functions that have the same type. Allow the
15636 subsequence to end with different bases in some circumstances.
15637 Record the chosen access functions in SUB_ACCESS_FN.
15638 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15639 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15640 (subscript_dependence_tester_1): Likewise dra and drb.
15641 (build_classic_dist_vector): Update calls accordingly.
15642 (subscript_dependence_tester): Likewise.
15643 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15644 DDR_COULD_BE_INDEPENDENT_P.
15645 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15646 comp_alias_ddrs instead of may_alias_ddrs.
15647 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15649 (vect_analyze_data_ref_dependence): Use it if
15650 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15651 distance vectors if that fails.
15652 (dependence_distance_ge_vf): New function.
15653 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15654 LOOP_VINFO_MAY_ALIAS_DDRS.
15656 2017-08-04 Richard Biener <rguenther@suse.de>
15658 PR middle-end/81705
15659 * fold-const.c (fold_binary_loc): Properly restrict
15660 minus_var0 && minus_var1 case when associating undefined overflow
15663 2017-08-04 Tom de Vries <tom@codesourcery.com>
15665 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15667 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15669 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15670 Don't start diagnostic messages with a capital letter.
15671 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15673 (rs6000_invalid_builtin): Likewise.
15674 (rs6000_trampoline_init): Likewise.
15676 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15679 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15680 after setting changeable df flags.
15682 2017-08-03 Richard Biener <rguenther@suse.de>
15684 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15685 up if the use is in USE - X.
15687 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15689 * toplev.c (dumpfile.h): New include.
15690 (internal_error_reentered): New static function. Use it...
15691 (internal_error_function): ...here to handle reentered internal_error.
15693 2017-08-03 Richard Biener <rguenther@suse.de>
15695 PR middle-end/81148
15696 * fold-const.c (split_tree): Add minus_var and minus_con
15697 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15698 here but always use minus_*.
15699 (associate_trees): Assert we never associate with MINUS_EXPR
15700 and NULL first operand. Do not recurse for PLUS_EXPR operands
15701 when associating as MINUS_EXPR either.
15702 (fold_binary_loc): Track minus_var and minus_con.
15704 2017-08-03 Tom de Vries <tom@codesourcery.com>
15707 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15708 ACCEL_COMPILER, apply finish_options on
15709 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15711 2017-08-03 Tom de Vries <tom@codesourcery.com>
15714 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15715 function_entry_patch_area_size > 0.
15717 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15720 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15721 instead of 10??LU, perform unit multiplication in wide_int,
15722 don't change alloc_object_size_limit if the limit is larger
15725 PR tree-optimization/81655
15726 PR tree-optimization/81588
15727 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15728 the case when ranges[i].low and high are 1 for unsigned type with
15731 PR middle-end/81052
15732 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15733 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15735 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15737 * tree-vrp.h: Add include guard.
15739 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15742 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15743 (ud2): New insn pattern.
15744 * config/i386/i386.c (ix86_expand_epilogue):
15745 For naked functions, generate ud2 instead of trap insn.
15747 2017-08-02 Marek Polacek <polacek@redhat.com>
15750 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15752 2017-08-02 Tom de Vries <tom@codesourcery.com>
15753 Cesar Philippidis <cesar@codesourcery.com>
15755 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15756 Add missing edge probabilities.
15758 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15760 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15761 Correct endianness.
15763 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15765 PR middle-end/79499
15766 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15767 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15768 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15769 prologue_seq sequences - if any.
15771 2017-08-02 Richard Biener <rguenther@suse.de>
15773 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15774 via vectors if supported, integer extracts via punning if supported
15775 or otherwise vector extracts.
15777 2017-08-02 Richard Biener <rguenther@suse.de>
15779 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15781 (bitmap_insert_into_set): ... this.
15783 2017-08-02 Richard Biener <rguenther@suse.de>
15785 PR tree-optimization/81633
15787 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15789 PR tree-optimization/71752
15790 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15792 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15794 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15795 (machine_function::call_ms2sysv_pad_out): Remove field.
15796 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15797 (ix86_compute_frame_layout): Likewise.
15799 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15802 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15803 attribute with interrupt attribute.
15805 2017-08-01 Andrew Pinski <apinski@cavium.com>
15807 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15808 BIT_INSERT_EXPR's operand 1
15809 to see if the types precision matches.
15811 2017-08-01 Martin Liska <mliska@suse.cz>
15813 PR middle-end/70140
15814 * builtins.c (expand_builtin_memcpy_args): Remove.
15815 (expand_builtin_memcpy): Call newly added function
15816 expand_builtin_memory_copy_args.
15817 (expand_builtin_memcpy_with_bounds): Likewise.
15818 (expand_builtin_mempcpy): Remove last argument.
15819 (expand_builtin_mempcpy_with_bounds): Likewise.
15820 (expand_builtin_memory_copy_args): New function created from
15821 expand_builtin_mempcpy_args with small modifications.
15822 (expand_builtin_mempcpy_args): Remove.
15823 (expand_builtin_stpcpy): Remove unused argument.
15824 (expand_builtin): Likewise.
15825 (expand_builtin_with_bounds): Likewise.
15827 2017-08-01 Martin Liska <mliska@suse.cz>
15830 Make mempcpy more optimal (PR middle-end/70140).
15832 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15835 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15836 __builtin_vec_cmpne verify both arguments are compatible vectors
15837 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15838 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15839 move computation of aligned to after checking the argument types.
15843 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15844 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15847 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15848 Doug Rupp <rupp@adacore.com>
15849 Olivier Hainque <hainque@adacore.com>
15851 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15852 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15854 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15855 for TARGET_OS_CPP_BUILTIN.
15856 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15857 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15859 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15860 passing required abi options to the assembler for EABI configurations.
15861 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15862 of .text.hot and .text.unlikely sections for kernel modules when
15863 using ARM style exceptions.
15864 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15865 options. Add EXTRA_CC1_SPEC.
15866 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15868 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15870 (ARM_TARGET2_DWARF_FORMAT): Define.
15871 * config/arm/t-vxworks: Adjust multilib control to removal of the
15872 Diab command line options.
15874 2017-08-01 Martin Liska <mliska@suse.cz>
15876 PR gcov-profile/81561
15877 * gcov.c (unblock): Make unblocking safe as we need to preserve
15878 index correspondence of blocks and block_lists.
15880 2017-08-01 Richard Biener <rguenther@suse.de>
15882 PR tree-optimization/81181
15883 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15884 (compute_antic): ... end of iteration here.
15886 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15888 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15889 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15890 (ftree-slp-vectorize): Likewise.
15891 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15892 can no longer be set independent of flag_tree_loop_vectorize.
15893 * omp-general.c (emp_max_vf): Likewise.
15894 * opts.c (enable_fdo_optimizations): Remove references to
15895 flag_tree_vectorize, these are now implicit.
15896 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15897 and leave it for the options machinery.
15899 2017-08-01 Martin Liska <mliska@suse.cz>
15901 PR middle-end/70140
15902 * builtins.c (expand_builtin_memcpy_args): Remove.
15903 (expand_builtin_memcpy): Call newly added function
15904 expand_builtin_memory_copy_args.
15905 (expand_builtin_memcpy_with_bounds): Likewise.
15906 (expand_builtin_mempcpy): Remove last argument.
15907 (expand_builtin_mempcpy_with_bounds): Likewise.
15908 (expand_builtin_memory_copy_args): New function created from
15909 expand_builtin_mempcpy_args with small modifications.
15910 (expand_builtin_mempcpy_args): Remove.
15911 (expand_builtin_stpcpy): Remove unused argument.
15912 (expand_builtin): Likewise.
15913 (expand_builtin_with_bounds): Likewise.
15915 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15918 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15919 print "ds:" only for immediates in generic address space.
15921 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15924 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15925 (ix86_function_ok_for_sibcall): Return false for naked functions.
15927 2017-08-01 Richard Biener <rguenther@suse.de>
15929 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15930 (compute_antic): Seed worklist with exit block predecessors.
15931 * cfganal.c (dfs_find_deadend): For a cycle return the source
15932 of the edge closing it.
15934 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15936 * config/aarch64/aarch64.c
15937 (aarch64_can_const_movi_rtx_p): Move 0 check.
15939 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15941 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15942 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15944 * match.pd: Ditto in address comparison pattern.
15946 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15948 PR tree-optimization/81627
15949 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15950 closed ssa form for store-store chain.
15952 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15954 PR tree-optimization/81620
15955 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15956 for store-store chain.
15958 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15960 PR tree-optimization/81588
15961 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15962 ranges[i].in_p, invert comparison code ccode. For >/>=,
15963 swap rhs1 and rhs2 and comparison code unconditionally,
15964 for </<= don't do that. Don't swap rhs1/rhs2 again if
15965 ranges[i].in_p, instead invert comparison code ccode if
15966 opcode or oe->rank is BIT_IOR_EXPR.
15969 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15970 a direct optab to conversion optab.
15971 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15972 with GET_MODE_INNER as last argument instead of optab_handler.
15973 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15974 vector extraction if possible and optab is available.
15975 * expr.c (store_constructor): Use convert_optab_handler instead
15976 of optab_handler. Use vector initialization from smaller
15977 vectors if possible and optab is available.
15978 * tree-vect-stmts.c (vectorizable_load): Likewise.
15979 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15980 now have two modes.
15981 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15982 of vec_init from half-sized vectors with the same element mode.
15983 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15984 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15985 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15986 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15987 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15988 after mode in gen_vec_extract* calls.
15989 (vec_extract<mode>): Renamed to ...
15990 (vec_extract<mode><ssescalarmodelower>): ... this.
15991 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15992 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15993 element mode after mode in gen_vec_init* calls.
15994 (VEC_INIT_HALF_MODE): New mode iterator.
15995 (vec_init<mode>): Renamed to ...
15996 (vec_init<mode><ssescalarmodelower>): ... this.
15997 (vec_init<mode><ssehalfvecmodelower>): New expander.
15998 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15999 (vec_extractv2sfsf): ... this.
16000 (vec_initv2sf): Renamed to ...
16001 (vec_initv2sfsf): ... this.
16002 (vec_extractv2si): Renamed to ...
16003 (vec_extractv2sisi): ... this.
16004 (vec_initv2si): Renamed to ...
16005 (vec_initv2sisi): ... this.
16006 (vec_extractv4hi): Renamed to ...
16007 (vec_extractv4hihi): ... this.
16008 (vec_initv4hi): Renamed to ...
16009 (vec_initv4hihi): ... this.
16010 (vec_extractv8qi): Renamed to ...
16011 (vec_extractv8qiqi): ... this.
16012 (vec_initv8qi): Renamed to ...
16013 (vec_initv8qiqi): ... this.
16014 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
16015 (vec_init<mode>): Renamed to ...
16016 (vec_init<mode><VEC_base_l>): ... this.
16017 (vec_extract<mode>): Renamed to ...
16018 (vec_extract<mode><VEC_base_l>): ... this.
16019 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
16020 (vec_initv2sfsf): ... this.
16021 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
16022 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16023 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
16024 element mode after mode in gen_vec_init* calls.
16025 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
16026 (vec_init<mode><Vel>): ... this.
16027 (vec_extract<mode>): Renamed to ...
16028 (vec_extract<mode><Vel>): ... this.
16029 * config/aarch64/iterators.md (Vel): New mode attribute.
16030 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
16031 Add element mode after mode in gen_vec_extract* calls.
16032 * config/s390/vector.md (non_vec_l): New mode attribute.
16033 (vec_extract<mode>): Renamed to ...
16034 (vec_extract<mode><non_vec_l>): ... this.
16035 (vec_init<mode>): Renamed to ...
16036 (vec_init<mode><non_vec_l>): ... this.
16037 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16038 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16040 * config/arm/iterators.md (V_elem_l): New mode attribute.
16041 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16042 (vec_extract<mode><V_elem_l>): ... this.
16043 (vec_extractv2di): Renamed to ...
16044 (vec_extractv2didi): ... this.
16045 (vec_init<mode>): Renamed to ...
16046 (vec_init<mode><V_elem_l>): ... this.
16047 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16048 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16049 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16050 Add element mode after gen_vec_extract* calls.
16051 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16052 (vec_init<mode><unitmode>): ... this.
16053 (vec_extract<mode>): Renamed to ...
16054 (vec_extract<mode><unitmode>): ... this.
16055 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16056 (vec_init<mode><unitmode>): ... this.
16057 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16058 (vec_initv2sfsf): ... this.
16059 (vec_extractv2sf): Renamed to ...
16060 (vec_extractv2sfsf): ... this.
16061 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16062 Add element mode after gen_vec_extract* calls.
16063 * config/mips/mips.md (unitmode): New mode iterator.
16064 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16065 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16066 * config/spu/spu.md (inner_l): New mode attribute.
16067 (vec_init<mode>): Renamed to ...
16068 (vec_init<mode><inner_l>): ... this.
16069 (vec_extract<mode>): Renamed to ...
16070 (vec_extract<mode><inner_l>): ... this.
16071 * config/sparc/sparc.md (veltmode): New mode iterator.
16072 (vec_init<VMALL:mode>): Renamed to ...
16073 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16074 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16075 (vec_initv2sisi): ... this.
16076 (vec_initv2sf): Renamed to ...
16077 (vec_initv2sfsf): ... this.
16078 (vec_extractv2sf): Renamed to ...
16079 (vec_extractv2sfsf): ... this.
16080 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16081 (vec_init<mode>): Renamed to ...
16082 (vec_init<mode><VEC_base_l>): ... this.
16083 (vec_extract<mode>): Renamed to ...
16084 (vec_extract<mode><VEC_base_l>): ... this.
16085 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16086 (vec_initv2sfsf): ... this.
16087 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16088 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16089 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16090 gen_vec_init* calls.
16092 2017-08-01 Richard Biener <rguenther@suse.de>
16094 PR tree-optimization/81297
16095 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16096 TREE_OVERFLOW from INTEGER_CSTs.
16098 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16100 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16102 2017-07-31 Carl Love <cel@us.ibm.com>
16104 * config/rs6000/rs6000-c: Add support for built-in functions
16105 vector signed char vec_xl_be (signed long long, signed char *);
16106 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16107 vector signed int vec_xl_be (signed long long, signed int *);
16108 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16109 vector signed long long vec_xl_be (signed long long, signed long long *);
16110 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16111 vector signed short vec_xl_be (signed long long, signed short *);
16112 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16113 vector double vec_xl_be (signed long long, double *);
16114 vector float vec_xl_be (signed long long, float *);
16115 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16116 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16117 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16119 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16120 (altivec_expand_builtin): Add switch statement to call
16121 altivec_expand_xl_be for each builtin.
16122 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16123 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16124 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16125 __builtin_vsx_le_be_v16qi.
16126 * doc/extend.texi: Update the built-in documentation file for the
16127 new built-in functions.
16129 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16132 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16134 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16136 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16138 * config.gcc: Add z14.
16139 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16140 CPU model numbers for z13s and z14.
16141 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16143 * config/s390/s390-opts.h (enum processor_type): Rename
16144 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16145 * config/s390/s390.c (processor_table): Add field for CPU name to
16146 be passed to Binutils.
16147 (s390_asm_output_machine_for_arch): Use the new field in
16148 processor_table for Binutils.
16149 (s390_expand_builtin): Replace arch12 with z14.
16150 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16151 (s390_get_sched_attrmask): Likewise.
16152 (s390_get_unit_mask): Likewise.
16153 * config/s390/s390.opt: Add z14 to processor_type enum.
16155 2017-07-31 Martin Jambor <mjambor@suse.cz>
16158 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16159 regardless of optimization level.
16161 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16162 Martin Liska <mliska@suse.cz>
16164 * predict.def: Remove old comment and adjust probability.
16165 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16166 PREDICT statements.
16168 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16171 * config/i386/i386.c (ix86_function_naked): New function.
16172 (ix86_can_use_return_insn_p): Return false for naked functions.
16173 (ix86_expand_prologue): Skip prologue for naked functions.
16174 (ix86_expand_epilogue): Skip epilogue for naked functions
16175 and emit trap instruction.
16176 (ix86_warn_func_return): New function.
16177 (ix86_attribute_table): Add "naked" attribute specification.
16178 (TARGET_WARN_FUNC_RETURN): Define.
16179 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16181 2017-07-31 Martin Liska <mliska@suse.cz>
16183 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16184 (dump_gimple_bb_header): Always dump BB info.
16185 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16187 2017-07-31 Martin Liska <mliska@suse.cz>
16190 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16191 also with current_function_decl non-null equality.
16193 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16196 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16197 change type to the element type, instead add eltype variable and
16198 use it where we are interested in the element type.
16200 PR tree-optimization/81603
16201 * ipa-polymorphic-call.c
16202 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16203 offset arithmetic in offset_int, bail out if the resulting bit offset
16204 doesn't fit into shwi.
16206 2017-07-31 Martin Liska <mliska@suse.cz>
16208 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16209 (gimplify_save_expr): Fix comment.
16211 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16214 * config/i386/i386.c (ix86_function_arg): Update arguments for
16216 (ix86_compute_frame_layout): Set the initial stack offset to
16217 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16218 INCOMING_FRAME_SP_OFFSET.
16219 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16220 stack before exception handler returns.
16221 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16222 the 'ERROR_CODE' for exception handler.
16224 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16226 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16227 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16228 (ASM_OUTPUT_REG_POP): Ditto.
16229 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16230 instead of asm_fprintf to output pure string.
16232 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16234 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16235 to imported_module_or_decl hook.
16236 (debug_nothing_tree_tree_tree_bool): Remove.
16237 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16238 * debug.c (do_nothing_debug_hooks): Use
16239 debug_nothing_tree_tree_tree_bool_bool instead of
16240 debug_nothing_tree_tree_tree_bool.
16241 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16242 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16243 * sdbout.c (sdb_debug_hooks): Likewise.
16244 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16245 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16247 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16248 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16249 attribute, don't add anything.
16251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16253 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16254 (fold_build2_stat_loc): Likewise.
16255 (fold_build3_stat_loc): Likewise.
16256 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16257 (fold_build1_loc): Remove macro.
16258 (fold_build2_loc): Likewise.
16259 (fold_build3_loc): Likewise.
16261 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16263 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16264 (gimple_build_debug_bind_source_stat): Likewise.
16265 * gimple.h (gimple_build_debug_bind): Remove macro.
16266 (gimple_build_debug_bind_source): Likewise.
16268 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16270 * bitmap.c (bitmap_alloc): Adjust.
16271 (bitmap_gc_alloc): Likewise.
16272 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16274 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16276 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16277 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16278 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16279 (bitmap_gc_alloc_stat): Likewise.
16280 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16282 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16284 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16285 * rtl.h (shallow_copy_rtx): Remove macro.
16287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16289 * emit-rtl.c (gen_raw_REG): Adjust.
16290 * gengenrtl.c (gendef): Likewise.
16291 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16292 * rtl.h (rtx_alloc): Remove macro.
16294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16296 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16297 (build_tree_list_stat): Likewise.
16298 * tree.h (build_tree_list): Remove macro.
16299 (build_tree_list_vec): Likewise.
16301 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16303 * tree.c (make_vector_stat): Remove _stat from name.
16304 (build_vector_stat): Likewise.
16305 * tree.h (make_vector_stat): Remove macro.
16306 (build_vector_stat): Likewise.
16308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16310 * tree.h (build_var_debug_value): Remove prototype.
16312 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16314 * tree.c (tree_cons_stat): Remove _stat from name.
16315 * tree.h (tree_cons): Remove macro.
16317 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16319 * tree.c (build_vl_exp_stat): Remove _stat from name.
16320 * tree.h (build_vl_exp): Remove macro.
16322 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16324 * tree.c (build_decl_stat): Remove _stat from name.
16325 * tree.h (build_decl): Remove macro.
16327 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16329 * gimple.c (gimple_build_with_ops_stat): Adjust.
16330 (gimple_alloc_stat): Remove _stat from name.
16331 * gimple.h (gimple_alloc): Remove macro.
16333 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16335 * tree.c (make_tree_vec_stat): Remove _stat from name.
16336 (grow_tree_vec_stat): Likewise.
16337 * tree.h (make_tree_vec_stat): Adjust prototype.
16338 (grow_tree_vec_stat): Likewise.
16339 (make_tree_vec): Remove macro.
16340 (grow_tree_vec): Likewise.
16342 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16344 * fold-const.c (fold_build1_stat_loc): Adjust.
16345 (fold_build2_stat_loc): Likewise.
16346 (fold_build3_stat_loc): Likewise.
16347 * tree.c (build0_stat): Remove _stat from name.
16348 (build1_stat): Likewise.
16349 (build2_stat): Likewise.
16350 (build3_stat): Likewise.
16351 (build4_stat): Likewise.
16352 (build5_stat): Likewise.
16353 * tree.h (build1_loc): Remove macro, and rename _stat function
16355 (build2_loc): Likewise.
16356 (build3_loc): Likewise.
16357 (build4_loc): Likewise.
16358 (build5_loc): Likewise.
16360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16362 * tree.c (make_int_cst_stat): Remove _stat from name.
16363 * tree.h (make_int_cst_stat): Adjust prototype.
16364 (make_int_cst): Remove macro.
16366 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16368 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16369 * tree.h (make_tree_binfo_stat): Adjust prototype.
16370 (make_tree_binfo): Remove.
16372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16374 * tree.c (copy_node_stat): Rename to copy_node.
16375 (build_distinct_type_copy): Adjust.
16376 * tree.h (copy_node_stat): Adjust prototype.
16377 (copy_node): Remove macro.
16379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16381 * tree.c (make_node_stat): rename to make_node.
16382 (build_tree_list_stat): Adjust.
16383 (build0_stat): Likewise.
16384 (build2_stat): Likewise.
16385 (build3_stat): Likewise.
16386 (build4_stat): Likewise.
16387 (build5_stat): Likewise.
16388 (build_decl_stat): Likewise.
16389 * tree.h (make_node_stat): Adjust prototype.
16390 (make_node): remove macro.
16392 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16394 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16395 (PPC_FEATURE2_SCV): Likewise.
16396 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16398 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16400 * config/aarch64/aarch64.c
16401 (aarch64_internal_mov_immediate): Add new special pattern.
16402 * config/aarch64/aarch64.md (*movdi_aarch64):
16403 Add reg/32bit const mov case.
16405 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16406 Richard Sandiford <richard.sandiford@linaro.org>
16408 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16409 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16410 Add integer and movi cases.
16411 (movi-split-hf-df-sf split, fp16): New.
16412 (enabled): Added TARGET_FP_F16INST.
16413 * config/aarch64/iterators.md (GPF_HF): New.
16414 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16416 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16418 * config/aarch64/aarch64.c
16419 (aarch64_simd_container_mode): Add prototype.
16420 (aarch64_expand_mov_immediate): Add HI support.
16421 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16422 (aarch64_can_const_movi_rtx_p): New.
16423 (aarch64_preferred_reload_class):
16424 Remove restrictions of using FP registers for certain SIMD operations.
16425 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16426 (aarch64_valid_floating_const): Add integer move validation.
16427 (aarch64_simd_imm_scalar_p): Remove.
16428 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16429 (aarch64_legitimate_constant_p): Expand list of supported cases.
16430 * config/aarch64/aarch64-protos.h
16431 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16432 (aarch64_reinterpret_float_as_int): New.
16433 (aarch64_simd_imm_scalar_p): Remove.
16434 * config/aarch64/constraints.md (Uvi): New.
16435 (Dd): Split into Ds and new Dd.
16436 * config/aarch64/aarch64.md (*movsi_aarch64):
16438 (*movdi_aarch64): Add SIMD mov case.
16440 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16442 * tree-predcom.c: (struct chain): Handle store-store chain in which
16443 stores for elimination only store loop invariant values.
16444 (execute_pred_commoning_chain): Ditto.
16445 (prepare_initializers_chain_store_elim): Ditto.
16446 (prepare_finalizers): Ditto.
16447 (is_inv_store_elimination_chain): New function.
16448 (initialize_root_vars_store_elim_1): New function.
16450 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16452 * tree-predcom.c: Revise general description of the pass.
16453 (enum chain_type): New enum type for store elimination.
16454 (struct chain): New field supporting store elimination.
16455 (struct component): Ditto.
16456 (dump_chain): Dump store-stores chain.
16457 (release_chain): Release resources.
16458 (split_data_refs_to_components): Compute and create component
16459 contains only stores for elimination.
16460 (get_chain_last_ref_at): New function.
16461 (make_invariant_chain): Initialization.
16462 (make_rooted_chain): Specify chain type in parameter and record it.
16463 (add_looparound_copies): Skip for store-stores chain.
16464 (determine_roots_comp): Compute type of chain and pass it to
16466 (initialize_root_vars_store_elim_2): New function.
16467 (finalize_eliminated_stores): New function.
16468 (remove_stmt): Handle store for elimination.
16469 (execute_pred_commoning_chain): Execute predictive commoning on
16470 store-store chains.
16471 (determine_unroll_factor): Skip unroll for store-stores chain.
16472 (prepare_initializers_chain_store_elim): New function.
16473 (prepare_initializers_chain): Hanlde store-store chain.
16474 (prepare_finalizers_chain, prepare_finalizers): New function.
16475 (tree_predictive_commoning_loop): Return integer value indicating
16476 if loop is unrolled or lcssa form is corrupted.
16477 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16479 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16481 * tree-predcom.c (initialize_root): Delete.
16482 (execute_pred_commoning_chain): Initialize root vars and replace
16483 reference of non-combined chain directly, rather than call above
16486 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16488 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16489 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16491 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16493 * tree-predcom.c (struct chain): New field init_seq.
16494 (release_chain): Release init_seq.
16495 (prepare_initializers_chain): Record intialization stmts in above
16497 (insert_init_seqs): New function.
16498 (tree_predictive_commoning_loop): Call insert_init_seqs.
16500 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16502 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16504 2017-07-28 Richard Biener <rguenther@suse.de>
16506 * match.pd: Remove superfluous :c.
16507 * genmatch.c (simplify::id): Add member.
16508 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16510 (current_id): New global.
16511 (dt_node::parent): Move from ...
16512 (dt_operand::parent): ... here. Add for_id member.
16513 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16514 (decision_tree::find_node): Relax order requirement when
16515 merging DT_TRUE nodes to ones inbetween the current simplify
16516 and the one we try to merge with. Add diagnostic whenever
16517 we need to enforce pattern order by not merging.
16518 (decision_tree::insert): Set current_id.
16519 (decision_tree::print_node): Dump parent node and for_id.
16520 (parser::last_id): Add member.
16521 (parser::push_simplify): Assign unique id.
16522 (parser::parser): Initialize last_id.
16524 2017-07-28 Martin Liska <mliska@suse.cz>
16527 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16528 gimple_build_debug_bind.
16530 2017-07-28 Richard Biener <rguenther@suse.de>
16532 PR tree-optimization/81502
16533 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16535 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16537 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16538 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16539 for BIT_FIELD_REF args.
16540 * fold-const.c (make_bit_field_ref): Likewise.
16541 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16543 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16546 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16547 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16548 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16549 Or it into SANITIZER_UNDEFINED.
16550 * ubsan.c: Include gimple-fold.h and varasm.h.
16551 (ubsan_expand_ptr_ifn): New function.
16552 (instrument_pointer_overflow): New function.
16553 (maybe_instrument_pointer_overflow): New function.
16554 (instrument_object_size): Formatting fix.
16555 (pass_ubsan::execute): Call instrument_pointer_overflow
16556 and maybe_instrument_pointer_overflow.
16557 * internal-fn.c (expand_UBSAN_PTR): New function.
16558 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16559 * sanitizer.def (__ubsan_handle_pointer_overflow,
16560 __ubsan_handle_pointer_overflow_abort): New builtins.
16561 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16562 * internal-fn.def (UBSAN_PTR): New internal function.
16563 * opts.c (sanitizer_opts): Add pointer-overflow.
16564 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16565 * fold-const.c (build_range_check): Compute pointer range check in
16566 integral type if pointer arithmetics would be needed. Formatting
16569 2017-07-28 Martin Liska <mliska@suse.cz>
16572 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16573 parameters that are of a variable-length.
16575 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16577 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16579 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16580 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16581 (CRT_CALL_STATIC_FUNCTION): Likewise.
16582 (ASM_DEFAULT_SPEC): New define.
16583 (ASM_SPEC32): Likewise.
16584 (ASM_SPEC64): Likewise.
16585 (ASM_SPEC_COMMON): Likewise.
16586 (ASM_SPEC): Likewise.
16587 (INVALID_64BIT): Likewise.
16588 (LINK_OS_DEFAULT_SPEC): Likewise.
16589 (LINK_OS_SPEC32): Likewise.
16590 (LINK_OS_SPEC64): Likewise.
16591 (POWERPC_LINUX): Likewise.
16592 (PTRDIFF_TYPE): Likewise.
16593 (RESTORE_FP_PREFIX): Likewise.
16594 (RESTORE_FP_SUFFIX): Likewise.
16595 (SAVE_FP_PREFIX): Likewise.
16596 (SAVE_FP_SUFFIX): Likewise.
16597 (SIZE_TYPE): Likewise.
16598 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16599 (TARGET_64BIT): Likewise.
16600 (TARGET_64BIT): Likewise.
16601 (TARGET_AIX): Likewise.
16602 (WCHAR_TYPE_SIZE): Likewise.
16603 (WCHAR_TYPE): Undefine.
16604 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16605 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16606 (CPP_OS_RTEMS_SPEC): Delete.
16607 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16608 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16610 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16612 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16614 PR tree-optimization/81578
16615 * tree-parloops.c (build_new_reduction): Bail out if
16616 reduction_code isn't one of the standard OpenMP reductions.
16617 Move the details printing after that decision.
16619 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16621 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16622 related to reload_in_progress.
16623 (splat_input_operand): Likewise.
16624 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16626 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16628 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16629 (TARGET_INSTANTIATE_DECLS): Likewise.
16630 (legitimate_indexed_address_p): Delete reload_in_progress code.
16631 (rs6000_debug_legitimate_address_p): Likewise.
16632 (rs6000_eliminate_indexed_memrefs): Likewise.
16633 (rs6000_emit_le_vsx_store): Likewise.
16634 (rs6000_emit_move_si_sf_subreg): Likewise.
16635 (rs6000_emit_move): Likewise.
16636 (register_to_reg_type): Likewise.
16637 (rs6000_pre_atomic_barrier): Likewise.
16638 (rs6000_machopic_legitimize_pic_address): Likewise.
16639 (rs6000_allocate_stack_temp): Likewise.
16640 (rs6000_address_for_fpconvert): Likewise.
16641 (rs6000_address_for_altivec): Likewise.
16642 (rs6000_secondary_memory_needed_rtx): Delete function.
16643 (rs6000_check_sdmode): Likewise.
16644 (rs6000_alloc_sdmode_stack_slot): Likewise.
16645 (rs6000_instantiate_decls): Likewise.
16646 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16647 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16648 Delete reload_in_progress.
16649 (*vec_reload_and_plus_<mptrsize>): Likewise.
16650 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16651 (vsx_div_v2di): Likewise.
16652 (vsx_udiv_v2di): Likewise.
16654 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16656 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16657 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16658 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16659 (rs6000_debug_reg_global): Delete print of LRA status.
16660 (rs6000_option_override_internal): Delete dead LRA related code.
16661 (rs6000_lra_p): Delete function.
16662 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16664 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16666 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16667 * config/riscv/rtems.h: New file.
16669 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16670 Sudakshina Das <sudi.das@arm.com>
16672 * config/aarch64/aarch64.md
16673 (define_split for and<mode>3nr_compare): Move
16674 non aarch64_logical_operand to a register.
16675 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16676 register immediate operand to a register.
16677 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16679 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16681 PR middle-end/81564
16682 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16684 2017-07-27 Richard Biener <rguenther@suse.de>
16686 PR tree-optimization/81573
16687 PR tree-optimization/81494
16688 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16689 multi defuse cycle case.
16691 2017-07-27 Richard Biener <rguenther@suse.de>
16693 PR tree-optimization/81571
16694 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16697 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16699 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16700 earlier and only if MASK_FPU is set. Adjust formatting.
16702 2017-07-27 Martin Liska <mliska@suse.cz>
16704 * opt-functions.awk: Add validation of value of Init.
16705 * optc-gen.awk: Pass new argument.
16707 2017-07-27 Martin Liska <mliska@suse.cz>
16709 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16710 Fix wrong condition.
16712 2017-07-27 Martin Liska <mliska@suse.cz>
16714 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16715 BBs and edges seen by autoFDO.
16717 2017-07-27 Richard Biener <rguenther@suse.de>
16719 PR tree-optimization/81502
16720 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16721 with incompatible but same sized type.
16722 (execute_update_addresses_taken): Likewise.
16724 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16726 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16727 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16729 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16732 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16733 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16734 Change s_operand to memory_operand.
16736 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16738 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16739 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16740 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16741 Emit instructions rather than returning an expression. Handle TFmode
16742 and KFmode by casting to TImode.
16743 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16744 (rs6000_emit_le_vsx_store): Likewise.
16745 * config/rs6000/vsx.md (VSX_TI): New iterator.
16746 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16747 (*vsx_le_undo_permute_<mode>): Likewise.
16748 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16749 emit the split sequence.
16750 (*vsx_le_perm_store_<mode>): Likewise.
16752 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16754 PR tree-optimization/81555
16755 PR tree-optimization/81556
16756 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16757 if true, force CHANGED for the recursive invocation.
16758 (reassociate_bb): Remember original length of ops array, pass
16759 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16761 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16762 attributes for noipa attribute. For naked attribute use
16763 lookup_attribute first before lookup_attribute_spec.
16764 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16766 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16767 for functions with noipa attribute.
16768 (cgraph_externally_visible_p): Return true for functions with noipa
16770 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16771 for functions with noipa attribute.
16772 * doc/extend.texi: Document noipa function attribute.
16773 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16774 also for functions with noipa attribute.
16775 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16777 2017-07-26 Andrew Pinski <apinski@cavium.com>
16779 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16780 vec_unalign_load_cost and vec_unalign_store_cost.
16782 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16784 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16785 -mvsx-small-integer option.
16786 (ISA_3_0_MASKS_IEEE): Likewise.
16787 (OTHER_VSX_VECTOR_MASKS): Likewise.
16788 (POWERPC_MASKS): Likewise.
16789 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16790 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16791 code, only testing for DImode being allowed in non-VSX floating
16793 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16794 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16796 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16797 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16798 TARGET_P8_VECTOR test.
16799 (rs6000_secondary_reload_simple_move): Likewise.
16800 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16801 since TARGET_P9_VECTOR was already tested.
16802 (rs6000_opt_masks): Remove -mvsx-small-integer.
16803 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16804 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16806 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16807 test for TARGET_VEXTRACTUB was used, and that uses
16809 (p9 extract splitter): Likewise.
16810 (vsx_extract_<mode>_di_p9): Likewise.
16811 (vsx_extract_<mode>_store_p9): Likewise.
16812 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16813 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16814 the elimination of TARGET_VSX_SMALL_INTEGER.
16815 (vsx_extract_<mode>_p8): Likewise.
16816 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16817 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16818 (vsx_set_<mode>_p9): Likewise.
16819 (vsx_set_v4sf_p9): Likewise.
16820 (vsx_set_v4sf_p9_zero): Likewise.
16821 (vsx_insert_extract_v4sf_p9): Likewise.
16822 (vsx_insert_extract_v4sf_p9_2): Likewise.
16823 * config/rs6000/rs6000.md (sign extend splitter): Change
16824 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16825 (floatsi<mode>2_lfiwax_mem): Likewise.
16826 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16827 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16828 since a test for TARGET_P9_VECTOR was used.
16829 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16830 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16831 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16832 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16833 TARGET_P8_VECTOR test.
16834 (fix_trunc<mode>si2_stfiwx): Likewise.
16835 (fix_trunc<mode>si2_internal): Likewise.
16836 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16837 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16839 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16840 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16841 TARGET_P8_VECTOR test.
16842 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16843 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16844 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16846 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16847 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16848 since a test for TARGET_P9_VECTOR was used.
16849 (splitter for loading small constants): Likewise.
16851 2017-07-26 Andrew Pinski <apinski@cavium.com>
16853 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16856 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16859 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16860 (fp_valid_at): Likewise.
16862 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16864 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16865 (qdf24xx_addrcost_table): Likewise.
16866 (cortexa57_tunings): Update to use generic_branch_cost.
16867 (cortexa72_tunings): Likewise.
16868 (cortexa73_tunings): Likewise.
16869 (qdf24xx_tunings): Likewise.
16871 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16873 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16874 (thunderx2t99_branch_cost): Likewise.
16875 (cortexa35_tunings): Update to use generic_branch_cost.
16876 (cortexa53_tunings): Likewise.
16877 (cortexa57_tunings): Likewise.
16878 (cortexa72_tunings): Likewise.
16879 (cortexa73_tunings): Likewise.
16880 (thunderx2t99_tunings): Likewise.
16882 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16884 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16885 (sparc_option_override): Honour MASK_FSMULD.
16886 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16887 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16888 * config/sparc/sparc.opt (mfsmuld): New option.
16889 * doc/invoke.texi (mfsmuld): Document option.
16891 2017-07-26 Marek Polacek <polacek@redhat.com>
16893 PR middle-end/70992
16894 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16896 2017-07-26 Richard Biener <rguenther@suse.de>
16898 * gimple-match-head.c (do_valueize): Return OP if valueize
16900 (get_def): New helper to get at the def stmt of a SSA name
16901 if valueize allows.
16902 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16903 do_valueize to get at the def stmt.
16904 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16906 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16908 PR middle-end/46932
16909 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16911 2017-07-26 Martin Liska <mliska@suse.cz>
16914 * function.c (expand_function_start): Make expansion of
16915 nonlocal_goto_save_area after parm_birth_insn.
16917 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16919 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16920 from all CPU target flags enable members.
16922 2017-07-26 Richard Biener <rguenther@suse.de>
16924 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16925 (decision_tree::gen): Make 'type' const.
16926 (write_predicate): Likewise.
16928 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16930 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16931 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16932 (rs6000_option_override_internal): Likewise.
16933 (rs6000_expand_vector_set): Likewise.
16934 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16935 (TARGET_UPPER_REGS_SF): Likewise.
16936 (TARGET_UPPER_REGS_DI): Likewise.
16937 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16938 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16939 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16940 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16941 (Splitters for DI constants in Altivec registers): Likewise.
16942 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16943 (vsx_set_v4sf_p9): Likewise.
16944 (vsx_set_v4sf_p9_zero): Likewise.
16945 (vsx_insert_extract_v4sf_p9): Likewise.
16946 (vsx_insert_extract_v4sf_p9_2): Likewise.
16948 2017-07-25 Carl Love <cel@us.ibm.com>
16950 * doc/extend.texi: Update the built-in documentation file for the
16951 existing built-in functions
16952 vector signed char vec_cnttz (vector signed char);
16953 vector unsigned char vec_cnttz (vector unsigned char);
16954 vector signed short vec_cnttz (vector signed short);
16955 vector unsigned short vec_cnttz (vector unsigned short);
16956 vector signed int vec_cnttz (vector signed int);
16957 vector unsigned int vec_cnttz (vector unsigned int);
16958 vector signed long long vec_cnttz (vector signed long long);
16959 vector unsigned long long vec_cnttz (vector unsigned long long);
16961 2017-07-25 Andrew Pinski <apinski@cavium.com>
16963 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16964 accesses where the use is for the first operand of a BIT_INSERT.
16966 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16969 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16970 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16972 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16974 * config/i386/gstabs.h: Delete.
16975 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16977 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16979 * config/i386/i386.c (ix86_decompose_address): Do not check for
16980 register RTX when looking at index_reg or base_reg.
16981 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16983 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16985 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16986 to update EH info here.
16988 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16990 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16992 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16994 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16996 2017-07-25 Torsten Duwe <duwe@suse.de>
16998 * common.opt: Introduce -fpatchable-function-entry
16999 command line option, and its variables function_entry_patch_area_size
17000 and function_entry_patch_area_start.
17001 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
17002 including a two-value parser.
17003 * target.def (print_patchable_function_entry): New target hook.
17004 * targhooks.h (default_print_patchable_function_entry): New function.
17005 * targhooks.c (default_print_patchable_function_entry): Likewise.
17006 * toplev.c (process_options): Switch off IPA-RA if
17007 patchable function entries are being generated.
17008 * varasm.c (assemble_start_function): Look at the
17009 patchable-function-entry command line switch and current
17010 function attributes and maybe generate NOP instructions by
17011 calling the print_patchable_function_entry hook.
17012 * doc/extend.texi: Document patchable_function_entry attribute.
17013 * doc/invoke.texi: Document -fpatchable_function_entry
17014 command line option.
17015 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
17017 * doc/tm.texi: Re-generate.
17019 2017-07-25 Jakub Jelinek <jakub@redhat.com>
17022 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
17023 TARGET_AVX512DQ rather than TARGET_AVX512BW.
17025 2017-07-25 Tamar Christina <tamar.christina@arm.com>
17027 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
17029 2017-07-25 Richard Biener <rguenther@suse.de>
17031 PR tree-optimization/81455
17032 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
17033 not walk in cycles when looking for guards.
17035 2017-07-25 Richard Biener <rguenther@suse.de>
17037 PR tree-optimization/81529
17038 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17039 when optimizing backedge uses.
17041 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17043 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17045 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17046 to dl_section_ref. On AIX, append an expression to subtract
17047 the size of the section length to dl_section_ref.
17049 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17051 * configure.ac: If any of the config.* scripts fail, exit 1.
17052 * configure: Regenerate.
17054 2017-07-25 Richard Biener <rguenther@suse.de>
17056 PR middle-end/81546
17057 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17058 of immediate uses, be more verbose on errors.
17060 2017-07-25 Richard Biener <rguenther@suse.de>
17062 PR tree-optimization/81510
17063 * tree-vect-loop.c (vect_is_simple_reduction): When the
17064 reduction stmt is not inside the loop bail out.
17066 2017-07-25 Richard Biener <rguenther@suse.de>
17068 PR tree-optimization/81303
17069 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17070 profitability check against LOOP_VINFO_NITERSM1.
17072 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17074 * domwalk.c (cmp_bb_postorder): Simplify.
17075 (sort_bbs_postorder): New function. Use it...
17076 (dom_walker::walk): ...here to optimize common cases.
17078 2017-07-25 Martin Liska <mliska@suse.cz>
17081 * ipa-visibility.c (function_and_variable_visibility): Make the
17082 redirection just on target that supports aliasing.
17083 Fix GNU coding style.
17085 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17088 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17090 * config/arm/rtems.h: Likewise.
17091 * config/bfin/rtems.h: Likewise.
17092 * config/i386/rtemself.h: Likewise.
17093 * config/lm32/rtems.h: Likewise.
17094 * config/m32c/rtems.h: Likewise.
17095 * config/m68k/rtemself.h: Likewise.
17096 * config/microblaze/rtems.h: Likewise.
17097 * config/mips/rtems.h: Likewise.
17098 * config/moxie/rtems.h: Likewise.
17099 * config/nios2/rtems.h: Likewise.
17100 * config/powerpcspe/rtems.h: Likewise.
17101 * config/rs6000/rtems.h: Likewise.
17102 * config/rtems.h: Likewise.
17103 * config/sh/rtems.h: Likewise.
17104 * config/sh/rtemself.h: Likewise.
17105 * config/sparc/rtemself.h: Likewise.
17107 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17110 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17111 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17112 * tree-ssa-structalias.c (alias_get_name): Same.
17114 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17117 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17118 instructions if no du chain is found.
17120 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17122 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17124 2017-07-25 Richard Biener <rguenther@suse.de>
17126 PR middle-end/81505
17127 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17130 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17132 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17133 upper-regs options.
17134 (ISA_2_7_MASKS_SERVER): Likewise.
17135 (ISA_3_0_MASKS_IEEE): Likewise.
17136 (OTHER_P8_VECTOR_MASKS): Likewise.
17137 (OTHER_VSX_VECTOR_MASKS): Likewise.
17138 (POWERPC_MASKS): Likewise.
17139 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17140 duplicate list of options.
17141 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17142 explicit -mupper-regs options.
17143 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17144 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17145 alias for -mupper-regs-df.
17146 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17147 (rs6000_init_hard_regno_mode_ok): Likewise.
17148 (rs6000_option_override_internal): Likewise.
17149 (rs6000_opt_masks): Likewise.
17150 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17151 options in terms of whether -mvsx or -mpower8-vector was used.
17152 (TARGET_UPPER_REGS_DI): Likewise.
17153 (TARGET_UPPER_REGS_SF): Likewise.
17154 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17155 -mupper-regs-* options.
17157 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17159 * passes.c (emergency_dump_function): Print some empty lines and a
17160 header before the RTL dump.
17162 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17164 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17166 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17169 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17170 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17172 2017-07-24 Carl Love <cel@us.ibm.com>
17174 * config/rs6000/rs6000-c.c: Add support for built-in functions
17175 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17176 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17177 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17178 vec_extract_fp_from_shortl): Add defines for the two builtins.
17179 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17180 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17182 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17183 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17184 * doc/extend.texi: Update the built-in documentation file for the
17185 new built-in function.
17187 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17190 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17192 * doc/generic.texi: Likewise.
17193 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17194 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17196 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17198 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17199 (aarch64_mls_elt_merge<mode>): Likewise.
17201 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17203 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17204 having __cxa_atexit.
17206 2017-07-23 Michael Collison <michael.collison@arm.com>
17208 * config/arm/arm.c (arm_option_override): Deprecate
17209 use of -mstructure-size-boundary.
17210 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17211 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17213 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17216 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17217 Reduce cost estimate for direct moves.
17219 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17222 * config/i386/i386.c (ix86_option_override_internal): Disable
17223 BMI, BMI2 and TBM instructions for -m16.
17225 2017-07-21 Carl Love <cel@us.ibm.com>
17227 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17228 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17229 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17230 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17231 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17232 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17233 VMULOSW): New enum "unspec" values.
17234 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17235 altivec_vmulosw): New patterns.
17236 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17237 VMULOSW): Add definitions.
17239 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17241 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17242 (qdf24xx): Likewise.
17243 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17244 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17245 (AARCH64_FL_V8_1): Renumber.
17246 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17247 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17248 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17249 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17250 rdma to feature modifiers list.
17252 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17254 PR middle-end/56727
17255 * ipa-visibility (function_and_variable_visibility): Convert
17256 recursive PLT call to direct call if appropriate.
17258 2017-07-21 Andrew Pinski <apinski@cavium.com>
17260 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17261 operand 1 to see if the types precision matches.
17262 * fold-const.c (operand_equal_p): Likewise.
17264 2017-07-21 Richard Biener <rguenther@suse.de>
17266 PR tree-optimization/81303
17267 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17268 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17269 (vect_peeling_hash_get_lowest_cost): Adjust.
17270 (vect_enhance_data_refs_alignment): Likewise. Use
17271 vect_get_peeling_costs_all_drs to compute the penalty for no
17272 peeling to match up costs.
17274 2017-07-21 Richard Biener <rguenther@suse.de>
17276 PR tree-optimization/81500
17277 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17278 we didn't identify a reduction path.
17280 2017-07-21 Tom de Vries <tom@codesourcery.com>
17281 Cesar Philippidis <cesar@codesourcery.com>
17283 PR gcov-profile/81442
17284 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17287 2017-07-21 Tom de Vries <tom@codesourcery.com>
17290 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17292 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17293 nvptx_override_options_after_change.
17295 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17297 * dwarf2out.c (output_file_names): Avoid double testing for
17298 dwarf_version >= 5.
17300 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17302 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17304 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17306 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17308 (try_crossjump_to_edge): Do not punt on partitioned functions.
17310 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17312 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17313 Put all BBs reachable only via paths crossing cold region to cold
17315 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17317 2016-07-21 Richard Biener <rguenther@suse.de>
17319 PR tree-optimization/81303
17320 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17321 into account prologue and epilogue iterations when raising
17322 min_profitable_iters to sth at least covering one vector iteration.
17324 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17326 * config/arm/arm.c (arm_test_cpu_arch_dat):
17329 2017-07-20 Nathan Sidwell <nathan@acm.org>
17331 Remove TYPE_METHODS.
17332 * tree.h (TYPE_METHODS): Delete.
17333 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17334 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17335 (dbxout_type_methods): Scan TYPE_FIELDS.
17336 (dbxout_type): Don't check TYPE_METHODS here.
17337 * function.c (use_register_for_decl): Always ignore register for
17338 class types when not optimizing.
17339 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17340 * tree.c (free_lang_data_in_type): Stitch out member functions and
17341 templates from TYPE_FIELDS.
17342 (build_distinct_type_copy, verify_type_variant,
17343 verify_type): Member fns are on TYPE_FIELDS.
17344 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17345 * tree-pretty-print.c (dump_generic_node): Likewise.
17347 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17350 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17351 V2TImode and V4TImode.
17352 (ix86_expand_vector_extract): Likewise.
17353 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17354 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17355 (ssescalarmode): Handle V4TImode and V2TImode.
17356 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17357 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17358 (VEXTRACTI128_MODE): New mode iterator.
17359 (splitter for *vec_extractv?ti first element): New.
17360 (VEC_INIT_MODE): New mode iterator.
17361 (vec_init<mode>): Consolidate 3 expanders into one using
17362 VEC_INIT_MODE mode iterator.
17364 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17366 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17367 non_spilled_static_chain_regno_p.
17369 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17371 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17373 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17375 * bb-reorder.c (connect_traces): Allow copying of blocks within
17378 2017-07-20 Richard Biener <rguenther@suse.de>
17380 * gimple.h (gimple_phi_result): Add gphi * overload.
17381 (gimple_phi_result_ptr): Likewise.
17382 (gimple_phi_arg): Likewise. Adjust index assert to only
17383 allow actual argument accesses rather than all slots available
17385 (gimple_phi_arg_def): Add gphi * overload.
17386 * tree-phinodes.c (make_phi_node): Initialize only actual
17388 (resize_phi_node): Clear memory not covered by old node,
17389 do not initialize excess argument slots.
17390 (reserve_phi_args_for_new_edge): Initialize new argument slot
17393 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17395 PR tree-optimization/81388
17397 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17399 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17400 by removing computation of may_be_zero.
17402 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17403 Tom de Vries <tom@codesourcery.com>
17405 PR middle-end/81030
17406 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17407 when gimple level profile disagrees with what RTL expander did.
17409 2017-07-20 Richard Biener <rguenther@suse.de>
17411 PR tree-optimization/61171
17412 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17413 (vect_analyze_stmt): Add slp instance parameter.
17414 (vectorizable_reduction): Likewise.
17415 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17416 (vect_is_simple_reduction): Deal with chains not detected
17417 as SLP reduction chain, specifically not properly associated
17418 chains containing a mix of plus/minus.
17419 (get_reduction_op): Remove.
17420 (get_initial_defs_for_reduction): Simplify, pass in whether
17421 this is a reduction chain, pass in the SLP node for the PHIs.
17422 (vect_create_epilog_for_reduction): Get the SLP instance as
17424 (vectorizable_reduction): Get the SLP instance as arg.
17425 During analysis remember the SLP node with the PHIs in the
17426 instance. Simplify getting at the vectorized reduction PHIs.
17427 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17428 through SLP instance.
17429 (vect_slp_analyze_operations): Likewise.
17430 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17431 (vect_transform_stmt): Likewise.
17433 2017-07-20 Tom de Vries <tom@codesourcery.com>
17435 PR tree-optimization/81489
17436 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17437 read of phi arg location to before loop that modifies phi.
17439 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17441 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17444 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17446 PR middle-end/81331
17447 * except.c (execute): Fix ordering issue.
17449 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17451 PR rtl-optimization/81423
17452 * combine.c (make_compound_operation_int): Don't try to optimize
17453 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17455 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17457 PR rtl-optimization/81423
17458 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17459 with a constant that is -1 in the truncated to mode.
17461 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17463 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17464 (determine_unlikely_bbs): ... here.
17465 * predict.h (propagate_unlikely_bbs_forward): Declare.
17466 * cfgexpand.c (pass_expand::execute): Use it.
17467 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17469 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17470 propagate_unlikely_bbs_forward.
17472 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17474 PR middle-end/81331
17475 * except.c (maybe_add_nop_after_section_switch): New function.
17478 2017-07-19 Tom de Vries <tom@codesourcery.com>
17480 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17482 2017-07-19 Tom de Vries <tom@codesourcery.com>
17484 * gimple.h (gimple_phi_arg): Make assert more strict.
17486 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17488 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17489 * config/rs6000/mmintrin.h: New file.
17490 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17492 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17494 PR tree-optimization/81346
17495 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17497 2017-07-19 Tom de Vries <tom@codesourcery.com>
17499 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17501 2017-07-19 Tom de Vries <tom@codesourcery.com>
17503 * config/nvptx/nvptx-modes.def: Add V2DImode.
17504 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17505 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17506 (nvptx_output_mov_insn): Handle lack of mov.b128.
17507 (nvptx_print_operand): Handle 'H' and 'L' codes.
17508 (nvptx_vector_mode_supported): Allow V2DImode.
17509 (nvptx_preferred_simd_mode): New function.
17510 (nvptx_data_alignment): New function.
17511 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17512 nvptx_preferred_simd_mode.
17513 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17515 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17517 2017-07-19 Tom de Vries <tom@codesourcery.com>
17519 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17520 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17521 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17522 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17523 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17524 (mov<VECIM>_insn): New define_insn.
17525 (define_expand "mov<VECIM>): New define_expand.
17527 2017-07-19 Tom de Vries <tom@codesourcery.com>
17529 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17531 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17533 PR tree-optimization/81346
17534 * fold-const.h (fold_div_compare, range_check_type): Declare.
17535 * fold-const.c (range_check_type): New function.
17536 (build_range_check): Use range_check_type.
17537 (fold_div_compare): No longer static, rewritten into
17538 a match.pd helper function.
17539 (fold_comparison): Don't call fold_div_compare here.
17540 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17541 as helper function.
17543 2017-07-19 Nathan Sidwell <nathan@acm.org>
17545 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17546 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17547 * tree.c (find_decls_types_r, verify_type): Use
17548 TYPE_{MIN,MAX}_VALUE_RAW.
17549 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17550 (hash_tree): Likewise.
17551 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17553 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17556 2017-07-18 Tom de Vries <tom@codesourcery.com>
17558 PR middle-end/81464
17559 * omp-expand.c (expand_omp_for_static_chunk): Handle
17560 equal-argument loop exit phi.
17562 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17565 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17566 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17567 operand 2 predicate.
17568 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17569 operand 2 predicate.
17570 (ror,rol -> rorx splitters): Use const_int_operand as
17571 operand 2 predicate.
17573 2017-06-18 Richard Biener <rguenther@suse.de>
17575 PR tree-optimization/81410
17576 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17577 the gap in the ! slp_perm SLP case after each group.
17579 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17581 PR middle-end/81463
17582 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17585 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17587 PR middle-end/81462
17588 * predict.c (set_even_probabilities): Cleanup; do not affect
17589 probabilities that are already known.
17590 (combine_predictions_for_bb): Call even when count is set.
17592 2017-07-18 Nathan Sidwell <nathan@acm.org>
17594 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17597 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17600 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17601 optimization for loop niter analysis.
17603 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17606 * config/avr/avr.c (avr_optimize_casesi): Don't use
17607 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17609 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17611 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17612 body_cost_vec from _vect_peel_extended_info.
17613 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17614 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17617 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17619 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17621 2017-07-18 Richard Biener <rguenther@suse.de>
17623 PR tree-optimization/80620
17624 PR tree-optimization/81403
17625 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17626 info when re-using a VN table entry.
17628 2017-07-18 Richard Biener <rguenther@suse.de>
17630 PR tree-optimization/81418
17631 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17632 vectype_in. Verify that with lane-reducing reduction operations
17633 we have a single def-use cycle.
17635 2017-07-17 Carl Love <cel@us.ibm.com>
17637 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17639 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17640 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17641 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17642 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17643 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17644 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17645 VMULOSW): New enum "unspec" values.
17646 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17647 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17648 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17649 altivec_vmulosw): New patterns.
17650 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17651 VMULOSW): Add definitions.
17652 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17654 * config/alpha/alpha.c: Include predict.h.
17656 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17658 * tree-vrp.c (compare_assert_loc): Fix comparison function
17659 to return predictable results.
17661 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17663 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17665 (subdi3): Likewise.
17666 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17667 * doc/invoke.texi (mexpand-adddi): Update text.
17669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17671 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17672 that also clobbers the CC register. The old expand code is moved
17674 (*arc_clzsi2): ... here.
17675 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17676 the CC register. The old expand code is moved to ...
17677 (arc_ctzsi2): ... here.
17679 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17681 * config/arc/arc.opt (mindexed-loads): Use initial value
17682 TARGET_INDEXED_LOADS_DEFAULT.
17683 (mauto-modify-reg): Use initial value
17684 TARGET_AUTO_MODIFY_REG_DEFAULT.
17685 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17686 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17687 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17688 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17690 2017-07-17 Martin Liska <mliska@suse.cz>
17693 * opts.c (finish_options): Do not allow -fgnu-tm
17694 w/ -fsanitize={kernel-,}address. Say sorry.
17696 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17699 * tree-loop-distribution.c (classify_partition): Only assert on
17700 numer of iterations.
17701 (merge_dep_scc_partitions): Delete prameter. Update function call.
17702 (distribute_loop): Remove code handling loop with unknown niters.
17703 (pass_loop_distribution::execute): Skip loop with unknown niters.
17705 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17708 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17709 function sort_partitions_by_post_order.
17711 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17713 PR tree-optimization/81374
17714 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17715 the max index of basic blocks, rather than number of basic blocks.
17717 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17719 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17721 (arc_legitimate_pic_operand_p): Likewise.
17722 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17724 (arc_needs_pcl_p): Likewise.
17725 (arc_legitimate_pc_offset_p): Likewise.
17726 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17727 function is also used in constrains.md.
17728 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17729 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17730 PLUS. Only return true/false in known cases, otherwise assert.
17731 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17732 is already called in arc_legitimate_constant_p.
17733 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17735 (LEGITIMATE_PIC_OPERAND_P): Use
17736 arc_raw_symbolic_reference_mentioned_p function.
17737 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17742 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17743 Andrew Burgess <andrew.burgess@embecosm.com>
17745 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17746 (arc_return_address_register): New function.
17747 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17748 (arc_handle_fndecl_attribute): Add naked attribute.
17749 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17750 (TARGET_WARN_FUNC_RETURN): Likewise.
17751 (arc_allocate_stack_slots_for_args): New function.
17752 (arc_warn_func_return): Likewise.
17753 (machine_function): Change type fn_type.
17754 (arc_compute_function_type): Consider new naked function type,
17755 change function return type.
17756 (arc_must_save_register): Adapt to handle new
17757 arc_compute_function_type's return type.
17758 (arc_expand_prologue): Likewise.
17759 (arc_expand_epilogue): Likewise.
17760 (arc_return_address_regs): Delete.
17761 (arc_return_address_register): New function.
17762 (arc_epilogue_uses): Use above function.
17763 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17764 (arc_function_type): Change encoding, add naked type.
17765 (ARC_INTERRUPT_P): Change to handle the new encoding.
17766 (ARC_FAST_INTERRUPT_P): Likewise.
17767 (ARC_NORMAL_P): Define.
17768 (ARC_NAKED_P): Likewise.
17769 (arc_compute_function_type): Delete prototype.
17770 * config/arc/arc.md (in_ret_delay_slot): Use
17771 arc_return_address_register function.
17772 (simple_return): Likewise.
17773 (p_return_i): Likewise.
17775 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17777 PR tree-optimization/81428
17778 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17779 can't be built for those types.
17781 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17783 Remove stuff dead since r239246.
17785 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17786 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17787 (avr_inform_devices): Remove dead stuff.
17789 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17791 * config/arm/arm_neon.h: Fix softp typo.
17793 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17795 PR tree-optimization/81365
17796 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17797 aggregate moves onto bb predecessor edges, make sure there are no
17798 loads that could alias the lhs in between the start of bb and the
17801 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17804 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17805 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17806 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17808 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17810 PR tree-optimization/81396
17811 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17812 (init_symbolic_number): Initialize it to 1.
17813 (perform_symbolic_merge): Add n_ops from both operands into the new
17815 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17816 without base_addr as useless if they need more than one operation.
17817 (bswap_replace): Handle !bswap case for NULL base_addr.
17819 2017-07-17 Tom de Vries <tom@codesourcery.com>
17822 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17825 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17827 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17828 conditional builtin define __FIX_LEON3FT_B2BST.
17830 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17832 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17833 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17836 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17838 PR rtl-optimization/81424
17839 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17840 to remove potential trapping from operands if -fnon-call-exceptions.
17842 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17844 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17845 profile_proability for scalling.
17846 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17848 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17850 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17852 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17854 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17855 fixpoint arithmetics.
17857 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17859 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17860 fixpoint arithmetics.
17862 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17864 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17865 fixpoint arithmetics.
17867 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17869 * profile-count.h (profile_probability::from_reg_br_prob_note,
17870 profile_probability::to_reg_br_prob_note): New functions.
17871 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17872 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17873 * predict.c (probability_reliable_p): Update.
17874 (edge_probability_reliable_p): Update.
17875 (br_prob_note_reliable_p): Update.
17876 (invert_br_probabilities): Update.
17877 (add_reg_br_prob_note): New function.
17878 (combine_predictions_for_insn): Update.
17879 * asan.c (asan_clear_shadow): Update.
17880 * cfgbuild.c (compute_outgoing_frequencies): Update.
17881 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17882 (update_br_prob_note): Update.
17883 (rtl_verify_edges): Update.
17884 (purge_dead_edges): Update.
17885 (fixup_reorder_chain): Update.
17886 * emit-rtl.c (try_split): Update.
17887 * ifcvt.c (cond_exec_process_insns): Update.
17888 (cond_exec_process_if_block): Update.
17889 (dead_or_predicable): Update.
17890 * internal-fn.c (expand_addsub_overflow): Update.
17891 (expand_neg_overflow): Update.
17892 (expand_mul_overflow): Update.
17893 * loop-doloop.c (doloop_modify): Update.
17894 * loop-unroll.c (compare_and_jump_seq): Update.
17895 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17896 * predict.h: Update.
17897 * reorg.c (mostly_true_jump): Update.
17899 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17900 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17901 * config/arc/arc.c: (emit_unlikely_jump): Update.
17902 * config/arm/arm.c: (emit_unlikely_jump): Update.
17903 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17904 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17905 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17906 (ix86_print_operand): Update.
17907 (ix86_split_fp_branch): Update.
17908 (predict_jump): Update.
17909 * config/ia64/ia64.c (ia64_print_operand): Update.
17910 * config/mmix/mmix.c (mmix_print_operand): Update.
17911 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17912 (rs6000_expand_split_stack_prologue): Update.
17913 * config/rs6000/rs6000.c: Update.
17914 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17915 (s390_expand_vec_movstr): Update.
17916 (s390_expand_cs_tdsi): Update.
17917 (s390_expand_split_stack_prologue): Update.
17918 * config/sh/sh.c (sh_print_operand): Update.
17919 (expand_cbranchsi4): Update.
17920 (expand_cbranchdi4): Update.
17921 * config/sparc/sparc.c (output_v9branch): Update.
17922 * config/spu/spu.c (get_branch_target): Update.
17923 (ea_load_store_inline): Update.
17924 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17925 * config/tilepro/tilepro.c: Update.
17927 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17929 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17930 (gimplify_save_expr): Likewise.
17932 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17934 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17936 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17938 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17940 * timevar.def (TV_IPA_FNSUMMARY): Define.
17942 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17944 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17945 to back store errata sensitive sequence from being generated.
17946 (sqrtdf2_fix): Likewise.
17948 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17950 * tree-ssa-threadupdate.c (compute_path_counts,
17951 update_joiner_offpath_counts): Use profile_probability.
17953 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17956 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17958 * config/arm/arm-c.c (arm_cpu_builtins): Define
17959 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17961 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17963 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17964 array entries to represent __ieee128 versions of the
17965 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17966 scalar_extract_sig, and scalar_insert_exp built-in functions.
17967 (altivec_resolve_overloaded_builtin): Add special case handling
17968 for the __builtin_scalar_insert_exp function, as represented by
17969 the P9V_BUILTIN_VEC_VSIEDP constant.
17970 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17971 exponent support for __ieee128 argument.
17972 (VSESQP): Add scalar extract signature support for __ieee128
17974 (VSTDCNQP): Add scalar test negative support for __ieee128
17976 (VSIEQP): Add scalar insert exponent support for __int128 argument
17977 with __ieee128 result.
17978 (VSIEQPF): Add scalar insert exponent support for __ieee128
17979 argument with __ieee128 result.
17980 (VSTDCQP): Add scalar test data class support for __ieee128
17982 (VSTDCNQP): Add overload support for scalar test negative with
17983 __ieee128 argument.
17984 (VSTDCQP): Add overload support for scalar test data class
17985 __ieee128 argument.
17986 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17987 UNSPEC_VSX_SXSIGDP.
17988 (UNSPEC_VSX_SIEXPQP): New constant.
17989 (xsxexpqp): New insn for VSX scalar extract exponent quad
17991 (xsxsigqp): New insn for VSX scalar extract significand quad
17993 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17994 precision with floating point argument.
17995 (xststdcqp): New expand for VSX scalar test data class quad
17997 (xststdcnegqp): New expand for VSX scalar test negative quad
17999 (xststdcqp): New insn to match expansions for VSX scalar test data
18000 class quad precision and VSX scalar test negative quad precision.
18001 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
18002 special case operand checking to enforce that second operand of
18003 VSX scalar test data class with quad precision argument is a 7-bit
18005 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18006 prototypes and descriptions of __ieee128 versions of
18007 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
18008 scalar_test_data_class, and scalar_test_neg built-in functions.
18010 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18012 PR tree-optimization/81162
18013 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
18014 replace a negate with an add.
18016 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
18018 * doc/invoke.texi (arm/-mcpu): Document +crypto.
18020 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18022 * config/arm/arm-c.c (arm_cpu_builtins): Define
18023 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
18025 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18027 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
18028 (armv8-r): Set ARM Cortex-R52 as default CPU.
18029 * config/arm/arm-tables.opt: Regenerate.
18030 * config/arm/arm-tune.md: Regenerate.
18031 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18033 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
18034 extension for -mcpu=cortex-r52.
18036 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18038 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18039 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18040 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18041 (fp-armv8): Define it as FP_ARMv8 only.
18042 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18043 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18045 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18046 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18047 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18048 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18049 than TARGET_FPU_ARMV8.
18050 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18051 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18052 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18054 * config/arm/neon.md (neon_vrint): Likewise.
18055 (neon_vcvt): Likewise.
18056 (neon_<fmaxmin_op><mode>): Likewise.
18057 (<fmaxmin><mode>3): Likewise.
18058 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18059 * config/arm/predicates.md (arm_cond_move_operator): Check against
18060 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18062 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18064 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18065 to top of function.
18067 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18069 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18070 loop in comment with memset.
18072 2017-07-14 Martin Liska <mliska@suse.cz>
18074 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18075 * dwarf2out.c (is_java): Remove the function.
18076 (output_pubname): Remove usage of the function.
18077 (lower_bound_default): Remove usage of DW_LANG_Java.
18078 (gen_compile_unit_die): Likewise.
18079 * gcc.c: Remove compiler defaults for .java and .zip files.
18080 * gimple-expr.c (remove_suffix): Change as there's no longer
18081 extension than 4-letter one.
18082 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18083 (gimplify_save_expr): Likewise.
18084 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18085 as it's possible even for other languages than Java.
18086 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18087 * lto-opts.c (lto_write_options): Remove reference to Java.
18088 * opts.c (strip_off_ending): Update file extension handling.
18089 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18090 * tree-eh.c (lower_resx): Likewise.
18091 * tree.c (free_lang_data_in_type): Remove dead code.
18092 (find_decls_types_r): Likewise.
18093 (build_common_builtin_nodes): Remove Java from a comment.
18094 (verify_type): Remove dead code.
18095 * varasm.c (assemble_external): Remove Java from a comment.
18097 2017-07-14 Martin Liska <mliska@suse.cz>
18099 * opts.c (finish_options): Add quotes.
18100 (common_handle_option): Likewise.
18102 2017-07-14 Martin Liska <mliska@suse.cz>
18104 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18105 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18106 Remove N_SO_PASCAL.
18107 * dwarf2out.c (lower_bound_default): Do not handle
18109 (gen_compile_unit_die): Likewise.
18110 * gcc.c: Remove default extension binding for GNU Pascal.
18111 * stmt.c: Remove Pascal language from a comment.
18112 * xcoffout.c: Likewise.
18114 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18117 * diagnostic-show-locus.c (fixit_cmp): New function.
18118 (layout::layout): Sort m_fixit_hints.
18119 (column_range::column_range): Assert that the values are valid.
18120 (struct char_span): New struct.
18121 (correction::overwrite): New method.
18122 (struct source_line): New struct.
18123 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18124 calls in terms of classes source_line and char_span, and
18125 correction::overwrite.
18126 (selftest::test_overlapped_fixit_printing_2): New function.
18127 (selftest::diagnostic_show_locus_c_tests): Call it.
18129 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18131 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18132 early if there is no lhs.
18134 2017-07-13 Martin Liska <mliska@suse.cz>
18136 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18137 (gen_reference_type_die): Likewise.
18138 * stor-layout.c: Remove Pascal-related comment.
18140 2017-07-13 Martin Liska <mliska@suse.cz>
18142 * opts.c (finish_options): Add quotes to error messages.
18143 (parse_sanitizer_options): Likewise.
18145 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18147 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18149 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18151 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18153 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18155 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18157 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18159 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18162 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18163 provides the hardware capability bits, define the macro
18164 __BUILTIN_CPU_SUPPORTS__.
18165 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18166 if GLIBC does not provide the hardware capability bits. Add a
18167 gcc_unreachable call if the built-in cpu function is neither
18168 __builtin_cpu_is nor __builtin_cpu_supports.
18169 (rs6000_get_function_versions_dispatcher): Change the warning
18170 that an old GLIBC is used which does not export the capability
18171 bits to be an error.
18172 * doc/extend.texi (target_clones attribute): Document the
18173 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18174 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18175 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18176 the macros defined by GCC if the newer GLIBC is available.
18178 2017-07-12 Jeff Law <law@redhat.com>
18180 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18181 remaining includes slightly.
18182 * config/riscv/riscv-builtins.c: Include profile-count.h.
18184 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18187 * config/avr/avr.c (avr_set_current_function): In diagnostic
18188 messages: Quote keywords and (parts of) identifiers.
18189 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18192 2017-07-12 Carl Love <cel@us.ibm.com>
18194 * config/rs6000/rs6000-c.c: Add support for built-in functions
18195 vector bool char vec_revb (vector bool char);
18196 vector bool short vec_revb (vector short char);
18197 vector bool int vec_revb (vector bool int);
18198 vector bool long long vec_revb (vector bool long long);
18199 * doc/extend.texi: Update the built-in documentation file for the
18200 new built-in functions.
18202 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18204 * config/s390/s390.md: Remove movcc splitter.
18206 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18208 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18209 load/store on condition.
18211 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18214 * config/avr/avr.c (avr_encode_section_info)
18215 [progmem && !TREE_READONLY]: Error if progmem object needs
18218 2017-07-11 Michael Collison <michael.collison@arm.com>
18220 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18223 2017-07-11 Carl Love <cel@us.ibm.com>
18225 * config/rs6000/rs6000-c.c: Add support for builtins
18226 vector unsigned int vec_parity_lsbb (vector signed int);
18227 vector unsigned int vec_parity_lsbb (vector unsigned int);
18228 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18229 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18230 vector unsigned long long vec_parity_lsbb (vector signed long long);
18231 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18232 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18233 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18234 * doc/extend.texi: Update the built-in documentation file for the
18235 new built-in functions.
18237 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18239 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18240 (layout::m_primary_loc): New field.
18241 (layout::layout): Initialize new field. Move location filtering
18242 logic from here to...
18243 (layout::maybe_add_location_range): ...this new method. Add
18244 support for filtering to just the lines already specified by other
18246 (layout::will_show_line_p): New method.
18247 (gcc_rich_location::add_location_if_nearby): New method.
18248 (selftest::test_add_location_if_nearby): New test function.
18249 (selftest::diagnostic_show_locus_c_tests): Call it.
18250 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18253 2017-07-11 Tom de Vries <tom@codesourcery.com>
18255 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18256 (bb_first_real_insn): New function.
18257 (nvptx_single): Add extra initialization of broadcasted condition
18260 2017-07-11 Nathan Sidwell <nathan@acm.org>
18262 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18264 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18266 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18267 Binutils doc as TEXI will mess them up.
18268 * doc/invoke.texi (AVR Options): Same here.
18270 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18272 * config/sparc/sparc.opt (mfix-ut700): New option.
18273 (mfix-gr712rc): Likewise.
18274 (sparc_fix_b2bst): New variable.
18275 * doc/invoke.texi (SPARC options): Document them.
18276 (ARM options): Fix warnings.
18277 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18278 instructions to prevent sequences that can trigger the store-store
18279 errata for certain LEON3FT processors.
18280 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18281 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18282 * config/sparc/sparc.md (fix_b2bst): New attribute.
18283 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18285 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18288 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18290 (*rsqrtsf2_sse): Ditto.
18292 (div<mode>3): Macroize insn from divdf3 and divsf3
18293 using MODEF mode iterator.
18295 2017-07-10 Martin Sebor <msebor@redhat.com>
18297 PR tree-optimization/80397
18298 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18299 instead of testing for equality to INTEGER_TYPE.
18301 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18303 * config.gcc: Remove uclibc from arc target spec.
18305 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18307 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18309 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18312 * lto-wrapper.c (remove_option): New function.
18313 (merge_and_complain): Merge PIC/PIE options more realistically.
18315 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18317 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18321 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18322 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18323 * config.in: Regenerate.
18324 * configure: Regenerate.
18325 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18326 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18327 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18328 (TARGET_GASISR_PROLOGUES): ...target mask.
18329 * common/config/avr/avr-common.c
18330 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18331 Set -mgas-isr-prologues.
18332 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18333 INSERT_PASS_BEFORE for it.
18334 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18335 * config/avr/avr.c (avr_option_override)
18336 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18337 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18338 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18339 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18340 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18342 (make_avr_pass_pre_proep): New function.
18343 (emit_push_sfr) <treg>: Add argument to function and use it
18344 instead of TMP_REG.
18345 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18346 and set machine->gasisr.yes.
18347 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18348 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18349 __gcc_isr.n_pushed to .L__stack_usage.
18350 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18351 (avr_asm_final_postscan_insn): ...this new static function.
18352 * config/avr/avr.h (machine_function)
18353 <is_no_gccisr, use_L__stack_usage>: New fields.
18354 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18355 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18356 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18357 (gasisr, *gasisr): New expander and insn.
18358 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18359 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18360 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18362 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18364 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18367 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18369 Move jump-tables out of .text again.
18372 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18373 (ASM_OUTPUT_ADDR_VEC): New function.
18374 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18375 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18376 INSN_ADDRESSes as asm comment.
18377 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18378 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18379 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18380 * config/avr/avr.md (*tablejump): Adjust comment.
18381 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18382 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18384 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18385 (avr_output_addr_vec): New proto.
18386 (avr_log_t) <insn_addresses>: New field.
18388 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18391 * config/i386/i386.c (ix86_function_arg_advance): Set
18392 outgoing_args_on_stack to true if there are outgoing arguments
18394 (ix86_function_arg): Likewise.
18395 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18396 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18397 * config/i386/i386.h (machine_function): Add
18398 outgoing_args_on_stack.
18400 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18402 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18403 supporting pthreds.
18404 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18406 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18408 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18409 (REAL_H): Remove $(MACHMODE_H).
18410 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18412 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18413 $(MACHMODE_H) and double-int.h.
18414 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18416 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18419 2017-07-07 Andrew Pinski <apinski@cavium.com>
18421 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18422 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18424 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18426 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18427 Add warning if GCC was not configured to link against a GLIBC that
18428 exports the hardware capability bits.
18429 (make_resolver_func): Make resolver function private and not a
18430 COMDAT function. Create the name with clone_function_name instead
18431 of make_unique_name.
18434 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18435 correct operand in doing the split.
18437 2017-07-07 Carl Love <cel@us.ibm.com>
18439 * config/rs6000/rs6000-c: Add support for built-in function
18440 vector unsigned short vec_pack_to_short_fp32 (vector float,
18442 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18443 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18444 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18445 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18446 (convert_4f32_8i16): Add define_expand.
18447 * doc/extend.texi: Update the built-in documentation file for the
18448 new built-in function.
18450 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18452 * config/sparc/m8.md: New file.
18453 * config/sparc/sparc.md: Include m8.md.
18455 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18457 * config/sparc/sparc.opt: New option -mvis4b.
18458 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18459 (sparc_option_override): Handle VIS4B.
18460 (enum sparc_builtins): Define
18461 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18462 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18463 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18464 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18465 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18466 (check_constant_argument): New function.
18467 (sparc_vis_init_builtins): Define builtins
18468 __builtin_vis_dictunpack{8,16,32},
18469 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18470 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18471 __builtin_vis_fpcmpde{8,16,32}shl and
18472 __builtin_vis_fpcmpur{8,16,32}shl.
18473 (sparc_expand_builtin): Check that the constant operands to
18474 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18475 constant and in range.
18476 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18478 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18479 (SPARC_IMM5_P): Likewise.
18480 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18481 (enabled): Handle vis4b.
18482 (UNSPEC_DICTUNPACK): New unspec.
18483 (UNSPEC_FPCMPSHL): Likewise.
18484 (UNSPEC_FPUCMPSHL): Likewise.
18485 (UNSPEC_FPCMPDESHL): Likewise.
18486 (UNSPEC_FPCMPURSHL): Likewise.
18487 (cpu_feature): New CPU feature `vis4b'.
18488 (dictunpack{8,16,32}): New insns.
18489 (FPCSMODE): New mode iterator.
18490 (fpcscond): New code iterator.
18491 (fpcsucond): Likewise.
18492 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18493 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18494 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18495 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18496 * config/sparc/constraints.md: Define constraints `q' for unsigned
18497 2-bit integer constants and `t' for unsigned 5-bit integer
18499 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18501 (imm5_operand_dictunpack16): Likewise.
18502 (imm5_operand_dictunpack32): Likewise.
18503 (imm2_operand): Likewise.
18504 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18505 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18506 ditunpack* and fpcmp*shl builtins.
18508 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18510 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18511 * config.in: Add HAVE_AS_SPARC6 define.
18512 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18514 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18516 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18517 (CPP_CPU_SPEC): Handle m8.
18518 (ASM_CPU_SPEC): Likewise.
18519 * config/sparc/sparc-opts.h (enum processor_type): Add
18521 * config/sparc/sparc.c (m8_costs): New struct.
18522 (sparc_option_override): Handle TARGET_CPU_m8.
18523 (sparc32_initialize_trampoline): Likewise.
18524 (sparc64_initialize_trampoline): Likewise.
18525 (sparc_issue_rate): Likewise.
18526 (sparc_register_move_cost): Likewise.
18527 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18528 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18529 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18530 (CPP_CPU_SPEC): Handle M8.
18531 (ASM_CPU_SPEC): Likewise.
18532 (AS_M8_FLAG): Define.
18533 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18534 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18535 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18537 * configure: Regenerate.
18538 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18541 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18543 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18545 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18546 ("*movdi_insn_sp32"): Do not set v3pipe.
18547 ("*movsi_insn"): Likewise.
18548 ("*movdi_insn_sp64"): Likewise.
18549 ("*movsf_insn"): Likewise.
18550 ("*movdf_insn_sp32"): Likewise.
18551 ("*movdf_insn_sp64"): Likewise.
18552 ("*zero_extendsidi2_insn_sp64"): Likewise.
18553 ("*sign_extendsidi2_insn"): Likewise.
18554 ("*mov<VM32:mode>_insn"): Likewise.
18555 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18556 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18557 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18558 ("<vlop:code><VL:mode>3"): Likewise.
18559 ("*not_<vlop:code><VL:mode>3"): Likewise.
18560 ("*nand<VL:mode>_vis"): Likewise.
18561 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18562 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18563 ("one_cmpl<VL:mode>2"): Likewise.
18564 ("faligndata<VM64:mode>_vis"): Likewise.
18565 ("alignaddrsi_vis"): Likewise.
18566 ("alignaddrdi_vis"): Likweise.
18567 ("alignaddrlsi_vis"): Likewise.
18568 ("alignaddrldi_vis"): Likewise.
18569 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18570 ("bmaskdi_vis"): Likewise.
18571 ("bmasksi_vis"): Likewise.
18572 ("bshuffle<VM64:mode>_vis"): Likewise.
18573 ("cmask8<P:mode>_vis"): Likewise.
18574 ("cmask16<P:mode>_vis"): Likewise.
18575 ("cmask32<P:mode>_vis"): Likewise.
18576 ("pdistn<P:mode>_vis"): Likewise.
18577 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18579 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18581 * config/sparc/sparc.md ("subtype"): New insn attribute.
18582 ("*wrgsr_sp64"): Set insn subtype.
18583 ("*rdgsr_sp64"): Likewise.
18584 ("alignaddrsi_vis"): Likewise.
18585 ("alignaddrdi_vis"): Likewise.
18586 ("alignaddrlsi_vis"): Likewise.
18587 ("alignaddrldi_vis"): Likewise.
18588 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18589 ("fexpand_vis"): Likewise.
18590 ("fpmerge_vis"): Likewise.
18591 ("faligndata<VM64:mode>_vis"): Likewise.
18592 ("bshuffle<VM64:mode>_vis"): Likewise.
18593 ("cmask8<P:mode>_vis"): Likewise.
18594 ("cmask16<P:mode>_vis"): Likewise.
18595 ("cmask32<P:mode>_vis"): Likewise.
18596 ("fchksm16_vis"): Likewise.
18597 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18598 ("fmean16_vis"): Likewise.
18599 ("fp<plusminus_insn>64_vis"): Likewise.
18600 ("<plusminus_insn>v8qi3"): Likewise.
18601 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18602 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18603 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18604 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18605 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18606 ("*movqi_insn"): Likewise.
18607 ("*movhi_insn"): Likewise.
18608 ("*movsi_insn"): Likewise.
18609 ("movsi_pic_gotdata_op"): Likewise.
18610 ("*movdi_insn_sp32"): Likewise.
18611 ("*movdi_insn_sp64"): Likewise.
18612 ("movdi_pic_gotdata_op"): Likewise.
18613 ("*movsf_insn"): Likewise.
18614 ("*movdf_insn_sp32"): Likewise.
18615 ("*movdf_insn_sp64"): Likewise.
18616 ("*zero_extendhisi2_insn"): Likewise.
18617 ("*zero_extendqihi2_insn"): Likewise.
18618 ("*zero_extendqisi2_insn"): Likewise.
18619 ("*zero_extendqidi2_insn"): Likewise.
18620 ("*zero_extendhidi2_insn"): Likewise.
18621 ("*zero_extendsidi2_insn_sp64"): Likewise.
18622 ("ldfsr"): Likewise.
18623 ("prefetch_64"): Likewise.
18624 ("prefetch_32"): Likewise.
18625 ("tie_ld32"): Likewise.
18626 ("tie_ld64"): Likewise.
18627 ("*tldo_ldub_sp32"): Likewise.
18628 ("*tldo_ldub1_sp32"): Likewise.
18629 ("*tldo_ldub2_sp32"): Likewise.
18630 ("*tldo_ldub_sp64"): Likewise.
18631 ("*tldo_ldub1_sp64"): Likewise.
18632 ("*tldo_ldub2_sp64"): Likewise.
18633 ("*tldo_ldub3_sp64"): Likewise.
18634 ("*tldo_lduh_sp32"): Likewise.
18635 ("*tldo_lduh1_sp32"): Likewise.
18636 ("*tldo_lduh_sp64"): Likewise.
18637 ("*tldo_lduh1_sp64"): Likewise.
18638 ("*tldo_lduh2_sp64"): Likewise.
18639 ("*tldo_lduw_sp32"): Likewise.
18640 ("*tldo_lduw_sp64"): Likewise.
18641 ("*tldo_lduw1_sp64"): Likewise.
18642 ("*tldo_ldx_sp64"): Likewise.
18643 ("*mov<VM32:mode>_insn"): Likewise.
18644 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18645 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18647 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18649 * config/sparc/sparc.md ("type"): New insn type viscmp.
18650 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18652 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18653 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18654 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18655 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18657 ("n7_vis_logical_11cycle"): Likewise.
18658 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18659 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18660 * config/sparc/niagara.md ("niag_vis"): Likewise.
18661 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18662 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18664 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18666 * config/sparc/sparc.md: New instruction type `bmask'.
18667 (bmaskdi_vis): Use the `bmask' type.
18668 (bmasksi_vis): Likewise.
18669 * config/sparc/ultra3.md (us3_array): Likewise.
18670 * config/sparc/niagara7.md (n7_array): Likewise.
18671 * config/sparc/niagara4.md (n4_array): Likewise.
18672 * config/sparc/niagara2.md (niag2_vis): Likewise.
18673 (niag3_vis): Likewise.
18674 * config/sparc/niagara.md (niag_vis): Likewise.
18676 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18678 * ipa-comdats.c: Remove optimize check from gate.
18679 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18680 for functions not optimized.
18681 (ipa_fn_summary_read): Skip optimize check.
18682 (ipa_fn_summary_write): Likewise.
18683 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18685 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18687 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18688 (check_callers): Give up on caller which is not optimized.
18689 (inline_small_functions): Likewise.
18690 (ipa_inline): Do not give up when not optimizing.
18691 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18692 away unoptimizes cdtors.
18693 (whole_program_function_and_variable_visibility): Do
18694 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18695 * ipa.c (process_references): Do not check optimize.
18696 (symbol_table::remove_unreachable_nodes): Update optimize check.
18697 (set_writeonly_bit): Update optimize check.
18698 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18699 (pass_ipa_single_use::gate): Remove.
18701 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18703 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18704 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18705 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18706 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18707 permute_load, permute_store, adjust_extract, adjust_splat,
18708 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18709 replace_swap_with_copy, dump_swap_insn_table,
18710 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18711 recombine_lvx_pattern, recombine_stvx_pattern,
18712 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18713 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18714 to file rs6000-p8swap.c.
18715 * config/rs6000/rs6000-p8swap.c: New file.
18716 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18717 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18718 and rs6000*-*-* targets.
18720 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18722 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18724 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18726 * lto-wrapper.c (merge_and_complain): Do not merge
18727 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18728 fsigned_zeros, ftrapping_math, fwrapv.
18729 (append_compiler_options): Do not track these options.
18730 (append_linker_options): Likewie
18732 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18734 * cgraphunit.c (cgraph_node::finalize_function): When
18735 !flag_toplevel_reorde set no_reorder flag.
18736 (varpool_node::finalize_decl): Likewise.
18737 (symbol_table::compile): Drop no toplevel reorder path.
18739 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18741 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18742 edges; zero probability is not better than uninitialized.
18744 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18746 * asan.h (asan_sanitize_allocas_p): Declare.
18747 * asan.c (asan_sanitize_allocas_p): New function.
18748 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18749 (handle_builtin_alloca): Likewise.
18750 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18751 if !asan_sanitize_allocas_p.
18752 * params.def (asan-instrument-allocas): Add new option.
18753 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18754 * opts.c (common_handle_option): Disable allocas sanitization for
18757 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18759 * asan.c: Include gimple-fold.h.
18760 (get_last_alloca_addr): New function.
18761 (handle_builtin_stackrestore): Likewise.
18762 (handle_builtin_alloca): Likewise.
18763 (asan_emit_allocas_unpoison): Likewise.
18764 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18765 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18766 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18767 (instrument_builtin_call): Pass gimple iterator to
18768 get_mem_refs_of_builtin_call.
18769 (last_alloca_addr): New global.
18770 * asan.h (asan_emit_allocas_unpoison): Declare.
18771 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18772 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18773 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18774 if function calls alloca.
18775 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18776 * gimple-fold.h (replace_call_with_value): Declare.
18777 * internal-fn.c: Include asan.h.
18778 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18779 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18781 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18783 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18784 (C_SELFTEST_FLAGS): New.
18785 (CPP_SELFTEST_FLAGS): New.
18786 (SELFTEST_DEPS): New, from deps of s-selftest.
18787 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18788 (CPP_SELFTEST_DEPS): New.
18789 (selftest): Add dependency on s-selftest-c++.
18790 (s-selftest): Rename to...
18791 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18792 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18793 than SELFTEST_FLAGS.
18794 (selftest-gdb): Rename to...
18795 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18797 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18798 (selftest-valgrind): Rename to...
18799 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18801 (selftest-valgrind): Reintroduce as an alias for
18802 selftest-c-valgrind.
18803 (s-selftest-c++): New.
18804 (selftest-c++-gdb): New.
18805 (selftest-c++-valgrind): New.
18807 2017-07-06 Olivier Hainque <hainque@adacore.com>
18809 * gcc.c (process_command): When deciding if undefined variables
18810 should be ignored when processing specs, accept "gcc -v" as well.
18812 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18814 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18815 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18817 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18819 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18820 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18821 * config/arm/arm-tables.opt: Regenerate.
18822 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18824 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18826 2017-07-06 Carl Love <cel@us.ibm.com>
18828 * ChangeLog: Clean up from mid air collision
18830 2017-07-06 Carl Love <cel@us.ibm.com>
18832 * config/rs6000/rs6000-c.c: Add support for built-in functions
18833 vector signed int vec_subc (vector signed int, vector signed int);
18834 vector signed __int128 vec_subc (vector signed __int128,
18835 vector signed __int128);
18836 vector unsigned __int128 vec_subc (vector unsigned __int128,
18837 vector unsigned __int128);
18838 vector signed int vec_sube (vector signed int, vector signed int,
18839 vector signed int);
18840 vector unsigned int vec_sube (vector unsigned int,
18841 vector unsigned int,
18842 vector unsigned int);
18843 vector signed __int128 vec_sube (vector signed __int128,
18844 vector signed __int128,
18845 vector signed__int128);
18846 vector unsigned __int128 vec_sube (vector unsigned __int128,
18847 vector unsigned __int128,
18848 vector unsigned __int128);
18849 vector signed int vec_subec (vector signed int, vector signed int,
18850 vector signed int);
18851 vector unsigned int vec_subec (vector unsigned int,
18852 vector unsigned int,
18853 vector unsigned int);
18854 vector signed __int128 vec_subec (vector signed __int128,
18855 vector signed __int128,
18856 vector signed__int128);
18857 vector unsigned __int128 vec_subec (vector unsigned __int128,
18858 vector unsigned __int128,
18859 vector unsigned __int128);
18860 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18861 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18862 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18863 BU_ALTIVEC_OVERLOAD_X definitions.
18864 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18865 * doc/extend.texi: Update the built-in documentation file for the new
18866 built-in functions.
18868 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18871 * diagnostic-show-locus.c (layout::layout): Use start and finish
18872 spelling location for the start and finish of each range.
18873 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18874 Add unused aspect param.
18875 * input.c (expand_location_1): Add "aspect" param, and use it
18876 to access the correct part of the location.
18877 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18879 (expand_location_to_spelling_point): Likewise.
18880 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18881 param, and pass it to expand_location_1.
18883 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18885 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18886 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18887 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18888 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18889 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18890 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18891 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18892 _mm_maskz_getmant_ss): New intrinsics.
18893 (__builtin_ia32_getexpss128_mask): Changed to ...
18894 __builtin_ia32_getexpss128_round ... this.
18895 (__builtin_ia32_getexpsd128_mask): Changed to ...
18896 __builtin_ia32_getexpsd128_round ... this.
18897 * config/i386/i386-builtin-types.def
18898 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18899 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18900 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18901 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18902 __builtin_ia32_getmantss_mask_round): New builtins.
18903 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18904 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18905 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18906 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18907 * config/i386/sse.md
18908 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18909 avx512f_sgetexp<mode><mask_scalar_name>
18910 <round_saeonly_scalar_name> ... this.
18911 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18912 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18913 vgetexp<ssescalarmodesuffix>
18914 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18915 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18916 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18917 avx512f_vgetmant<mode><mask_scalar_name>
18918 <round_saeonly_scalar_name> ... this.
18919 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18920 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18921 vgetmant<ssescalarmodesuffix>
18922 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18923 %0<mask_scalar_operand4>, %1, %2
18924 <round_saeonly_scalar_mask_op4>, %3} ... this.
18925 * config/i386/subst.md (mask_scalar_operand4,
18926 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18927 round_saeonly_scalar_nimm_predicate): New subst attributes.
18929 2017-07-06 Julia Koval <julia.koval@intel.com>
18931 * config/i386/i386.c (ix86_erase_embedded_rounding):
18932 Remove code for old rounding pattern.
18934 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18936 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18938 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18940 * doc/sourcebuild.texi (Test Directives, Variants of
18941 dg-require-support): Add documentation for dg-require-stack-check.
18943 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18945 * config/i386/subst.md (mask_scalar, round_scalar,
18946 round_saeonly_scalar): New meta-templates.
18947 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18948 round_scalar_mask_operand3, round_scalar_mask_op3,
18949 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18950 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18951 round_saeonly_scalar_constraint,
18952 round_saeonly_scalar_prefix): New subst attribute.
18953 * config/i386/sse.md
18954 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18955 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18956 <round_scalar_name> ... this.
18957 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18958 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18959 <round_scalar_name> ... this.
18960 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18961 <sse>_vm<code><mode>3<mask_scalar_name>
18962 <round_saeonly_scalar_name> ... this.
18963 (v<plusminus_mnemonic><ssescalarmodesuffix>
18964 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18965 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18966 v<plusminus_mnemonic><ssescalarmodesuffix>
18967 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18968 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18969 (v<multdiv_mnemonic><ssescalarmodesuffix>
18970 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18971 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18972 v<multdiv_mnemonic><ssescalarmodesuffix>
18973 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18974 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18975 (v<maxmin_float><ssescalarmodesuffix>
18976 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18977 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18978 v<maxmin_float><ssescalarmodesuffix>
18979 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18980 %0<mask_scalar_operand3>, %1, %<iptr>2
18981 <round_saeonly_scalar_mask_op3>} ... this.
18983 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18985 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18986 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18988 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18989 Alan Hayward <alan.hayward@arm.com>
18990 David Sherwood <david.sherwood@arm.com>
18992 * combine.c (simplify_if_then_else): Remove "enum" before
18994 * compare-elim.c (can_eliminate_compare): Likewise.
18995 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18997 (aarch64_lookup_simd_builtin_type): Likewise.
18998 (aarch64_simd_builtin_type): Likewise.
18999 (aarch64_init_simd_builtin_types): Likewise.
19000 (aarch64_simd_expand_args): Likewise.
19001 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
19003 (aarch64_reverse_mask): Likewise.
19004 (aarch64_simd_emit_reg_reg_move): Likewise.
19005 (aarch64_gen_adjusted_ldpstp): Likewise.
19006 (aarch64_ccmp_mode_to_code): Likewise.
19007 (aarch64_operands_ok_for_ldpstp): Likewise.
19008 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19009 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19011 (aarch64_min_divisions_for_recip_mul): Likewise.
19012 (aarch64_reassociation_width): Likewise.
19013 (aarch64_get_condition_code_1): Likewise.
19014 (aarch64_simd_emit_reg_reg_move): Likewise.
19015 (aarch64_simd_attr_length_rglist): Likewise.
19016 (aarch64_reverse_mask): Likewise.
19017 (aarch64_operands_ok_for_ldpstp): Likewise.
19018 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
19019 (aarch64_gen_adjusted_ldpstp): Likewise.
19020 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
19022 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
19023 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19024 (arm_lookup_simd_builtin_type): Likewise.
19025 (arm_simd_builtin_type): Likewise.
19026 (arm_init_simd_builtin_types): Likewise.
19027 (arm_expand_builtin_args): Likewise.
19028 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
19029 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
19030 (ft32_setup_incoming_varargs): Likewise.
19031 (ft32_function_arg): Likewise.
19032 (ft32_function_arg_advance): Likewise.
19033 (ft32_pass_by_reference): Likewise.
19034 (ft32_arg_partial_bytes): Likewise.
19035 (ft32_valid_pointer_mode): Likewise.
19036 (ft32_addr_space_pointer_mode): Likewise.
19037 (ft32_addr_space_legitimate_address_p): Likewise.
19038 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19040 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19041 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19042 (iamcu_alignment): Likewise.
19043 (canonicalize_vector_int_perm): Likewise.
19044 (ix86_noce_conversion_profitable_p): Likewise.
19045 (ix86_mpx_bound_mode): Likewise.
19046 (ix86_operands_ok_for_move_multiple): Likewise.
19047 * config/microblaze/microblaze-protos.h
19048 (microblaze_expand_conditional_branch_reg): Likewise.
19049 * config/microblaze/microblaze.c
19050 (microblaze_expand_conditional_branch_reg): Likewise.
19051 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19053 (rs6000_reassociation_width): Likewise.
19054 (rs6000_invalid_binary_op): Likewise.
19055 (fusion_p9_p): Likewise.
19056 (emit_fusion_p9_load): Likewise.
19057 (emit_fusion_p9_store): Likewise.
19058 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19060 (riscv_hard_regno_mode_ok_p): Likewise.
19061 (riscv_address_insns): Likewise.
19062 (riscv_split_symbol): Likewise.
19063 (riscv_legitimize_move): Likewise.
19064 (riscv_function_value): Likewise.
19065 (riscv_hard_regno_nregs): Likewise.
19066 (riscv_expand_builtin): Likewise.
19067 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19068 (riscv_build_integer): Likewise.
19069 (riscv_split_integer): Likewise.
19070 (riscv_legitimate_constant_p): Likewise.
19071 (riscv_cannot_force_const_mem): Likewise.
19072 (riscv_regno_mode_ok_for_base_p): Likewise.
19073 (riscv_valid_base_register_p): Likewise.
19074 (riscv_valid_offset_p): Likewise.
19075 (riscv_valid_lo_sum_p): Likewise.
19076 (riscv_classify_address): Likewise.
19077 (riscv_legitimate_address_p): Likewise.
19078 (riscv_address_insns): Likewise.
19079 (riscv_load_store_insns): Likewise.
19080 (riscv_force_binary): Likewise.
19081 (riscv_split_symbol): Likewise.
19082 (riscv_force_address): Likewise.
19083 (riscv_legitimize_address): Likewise.
19084 (riscv_move_integer): Likewise.
19085 (riscv_legitimize_const_move): Likewise.
19086 (riscv_legitimize_move): Likewise.
19087 (riscv_address_cost): Likewise.
19088 (riscv_subword): Likewise.
19089 (riscv_output_move): Likewise.
19090 (riscv_canonicalize_int_order_test): Likewise.
19091 (riscv_emit_int_order_test): Likewise.
19092 (riscv_function_arg_boundary): Likewise.
19093 (riscv_pass_mode_in_fpr_p): Likewise.
19094 (riscv_pass_fpr_single): Likewise.
19095 (riscv_pass_fpr_pair): Likewise.
19096 (riscv_get_arg_info): Likewise.
19097 (riscv_function_arg): Likewise.
19098 (riscv_function_arg_advance): Likewise.
19099 (riscv_arg_partial_bytes): Likewise.
19100 (riscv_function_value): Likewise.
19101 (riscv_pass_by_reference): Likewise.
19102 (riscv_setup_incoming_varargs): Likewise.
19103 (riscv_print_operand): Likewise.
19104 (riscv_elf_select_rtx_section): Likewise.
19105 (riscv_save_restore_reg): Likewise.
19106 (riscv_for_each_saved_reg): Likewise.
19107 (riscv_register_move_cost): Likewise.
19108 (riscv_hard_regno_mode_ok_p): Likewise.
19109 (riscv_hard_regno_nregs): Likewise.
19110 (riscv_class_max_nregs): Likewise.
19111 (riscv_memory_move_cost): Likewise.
19112 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19113 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19114 (rl78_addr_space_address_mode): Likewise.
19115 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19117 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19118 (rs6000_reassociation_width): Likewise.
19119 (rs6000_invalid_binary_op): Likewise.
19120 (fusion_p9_p): Likewise.
19121 (emit_fusion_p9_load): Likewise.
19122 (emit_fusion_p9_store): Likewise.
19123 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19124 (ok_for_simple_move_operands): Likewise.
19125 (ok_for_simple_move_strict_operands): Likewise.
19126 (ok_for_simple_arith_logic_operands): Likewise.
19127 (visium_legitimize_reload_address): Likewise.
19128 (visium_select_cc_mode): Likewise.
19129 (output_cbranch): Likewise.
19130 (visium_split_double_move): Likewise.
19131 (visium_expand_copysign): Likewise.
19132 (visium_expand_int_cstore): Likewise.
19133 (visium_expand_fp_cstore): Likewise.
19134 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19135 (visium_function_arg): Likewise.
19136 (visium_function_arg_advance): Likewise.
19137 (visium_libcall_value): Likewise.
19138 (visium_setup_incoming_varargs): Likewise.
19139 (visium_legitimate_constant_p): Likewise.
19140 (visium_legitimate_address_p): Likewise.
19141 (visium_legitimize_address): Likewise.
19142 (visium_secondary_reload): Likewise.
19143 (visium_register_move_cost): Likewise.
19144 (visium_memory_move_cost): Likewise.
19145 (prepare_move_operands): Likewise.
19146 (ok_for_simple_move_operands): Likewise.
19147 (ok_for_simple_move_strict_operands): Likewise.
19148 (ok_for_simple_arith_logic_operands): Likewise.
19149 (visium_function_value_1): Likewise.
19150 (rtx_ok_for_offset_p): Likewise.
19151 (visium_legitimize_reload_address): Likewise.
19152 (visium_split_double_move): Likewise.
19153 (visium_expand_copysign): Likewise.
19154 (visium_expand_int_cstore): Likewise.
19155 (visium_expand_fp_cstore): Likewise.
19156 (visium_split_cstore): Likewise.
19157 (visium_select_cc_mode): Likewise.
19158 (visium_split_cbranch): Likewise.
19159 (output_cbranch): Likewise.
19160 (visium_print_operand_address): Likewise.
19161 * expmed.c (flip_storage_order): Likewise.
19162 * expmed.h (emit_cstore): Likewise.
19163 (flip_storage_order): Likewise.
19164 * genrecog.c (validate_pattern): Likewise.
19165 * hsa-gen.c (gen_hsa_addr): Likewise.
19166 * internal-fn.c (expand_arith_overflow): Likewise.
19167 * ira-color.c (allocno_copy_cost_saving): Likewise.
19168 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19169 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19170 (process_invariant_for_inheritance): Likewise.
19171 * lra-eliminations.c (move_plus_up): Likewise.
19172 * omp-low.c (lower_oacc_reductions): Likewise.
19173 * simplify-rtx.c (simplify_subreg): Likewise.
19174 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19175 (TARGET_CHKP_BOUND_MODE): Likewise..
19176 * targhooks.c (default_chkp_bound_mode): Likewise.
19177 (default_setup_incoming_vararg_bounds): Likewise.
19178 * targhooks.h (default_chkp_bound_mode): Likewise.
19179 (default_setup_incoming_vararg_bounds): Likewise.
19180 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19181 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19182 (have_whole_vector_shift): Likewise.
19183 * tree-vect-stmts.c (vectorizable_load): Likewise.
19184 * doc/tm.texi: Regenerate.
19186 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19188 Graceful degrade if Binutils PR21472 is not available.
19191 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19192 .rodata in flash test fails.
19193 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19194 * confgure: Regenerate.
19195 * config.in: Regenerate.
19196 * config/avr/avr.c (avr_asm_named_section)
19197 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19198 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19199 (avr_asm_init_sections): Same.
19201 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19203 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19204 (fma<VH:mode>4_intrinsic): Likewise.
19205 (*fmsub<VCVTF:mode>4): Likewise.
19206 (*fmsub<VH:mode>4_intrinsic): Likewise.
19208 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19211 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19212 Don't depend on "optimize > 0".
19213 (out_movhi_r_mr, out_movqi_mr_r): Same.
19214 (out_movhi_mr_r, out_movqi_r_mr): Same.
19215 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19216 io_address_operand on "optimize > 0".
19218 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19220 * tree-loop-distribution.c: Add general explanantion on the pass.
19221 (generate_loops_for_partition): Mark distributed loop.
19222 (pg_add_dependence_edges): New parameter. Handle alias data
19223 dependence specially and record it in the parameter if asked.
19224 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19225 (init_partition_graph_vertices, add_partition_graph_edge): New.
19226 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19227 (free_partition_graph_vdata, build_partition_graph): New.
19228 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19229 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19230 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19231 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19232 (version_for_distribution_p, finalize_partitions): New.
19233 (distribute_loop): Handle alias data dependence specially. Factor
19234 out loop fusion code as functions and call these functions.
19236 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19238 * tree-loop-distribution.c (classify_partition): New parameter and
19239 better handle reduction statement.
19240 (rdg_build_partitions): Revise comment.
19241 (distribute_loop): Compute statements in all partitions and pass it
19242 to classify_partition.
19244 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19246 * tree-loop-distribution.c (enum partition_type): New.
19247 (struct partition): New field type.
19248 (partition_merge_into): Add parameter. Update partition type.
19249 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19250 (build_rdg_partition_for_vertex): Compute partition type.
19251 (rdg_build_partitions): Dump partition type.
19252 (distribute_loop): Update calls to partition_merge_into.
19254 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19256 * tree-loop-distribution.c (struct ddr_hasher): New.
19257 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19259 (classify_partition): Call get_data_dependence.
19260 (pg_add_dependence_edges): Ditto.
19261 (distribute_loop): Release data dependence hash table.
19263 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19265 * tree-loop-distribution.c (ref_base_address): Delete.
19266 (similar_memory_accesses): Rename ...
19267 (share_memory_accesses): ... to this. Check if partitions access
19268 the same memory reference.
19269 (distribute_loop): Call share_memory_accesses.
19271 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19273 * tree-loop-distribution.c (struct partition): New field recording
19274 its data reference.
19275 (partition_alloc, partition_free): Init and release data refs.
19276 (partition_merge_into): Merge data refs.
19277 (build_rdg_partition_for_vertex): Collect data refs for partition.
19278 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19280 (distribute_loop): Remve data refs from vertice data of partition
19283 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19285 * tree-loop-distribution.c (params.h): Include header file.
19286 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19287 (datarefs_vec): New global var.
19288 (create_rdg_vertices): Use datarefs_vec directly.
19289 (free_rdg): Don't free data references.
19290 (build_rdg): Update use. Don't free data references.
19291 (distribute_loop): Compute global variable for data references.
19292 Bail out if there are too many data references.
19294 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19296 * tree-loop-distribution.c (loop_nest): New global var.
19297 (build_rdg): Use loop directly, rather than loop nest.
19298 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19300 (distribute_loop): Compute global variable loop nest. Update use.
19302 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19304 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19305 (partition_merge_into): New parameter. Dump reason for fusion.
19306 (distribute_loop): Update use of partition_merge_into.
19308 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19310 * tree-loop-distribution.c (bb_top_order_index): New.
19311 (bb_top_order_index_size, bb_top_order_cmp): New.
19312 (stmts_from_loop): Use topological order.
19313 (pass_loop_distribution::execute): Compute and release topological
19314 order for basic blocks.
19316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19318 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19321 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19323 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19324 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19325 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19326 * internal-fn.def (LOOP_DIST_ALIAS): New.
19327 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19328 (fold_loop_internal_call): ... this.
19329 (vect_loop_dist_alias_call): New function.
19330 (set_uid_loop_bbs): Call fold_loop_internal_call.
19331 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19334 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19337 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19338 Require dead FLAGS_REG at the beginning of a peephole.
19340 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19343 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19344 arguments in the call to __builtin_ia32_sbb_u32.
19345 (_subborrow_u64): Swap _X and _Y arguments in the call to
19346 __builtin_ia32_sbb_u64.
19348 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19351 * tree-vrp.c (compare_assert_loc): Turn into a function template
19352 with stable template parameter. Only test if a->e is NULL,
19353 !a->e == !b->e has been verified already. Use e == NULL or
19354 e != NULL instead of e or ! e tests. If stable is true, don't use
19355 iterative_hash_expr, on the other side allow a or b or both NULL
19356 and sort the NULLs last.
19357 (process_assert_insertions): Sort using compare_assert_loc<false>
19358 instead of compare_assert_loc, later sort using
19359 compare_assert_loc<true> before calling process_assert_insertions_for
19360 in a loop. Use break instead of continue once seen NULL pointer.
19362 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19364 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19365 Cortex-R7 and Cortex-R8 processors.
19367 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19369 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19370 uninitialized while src is not.
19372 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19374 * common/config/arm/arm-common.c: Adjust include path for
19376 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19377 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19378 (arm-cpu-data.h): Likewise.
19379 (arm-cpu-cdata.h): Likewise.
19380 * config/arm/arm-cpu.h: Delete.
19381 * config/arm/arm-cpu-cdata.h: Delete.
19382 * config/arm/arm-cpu-data.h: Delete.
19384 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19386 * config/arm/arm-cpus.in (cortex-a55): New.
19387 (cortex-a75): Likewise.
19388 (cortex-a75.cortex-a55): Likewise.
19389 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19391 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19392 * config/arm/arm-cpu-cdata.h: Regenerate.
19393 * config/arm/arm-cpu-data.h: Regenerate.
19394 * config/arm/arm-cpu.h: Regenerate.
19395 * config/arm/arm-tables.opt: Regenerate.
19396 * config/arm/arm-tune.md: Regenerate.
19398 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19400 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19402 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19404 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19407 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19409 PR tree-optimization/81292
19410 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19411 full_string_p, also call adjust_related_strinfos if the adjustment
19412 is simple, otherwise invalidate related strinfos.
19414 2017-07-04 Martin Liska <mliska@suse.cz>
19417 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19418 newly created variable as DECL_IGNORED_P.
19420 2017-07-04 Martin Liska <mliska@suse.cz>
19423 * ipa-inline.c (inline_small_functions):
19424 Use xstrdup_for_dump.
19426 2017-07-04 Tom de Vries <tom@codesourcery.com>
19428 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19430 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19433 * config/darwin.c (darwin_function_switched_text_sections):
19436 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19438 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19440 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19442 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19444 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19446 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19447 min_profitable_iters, and th as inclusive lower bounds.
19448 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19449 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19450 for min_profitable_iters and min_profitable_estimate.
19451 (vect_transform_loop): Treat th as an inclusive lower bound.
19452 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19454 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19457 * config/darwin.c (darwin_function_switched_text_sections):
19458 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19459 in two pieces, and suppress the use of buf.
19461 2017-07-03 Nathan Sidwell <nathan@acm.org>
19463 * hash-table.h (hash_table_mod1): Fix indentation.
19465 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19467 PR middle-end/81290
19468 * predict.c (force_edge_cold): Be more careful about propagation
19470 * profile-count.h (profile_probability::guessed,
19471 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19473 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19475 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19477 * doc/invoke.texi (rcpc architecture extension): Document it.
19479 2017-07-03 Richard Biener <rguenther@suse.de>
19481 PR tree-optimization/60510
19482 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19483 the scalar reduction PHI and use it.
19484 (vectorizable_reduction): Properly guard the single_defuse_cycle
19485 path for non-SLP reduction chains where we cannot use it.
19486 Rework reduc_def/index and vector type deduction. Rework
19487 vector operand gathering during reduction op code-gen.
19488 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19489 chains dissolve the chain and leave it to non-SLP reduction
19492 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19494 * tree-data-ref.h (dr_alignment): Declare.
19495 * tree-data-ref.c (dr_alignment): New function.
19496 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19497 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19499 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19501 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19503 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19504 and base_misalignment fields.
19505 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19506 * tree-data-ref.c: Include builtins.h.
19507 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19508 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19509 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19510 * tree-vect-data-refs.c: Include tree-cfg.h.
19511 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19512 fields instead of calculating an alignment here.
19513 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19514 innermost_loop_behavior fields.
19516 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19518 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19520 (DR_STEP_ALIGNMENT): New macro.
19521 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19522 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19523 (create_data_ref): Print it.
19524 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19525 to tell whether the step preserves vector (mis)alignment.
19526 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19527 Move the check for an integer step and generalise to all INTEGER_CST.
19528 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19529 Print the outer step alignment.
19531 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19533 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19534 with offset_alignment.
19535 (DR_ALIGNED_TO): Delete.
19536 (DR_OFFSET_ALIGNMENT): New macro.
19537 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19538 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19539 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19540 (create_data_ref): Likewise.
19541 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19542 (vect_analyze_data_refs): Likewise.
19543 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19544 creating dummy innermost behavior.
19546 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19548 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19549 with a "innermost_loop_behavior *" and refeence tree.
19550 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19551 (create_data_ref): Update call accordingly.
19552 * tree-predcom.c (find_looparound_phi): Likewise.
19554 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19556 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19557 fields with dr_wrt_vec_loop.
19558 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19559 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19560 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19561 (vect_dr_behavior): New function.
19562 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19563 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19564 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19565 track whether the step preserves the misalignment.
19566 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19567 Use vect_dr_behavior.
19568 (vect_setup_realignment): Update call accordingly.
19569 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19570 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19571 call to vect_create_addr_base_for_vector_ref.
19572 (vect_create_cond_for_align_checks): Likewise.
19573 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19574 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19575 (vect_recog_mask_conversion_pattern): Likewise.
19576 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19577 (new_stmt_vec_info): Remove redundant zeroing.
19579 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19581 * common/config/arm/arm-common.c (arm_be8_option): New function.
19582 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19583 (ISA_ARMv6): Add isa_bit_be8.
19584 * config/arm/arm.h (arm_be8_option): Add prototype.
19585 (BE8_SPEC_FUNCTION): New define.
19586 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19587 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19588 (mlittle-endian): Similarly.
19589 (mbe8, mbe32): New options.
19590 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19591 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19593 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19595 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19597 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19599 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19600 (cleanup_tree_cfg_bb): Use it.
19601 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19603 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19605 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19608 * loop-doloop.c (add_test): Update profile.
19610 2017-07-03 Martin Liska <mliska@suse.cz>
19613 * sanopt.c (rewrite_usage_of_param): New function.
19614 (sanitize_rewrite_addressable_params): Likewise.
19615 (pass_sanopt::execute): Call rewrite_usage_of_param.
19617 2017-07-03 Richard Biener <rguenther@suse.de>
19619 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19620 back to using VIEW_CONVERT_EXPR.
19622 2017-07-03 Martin Liska <mliska@suse.cz>
19625 * doc/extend.texi: Document when a resolver function is
19626 generated for target_clones.
19628 2017-07-03 Martin Liska <mliska@suse.cz>
19630 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19631 and shadow memory of auto variables which are subject of
19632 use-after-scope sanitization.
19633 (asan_expand_mark_ifn): Add do set only when is_poison.
19635 2016-07-03 Richard Biener <rguenther@suse.de>
19637 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19639 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19640 (vectorizable_reduction): Perform reduction PHI creation when
19641 visiting a reduction PHI and adjust and simplify code generation
19642 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19643 (vect_transform_loop): Visit reduction PHIs.
19644 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19645 defs into the SLP tree.
19646 (vect_build_slp_tree): Reduction defs terminate the recursion.
19647 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19649 (vect_get_vec_defs_for_stmt_copy): Export.
19650 (vect_get_vec_defs): Likewise.
19651 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19653 (vect_get_vec_defs_for_stmt_copy): Declare.
19654 (vect_get_vec_defs): Likewise.
19656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19658 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19659 parameter with a "loop" parameter and use it instead of the
19660 loop containing DR_STMT. Don't check simple_iv when doing
19661 BB analysis. Describe the two analysis modes in the comment.
19663 2017-07-03 Tom de Vries <tom@codesourcery.com>
19665 PR tree-optimization/69468
19666 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19667 (find_same_succ_bb): Handle ignore_edge_flags.
19669 2017-07-03 Tom de Vries <tom@codesourcery.com>
19671 PR tree-optimization/81192
19672 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19674 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19676 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19678 2017-07-03 Tom de Vries <tom@codesourcery.com>
19680 PR tree-optimization/81192
19681 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19682 BB_SAME_SUCC (bb) == NULL.
19684 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19686 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19689 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19691 * dumpfile.c: Include profile-count.h
19692 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19694 (insert_cond_bb): Update profile.
19695 * tree-cfg.h (insert_cond_bb): Update prototype.
19696 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19697 * tree-dump.c: Do not include tree-cfg.
19699 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19701 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19703 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19705 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19708 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19710 * tree-complex.c (expand_complex_div_wide): update profile.
19712 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19713 Alan Hayward <alan.hayward@arm.com>
19714 David Sherwood <david.sherwood@arm.com>
19716 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19717 (CORETYPES_H): New define.
19718 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19719 (insn-modes-inline.h, s-modes-inline-h): New rules.
19720 (generated_files): Add insn-modes-inline.h.
19721 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19722 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19723 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19724 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19725 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19726 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19727 (build/gencodes.o, build/genconditions.o): Likewise.
19728 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19729 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19730 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19731 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19732 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19733 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19734 * coretypes.h: Include everything up to real.h for generators.
19735 Include insn-modes.h first. Include wide-int-print.h after
19736 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19737 * machmode.h: Don't include insn-modes.h here.
19738 * function-tests.c: Remove includes of signop.h, machmode.h,
19739 double-int.h and wide-int.h.
19741 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19743 * optc-save-gen.awk: Likewise.
19744 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19745 * godump.c: Remove include of wide-int-print.h.
19746 * pretty-print.h: Likewise.
19747 * wide-int-print.cc: Likewise.
19748 * wide-int.cc: Likewise.
19749 * hash-map-tests.c: Remove include of signop.h.
19750 * hash-set-tests.c: Likewise.
19751 * rtl-tests.c: Likewise.
19752 * mkconfig.sh: Remove include of machmode.h.
19753 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19755 (emit_insn_modes_inline_h): ...this new function. Emit the code
19756 into an insn-modes-inline.h header file, adding appropriate
19757 include guards and end comments.
19758 (emit_insn_modes_c_header): Remove include of machmode.h.
19759 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19761 (main): Handle -i flag and call emit_insn_modes_inline_h when
19764 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19766 * tree-ssa-strlen.c (strinfo): Rename the length field to
19767 nonzero_chars. Add a full_string_p field.
19768 (compare_nonzero_chars, zero_length_string_p): New functions.
19769 (get_addr_stridx): Add an offset_out parameter.
19770 Use compare_nonzero_chars.
19771 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19772 (new_strinfo): Update after above changes to strinfo.
19773 (set_endptr_and_length): Set full_string_p.
19774 (get_string_length): Update after above changes to strinfo.
19775 (unshare_strinfo): Update call to new_strinfo.
19776 (maybe_invalidate): Likewise.
19777 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19778 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19779 as a uhwi instead of an shwi. Update after above changes to
19780 strinfo and new_strinfo.
19781 (zero_length_string): Assert that chainsi contains full strings.
19782 Use zero_length_string_p. Update call to new_strinfo.
19783 (adjust_related_strinfos): Update after above changes to strinfo.
19784 Copy full_string_p from origsi.
19785 (adjust_last_stmt): Use zero_length_string_p.
19786 (handle_builtin_strlen): Update after above changes to strinfo and
19787 new_strinfo. Install the lhs as the string length if the previous
19788 entry didn't describe a full string.
19789 (handle_builtin_strchr): Update after above changes to strinfo
19791 (handle_builtin_strcpy): Likewise.
19792 (handle_builtin_strcat): Likewise.
19793 (handle_builtin_malloc): Likewise.
19794 (handle_pointer_plus): Likewise.
19795 (handle_builtin_memcpy): Likewise. Track nonzero characters
19796 that aren't necessarily followed by a nul terminator.
19797 (handle_char_store): Likewise.
19799 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19801 PR tree-optimization/80769
19802 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19803 for malloc and calloc. Document the new invariant that all related
19804 strinfos have delayed lengths or none do.
19805 (verify_related_strinfos): Move earlier in file.
19806 (set_endptr_and_length): New function, split out from...
19807 (get_string_length): ...here. Also set the lengths of related
19809 (zero_length_string): Assert that chainsi has known (rather than
19811 (adjust_related_strinfos): Likewise.
19813 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19815 PR tree-optimization/81136
19816 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19817 assert that two references with the same misalignment have the same
19818 compile-time misalignment if those compile-time misalignments
19821 2017-07-01 Andi Kleen <ak@linux.intel.com>
19823 * print-tree.c (print_node): Print all attributes.
19825 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19827 * cfg.c (scale_bbs_frequencies): New function.
19828 * cfg.h (scale_bbs_frequencies): Declare it.
19829 * cfgloopanal.c (single_likely_exit): Cleanup.
19830 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19832 (scale_loop_profile): Likewise.
19833 (loop_version): Likewise.
19834 (create_empty_loop_on_edge): Update.
19835 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19836 scale_loop_frequencies, scale_loop_profile, loopify,
19837 loop_version): Update prototypes.
19838 * modulo-sched.c (sms_schedule): Update.
19839 * predict.c (unlikely_executed_edge_p): Also check probability.
19840 (probably_never_executed_edge_p): Fix typo.
19841 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19842 * tree-parloops.c (gen_parallel_loop): Update.
19843 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19844 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19845 * tree-ssa-loop-split.c (split_loop): Update.
19846 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19847 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19848 (vect_loop_versioning): Update.
19849 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19851 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19853 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19855 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19857 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19858 to keep profile consistent.
19860 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19862 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19863 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19864 * profile-count.h (max_safe_multiplier): Make unsigned.
19865 (profile_count::guessed_zero): New.
19867 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19869 * bb-reorder.c (fix_up_crossing_landing_pad,
19870 fix_crossing_conditional_branches): Use make_single_succ_edge
19871 to keep profile consistent.
19873 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19875 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19878 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19881 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19882 the right scopes, make sure cond_jump isn't preserved between multiple
19883 iterations. Search for fallthru edge whenever there are 3+ edges and
19884 use find_fallthru_edge for it.
19886 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19888 Patch by Alexander Monakov <amonakov@ispras.ru>
19889 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19890 probabilities consistently.
19892 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19894 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19895 emit_cmp_and_jump_insns.
19897 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19900 * tree-inline.c (expand_call_inline): Combine profile statuses.
19902 2017-06-30 Andrew Pinski <apinski@cavium.com>
19904 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19905 fold_stmt returned true.
19907 2017-06-30 Nathan Sidwell <nathan@acm.org>
19909 * ggc.h (empty_string): Delete.
19910 * cfgexpand.c (expand_asm_stmt): Use plain "".
19911 * optabs.c (expand_asm_memory_barrier): Likewise.
19912 * stringpool.c (empty_string): Delete.
19913 (digit_vector, digit_string): Delete.
19914 (ggc_alloc_string): Use plain "", don't optimize single digit
19915 strings. Use ggc_alloc_atomic.
19917 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19919 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19920 comparison set and one other set, use the cost of the non-comparison
19923 2017-06-30 Nathan Sidwell <nathan@acm.org>
19925 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19928 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19930 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19931 loops. Remove now unneeded calls to gimple_switch_set_label() that
19932 just set removed labels to NULL_TREE.
19934 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19936 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19937 (set_range_info): ...here. Only call set_range_info_raw if domain
19939 (set_nonzero_bits): Call set_range_info_raw.
19940 * tree-ssanames.h (set_range_info_raw): New.
19942 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19945 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19946 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19947 of nonimmediate_operand and <store_mask_constraint> instead of m
19948 for the input operand. For V8FI iterator, always split if input
19949 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19950 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19951 <store_mask_predicate> instead of register_operand and
19952 <store_mask_constraint> instead of v for the input operand. Make
19953 sure both operands aren't MEMs for if not <mask_applied>.
19955 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19957 * lto-wrapper.c (copy_file) Close both file descriptors before
19960 2017-06-30 Martin Liska <mliska@suse.cz>
19963 * multiple_target.c (create_dispatcher_calls): Make ifunc
19964 also for function that don't have calls or are not referenced.
19966 2017-06-30 Richard Biener <rguenther@suse.de>
19968 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19969 analyze the first scalar stmt. Move vector type computation
19970 for the BB case here from ...
19971 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19972 live operation processing in the SLP case properly.
19974 2017-06-30 Richard Biener <rguenther@suse.de>
19976 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19978 2017-06-30 Martin Liska <mliska@suse.cz>
19981 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19982 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19984 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19986 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19989 2017-06-30 Richard Biener <rguenther@suse.de>
19991 PR tree-optimization/81249
19992 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19993 condition reduction result to original scalar type.
19995 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19997 * profile-count.h (enum profile_quality): Fix typos and whitespace
20000 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20002 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
20003 type for branch probabilities.
20005 2017-06-29 Julian Brown <julian@codesourcery.com>
20006 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20008 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
20009 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
20010 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
20011 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
20013 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20015 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
20016 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
20017 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
20018 CC usage from generic code to here.
20019 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
20020 CC usage into the target macros.
20022 2017-06-29 Maya Rashish <coypu@sdf.org>
20024 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
20027 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20029 * arm/arm-builtins.c: Include profile-count.h
20030 * except.c (sjlj_emit_function_enter): Use
20031 profile_probability::unlikely.
20033 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20035 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20036 and tocrel_offset be pointer args rather than implicitly using
20038 (legitimate_constant_pool_address_p, rs6000_emit_move,
20039 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20040 tocrel_offset and use in toc_relative_expr_p call.
20041 (print_operand, print_operand_address): Use static tocrel_base_oac
20042 and tocrel_offset_oac.
20043 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20046 2017-06-29 Maya Rashish <coypu@sdf.org>
20048 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20050 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20052 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20053 objects, take into account only the alignment of 'op0' and 'mode1' if
20056 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20058 * ccmp.c (ccmp_tree_comparison_p): New function.
20059 (ccmp_candidate_p): Update to use above function.
20060 (get_compare_parts): New function.
20061 (expand_ccmp_next): Update to use new functions.
20062 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20064 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20065 take mode as argument.
20066 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20067 * expr.c (expand_expr_real_1): Pass mode as argument.
20069 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20071 * combine.c (combine_instructions): Print insns to dump_file, together
20074 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20076 * asan.c (asan_emit_stack_protection): Update.
20077 (create_cond_insert_point): Update.
20078 * auto-profile.c (afdo_propagate_circuit): Update.
20079 * basic-block.h (struct edge_def): Turn probability to
20080 profile_probability.
20081 (EDGE_FREQUENCY): Update.
20082 * bb-reorder.c (find_traces_1_round): Update.
20083 (better_edge_p): Update.
20084 (sanitize_hot_paths): Update.
20085 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20086 (make_single_succ_edge): Update.
20087 (check_bb_profile): Update.
20088 (dump_edge_info): Update.
20089 (update_bb_profile_for_threading): Update.
20090 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20091 probabilitycount to 0.
20092 * cfgbuild.c (compute_outgoing_frequencies): Update.
20093 * cfgcleanup.c (try_forward_edges): Update.
20094 (outgoing_edges_match): Update.
20095 (try_crossjump_to_edge): Update.
20096 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20097 (expand_gimple_tailcall): Update.
20098 (construct_init_block): Use make_single_succ_edge.
20099 (construct_exit_block): Use make_single_succ_edge.
20100 * cfghooks.c (verify_flow_info): Update.
20101 (redirect_edge_succ_nodup): Update.
20102 (split_edge): Update.
20103 (account_profile_record): Update.
20104 * cfgloopanal.c (single_likely_exit): Update.
20105 * cfgloopmanip.c (scale_loop_profile): Update.
20106 (set_zero_probability): Remove.
20107 (duplicate_loop_to_header_edge): Update.
20108 * cfgloopmanip.h (loop_version): Update prototype.
20109 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20110 (force_nonfallthru_and_redirect): Update.
20111 (update_br_prob_note): Update.
20112 (rtl_verify_edges): Update.
20113 (purge_dead_edges): Update.
20114 (rtl_lv_add_condition_to_bb): Update.
20115 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20116 * cgraphunit.c (init_lowered_empty_function): Update.
20117 (cgraph_node::expand_thunk): Update.
20118 * cilk-common.c: Include profile-count.h
20119 * dojump.c (inv): Remove.
20120 (jumpifnot): Update.
20121 (jumpifnot_1): Update.
20122 (do_jump_1): Update.
20124 (do_jump_by_parts_greater_rtx): Update.
20125 (do_compare_rtx_and_jump): Update.
20126 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20127 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20128 * dwarf2cfi.c: Include profile-count.h
20129 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20130 (sjlj_emit_dispatch_table): Likewise.
20131 * explow.c: Include profile-count.h
20132 * expmed.c (emit_store_flag_force): Update.
20133 (do_cmp_and_jump): Update.
20134 * expr.c (compare_by_pieces_d::generate): Update.
20135 (compare_by_pieces_d::finish_mode): Update.
20136 (emit_block_move_via_loop): Update.
20137 (store_expr_with_bounds): Update.
20138 (store_constructor): Update.
20139 (expand_expr_real_2): Update.
20140 (expand_expr_real_1): Update.
20141 * expr.h (try_casesi, try_tablejump): Update prototypes.
20142 * gimple-pretty-print.c (dump_probability): Update.
20143 (dump_profile): New.
20144 (dump_gimple_label): Update.
20145 (dump_gimple_bb_header): Update.
20146 * graph.c (draw_cfg_node_succ_edges): Update.
20147 * hsa-gen.c (convert_switch_statements): Update.
20148 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20149 (find_if_case_1): Update.
20150 (find_if_case_2): Update.
20151 * internal-fn.c (expand_arith_overflow_result_store): Update.
20152 (expand_addsub_overflow): Update.
20153 (expand_neg_overflow): Update.
20154 (expand_mul_overflow): Update.
20155 (expand_vector_ubsan_overflow): Update.
20156 * ipa-cp.c (good_cloning_opportunity_p): Update.
20157 * ipa-split.c (split_function): Use make_single_succ_edge.
20158 * ipa-utils.c (ipa_merge_profiles): Update.
20159 * loop-doloop.c (add_test): Update.
20160 (doloop_modify): Update.
20161 * loop-unroll.c (compare_and_jump_seq): Update.
20162 (unroll_loop_runtime_iterations): Update.
20163 * lra-constraints.c (lra_inheritance): Update.
20164 * lto-streamer-in.c (input_cfg): Update.
20165 * lto-streamer-out.c (output_cfg): Update.
20166 * mcf.c (adjust_cfg_counts): Update.
20167 * modulo-sched.c (sms_schedule): Update.
20168 * omp-expand.c (expand_omp_for_init_counts): Update.
20169 (extract_omp_for_update_vars): Update.
20170 (expand_omp_ordered_sink): Update.
20171 (expand_omp_for_ordered_loops): Update.
20172 (expand_omp_for_generic): Update.
20173 (expand_omp_for_static_nochunk): Update.
20174 (expand_omp_for_static_chunk): Update.
20175 (expand_cilk_for): Update.
20176 (expand_omp_simd): Update.
20177 (expand_omp_taskloop_for_outer): Update.
20178 (expand_omp_taskloop_for_inner): Update.
20179 * omp-simd-clone.c (simd_clone_adjust): Update.
20180 * optabs.c (expand_doubleword_shift): Update.
20181 (expand_abs): Update.
20182 (emit_cmp_and_jump_insn_1): Update.
20183 (expand_compare_and_swap_loop): Update.
20184 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20185 * predict.c (predictable_edge_p): Update.
20186 (edge_probability_reliable_p): Update.
20187 (set_even_probabilities): Update.
20188 (combine_predictions_for_insn): Update.
20189 (combine_predictions_for_bb): Update.
20190 (propagate_freq): Update.
20191 (estimate_bb_frequencies): Update.
20192 (force_edge_cold): Update.
20193 * profile-count.c (profile_count::dump): Add missing space into dump.
20194 (profile_count::debug): Add newline.
20195 (profile_count::differs_from_p): Explicitly convert to unsigned.
20196 (profile_count::stream_in): Update.
20197 (profile_probability::dump): New member function.
20198 (profile_probability::debug): New member function.
20199 (profile_probability::differs_from_p): New member function.
20200 (profile_probability::differs_lot_from_p): New member function.
20201 (profile_probability::stream_in): New member function.
20202 (profile_probability::stream_out): New member function.
20203 * profile-count.h (profile_count_quality): Rename to ...
20204 (profile_quality): ... this one.
20205 (profile_probability): New.
20206 (profile_count): Update.
20207 * profile.c (compute_branch_probabilities): Update.
20208 * recog.c (peep2_attempt): Update.
20209 * sched-ebb.c (schedule_ebbs): Update.
20210 * sched-rgn.c (find_single_block_region): Update.
20211 (compute_dom_prob_ps): Update.
20212 (schedule_region): Update.
20213 * sel-sched-ir.c (compute_succs_info): Update.
20214 * stmt.c (struct case_node): Update.
20215 (do_jump_if_equal): Update.
20216 (get_outgoing_edge_probs): Update.
20217 (conditional_probability): Update.
20218 (emit_case_dispatch_table): Update.
20219 (expand_case): Update.
20220 (expand_sjlj_dispatch_table): Update.
20221 (emit_case_nodes): Update.
20222 * targhooks.c: Update.
20223 * tracer.c (better_p): Update.
20224 (find_best_successor): Update.
20225 * trans-mem.c (expand_transaction): Update.
20226 * tree-call-cdce.c: Update.
20227 * tree-cfg.c (gimple_split_edge): Upate.
20228 (move_sese_region_to_fn): Upate.
20229 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20230 * tree-eh.c (lower_resx): Upate.
20231 (cleanup_empty_eh_move_lp): Upate.
20232 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20233 * tree-inline.c (copy_edges_for_bb): Update.
20234 (copy_cfg_body): Update.
20235 * tree-parloops.c (gen_parallel_loop): Update.
20236 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20237 (gimple_gen_time_profiler): Update.
20238 * tree-ssa-dce.c (remove_dead_stmt): Update.
20239 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20240 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20241 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20242 (unloop_loops): Update.
20243 (try_peel_loop): Update.
20244 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20245 * tree-ssa-loop-split.c (connect_loops): Update.
20246 (split_loop): Update.
20247 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20248 (hoist_guard): Update.
20249 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20250 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20251 (value_replacement): Update.
20252 * tree-ssa-reassoc.c (branch_fixup): Update.
20253 * tree-ssa-tail-merge.c (replace_block_by): Update.
20254 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20255 (create_edge_and_update_destination_phis): Update.
20256 (compute_path_counts): Update.
20257 (recompute_probabilities): Update.
20258 (update_joiner_offpath_counts): Update.
20259 (freqs_to_counts_path): Update.
20260 (duplicate_thread_path): Update.
20261 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20262 (struct switch_conv_info): Update.
20263 (gen_inbound_check): Update.
20264 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20265 (vect_do_peeling): Update.
20266 (vect_loop_versioning): Update.
20267 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20268 (optimize_mask_stores): Update.
20269 * ubsan.c (ubsan_expand_null_ifn): Update.
20270 * value-prof.c (gimple_divmod_fixed_value): Update.
20271 (gimple_divmod_fixed_value_transform): Update.
20272 (gimple_mod_pow2): Update.
20273 (gimple_mod_pow2_value_transform): Update.
20274 (gimple_mod_subtract): Update.
20275 (gimple_mod_subtract_transform): Update.
20276 (gimple_ic): Update.
20277 (gimple_stringop_fixed_value): Update.
20278 (gimple_stringops_transform): Update.
20279 * value-prof.h: Update.
20281 2017-06-29 Carl Love <cel@us.ibm.com>
20283 * config/rs6000/rs6000-c.c: Add support for built-in functions
20284 vector signed int vec_signed (vector float);
20285 vector signed long long vec_signed (vector double);
20286 vector signed int vec_signed2 (vector double, vector double);
20287 vector signed int vec_signede (vector double);
20288 vector signed int vec_signedo (vector double);
20289 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20290 instruction generator.
20291 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20292 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20293 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20295 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20296 vunsignede_v2df): Add define_expands.
20297 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20298 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20299 VEC_UNSIGNEDO): Add definitions.
20300 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20301 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20302 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20303 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20304 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20305 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20306 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20307 vec_unsignede, vec_unsignedo): Add builtin defines.
20308 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20310 * doc/extend.texi: Update the built-in documentation file for the
20311 new built-in functions.
20313 2017-06-29 Richard Biener <rguenther@suse.de>
20315 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20316 reduction chains to LOOP_VINFO_REDUCTIONS.
20317 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20318 SLP reductions after processing reduction chains.
20320 2017-06-29 Nathan Sidwell <nathan@acm.org>
20322 * builtins.c (fold_builtin_FUNCTION): Use
20323 lang_hooks.decl_printable_name.
20325 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20327 PR middle-end/81194
20328 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20329 with only one label.
20330 * stmt.c (expand_case): Assert NCASES is greater than one.
20332 2017-06-29 Richard Biener <rguenther@suse.de>
20334 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20336 (group_case_labels): Likewise.
20337 (find_taken_edge): Push sanity checking on val to workers...
20338 (find_taken_edge_cond_expr): ... here
20339 (find_taken_edge_switch_expr): ... and here, handle cases
20340 with just a default label.
20341 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20342 (group_case_labels): Likewise.
20343 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20344 group_case_labels does anything cleanup the CFG again.
20346 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20348 PR tree-optimization/81196
20349 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20350 exit condition comparing two IVs.
20352 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20354 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20355 profile to the dummy entry at the end of the list of architectures.
20356 * config/arm/arm-cpu-cdata.h: Regenerated.
20358 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20359 Michael Collison <michael.collison@arm.com>
20362 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20364 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20365 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20366 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20367 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20368 of shift when the shift amount is masked with constant equal to
20369 the size of the mode.
20370 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20373 2017-06-29 Martin Liska <mliska@suse.cz>
20375 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20377 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20380 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20383 (edge_growth_cache_entry::edge_growth_cache_entry): New
20385 (reset_edge_growth_cache): Update to use constructor.
20387 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20389 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20390 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20391 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20393 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20395 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20396 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20398 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20400 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20401 (*-linux-uclibc*): Add t-uclibc tmake_file.
20402 * config/t-musl: New.
20403 * config/t-uclibc: New.
20405 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20407 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20409 (gen_comm_data): Emit architectural setting of arch_prof.
20410 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20412 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20413 (armv8-m.base, armv8-m.main): Likewise.
20414 * arm-protos.h (arm_build_target): Add profile field.
20415 (arch_option): Likewise.
20416 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20418 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20419 arm_active_target.profile.
20421 2017-06-28 Richard Biener <rguenther@suse.de>
20423 PR middle-end/81227
20424 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20425 TYPE_OVERFLOW_WRAPS.
20426 * match.pd (negate_expr_p): Likewise.
20427 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20428 fold_build2, not fold_binary.
20430 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20432 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20433 Convert memory address to Pmode.
20434 (aarch64_print_operand): Assert MEM operands are always Pmode.
20436 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20439 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20440 Remove redundant if.
20441 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20442 * config/arm/aarch-common-protos.h
20443 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20444 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20446 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20449 * multiple_target.c (create_dispatcher_calls): Set the default
20450 clone to be static, not public.
20452 2017-06-28 Richard Biener <rguenther@suse.de>
20454 * tree-vect-loop.c (vectorizable_reduction): Move special
20455 cond reduction IV var creation ...
20456 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20457 parameter. Use STMT_VINFO_VECTYPE.
20458 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20461 2017-06-28 Martin Liska <mliska@suse.cz>
20464 * ipa-visibility.c (non_local_p): Handle visibility.
20466 2017-06-28 Martin Liska <mliska@suse.cz>
20469 * common.opt: Add IntegerRange to various options.
20470 * opt-functions.awk (integer_range_info): New function.
20471 * optc-gen.awk: Add integer_range_info to cl_options struct.
20472 * opts-common.c (decode_cmdline_option): Handle
20473 CL_ERR_INT_RANGE_ARG.
20474 (cmdline_handle_error): Likewise.
20475 * opts.c (print_filtered_help): Show valid interval in
20476 when --help is provided.
20477 * opts.h (struct cl_option): Add range_min and range_max fields.
20478 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20480 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20482 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20483 (x * C EQ/NE y * C): New transformation.
20485 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20487 * genmultilib (combination_space): Accept '+' in option names.
20489 2017-06-28 Martin Liska <mliska@suse.cz>
20492 * asan.c (instrument_derefs): Bail out inner references
20493 that are hard register variables.
20495 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20498 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20499 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20501 2017-06-28 Richard Biener <rguenther@suse.de>
20503 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20504 (vect_get_slp_defs): Adjust.
20505 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20507 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20509 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20510 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20511 (vectorizable_reduction): Adjust.
20512 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20514 (vect_get_slp_defs): Likewise.
20515 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20516 (vectorizable_bswap): Adjust.
20517 (vectorizable_call): Likewise.
20518 (vectorizable_conversion): Likewise.
20519 (vectorizable_assignment): Likewise.
20520 (vectorizable_shift): Likewise.
20521 (vectorizable_operation): Likewise.
20522 (vectorizable_store): Likewise.
20523 (vectorizable_condition): Likewise.
20524 (vectorizable_comparison): Likewise.
20526 2017-06-28 Michael Collison <michael.collison@arm.com>
20529 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20531 (arm_gen_movmemqi): Removed unused variable 'i'.
20532 Convert 'for' loop into 'while' loop.
20533 (arm_expand_prologue): Remove last unnecessary set of insn.
20534 (thumb_pop): Remove unused variable 'pushed_words'.
20535 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20537 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20539 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20540 * config/s390/s390-protos.h: Add prototype of
20541 s390_rel_address_ok_p.
20542 * config/s390/s390.c (s390_got_symbol): New function.
20543 (s390_rel_address_ok_p): New function.
20544 (legitimize_pic_address): Use s390_rel_address_ok_p.
20545 (s390_load_got): Use s390_got_symbol.
20546 (s390_option_override): Issue error if
20547 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20548 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20550 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20552 2017-06-27 Andrew Pinski <apinski@cavium.com>
20554 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20555 (X * copysign (1.0, X)): New pattern.
20556 (X * copysign (1.0, -X)): New pattern.
20557 (copysign (-1.0, CST)): New pattern.
20559 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20561 * genmultilib (combination_space): Remove variable.
20562 Validate reuse rules against regular expression for any sequence
20563 of multilib options in any order.
20565 2017-06-27 Michael Collison <michael.collison@arm.com>
20567 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20568 call aarch64_split_simd_combine.
20569 * (aarch64_combine_internal<mode>): Delete pattern.
20570 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20571 Allow register and subreg operands.
20573 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20575 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20576 specific need, just fallback on defaults.
20577 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20579 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20580 Olivier Hainque <hainque@adacore.com>
20582 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20584 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20585 targets. Pick a default if no particular attempt applied.
20586 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20589 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20591 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20592 (x86_64-wrs-vxworks7): Likewise.
20594 2017-06-27 Marek Polacek <polacek@redhat.com>
20597 * ubsan.c (instrument_null): Check get_base_address's result for null.
20599 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20601 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20603 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20605 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20606 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20607 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20608 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20609 New function types.
20610 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20611 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20612 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20613 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20614 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20615 BUILT_IN_FEUPDATEENV): New builtins.
20616 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20617 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20618 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20619 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20621 (builtin_structptr_types): Adjust size.
20622 * tree.c (builtin_structptr_types): Add four entries.
20624 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20625 Olivier Hainque <hainque@adacore.com>
20627 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20628 (TLS_SYM): New local macro, forcing reference to __tls__ on
20629 link command lines for VxWorks 7 RTPs, triggering initialization
20631 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20632 OS features TLS support, true for RTPs on VxWorks 7.
20633 * config/vxworks.c (vxworks_override_options): Setup emutls
20636 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20638 * predict.c (test_prediction_value_range): Use -1U instead of -1
20639 to avoid narrowing conversion warning.
20640 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20641 to avoid narrowing conversion warning.
20642 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20644 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20646 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20648 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20649 64bit configurations.
20650 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20651 (SIZE_TYPE): Likewise.
20652 * config/vxworks.c (vxworks_emutls_var_fields): Use
20653 long_unsigned_type_node instead of unsigned_type_node as the offset
20654 field type, which is "pointer" mode in emutls.c.
20656 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20659 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20661 PR middle-end/81207
20662 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20663 gimple_vuse copying separately from gimple_vdef copying.
20665 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20667 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20670 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20671 Olivier Hainque <hainque@adacore.com>
20673 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20674 for all vxworks7 targets.
20675 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20676 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20677 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20678 variations for VX6/VX7 and 32/64bits later on in ...
20679 (VXWORKS_LIB_SPEC): Leverage new macros.
20680 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20681 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20683 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20685 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20686 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20688 2017-06-26 Carl Love <cel@us.ibm.com>
20690 * config/rs6000/rs6000-c.c: Add support for built-in functions
20691 vector bool char vec_reve (vector bool char);
20692 vector signed char vec_reve (vector signed char);
20693 vector unsigned char vec_reve (vector unsigned char);
20694 vector bool int vec_reve (vector bool int);
20695 vector signed int vec_reve (vector signed int);
20696 vector unsigned int vec_reve (vector unsigned int);
20697 vector bool long long vec_reve (vector bool long long);
20698 vector signed long long vec_reve (vector signed long long);
20699 vector unsigned long long vec_reve (vector unsigned long long);
20700 vector bool short vec_reve (vector bool short);
20701 vector signed short vec_reve (vector signed short);
20702 vector double vec_reve (vector double);
20703 vector float vec_reve (vector float);
20704 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20705 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20706 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20707 (altivec_vreve): New pattern.
20708 * config/rs6000/altivec.h (vec_reve): New define.
20709 * doc/extend.texi (vec_rev): Update the built-in documentation file
20710 for the new built-in functions.
20712 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20714 PR tree-optimization/71815
20715 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20717 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20719 (slsr_process_phi): Likewise.
20720 (replace_uncond_cands_and_profitable_phis): Don't replace a
20721 multiply candidate with a stride of 1 (copy or cast).
20722 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20724 (lowest_cost_path): Likewise.
20725 (total_savings): Likewise.
20727 2017-06-26 Richard Biener <rguenther@suse.de>
20730 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20731 Use def_builtin_pure for all gather builtins.
20733 2017-06-26 Richard Biener <rguenther@suse.de>
20735 PR tree-optimization/81203
20736 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20737 non-dominating BBs.
20739 2017-06-26 Marek Polacek <polacek@redhat.com>
20742 * doc/invoke.texi: Document -Wmultistatement-macros.
20744 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20746 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20747 arm_neon_ok_no_float_abi effective target.
20749 2017-06-26 Richard Biener <rguenther@suse.de>
20751 PR tree-optimization/80928
20752 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20753 (copy_bbs): Set BB_DUPLICATED flag early.
20754 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20756 (execute_on_shrinking_pred): Likewise.
20757 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20758 BB_DUPLICATED blocks.
20759 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20760 iterate over all PHIs considering removal of *gsi.
20762 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20764 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20767 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20769 * config/rs6000/rs6000-string.c: (expand_block_clear,
20770 do_load_for_compare, select_block_compare_mode,
20771 compute_current_alignment, expand_block_compare,
20772 expand_strncmp_align_check, expand_strn_compare,
20773 expand_block_move, rs6000_output_load_multiple)
20774 Move functions related to string/block move/compare
20775 to a separate file.
20776 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20777 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20778 for this function which is now used in two files.
20779 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20780 * config.gcc: Add rs6000-string.o to extra_objs for
20781 targets powerpc*-*-* and rs6000*-*-*.
20783 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20786 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20787 32-bit, since indexed is not valid for DImode.
20788 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20789 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20790 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20791 (define_peephole2 for Altivec d-form store): Likewise.
20794 * multiple_target.c (create_dispatcher_calls): Only create the
20795 dispatcher call if the function is the default clone of a
20796 versioned function.
20798 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20800 PR middle-end/80902
20801 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20802 a call, force the call to not be a tail call.
20804 2017-06-23 Jeff Law <law@redhat.com>
20806 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20809 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20811 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20812 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20813 handling for early expansion of vector shifts (sl,sr,sra,rl).
20814 (builtin_function_type): Add vector shift right instructions
20815 to the unsigned argument list.
20817 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20819 rtl-optimizatoin/79286
20820 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20821 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20822 trap. PIC register plus a const unspec without offset can never trap.
20824 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20826 * tree.h (builtin_structptr_type): New type.
20827 (builtin_structptr_types): Declare new array.
20828 * tree.c (builtin_structptr_types): New array.
20829 (free_lang_data, build_common_tree_nodes): Use it.
20831 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20834 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20837 2017-06-22 Matt Turner <mattst88@gmail.com>
20839 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20840 Lake models to skylake case. Assume skylake for unknown
20841 models with clflushopt.
20843 2017-06-22 Jeff Law <law@redhat.com>
20845 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20846 frame sizes that do not satisfy aarch64_uimm12_shift.
20848 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20850 * profile-count.h (apply_probability,
20851 apply_scale, probability_in): Fix checks for zero.
20853 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20855 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20856 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20858 2016-06-22 Richard Biener <rguenther@suse.de>
20860 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20861 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20862 REDUC_MAX_EXPR support.
20863 (vectorizable_reduction): Likewise.
20864 (vect_create_epilog_for_reduction): Likewise.
20866 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20868 * match.pd (A / (1 << B) -> A >> B): New.
20869 * generic-match-head.c: Include optabs-tree.h.
20870 * gimple-match-head.c: Likewise.
20871 * optabs-tree.h (target_supports_op_p): New.
20872 * optabs-tree.c (target_supports_op_p): New.
20874 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20876 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20877 $gcc_cv_ld --help output.
20878 (gcc_cv_ld_demangle): Likewise.
20879 (gcc_cv_ld_eh_frame_hdr): Likewise.
20880 (gcc_cv_ld_pie): Likewise.
20881 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20882 (gcc_cv_ld_buildid): Likewise.
20883 (gcc_cv_ld_sysroot): Likewise.
20884 (ld_bndplt_support): Likewise.
20885 (ld_pushpopstate_support): Likewise.
20886 * configure: Regenerate.
20887 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20889 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20892 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20893 operands indexes to avoid gap between operands and match_dups.
20895 2017-06-21 Andrew Pinski <apinski@cavium.com>
20897 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20898 Increment Arith_shift and Arith_shift_reg by 1.
20899 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20901 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20902 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20903 (aarch64_strip_extend): Add new argument and test for it.
20904 (aarch64_cheap_mult_shift_p): New function.
20905 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20906 add a cost if it is true.
20907 Update calls to aarch64_strip_extend.
20908 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20910 2017-06-21 Andrew Pinski <apinski@cavium.com>
20912 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20914 (thunderxt88): Likewise.
20915 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20916 (thunderx_prefetch_tune): New variable.
20917 (thunderx2t99_prefetch_tune): Update for the correct values.
20918 (thunderxt88_tunings): New variable.
20919 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20920 generic_prefetch_tune.
20921 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20923 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20925 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20926 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20927 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20928 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20929 (aarch64_atomic_cas<mode>, GPI): Likewise.
20931 2017-06-21 Martin Liska <mliska@suse.cz>
20933 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20934 statements on cold and hot labels.
20935 * predict.c (tree_estimate_probability_bb): Remove the
20936 prediction from this place.
20938 2017-06-21 Martin Liska <mliska@suse.cz>
20940 PR tree-optimization/79489
20941 * gimplify.c (maybe_add_early_return_predict_stmt): New
20943 (gimplify_return_expr): Call the function.
20944 * predict.c (tree_estimate_probability_bb): Remove handling
20946 * predict.def: Update comment about early return predictor.
20947 * gimple-predict.h (is_gimple_predict): New function.
20948 * predict.def: Change default value of early return to 66.
20949 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20951 * passes.def: Put pass_strip_predict_hints to the beginning of
20954 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20956 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20957 FUNCTION_DECL declarations.
20958 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20960 (dwaf2out_decl): Likewise.
20961 * godump.c (go_early_global_decl): Skip call to the real debug hook
20962 for FUNCTION_DECL declarations.
20963 * passes.c (rest_of_decl_compilation): Skip call to the
20964 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20965 -fdump-go-spec is passed.
20967 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20969 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20970 Reorder for compactness.
20971 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20972 (def_builtin_pure, def_builtin_pure2): New functions.
20973 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20975 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20977 * match.pd (nop_convert): New predicate.
20978 ((A +- CST1) +- CST2): Allow some NOP conversions.
20980 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20983 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20984 with ctors/dtors if GOVD_SHARED is set.
20986 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20988 * config/aarch64/aarch64.md (movti_aarch64):
20989 Emit mov rather than orr.
20990 (movtf_aarch64): Likewise.
20991 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20992 Emit mov rather than orr.
20994 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20996 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20997 Swap alternatives, make integer dup more expensive.
20999 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
21001 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21002 Return true for non-tls symbols.
21004 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
21006 * config/aarch64/aarch64-cores.def (cortex-a55): New.
21007 (cortex-a75): Likewise.
21008 (cortex-a75.cortex-a55): Likewise.
21009 * config/aarch64/aarch64-tune.md: Regenerate.
21010 * doc/invoke.texi (-mtune): Document new values for -mtune.
21012 2017-06-21 Tom de Vries <tom@codesourcery.com>
21014 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
21015 stack_size feature.
21016 (Effective-Target Keywords, Other attributes): Suggest using
21017 dg-add-options stack_size feature to get stack limit in stack_size
21018 effective target documentation.
21020 2017-06-21 Julian Brown <julian@codesourcery.com>
21021 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21023 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
21024 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
21025 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
21027 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
21028 attribute type list for neon_multiply.
21029 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
21030 attribute type list for neon_multiply.
21031 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
21032 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
21033 attribute type list for neon_multiply.
21034 * config/arm/types.md (crypto_pmull): Add.
21035 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21036 attribute type list.
21038 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21040 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21043 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21045 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21046 to make sure not to dereference a NULL cost_classes_ptr pointer.
21048 2017-06-20 Carl Love <cel@us.ibm.com>
21050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21051 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21052 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21053 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21054 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21055 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21056 VMULOSW): New enum "unspec" values.
21057 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21058 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21059 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21060 altivec_vmulosw): New patterns.
21061 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21062 VMULOSW): Add definitions.
21064 2017-06-20 Julia Koval <julia.koval@intel.com>
21066 * config/i386/i386.c: Fix rounding expand for new pattern.
21067 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21069 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21071 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21072 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21074 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21076 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21079 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21081 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21082 architecture, then by implementer ID.
21083 * config/aarch64/aarch64-tune.md: Regenerate.
21085 2017-06-20 Richard Biener <rguenther@suse.de>
21087 PR middle-end/81097
21088 * fold-const.c (split_tree): Fold to type before negating.
21090 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21092 * diagnostic-show-locus.c
21093 (selftest::test_fixit_deletion_affecting_newline): New function.
21094 (selftest::diagnostic_show_locus_c_tests): Call it.
21096 2017-06-20 Andreas Schwab <schwab@suse.de>
21099 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21102 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21104 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21105 __ARM_FEATURE_COPROC according to support.
21107 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21109 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21110 Rewritten to avoid overflow for > 32-bit pointers.
21113 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21114 by removing enum keyword.
21115 (ubsan_type_descriptor): Likewise. Formatting fix.
21118 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21119 splitter): Require TARGET_SSE2 in the condition.
21121 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21124 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21125 for doing vector set of SFmode on ISA 3.0.
21126 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21127 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21129 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21130 SFmode value into a V4SF variable that was extracted from another
21131 V4SF variable without converting the element to double precision
21132 and back to single precision vector format.
21133 (vsx_insert_extract_v4sf_p9_2): Likewise.
21135 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21137 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21138 in UWHI to avoid undefined overflow.
21141 * ubsan.h (enum ubsan_encode_value_phase): New.
21142 (ubsan_encode_value): Change second argument to
21143 enum ubsan_encode_value_phase with default value of
21144 UBSAN_ENCODE_VALUE_GENERIC.
21145 * ubsan.c (ubsan_encode_value): Change second argument to
21146 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21147 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21148 create_tmp_var_raw instead of create_tmp_var and use a
21150 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21151 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21152 ubsan_encode_value callers.
21155 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21156 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21157 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21159 2017-06-19 Richard Biener <rguenther@suse.de>
21161 PR middle-end/81118
21162 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21163 estimates if we changed anything.
21165 2017-06-19 Richard Biener <rguenther@suse.de>
21167 PR tree-optimization/80887
21168 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21169 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21170 simplified lookups, then reset mprts_hook.
21171 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21173 (try_to_simplify): Likewise.
21175 2017-06-19 Martin Liska <mliska@suse.cz>
21178 * gimplify.c (gimplify_switch_expr):
21179 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21181 2017-06-19 Martin Liska <mliska@suse.cz>
21183 * doc/install.texi: Document that PGO runs in 4 stages.
21185 2017-06-19 Martin Liska <mliska@suse.cz>
21188 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21189 to dispatcher function name.
21190 * multiple_target.c (replace_function_decl): New function.
21191 (create_dispatcher_calls): Redirect both edges and references.
21193 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21195 * profile-count.c (profile_count::dump): Dump quality.
21196 (profile_count::differs_from_p): Update for unsigned val.
21197 * profile-count.h (profile_count_quality): New enum.
21198 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21200 2017-06-19 Richard Biener <rguenther@suse.de>
21202 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21203 struct function as arg.
21204 (estimate_numbers_of_iterations): Export overload with loop arg.
21205 (free_numbers_of_iterations_estimates_loop): Use an overload of
21206 free_numbers_of_iterations_estimates instead.
21207 * tree-cfg.c (remove_bb): Adjust.
21208 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21209 * tree-parloops.c (gen_parallel_loop): Likewise.
21210 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21212 (tree_unroll_loops_completely): Likewise.
21213 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21214 Use an overload instead and export.
21215 (estimated_loop_iterations): Adjust.
21216 (max_loop_iterations): Likewise.
21217 (likely_max_loop_iterations): Likewise.
21218 (estimate_numbers_of_iterations): Take struct function as arg
21220 (loop_exits_before_overflow): Adjust.
21221 (free_numbers_of_iterations_estimates_loop): Use an overload.
21222 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21223 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21225 2017-06-19 Richard Biener <rguenther@suse.de>
21228 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21229 RANGE_EXPR conservatively.
21231 2017-06-16 Carl Love <cel@us.ibm.com>
21233 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21234 definitions for vec_float, vec_float2, vec_floato,
21235 vec_floate built-ins.
21236 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21237 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21239 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21240 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21241 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21242 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21243 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21244 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21245 vec_floato): Add builtin defines.
21246 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21247 Update the built-in documentation file for the new built-in
21250 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21252 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21253 (mthumb): Mark as the negative of -marm.
21255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21257 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21259 (ARM Options, -mtune): Document that this accepts the same
21260 extension options as -mcpu.
21261 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21265 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21266 permitted extensions.
21268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21270 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21271 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21272 (armv8-m.main): Add option +nodsp.
21273 * config/arm/arm-cpu-cdata.h: Regenerated.
21275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21277 * config/arm/t-fuchsia: New file.
21278 * config.gcc (arm*-*-fuchsia*): Use it.
21280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21282 * config/arm/t-symbian: Rewrite for new option infrastructure.
21284 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21286 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21287 (MULTILIB_REQUIRED): Likewise.
21289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21291 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21292 (MULTILIB_RESUE): Likewise.
21293 (MULTILIB_MATCHES): Likewise.
21294 (MULTLIB_REQUIRED): Likewise.
21296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21298 * config/arm/t-rtems: Rewrite for new option framework.
21300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21302 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21303 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21304 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21305 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21306 * config/arm/t-multilib: ... here.
21307 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21308 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21309 armv7-a and armv8*-a when A-profile libraries have not been built.
21310 * config/arm/t-rmprofile: Rewrite.
21312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21314 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21315 with a backslash. Remove the backslash after substituting unescaped
21317 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21321 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21322 the driver to pass through the default setting of -mfloat-abi.
21323 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21325 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21326 (all_feat_combs): New rule.
21327 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21329 * config/arm/t-aprofile: Rewrite.
21331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21333 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21334 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21335 fpu is not specified by the user/command-line.
21336 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21337 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21338 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21339 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21340 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21341 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21345 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21346 * config/arm/t-arm-elf: Rewritten.
21348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21350 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21351 have some floating-point instructions.
21352 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21353 (TARGET_MAYBE_HARD_FLOAT): New macro.
21354 * config/arm/arm-builtins.c (arm_init_builtins): Use
21355 TARGET_MAYBE_HARD_FLOAT.
21356 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21360 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21361 (configargs.h): Include it.
21362 (arm_print_hint_for_fpu_option): New function.
21363 (arm_parse_fpu_option): New function.
21364 (candidate_extension): New class.
21365 (arm_canon_for_multilib): New function.
21366 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21367 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21368 (ARCH_CANONICAL_SPECS): New macro.
21369 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21373 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21374 are set after handling multilib fragments. Set target_cpu_default2
21377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21379 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21381 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21382 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21384 (check_arch): Likewise.
21385 * config/arm/arm.c (arm_configure_build_target): Handle
21386 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21387 options in the default.
21389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21391 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21392 when an option is an alias of another.
21393 * config/arm/parsecpu.awk (optalias): New parser token.
21394 (gen_comm_data): Mark non-alias options as such. Emit entries
21395 for extension aliases.
21396 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21397 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21398 (armv6kz, armv6zk, armv6t2): Likewise.
21399 (armv7): Make vfpv3-d16 an alias.
21400 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21402 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21403 Sort in canonical order.
21404 (armv8-a): Sort in canonical order.
21405 (armv8.1-a, armv8.2-a): Likewise.
21406 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21408 (cortex-a9): Sort in canonical order.
21409 * config/arm/arm.c (selftests.h): Include it.
21410 (arm_test_cpu_arch_data): New function.
21411 (arm_run_self_tests): New function.
21412 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21413 (targetm): Move declaration to the end of the file.
21414 * arm-cpu-cdata.h: Regenerated.
21416 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21418 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21419 call to target_mode_check describing the type of option passed.
21420 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21421 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21422 arm_parse_cpu_option_name to match parameters against list of
21424 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21425 arm_arch_core_flags data structure.
21426 * config/arm/arm-cpu_cdata.h: Regenerated.
21428 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21430 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21432 (arm_print_hint_for_cpu_option): Likewise.
21433 (arm_print_hint_for_arch_option): Likewise.
21434 (arm_parse_cpu_option_name): Likewise.
21435 (arm_parse_arch_option_name): Likewise.
21436 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21437 of entries in the all_fpus list.
21438 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21439 (arm_parse_cpu_option_name): Declare.
21440 (arm_parse_arch_option_name): Declare.
21441 (arm_parse_option_features): Declare.
21442 (arm_intialize_isa): Declare.
21443 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21445 (gen_comm_data): ... here. Make definitions non-static.
21446 * config/arm/arm-cpu-data.h: Regenerated.
21447 * config/arm/arm-cpu-cdata.h: Regenerated.
21449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21451 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21452 (cpu_arch_extension): New structure.
21453 (cpu_arch_option, arch_option, cpu_option): New structures.
21454 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21455 architecture types.
21456 (gen_data): Generate new format data tables.
21457 * config/arm/arm.c (cpu_tune): New structure.
21458 (cpu_option, processors): Delete.
21459 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21460 (arm_print_hint_for_cpu_option): ... this and ...
21461 (arm_print_hint_for_arch_option): ... this.
21462 (arm_parse_arch_cpu_name): Delete. Replace with ...
21463 (arm_parse_cpu_option_name): ... this and ...
21464 (arm_parse_arch_option_name): ... this.
21465 (arm_unrecognized_feature): Change type of target parameter to
21467 (arm_parse_arch_cpu_features): Delete. Replace with ...
21468 (arm_parse_option_features): ... this.
21469 (arm_configure_build_target): Rework to use new configuration data
21471 (arm_print_tune_info): Rework for new configuration data tables.
21472 * config/arm/arm-cpu-data.h: Regenerated.
21473 * config/arm/arm-cpu.h: Regenerated.
21475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21477 * Makefile.in (OBJS): Move sbitmap.o from here ...
21478 (OBJS-libcommon): ... to here.
21480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21482 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21483 (ISA_ALL_CRYPTO): New macro.
21484 (ISA_ALL_SIMD): New macro
21485 (ISA_ALL_FP): New macro.
21486 * config/arm/arm.c (fpu_bitlist): Update initializer.
21487 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21489 (arm9e): Add fpu. Add option for nofp
21490 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21491 (arm926ej-s, arm1026ej-s): Likewise.
21492 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21493 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21494 neon-fp16, neon-vfpv4, nofp and nosimd.
21495 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21496 (cortex-a8): Add fpu. Add option for nofp.
21497 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21498 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21499 (cortex-r4f): Add fpu.
21500 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21501 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21503 (cortex-r8): Likewise.
21504 (cortex-m4): Add fpu. Add option for nofp.
21505 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21506 (cortex-a17.cortex-a7): Likewise.
21507 (cortex-a32): Add fpu. Add options for crypto and nofp.
21508 (cortex-a35, cortex-a53): Likewise.
21509 (cortex-a57): Add fpu. Add option for crypto.
21510 (cortex-a72, cortex-a73): Likewise.
21511 (exynos-m1): Likewise.
21512 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21513 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21514 (cortex-m33): Add fpu. Add option for nofp.
21515 * config/arm/arm-cpu-cdata.h: Regenerated
21516 * config/arm/arm-cpu-data.h: Regenerated.
21518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21520 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21521 (armv5te, armv5tej): Likewise.
21522 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21523 (armv7): Add options fp and vfpv3-d16.
21524 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21525 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21527 (armv7ve): Likewise.
21528 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21529 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21530 (armv8-a): Add nocrypto option.
21531 (armv8.1-a, armv8.2-a): Likewise.
21532 (armv8-m.main): add options fp, fp.dp and nofp.
21534 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21536 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21538 (armv8-a+crc): Delete.
21539 (armv8.1-a): Add options simd, crypto and nofp.
21540 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21541 (armv8.2-a+fp16): Delete.
21542 (armv8-m.main): Add option dsp.
21543 (armv8-m.main+dsp): Delete.
21544 (cortex-a8): Add fpu. Add nofp option.
21545 (cortex-a9): Add fpu. Add nofp and nosimd options.
21546 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21547 link to main cpu and architecture data structures.
21548 (gen_comm_data): Only put isa attributes from the main architecture
21550 (option): New statement for architecture and CPU entries.
21551 * arm.c (struct cpu_option): New structure.
21552 (struct processors): Add entry for options.
21553 (arm_unrecognized_feature): New function.
21554 (arm_parse_arch_cpu_name): Ignore any characters after the first
21556 (arm_parse_arch_cpu_feature): New function.
21557 (arm_configure_build_target): Separate out any CPU and architecture
21558 features and parse separately. Don't error out if -mfpu=auto is
21559 used with only an architecture string.
21560 (arm_print_asm_arch_directives): New function.
21561 (arm_file_start): Call it.
21562 * config/arm/arm-cpu-cdata.h: Regenerated.
21563 * config/arm/arm-cpu-data.h: Likewise.
21564 * config/arm/arm-tables.opt: Likewise.
21566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21568 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21569 assembler when it is not -mfpu=auto.
21571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21573 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21574 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21575 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21576 (ASM_CPU_SPEC): Rewrite.
21577 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21578 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21579 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21580 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21581 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21582 copied string is NUL-terminated. Also strip any characters prefixed
21584 (arm_rewrite_selected_arch): New function.
21585 (arm_rewrite_march): New function.
21587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21589 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21590 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21591 (march, mcpu, mtune): Convert to string-based options.
21592 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21593 (arm_parse_arch_cpu_name): New function.
21594 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21595 identify selected architecture or CPU.
21596 (arm_option_save): New function.
21597 (TARGET_OPTION_SAVE): Redefine.
21598 (arm_option_restore): Restore string options.
21599 (arm_option_print): Print string options.
21601 2017-06-16 Martin Sebor <msebor@redhat.com>
21603 PR tree-optimization/80933
21604 PR tree-optimization/80934
21605 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21606 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21607 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21608 (gimple_fold_builtin): Call them.
21610 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21612 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21613 as unlikely; update profile.
21615 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21617 * predict.c (force_edge_cold): Handle declaring edges impossible
21620 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21622 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21624 (try_unroll_loop_completely): Fix reporting.
21626 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21628 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21630 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21633 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21634 if given a non-constant argument for an intrinsic which requires a
21637 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21639 * profile.c (compare_freqs): New function.
21640 (branch_prob): Sort edge list.
21641 (find_spanning_tree): Assume that the list is priority sorted.
21643 2017-06-16 Richard Biener <rguenther@suse.de>
21645 PR tree-optimization/81090
21646 * passes.def (pass_record_bounds): Remove.
21647 * tree-pass.h (make_pass_record_bounds): Likewise.
21648 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21649 make_pass_record_bounds): Likewise.
21650 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21651 not free niter estimates at the beginning but at the end.
21652 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21654 2017-06-16 Richard Biener <rguenther@suse.de>
21656 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21657 initializer to workaround ICE in host GCC 4.8.
21659 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21661 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21663 (clone_inlined_nodes): Update.
21665 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21667 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21668 prefetch settings, and enable prefetching by default at -O3.
21670 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21672 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21673 Set flag_prefetch_loop_arrays according to tuning data.
21675 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21677 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21678 New tune structure.
21679 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21680 [Unrelated to main purpose of the patch] Place the pointer field last
21681 to enable type checking errors when tune structure are wrongly merged.
21682 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21683 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21684 (thunderx2t99_prefetch_tune): New tune constants.
21685 (tune_params *_tunings): Update all tunings (no functional change).
21686 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21687 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21688 from tunings structures.
21690 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21693 * ubsan.c (instrument_null): Add T argument, use it instead
21694 of computing it based on IS_LHS.
21695 (instrument_object_size): Likewise.
21696 (pass_ubsan::execute): Adjust instrument_null and
21697 instrument_object_size callers to pass gimple_get_lhs or
21698 gimple_assign_rhs1 result to it. Use instrument_null instead of
21699 calling get_base_address and instrument_mem_ref. Handle
21700 aggregate call arguments for object-size sanitization.
21702 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21704 PR tree-optimization/81089
21705 * tree-vrp.c (is_masked_range_test): Validate operands of
21708 2017-06-15 Martin Sebor <msebor@redhat.com>
21711 * dumpfile.c (dump_register): Avoid calling memset to initialize
21712 a class with a default ctor.
21713 * gcc.c (struct compiler): Remove const qualification.
21714 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21715 * hash-table.h: Ditto.
21716 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21718 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21719 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21721 * params.h (struct param_info): Make struct members non-const.
21722 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21723 with default initialization.
21724 * vec.h (vec_copy_construct, vec_default_construct): New helper
21726 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21727 with vec_copy_construct.
21728 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21729 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21730 * doc/invoke.texi (-Wclass-memaccess): Document.
21732 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21734 * emit-rtl.h (is_leaf): Update comment about local
21735 register allocator.
21737 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21740 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21741 for a variable to have a section before checking if the section has a
21743 Set section to.persistent if persistent attribute is set.
21744 Warn if .persistent attribute is used on an automatic variable.
21746 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21748 PR rtl-optimization/80474
21749 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21751 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21753 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21756 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21758 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21759 (casesi): Emit bounds checking as RTL.
21760 (casesi_internal_mips16_<mode>): Remove bounds checking.
21762 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21764 * config/xtensa/xtensa.c (xtensa_option_override): Append
21765 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21766 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21767 xtensa_doloop_hooks): Define unconditionally.
21768 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21770 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21771 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21772 for it in xtensa_option_override.
21773 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21774 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21776 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21778 * doc/cppopts.texi: Document '-' special value to -MF.
21780 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21782 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21783 (cortex_a53_fconst): Likewise.
21784 (cortex_a53_fpmul): Likewise.
21785 (cortex_a53_f_load_64): Likewise.
21786 (cortex_a53_f_load_many): Likewise.
21787 (cortex_a53_advsimd_alu): Likewise.
21788 (cortex_a53_advsimd_alu_q): Likewise.
21789 (cortex_a53_advsimd_mul): Likewise.
21790 (cortex_a53_advsimd_mul_q): Likewise.
21791 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21792 Add missing fmul, r2f_cvt and fconst cases.
21794 2017-06-14 Richard Biener <rguenther@suse.de>
21796 PR middle-end/81088
21797 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21799 (fold_binary_loc): When associating do not treat pre-existing
21800 TREE_OVERFLOW on literal constants as a reason to allow
21801 TREE_OVERFLOW on associated literal constants.
21803 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21805 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21806 (MASK_FEATURES): New macro.
21807 * config/sparc/sparc.c (sparc_option_override): Remove the special
21808 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21810 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21812 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21813 a division of 0 if non-call exceptions are enabled.
21815 2017-06-14 Andrew Pinski <apinski@cavium.com>
21816 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21819 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21820 Improve vector initialization code gen for only variable case.
21822 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21824 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21826 2017-06-14 Richard Biener <rguenther@suse.de>
21828 PR tree-optimization/81083
21829 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21832 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21834 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21835 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21836 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21837 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21838 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21839 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21841 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21843 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21844 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21846 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21848 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21850 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21852 * config/rs6000/t-rtems: Don't handle SPE.
21854 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21856 * config/rs6000/t-linux: Don't handle SPE.
21858 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21860 * config/rs6000/eabispe.h: Delete file.
21862 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21864 * config/rs6000/t-spe: Delete file.
21866 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21868 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21869 (rs6000_legitimate_offset_address_p): Return false for anything in
21870 V2SImode or V2SFmode.
21872 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21874 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21875 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21876 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21878 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21879 (rs6000_legitimate_offset_address_p): Ditto.
21880 (rs6000_emit_move): Ditto.
21881 (rs6000_init_builtins): Remove V4HI_type_node.
21883 2017-06-13 Martin Liska <mliska@suse.cz>
21886 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21887 (gate_asan): Likewise.
21888 * asan.h (asan_no_sanitize_address_p): Remove the function.
21889 (sanitize_flags_p): New function.
21890 * builtins.def: Fix coding style.
21891 * common.opt: Use renamed enum value.
21892 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21893 * doc/extend.texi: Document no_sanitize attribute.
21894 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21895 to SANITIZE_UNDEFINED_NONDEFAULT.
21896 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21897 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21898 Use sanitize_flags_p.
21899 * gimplify.c (gimplify_function_tree): Likewise.
21900 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21901 * opts.c (parse_no_sanitize_attribute): New function.
21902 (common_handle_option): Use renamed enum value.
21903 * opts.h (parse_no_sanitize_attribute): Declare.
21904 * tree.c (sanitize_flags_p): New function.
21905 * tree.h: Declared here.
21906 * tsan.c: Use sanitize_flags_p.
21907 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21908 (instrument_mem_ref): Likewise.
21909 (instrument_bool_enum_load): Likewise.
21910 (do_ubsan_in_current_function): Remove the function.
21911 (pass_ubsan::execute): Use sanitize_flags_p.
21912 * ubsan.h: Remove do_ubsan_in_current_function
21913 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21914 (dump_function_to_file): Use it here.
21916 2017-06-13 Martin Jambor <mjambor@suse.cz>
21918 PR tree-optimization/80803
21919 PR tree-optimization/81063
21920 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21921 (propagate_subaccesses_across_link): Enqueue subtree whenever
21922 necessary instead of relying on the caller.
21924 2017-06-13 Martin Jambor <mjambor@suse.cz>
21926 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21927 that have a first_link.
21928 (sort_and_splice_var_accesses): Do not check first_link before
21930 (subtree_mark_written_and_enqueue): Likewise.
21931 (propagate_all_subaccesses): Likewise and do not stop at first
21932 parent with a first_link.
21934 2017-06-13 Martin Jambor <mjambor@suse.cz>
21936 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21939 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21941 * match.pd: New pattern.
21943 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21945 * tree-vrp.c (is_masked_range_test): New function.
21946 (register_edge_assert_for): Determine ranges for
21949 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21951 PR tree-optimization/67328
21952 * fold-const.c (maskable_range_p): New function.
21953 (build_range_check): Generate bittests if possible.
21955 2017-06-13 Martin Liska <mliska@suse.cz>
21957 * gimple-pretty-print.c (dump_probability): Add new argument.
21958 (dump_edge_probability): Dump both probability and count.
21959 (dump_gimple_label): Likewise.
21960 (dump_gimple_bb_header): Likewise.
21962 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21965 * config/avr/avr-devices.c: Fix indentation.
21966 * config/avr/gen-avr-mmcu-specs.c: Dito.
21968 2017-06-13 Richard Biener <rguenther@suse.de>
21970 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21971 instead get vector type from stmt_info.
21972 (vectorizable_reduction): Adjust. Remove dead code.
21974 2017-06-13 Richard Biener <rguenther@suse.de>
21976 PR middle-end/81065
21977 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21978 case of C * (x * C2 + C3).
21979 (fold_addr_of_array_ref_difference): Properly fold index difference.
21981 2017-06-12 David S. Miller <davem@davemloft.net>
21984 * config/sparc/sparc.md (return expander): Emit frame blockage if
21985 function uses alloca.
21987 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21989 * combine.c (make_field_assignment): Check len rather than the mode
21990 precision when calling force_to_mode.
21992 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21994 Support multilibs and devices that see flash in RAM address range.
21997 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21998 (avr_mcu_t) <flash_pm_offset>: New field.
21999 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
22000 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
22001 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
22002 (AVR_TINY_PM_OFFSET): Remove macro.
22003 * config/avr/avr.opt (-mshort-calls): New option.
22004 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22005 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
22006 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
22007 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
22008 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
22009 instead of avr_arch->have_jmp_call.
22010 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
22011 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
22012 avr_arch->flash_pm_offset to define.
22013 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
22014 new field flash_pm_offset. Add entry for avrxmega3.
22015 (avr_texinfo): Add entry for avrxmega3.
22016 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
22017 attiny212, attiny214,
22018 attiny412, attiny414, attiny416, attiny417,
22019 attiny814, attiny816, attiny817,
22020 attiny1614, attiny1616, attiny1617,
22021 attiny3214, attiny3216, attiny3217.
22022 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
22023 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
22024 (avr_print_operand_address) [AVR_TINY]: Same.
22025 (avr_asm_init_sections) <readonly_data_section>: Only patch
22026 callback if avr_arch->flash_pm_offset = 0.
22027 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
22028 for rodata if avr_arch->flash_pm_offset != 0.
22029 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
22030 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
22031 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
22032 (m_options): Append opt_rcall.
22033 (m_dirnames): Append dir_rcall.
22034 * config/avr/t-multilib: Regenerate.
22036 * configure.ac [target=avr]: Check whether avrxmega3 default
22037 linker description file works as needed.
22038 * configure: Regenerate.
22039 * doc/avr-mmcu.texi: Regenerate.
22040 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22041 <__AVR_ARCH__>: Document avrxmega3 and 103.
22042 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22043 <__AVR_SHORT_CALLS__>: Document it.
22044 <__AVR_PM_BASE_ADDRESS__>: Document it.
22045 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22046 (AVR Variable Attributes) <progmem>: Document this is
22047 not needed for avrxmega3.
22048 (AVR Named Address Spaces) <__flash>: Dito.
22050 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22052 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22054 2017-06-12 Doug Rupp <rupp@adacore.com>
22056 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22057 Append vxworks-stdint.h to the tm_file list.
22058 * config/vxworks-stdint.h: New file.
22060 2017-06-12 Martin Liska <mliska@suse.cz>
22062 PR tree-optimization/81041
22063 * tree-profile.c (gimple_gen_ic_func_profiler):
22064 Create an extra BB in profile-generate
22065 (gimple_gen_time_profiler): Likewise.
22067 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22069 PR tree-optimization/81003
22070 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22071 (update_range_test): Use it instead of force_gimple_operand_gsi.
22073 2017-06-12 Richard Biener <rguenther@suse.de>
22075 PR tree-optimization/81053
22076 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22077 with backedge value not defined in loop. Simplify def stmt
22080 2017-06-11 Tom de Vries <tom@codesourcery.com>
22083 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22085 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22086 nvptx_cannot_force_const_mem.
22088 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22090 * opts.c (finish_options): Move test for flag_split_stack after
22091 it has been initialized.
22093 2017-06-11 Jason Merrill <jason@redhat.com>
22095 * tree.h (id_equal): New.
22096 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22097 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22098 instead of strcmp of IDENTIFIER_POINTER.
22100 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22102 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22103 (mark_all_inlined_calls_cdtor): Fix formating.
22104 (inline_transform): Rescale profile before inlining.
22106 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22108 * cgraph.h (cgraph_edge::clone): Update prototype.
22109 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22110 (cgraph_node::create_clone): Update.
22111 (cgraph_node::create_version_clone): Update.
22112 * tree-inline.c (copy_bb): Update.
22113 (expand_call_inline): Update.
22115 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22117 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22118 factored out from ...
22119 (rs6000_emit_prologue): ... here.
22121 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22123 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22124 factored out from ...
22125 (rs6000_emit_prologue): ... here.
22127 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22129 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22131 (handle_missing_profiles): Fix computation of tp_first_run.
22132 (counts_to_freqs): Do not touch freqs when count is 0.
22134 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22136 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22139 2017-06-10 Tom de Vries <tom@codesourcery.com>
22141 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22142 attributes): Document signal effective target.
22144 2017-06-10 Tom de Vries <tom@codesourcery.com>
22146 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22147 Document effective target stack_size.
22149 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22151 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22152 to the edit_context if they can be auto-applied.
22154 2017-06-9 Ian Lance Taylor <iant@golang.org>
22156 * opts.c (finish_options): If -fsplit-stack, disable implicit
22157 -forder-blocks-and-partition.
22158 * doc/invoke.texi (Optimize Options): Document that when using
22159 -fsplit-stack -forder-blocks-and-partition is not implicitly
22162 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22164 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22165 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22166 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22167 * builtins.def (abort, trap, unreachable): Declare cold.
22168 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22169 * tree-core.h (ECF_COLD): New.
22170 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22171 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22173 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22175 * predict.c (unlikely_executed_stmt_p): Cleanup.
22177 2017-06-09 Richard Biener <rguenther@suse.de>
22179 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22180 model if the ref is always written to.
22182 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22184 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22186 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22188 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22191 2017-06-09 Tom de Vries <tom@codesourcery.com>
22194 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22195 "target cannot support label values" when encountering LABEL_REF.
22197 2017-06-09 Martin Liska <mliska@suse.cz>
22199 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22200 (gimple_gen_ic_func_profiler): Emit direct comparison
22201 of __gcov_indirect_call_callee with NULL.
22202 (gimple_gen_time_profiler): Change probability from
22203 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22205 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22207 * profile.c (edge_gcov_counts): Turn to pointer.
22208 (compute_branch_probabilities, compute_branch_probabilities): Update.
22209 (branch_prob): Do not clear edge_gcov_count.
22210 * profile.h (edge_gcov_counts): Turn to pointer.
22211 (edge_gcov_count): Update.
22213 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22215 * gimple.h (gimple_check_failed): Mark cold.
22217 2017-06-09 Richard Biener <rguenther@suse.de>
22219 PR tree-optimization/66623
22220 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22221 refactor check_reduction into two parts, properly computing
22222 whether we have to check reduction validity for outer loop
22225 2017-06-09 Richard Biener <rguenther@suse.de>
22227 PR tree-optimization/79483
22228 * graphite-scop-detection.c (order): New global.
22229 (get_order): Compute bb to order mapping that satisfies code
22230 generation constraints.
22231 (cmp_pbbs): New helper.
22232 (build_scops): Start domwalk at entry block, sort generated
22235 2017-06-09 Richard Biener <rguenther@suse.de>
22237 PR middle-end/81007
22238 * ipa-polymorphic-call.c
22239 (ipa_polymorphic_call_context::restrict_to_inner_class):
22240 Skip FIELD_DECLs with error_mark_node type.
22241 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22244 2017-06-09 Martin Liska <mliska@suse.cz>
22246 * predict.c (struct branch_predictor): New struct.
22247 (test_prediction_value_range): New test.
22248 (predict_c_tests): New function.
22249 * selftest-run-tests.c (selftest::run_tests): Run the function.
22250 * selftest.h: Declare new tests.
22252 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22255 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22256 gen_add3_insn did not fail.
22257 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22258 r0, construct that number in a temporary reg and add that reg to r0.
22259 If asked to put the result in r0 as well, fail.
22261 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22263 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22264 for early expansion of vec_eqv.
22266 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22268 PR middle-end/81005
22269 * ubsan.c (instrument_null): Avoid pointless code temporary.
22270 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22272 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22276 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22279 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22282 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22284 * predict.c (unlikely_executed_edge_p): Move ahead.
22285 (probably_never_executed_edge_p): Use it.
22287 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22289 PR middle-end/79988
22290 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22291 gimple_call_builtin_p call.
22293 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22295 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22296 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22297 rtl_check_failed_type2, rtl_check_failed_code1,
22298 rtl_check_failed_code2, rtl_check_failed_code_mode,
22299 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22300 rtvec_check_failed_bounds, rtl_check_failed_flag,
22301 _fatal_insn_not_found, _fatal_insn): Likewise.
22302 * tree.h (tree_contains_struct_check_failed,
22303 tree_check_failed, tree_not_check_failed,
22304 tree_class_check_failed, tree_range_check_failed,
22305 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22306 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22307 tree_operand_check_failed, omp_clause_check_failed,
22308 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22311 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22313 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22314 flag_branch_probabilities.
22315 * ipa-inline.c (edge_badness): Likewise.
22316 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22317 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22318 * predict.c (maybe_hot_frequency_p): Likewise.
22319 (probably_never_executed): Likewise.
22320 * sched-ebb.c (schedule_ebbs): Likewise.
22321 * sched-rgn.c (find_single_block_region): Likewise.
22322 * tracer.c (tail_duplicate): Likewise.
22324 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22326 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22327 longer requires x_flag_profile_use.
22329 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22331 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22332 instead of flag_reorder_blocks_and_partition.
22333 * dbxout.c (dbxout_function_end): Likewise.
22334 * dwarf2out.c (gen_subprogram_die): Likewise.
22335 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22336 * hw-doloop.c (reorg_loops): Likewise.
22337 * varasm.c (assemble_start_function,
22338 assemble_end_function): Likewise.
22339 (decide_function_section): Do not check for
22340 flag_reorder_blocks_and_partition.
22342 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22344 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22346 (chkp_get_hard_register_fake_addr_expr): Ditto.
22347 (chkp_build_addr_expr): Add check for hard reg case.
22348 (chkp_parse_array_and_component_ref): Ditto.
22349 (chkp_find_bounds_1): Ditto.
22350 (chkp_process_stmt): Don't generate bounds store for
22353 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22355 * predict.c (maybe_hot_bb_p): Do not check profile status.
22356 (maybe_hot_edge_p): Likewise.
22357 (probably_never_executed): Check for zero counts even if profile
22359 (unlikely_executed_edge_p): New function.
22360 (unlikely_executed_stmt_p): New function.
22361 (unlikely_executed_bb_p): New function.
22362 (set_even_probabilities): Use unlikely predicates.
22363 (combine_predictions_for_bb): Likewise.
22364 (predict_paths_for_bb): Likewise.
22365 (predict_paths_leading_to_edge): Likewise.
22366 (determine_unlikely_bbs): New function.
22367 (estimate_bb_frequencies): Use it.
22368 (compute_function_frequency): Use zero counts even if profile is
22370 * profile-count.h: Fix typo.
22372 2017-08-08 Julia Koval <julia.koval@intel.com>
22374 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22375 _mm512_mask_cvtsepi16_storeu_epi8,
22376 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22377 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22378 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22379 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22380 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22381 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22382 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22383 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22384 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22385 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22386 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22387 __builtin_ia32_pmovuswb256mem_mask,
22388 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22389 __builtin_ia32_pmovwb512mem_mask): New builtins.
22391 2017-08-08 Julia Koval <julia.koval@intel.com>
22393 PR target/73350,80862
22394 * config/i386/subst.md (round): Fix round pattern.
22395 * config/i386/i386.c (ix86_erase_embedded_rounding):
22396 Fix erasing rounding for the fixed pattern.
22398 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22400 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22402 2017-06-08 Martin Liska <mliska@suse.cz>
22404 PR gcov-profile/80911
22405 * gcov.c (block_info::block_info): New constructor.
22407 2017-06-07 Carl Love <cel@us.ibm.com>
22409 * config/rs6000/rs6000-c: The return type of the following
22410 built-in functions was implemented as int not long long. Fix sign
22411 of return value for the unsigned version of vec_mulo and vec_mule.
22412 vector unsigned long long vec_bperm (vector unsigned long long,
22413 vector unsigned char)
22414 vector signed long long vec_mule (vector signed int,
22416 vector unsigned long long vec_mule (vector unsigned int,
22417 vector unsigned int)
22418 vector signed long long vec_mulo (vector signed int,
22420 vector unsigned long long vec_mulo (vector unsigned int,
22421 vector unsigned int)
22422 * doc/extend.texi: Fix the documentation for the built-in
22425 2017-06-07 Carl Love <cel@us.ibm.com>
22428 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22431 2017-06-07 Carl Love <cel@us.ibm.com>
22433 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22434 support, Generate doublehv for signed int/float for BE case only.
22436 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22438 * doc/invoke.texi (mcx16): Rewrite.
22440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22442 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22443 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22444 *mov<mode>_softfloat, and an anonymous splitter): Use
22445 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22449 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22451 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22452 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22453 (rs6000_debug_reg_global): Adjust.
22454 (rs6000_init_hard_regno_mode_ok): Adjust.
22455 (rs6000_dbx_register_number): Adjust.
22456 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22457 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22458 Remove SPE_ACC and SPEFSCR.
22459 (REG_ALLOC_ORDER): Ditto.
22460 (FRAME_POINTER_REGNUM): Change to 111.
22461 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22462 (REG_CLASS_NAMES): Ditto.
22463 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22464 (REGISTER_NAMES): Ditto.
22465 (ADDITIONAL_REG_NAMES): Ditto.
22466 (rs6000_reg_names): Ditto.
22467 * config/rs6000/rs6000.md: Renumber some register number
22470 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22472 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22474 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22475 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22477 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22478 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22479 Delete the SPE high registers.
22480 (REG_ALLOC_ORDER): Ditto.
22481 (enum reg_class): Remove SPE_HIGH_REGS.
22482 (REG_CLASS_NAMES): Ditto.
22483 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22484 (REGISTER_NAMES): Ditto.
22485 (rs6000_reg_names): Ditto.
22486 * doc/tm.texi.in: Remove SPE as example.
22487 * doc/tm.texi: Regenerate.
22489 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22491 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22492 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22493 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22494 * config/rs6000/rs6000.md (type): Remove "brinc".
22496 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22498 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22499 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22500 * config/rs6000/linuxspe.h: Delete file.
22501 * config/rs6000/rs6000.md: Don't include spe.md.
22502 * config/rs6000/spe.h: Delete file.
22503 * config/rs6000/spe.md: Delete file.
22504 * config/rs6000/t-rs6000: Remove spe.md.
22506 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22508 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22509 (reg_or_none500mem_operand): Delete.
22510 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22511 instead of reg_or_none500mem_operand.
22513 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22516 handling of SPE flags.
22517 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22519 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22521 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22523 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22524 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22525 paired_divv2sf3): Similar.
22526 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22527 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22528 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22530 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22531 Rename the paired_* instruction patterns.
22532 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22534 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22535 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22536 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22537 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22538 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22539 PAIRED_VECTOR_MODE.
22540 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22541 (spe_func_has_64bit_regs_p): Delete.
22542 (spe_expand_predicate_builtin): Delete.
22543 (spe_expand_evsel_builtin): Delete.
22544 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22545 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22546 (invalid_e500_subreg): Delete.
22547 (rs6000_legitimize_address): Always force_reg op2 as well, for
22548 paired single memory accesses.
22549 (rs6000_member_type_forces_blk): Delete.
22550 (rs6000_spe_function_arg): Delete.
22551 (rs6000_expand_unop_builtin): Delete SPE handling.
22552 (rs6000_expand_binop_builtin): Ditto.
22553 (spe_expand_stv_builtin): Delete.
22554 (bdesc_2arg_spe): Delete.
22555 (spe_expand_builtin): Delete.
22556 (spe_expand_predicate_builtin): Delete.
22557 (spe_expand_evsel_builtin): Delete.
22558 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22559 (spe_init_builtins): Delete.
22560 (spe_func_has_64bit_regs_p): Delete.
22561 (savres_routine_name): Delete "info" parameter. Adjust callers.
22562 (rs6000_emit_stack_reset): Ditto.
22563 (rs6000_dwarf_register_span): Delete.
22564 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22565 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22566 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22568 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22570 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22571 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22572 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22573 mulv2sf3, divv2sf3): Delete expanders.
22575 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22577 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22579 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22581 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22582 * config/rs6000/rs6000.c: Ditto.
22584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22586 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22587 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22588 comparison_operator.
22590 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22592 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22593 * config/rs6000/rs6000.opt: Ditto.
22594 * config/rs6000/t-rtems: Ditto.
22596 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22598 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22599 TARGET_E500_SINGLE by 0, simplify.
22600 * config/rs6000/rs6000.c: Ditto.
22601 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22602 (spe_build_register_parallel): Delete.
22603 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22604 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22605 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22606 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22607 (E500_CONVERT): Delete.
22608 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22612 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22613 * config/rs6000/dfp.md: Ditto.
22614 (negdd2, *negdd2_fpr): Merge.
22615 (absdd2, *absdd2_fpr): Merge.
22616 (negtd2, *negtd2_fpr): Merge.
22617 (abstd2, *abstd2_fpr): Merge.
22618 * config/rs6000/e500.h: Delete file.
22619 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22620 TARGET_FPRS by 1 and simplify.
22621 * config/rs6000/rs6000-c.c: Ditto.
22622 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22623 TARGET_DF_SPE by 0.
22624 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22626 * config/rs6000/rs6000.md: Ditto.
22627 (floatdidf2, *floatdidf2_fpr): Merge.
22628 (move_from_CR_gt_bit): Delete.
22629 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22630 (E500_CR_IOR_COMPARE): Delete.
22631 (All patterns that require !TARGET_FPRS): Delete.
22632 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22634 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22636 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22638 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22640 * graphds.c (add_edge): Intitialize edge's attached data.
22641 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22642 pointer parameter. Call pointed function on each edge during
22643 graph traversing. Skip traversing the edge when the function
22645 (graphds_dfs, graphds_scc): Ditto.
22646 (for_each_edge): New parameter. Pass the new parameter to callback
22648 * graphds.h (skip_edge_callback): New function pointer type.
22649 (graphds_dfs, graphds_scc): New function pointer parameter.
22650 (graphds_edge_callback, for_each_edge): New parameter.
22652 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22654 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22655 out code checking if runtime alias check is possible to below ...
22656 Call the new function.
22657 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22658 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22660 2017-06-07 Marek Polacek <polacek@redhat.com>
22663 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22664 TYPE_OVERFLOW_WRAPS check.
22666 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22668 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22669 if versioning is required.
22670 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22671 peeling with the check for versioning.
22673 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22675 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22676 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22677 Set true to new parameter if new ssa variable is defined.
22678 (vect_gen_vector_loop_niters): Refactor. Set range information
22679 for the new vector loop bound variable.
22680 (vect_do_peeling): Ditto.
22682 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22684 * tree-affine.c (ssa.h): Include header file.
22685 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22686 has wrapping overflow behavior.
22688 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22690 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22692 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22694 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22695 (tree_to_aff_combination): ... here.
22697 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22699 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22700 reg_pressure model function.
22701 (ivopts_global_cost_for_size): Delete.
22702 (determine_set_costs, iv_ca_recount_cost): Call new model function
22703 ivopts_estimate_reg_pressure.
22705 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22707 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22708 expensive than udiv. Remove floating point cases from mod.
22710 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22712 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22713 Increase idiv cost.
22715 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22717 * config/aarch64/aarch64.md
22718 (copysignsf3): Fix mask generation.
22720 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22722 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22724 (class dump_manager): Add register_dumps method.
22725 * dumpfile.c: Include langhooks.h.
22726 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22727 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22728 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22729 (dump_manager::dump_register): Start with 512 entries instead of 32.
22730 (dump_manager::register_dumps): New method.
22731 * toplev.c (general_init): Instead of invoking register_dumps
22732 langhook, invoke register_dumps method on the dump manager.
22733 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22736 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22738 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22739 Say that # requires an associated define_split to exist, and that
22740 the define_split must be suitable for use after register allocation.
22742 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22744 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22745 (compute_outgoing_frequencies): Also initialize zero counts.
22746 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22747 around loops; preserve more of profile when nothing changes.
22749 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22751 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22753 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22754 * config/arm/arm-cpu-cdata.h: Regenerate.
22755 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22756 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22757 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22758 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22759 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22761 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22763 * config/arm/t-rmprofile: Likewise.
22764 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22766 2017-06-06 David S. Miller <davem@davemloft.net>
22769 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22770 blockage if function uses alloca.
22772 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22774 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22775 New "uid" fields to hold pretty-print IDs of group and ref.
22776 Memory references are now identified as <group_id>:<ref_id>
22777 instead of using [random] addresses.
22778 (dump_mem_details): Simplify, no functional change.
22779 (dump_mem_ref): Simplify and make output more concise.
22780 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22781 (find_or_create_group): Initialize group uid.
22782 (record_ref): Initialize ref uid. Improve debug output.
22783 (prune_group_by_reuse, should_issue_prefetch_p,)
22784 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22785 (mark_nontemporal_store, determine_loop_nest_reuse):
22786 Improve debug output.
22788 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22790 * dbgcnt.def (prefetch): New debug counter.
22791 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22792 (schedule_prefetches): Stop issueing prefetches if debug counter
22795 2017-06-06 Tom de Vries <tom@codesourcery.com>
22797 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22798 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22800 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22802 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22803 Use aarch64_reg_or_zero predicate for operand 4.
22804 (aarch64_compare_and_swap<mode> define_insn_and_split):
22805 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22806 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22808 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22810 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22811 (arm_compute_save_core_reg_mask): This.
22812 (thumb1_compute_save_reg_mask): Rename into ...
22813 (thumb1_compute_save_core_reg_mask): This.
22814 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22815 (arm_compute_frame_layout): Likewise.
22817 2017-06-06 Richard Biener <rguenther@suse.de>
22819 PR tree-optimization/80974
22820 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22821 keep or clear leaders SSA info.
22823 2017-06-06 Tom de Vries <tom@codesourcery.com>
22825 * config/nvptx/nvptx.c (split_mode_p): New function.
22826 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22828 2017-06-06 Tom de Vries <tom@codesourcery.com>
22830 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22832 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22835 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22838 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22840 * shrink-wrap.c (handle_simple_exit): Update profile.
22841 (try_shrink_wrapping): Upate profile.
22843 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22845 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22846 (tree_guess_outgoing_edge_probabilities): New.
22847 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22848 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22850 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22852 * ipa-split.c (split_function): Initialize return bb profile.
22854 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22856 * profile.c (compute_branch_probabilities): Also initialize
22857 EXIT_BLOCK profile.
22859 2017-06-06 Richard Biener <rguenther@suse.de>
22861 PR tree-optimization/80928
22862 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22863 (vect_analyze_loop_operations): Properly guard analysis for
22865 (vect_transform_loop): Likewise.
22866 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22867 (vect_model_induction_cost): Do not cost for pure SLP.
22868 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22869 of induction in inner loop vectorization.
22870 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22871 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22872 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22874 (vect_analyze_slp_cost_1): Cost induction.
22875 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22876 (vect_get_slp_vect_defs): Likewise.
22877 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22878 (vect_transform_stmt): Handle SLP reductions.
22879 * tree-vectorizer.h (vectorizable_induction): Adjust.
22881 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22883 * config/rs6000/rs6000.c (make_resolver_func): Update
22884 init_lowered_empty_function call.
22886 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22888 * doc/invoke.texi: Document the -fprofile-abs-path option.
22889 * common.opt (fprofile-abs-path): New option.
22890 * gcov-io.h (gcov_write_filename): Declare.
22891 * gcov-io.c (gcov_write_filename): New function.
22892 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22893 * profile.c (output_location): Likewise.
22895 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22897 * shring-wrap.c: Revert accidental commit.
22899 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22901 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22903 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22905 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22907 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22908 profile in callgraph edge.
22909 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22910 (apply_scale): Likewise.
22911 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22912 Also scale profile when inlining function with zero profile.
22913 (initialize_cfun): Update exit block profile even when it is zero.
22914 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22915 when profile is read.
22917 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22919 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22920 (CLONE_*): New constants to define the processors we can generate
22921 code for with the target_clone attribute.
22922 (rs6000_clone_map): New array to identify which clone processors
22923 the current program is running on.
22924 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22925 target_clone attribute.
22926 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22927 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22928 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22929 (cpu_expand_builtin): Add support for target_clone attribute.
22930 (rs6000_valid_attribute_p): Allow "default" attribute.
22931 (get_decl_name): New debug function to simplify printing the
22932 current function name in debugging statements.
22933 (rs6000_clone_priority): New functions to support the target_clone
22934 attribute, and be able to generate code to switch between ISA 2.05
22935 through ISA 3.0 (power6 through power9).
22936 (rs6000_compare_version_priority): Likewise.
22937 (rs6000_get_function_versions_dispatcher): Likewise.
22938 (make_resolver_func): Likewise.
22939 (add_condition_to_bb): Likewise.
22940 (dispatch_function_versions): Likewise.
22941 (rs6000_generate_version_dispatcher_body): Likewise.
22942 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22943 (fusion_gpr_load_p): Fix a spacing issue.
22944 * doc/extend.texi (Common Function Attributes): Document that the
22945 PowerPC supports the target_clone attribute.
22947 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22949 * config/arm/arm.h: explain F symbol found in description of ARM
22950 register allocation in its legend.
22952 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22954 * config/mips/frame-header-opt.c: Include profile-count.h.
22955 * config/riscv/riscv.c: Include profile-count.h
22957 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22959 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22961 (sm_set_flag_if_changed): Add bbs field.
22962 (execute_sm_if_changed_flag_set): Pass BBS.
22963 (execute_sm): Update.
22965 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22967 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22970 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22972 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22973 (peephole2): New peephole2 to emit the above.
22974 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22976 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22978 * config/aarch64/aarch64.c (define_peephole2 above
22979 *sub_<shift>_<mode>): New peephole.
22981 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22983 * config/i386/i386.c (make_resolver_func): Update.
22984 * Makefile.in: Add profile-count.h and profile-count.o
22985 * auto-profile.c (afdo_indirect_call): Update to new API.
22986 (afdo_set_bb_count): Update.
22987 (afdo_propagate_edge): Update.
22988 (afdo_propagate_circuit): Update.
22989 (afdo_calculate_branch_prob): Update.
22990 (afdo_annotate_cfg): Update.
22991 * basic-block.h: Include profile-count.h
22992 (struct edge_def): Turn count to profile_count.
22993 (struct basic_block_def): Likewie.
22994 (REG_BR_PROB_BASE): Move to profile-count.h
22995 (RDIV): Move to profile-count.h
22996 * bb-reorder.c (max_entry_count): Turn to profile_count.
22997 (find_traces): Update.
22998 (rotate_loop):Update.
22999 (connect_traces):Update.
23000 (sanitize_hot_paths):Update.
23001 * bt-load.c (migrate_btr_defs): Update.
23002 * cfg.c (RDIV): Remove.
23003 (init_flow): Use alloc_block.
23004 (alloc_block): Uninitialize count.
23005 (unchecked_make_edge): Uninitialize count.
23006 (check_bb_profile): Update.
23007 (dump_edge_info): Update.
23008 (dump_bb_info): Update.
23009 (update_bb_profile_for_threading): Update.
23010 (scale_bbs_frequencies_int): Update.
23011 (scale_bbs_frequencies_gcov_type): Update.
23012 (scale_bbs_frequencies_profile_count): New.
23013 * cfg.h (update_bb_profile_for_threading): Update.
23014 (scale_bbs_frequencies_profile_count): Declare.
23015 * cfgbuild.c (compute_outgoing_frequencies): Update.
23016 (find_many_sub_basic_blocks): Update.
23017 * cfgcleanup.c (try_forward_edges): Update.
23018 (try_crossjump_to_edge): Update.
23019 * cfgexpand.c (expand_gimple_tailcall): Update.
23020 (construct_exit_block): Update.
23021 * cfghooks.c (verify_flow_info): Update.
23022 (dump_bb_for_graph): Update.
23023 (split_edge): Update.
23024 (make_forwarder_block): Update.
23025 (duplicate_block): Update.
23026 (account_profile_record): Update.
23027 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
23028 (get_estimated_loop_iterations): Update.
23029 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
23030 (single_likely_exit): Update.
23031 * cfgloopmanip.c (scale_loop_profile): Update.
23033 (set_zero_probability): Update.
23034 (lv_adjust_loop_entry_edge): Update.
23035 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23036 (purge_dead_edges): Update.
23037 (rtl_account_profile_record): Update.
23038 * cgraph.c (cgraph_node::create): Uninitialize count.
23039 (symbol_table::create_edge): Uninitialize count.
23040 (cgraph_update_edges_for_call_stmt_node): Update.
23041 (cgraph_edge::dump_edge_flags): Update.
23042 (cgraph_node::dump): Update.
23043 (cgraph_edge::maybe_hot_p): Update.
23044 * cgraph.h: Include profile-count.h
23045 (create_clone), create_edge, create_indirect_edge): Update.
23046 (cgraph_node): Turn count to profile_count.
23047 (cgraph_edge0: Likewise.
23048 (make_speculative, clone): Update.
23049 (create_edge): Update.
23050 (init_lowered_empty_function): Update.
23051 * cgraphclones.c (cgraph_edge::clone): Update.
23052 (duplicate_thunk_for_node): Update.
23053 (cgraph_node::create_clone): Update.
23054 * cgraphunit.c (cgraph_node::analyze): Update.
23055 (cgraph_node::expand_thunk): Update.
23056 * final.c (dump_basic_block_info): Update.
23057 * gimple-streamer-in.c (input_bb): Update.
23058 * gimple-streamer-out.c (output_bb): Update.
23059 * graphite.c (print_global_statistics): Update.
23060 (print_graphite_scop_statistics): Update.
23061 * hsa-brig.c: Include basic-block.h.
23062 * hsa-dump.c: Include basic-block.h.
23063 * hsa-gen.c (T sum_slice): Update.
23064 (convert_switch_statements):Update.
23065 * hsa-regalloc.c: Include basic-block.h.
23066 * ipa-chkp.c (chkp_produce_thunks): Update.
23067 * ipa-cp.c (struct caller_statistics): Update.
23068 (init_caller_stats): Update.
23069 (gather_caller_stats): Update.
23070 (ipcp_cloning_candidate_p): Update.
23071 (good_cloning_opportunity_p): Update.
23072 (get_info_about_necessary_edges): Update.
23073 (dump_profile_updates): Update.
23074 (update_profiling_info): Update.
23075 (update_specialized_profile): Update.
23076 (perhaps_add_new_callers): Update.
23077 (decide_about_value): Update.
23078 (ipa_cp_c_finalize): Update.
23079 * ipa-devirt.c (struct odr_type_warn_count): Update.
23080 (struct decl_warn_count): Update.
23081 (struct final_warning_record): Update.
23082 (possible_polymorphic_call_targets): Update.
23083 (ipa_devirt): Update.
23084 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23085 * ipa-icf.c (sem_function::merge): Update.
23086 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23087 * ipa-inline.c (compute_uninlined_call_time): Update.
23088 (compute_inlined_call_time): Update.
23089 (want_inline_small_function_p): Update.
23090 (want_inline_self_recursive_call_p): Update.
23091 (edge_badness): Update.
23092 (lookup_recursive_calls): Update.
23093 (recursive_inlining): Update.
23094 (inline_small_functions): Update.
23095 (dump_overall_stats): Update.
23096 (dump_inline_stats): Update.
23097 * ipa-profile.c (ipa_profile_generate_summary): Update.
23098 (ipa_propagate_frequency): Update.
23099 (ipa_profile): Update.
23100 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23101 * ipa-utils.c (ipa_merge_profiles): Update.
23102 * loop-doloop.c (doloop_modify): Update.
23103 * loop-unroll.c (report_unroll): Update.
23104 (unroll_loop_runtime_iterations): Update.
23105 * lto-cgraph.c (lto_output_edge): Update.
23106 (lto_output_node): Update.
23107 (input_node): Update.
23108 (input_edge): Update.
23109 (merge_profile_summaries): Update.
23110 * lto-streamer-in.c (input_cfg): Update.
23111 * lto-streamer-out.c (output_cfg): Update.
23112 * mcf.c (create_fixup_graph): Update.
23113 (adjust_cfg_counts): Update.
23114 (sum_edge_counts): Update.
23115 * modulo-sched.c (sms_schedule): Update.
23116 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23117 * predict.c (maybe_hot_count_p): Update.
23118 (probably_never_executed): Update.
23119 (dump_prediction): Update.
23120 (combine_predictions_for_bb): Update.
23121 (propagate_freq): Update.
23122 (handle_missing_profiles): Update.
23123 (counts_to_freqs): Update.
23124 (rebuild_frequencies): Update.
23125 (force_edge_cold): Update.
23126 * predict.h: Include profile-count.h
23127 (maybe_hot_count_p, counts_to_freqs): UPdate.
23128 * print-rtl-function.c: Do not include cfg.h
23129 * print-rtl.c: Include basic-block.h
23130 * profile-count.c: New file.
23131 * profile-count.h: New file.
23132 * profile.c (is_edge_inconsistent): Update.
23133 (correct_negative_edge_counts): Update.
23134 (is_inconsistent): Update.
23135 (set_bb_counts): Update.
23136 (read_profile_edge_counts): Update.
23137 (compute_frequency_overlap): Update.
23138 (compute_branch_probabilities): Update; Initialize and deinitialize
23140 (branch_prob): Update.
23141 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23142 (edge_gcov_count): New.
23143 (bb_gcov_count): New.
23144 * shrink-wrap.c (try_shrink_wrapping): Update.
23145 * tracer.c (better_p): Update.
23146 * trans-mem.c (expand_transaction): Update.
23147 (ipa_tm_insert_irr_call): Update.
23148 (ipa_tm_insert_gettmclone_call): Update.
23149 * tree-call-cdce.c: Update.
23150 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23151 (gimple_duplicate_sese_tail): Update.
23152 (gimple_account_profile_record): Update.
23153 (execute_fixup_cfg): Update.
23154 * tree-inline.c (copy_bb): Update.
23155 (copy_edges_for_bb): Update.
23156 (initialize_cfun): Update.
23157 (freqs_to_counts): Update.
23158 (copy_cfg_body): Update.
23159 (expand_call_inline): Update.
23160 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23161 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23162 (try_unroll_loop_completely): Update.
23163 (try_peel_loop): Update.
23164 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23165 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23166 * tree-ssa-loop-split.c (connect_loops): Update.
23167 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23168 * tree-ssa-reassoc.c (branch_fixup): Update.
23169 * tree-ssa-tail-merge.c (replace_block_by): Update.
23170 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23171 (compute_path_counts): Update.
23172 (update_profile): Update.
23173 (recompute_probabilities): Update.
23174 (update_joiner_offpath_counts): Update.
23175 (estimated_freqs_path): Update.
23176 (freqs_to_counts_path): Update.
23177 (clear_counts_path): Update.
23178 (ssa_fix_duplicate_block_edges): Update.
23179 (duplicate_thread_path): Update.
23180 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23181 (struct switch_conv_info): Update.
23182 * tree-tailcall.c (decrease_profile): Update.
23183 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23184 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23185 * value-prof.c (check_counter): Update.
23186 (gimple_divmod_fixed_value): Update.
23187 (gimple_mod_pow2): Update.
23188 (gimple_mod_subtract): Update.
23189 (gimple_ic_transform): Update.
23190 (gimple_stringop_fixed_value): Update.
23191 * value-prof.h (gimple_ic): Update.
23193 2017-06-02 Carl Love <cel@us.ibm.com>
23195 * config/rs6000/rs6000-c: Add support for built-in functions
23196 vector double vec_doublee (vector signed int);
23197 vector double vec_doublee (vector unsigned int);
23198 vector double vec_doublee (vector float);
23199 vector double vec_doubleh (vector signed int);
23200 vector double vec_doubleh (vector unsigned int);
23201 vector double vec_doubleh (vector float);
23202 vector double vec_doublel (vector signed int);
23203 vector double vec_doublel (vector unsigned int);
23204 vector double vec_doublel (vector float);
23205 vector double vec_doubleo (vector signed int);
23206 vector double vec_doubleo (vector unsigned int);
23207 vector double vec_doubleo (vector float);.
23208 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23209 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23211 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23212 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23213 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23215 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23216 vec_doublel, vec_doubleh.
23217 * doc/extend.texi: Update the built-in documentation file for the
23218 new built-in functions.
23220 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23223 * ipa-inline-analysis.c (free_growth_caches): Set
23224 edge_removal_hook_holder to NULL after removing it.
23226 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23228 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23229 comparision with zero.
23231 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23232 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23233 for early expansion of vec_min and vec_max builtins.
23234 (builtin_function_type): Add min/max unsigned variants to those
23235 identified as having unsigned arguments.
23237 2017-06-02 Olivier Hainque <hainque@adacore.com>
23239 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23241 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23243 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23244 Use VALL_F16 iterator rather than VALL.
23246 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23248 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23249 Emit CBNZ inside loop when doing a strong exchange and comparing
23250 against zero. Generate the CC flags after the loop.
23252 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23254 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23255 (dl_section_ref): New.
23256 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23257 On AIX, append an expression to subtract the size of the
23258 section length to dl_section_ref.
23260 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23262 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23263 for early expansion of vector absolute builtins.
23265 2017-06-02 Richard Biener <rguenther@suse.de>
23267 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23268 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23270 2017-06-02 Richard Biener <rguenther@suse.de>
23272 PR tree-optimization/80948
23273 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23274 stmt order as well.
23276 2017-06-02 Richard Biener <rguenther@suse.de>
23278 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23280 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23281 for inductions as relevant.
23283 2017-06-02 Richard Biener <rguenther@suse.de>
23285 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23286 (vectorizable_induction): ... this. Remove dead code.
23288 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23290 * builtins. (expand_builtin_alloca): Remove second parameter and
23291 infer its value from the first parameter instead.
23292 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23294 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23296 PR rtl-optimization/80903
23297 * loop-doloop.c (add_test): Unshare sequence.
23299 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23301 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23303 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23305 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23307 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23308 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23309 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23310 (xlogue_layout::get_stub_rtx): Make static.
23311 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23312 (xlogue_layout::compute_stub_managed_regs): Rename to...
23313 (xlogue_layout::count_stub_managed_regs): ...this.
23314 (xlogue_layout::is_stub_managed_reg): New function.
23315 (xlogue_layout::m_stub_names): Rename to...
23316 (xlogue_layout::s_stub_names): ...this, make static.
23317 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23318 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23319 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23320 xlogue_layout::s_stub_names): Instantiate statics.
23321 (stub_managed_regs): Remove.
23322 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23323 (disable_call_ms2sysv_xlogues): Rename to...
23324 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23325 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23327 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23328 change after reload_completed.
23329 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23331 (ix86_expand_prologue): Likewise.
23332 (ix86_expand_epilogue): Likewise.
23333 (ix86_expand_split_stack_prologue): Likewise.
23334 (ix86_compute_frame_layout): Remove frame parameter ...
23335 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23336 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23338 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23339 (ix86_frame): Move from here ...
23340 * config/i386/i386.h (ix86_frame): ... to here.
23341 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23342 complete ix86_frame data structure instead. Remove some_ld_name.
23344 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23346 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23347 symbols that hold a DECL_VALUE_EXPR.
23349 2017-06-01 Martin Jambor <mjambor@suse.cz>
23351 PR tree-optimization/80898
23352 * tree-sra.c (process_subtree_disqualification): Removed.
23353 (disqualify_candidate): Do not acll
23354 process_subtree_disqualification.
23355 (subtree_mark_written_and_enqueue): New function.
23356 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23357 RHS has been disqualified and re-queue LHS if necessary. Apart
23358 from that, ignore disqualified RHS.
23360 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23362 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23363 address fetch for z10 or later.
23365 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23367 * config/arc/arc.md (tst_movb): Add guard when splitting.
23369 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23371 * config/arc/arc.c (arc_can_eliminate): Test against
23372 arc_frame_pointer_needed.
23374 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23376 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23377 to prevent store reordering.
23378 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23379 (type): Add block type.
23380 (stack_tie): Define special instruction to be used in
23383 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23385 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23386 constraint. It is not valid for the pattern.
23387 (noncommutative_binary_comparison): Likewise.
23389 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23391 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23394 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23396 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23397 be used by the reg-alloc.
23399 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23401 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23402 reg-alloc when having mul64 or mul32x16 instructions.
23403 (mulsidi3): Likewise.
23404 (umulsidi3): Likewise.
23405 (mulsi32x16): New pattern.
23406 (mulsi64): Likewise.
23407 (mulsidi64): Likewise.
23408 (umulsidi64): Likewise.
23409 (MUL32x16_REG): Define.
23410 (mul64_600): Use MUL32x16_REG.
23411 (mac64_600): Likewise.
23412 (umul64_600): Likewise.
23413 (umac64_600): Likewise.
23415 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23417 * config/arc/arc.md (mulsi3_700): Make it commutative.
23419 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23421 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23423 (*sign_extendsidi2_insn): Likewise for movstosw.
23425 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23427 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23428 the type of the input discriminant value. Convert the
23429 discriminant value of signedness vary.
23431 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23433 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23434 Add to -Wall section.
23436 2017-06-01 Richard Biener <rguenther@suse.de>
23438 PR middle-end/66313
23439 * fold-const.c (fold_plusminus_mult_expr): If the factored
23440 factor may be zero use a wrapping type for the inner operation.
23441 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23442 and handle moved defs.
23443 (process_assignment): Properly guard the unary op case. Return a
23444 tri-state indicating that moving the stmt before the call may allow
23445 to continue. Pass through to_move.
23446 (find_tail_calls): Handle moving unrelated defs before
23449 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23452 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23453 splitter result in the canonical way.
23455 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23457 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23458 also for 32bit target. Update insn attributes.
23459 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23461 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23463 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23464 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23465 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23466 (_mm_maskz_min_ss): New intrinsics.
23468 2017-05-31 Martin Liska <mliska@suse.cz>
23470 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23471 Change comment style to one we normally use.
23472 (vectorizable_reduction): Likewise.
23473 (vectorizable_induction): Likewise.
23474 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23475 (vectorizable_call): Likewise.
23476 (vectorizable_simd_clone_call): Likewise.
23477 (vectorizable_conversion): Likewise.
23478 (vectorizable_assignment): Likewise.
23479 (vectorizable_shift): Likewise.
23480 (vectorizable_operation): Likewise.
23481 (vectorizable_store): Likewise.
23482 (vectorizable_load): Likewise.
23483 * tree-vectorizer.h: Likewise.
23485 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23487 * passes.c (emergency_dump_function): New.
23488 * tree-pass.h (emergency_dump_function): Declare.
23489 * plugin.c (plugins_internal_error_function): Remove.
23490 * plugin.h (plugins_internal_error_function): Remove declaration.
23491 * toplev.c (internal_error_function): New static function. Use it...
23492 (general_init): ...here.
23494 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23496 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23497 (arc_rtx_costs): Add costs for new patterns.
23498 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23499 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23501 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23503 * tree-ssa-strlen.c (get_next_strinfo): New function.
23504 (get_stridx_plus_constant): Use it.
23505 (zero_length_string): Likewise.
23506 (adjust_related_strinfos): Likewise.
23507 (adjust_last_stmt): Likewise.
23509 2017-05-31 Richard Biener <rguenther@suse.de>
23512 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23513 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23515 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23517 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23518 loop_vinfo argument and use of dependence distance vectors.
23519 Check instead whether the two references differ only in their
23520 initial value and assume that they have the same alignment if the
23521 difference is a multiple of the vector alignment.
23522 (vect_analyze_data_refs_alignment): Update call accordingly.
23524 2017-05-31 Martin Liska <mliska@suse.cz>
23527 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23529 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23531 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23532 (create_intersect_range_checks): Move from ...
23533 * tree-data-ref.c (create_intersect_range_checks_index)
23534 (create_intersect_range_checks): ... to here.
23535 (create_runtime_alias_checks): New function factored from ...
23536 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23537 here. Call above function.
23538 * tree-data-ref.h (create_runtime_alias_checks): New function.
23540 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23542 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23543 segment length for dr_b and compute it in wide_int.
23545 2017-05-31 Richard Biener <rguenther@suse.de>
23547 PR tree-optimization/80906
23548 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23549 and pass through iv_map.
23550 (copy_bb_and_scalar_dependences): Adjust.
23551 (translate_pending_phi_nodes): Likewise.
23552 (copy_loop_close_phi_args): Handle code-generating IVs instead
23555 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23557 * diagnostic-color.c (color_dict): Add "type-diff".
23558 (parse_gcc_colors): Update comment.
23559 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23560 -fdiagnostics-show-template-tree and -fno-elide-type.
23561 (GCC_COLORS): Add type-diff to example.
23563 (-fdiagnostics-show-template-tree): New.
23564 (-fno-elide-type): New.
23565 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23566 the pp_format_decoder callback. Call any m_format_postprocessor's
23568 (pretty_printer::pretty_printer): Initialize
23569 m_format_postprocessor.
23570 (pretty_printer::~pretty_printer): Delete any
23571 m_format_postprocessor.
23572 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23573 (class format_postprocessor): New class.
23574 (struct pretty_printer::format_decoder): Document the new parameters.
23575 (struct pretty_printer::m_format_postprocessor): New field.
23576 * tree-diagnostic.c (default_tree_printer): Update for new
23577 bool and const char ** params.
23578 * tree-diagnostic.h (default_tree_printer): Likewise.
23580 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23582 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23583 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23584 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23585 rs6000_gen_cell_microcode code.
23586 (rs6000_final_prescan_insn): Delete.
23587 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23588 "warn-cell-microcode" entries.
23589 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23590 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23591 throughout. Change cc_reg_not_micro_cr0_operand to
23592 cc_reg_not_cr0_operand throughout.
23593 (*extendhi<mode>2_noload): Delete.
23594 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23595 (mwarn-cell-microcode): Delete.
23596 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23597 -mgen-cell-microcode and -mwarn-cell-microcode.
23599 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23602 * config/i386/constraints.md (Yd): New constraint.
23604 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23605 and (?Yd, r) alternatives. Update insn attributes.
23606 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23607 and (?*Yd, r) alternatives. Update insn attributes.
23608 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23610 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23612 * gimplify.c (gimplify_modify_expr): Don't create a
23613 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23616 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23618 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23620 2017-05-30 Richard Biener <rguenther@suse.de>
23622 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23623 and reduc_def fields.
23624 (STMT_VINFO_REDUC_TYPE): New define.
23625 (STMT_VINFO_REDUC_DEF): Likewise.
23626 (vect_force_simple_reduction): Adjust prototype.
23627 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23628 (vect_is_simple_reduction): Remove check_reduction argument.
23629 (vect_force_simple_reduction): Adjust and set
23630 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23631 (vectorizable_reduction): Do not re-do reduction analysis
23632 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23633 * tree-parloops.c (gather_scalar_reductions): Adjust.
23635 2017-05-30 Richard Biener <rguenther@suse.de>
23637 PR middle-end/80901
23638 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23641 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23643 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23644 Introduce unknown_misalignment parameter and remove vf.
23645 (vect_peeling_hash_get_lowest_cost):
23646 Pass unknown_misalignment parameter.
23647 (vect_enhance_data_refs_alignment):
23648 Fix unsupportable data ref treatment.
23650 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23652 * tree-vect-data-refs.c (vect_get_data_access_cost):
23653 Workaround for SLP handling.
23654 (vect_enhance_data_refs_alignment):
23655 Compute costs for doing no peeling at all, compare to the best
23656 peeling costs so far and avoid peeling if cheaper.
23658 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23660 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23661 Return peeling info and set costs to zero for unlimited cost
23663 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23664 with unknown misalignment. Compute and costs for unknown
23665 misalignment, compare them to the costs for known misalignment
23666 and choose the cheapest for peeling.
23668 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23670 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23671 (vect_get_peeling_costs_all_drs): Create function.
23672 (vect_peeling_hash_get_lowest_cost):
23673 Use vect_get_peeling_costs_all_drs.
23674 (vect_peeling_supportable): Create function.
23675 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23677 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23679 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23680 DR_HAS_NEGATIVE_STEP.
23681 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23682 (vect_enhance_data_refs_alignment): Use.
23683 (vect_duplicate_ssa_name_ptr_info): Use.
23684 * tree-vectorizer.h (dr_misalignment): Use.
23685 (known_alignment_for_access_p): Use.
23687 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23690 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23692 (has_section_name): New function.
23694 2017-05-30 Martin Liska <mliska@suse.cz>
23697 * auto-profile.c (get_function_decl_from_block): Fix
23700 2017-05-30 Richard Biener <rguenther@suse.de>
23702 PR middle-end/80876
23703 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23705 2017-05-30 Martin Liska <mliska@suse.cz>
23707 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23708 * dumpfile.h (struct dump_file_info): Remove ctors.
23710 2017-05-30 Martin Liska <mliska@suse.cz>
23712 * predict.def: Fix GNU coding style.
23714 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23716 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23717 Mark 'to' argument with ATTRIBUTE_UNUSED.
23719 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23721 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23722 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23723 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23726 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23728 * doc/install.texi (Options specification): Restore entry of
23729 --enable-sjlj-exceptions.
23731 2017-05-27 Michael Eager <eager@eagercon.com>
23734 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23736 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23738 * config/microblaze/microblaze.h
23739 (FIXED_REGISTERS): Update in macro.
23740 (CALL_USED_REGISTERS): Update in macro.
23742 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23744 * doc/install.texi: Add links to macOS binary distributions.
23746 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23750 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23752 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23754 2017-05-26 Martin Liska <mliska@suse.cz>
23756 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23758 2017-05-26 Martin Liska <mliska@suse.cz>
23760 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23761 always leading ';; '.
23762 (dump_bb_info): Likewise.
23763 (brief_dump_cfg): Likewise.
23764 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23765 * dumpfile.c: Remove usage of TDF_VERBOSE.
23766 * dumpfile.h (enum dump_kind): Likewise.
23767 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23768 * print-tree.c (debug_verbose): Remove.
23769 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23770 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23771 * tree-diagnostic.c (default_tree_printer): Replace
23772 TDF_DIAGNOSTIC with TDF_SLIM.
23774 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23776 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23777 in parameter loop, rather than loop_vinfo.
23778 (create_intersect_range_checks): Ditto.
23779 (vect_create_cond_for_alias_checks): Update call to above functions.
23781 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23783 PR tree-optimization/80815
23784 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23785 for merging runtime alias checks. Handle negative DR_STEPs.
23787 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23789 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23791 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23792 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23793 out code pruning runtime alias checks.
23794 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23795 factored out from above.
23796 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23798 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23800 (prune_runtime_alias_test_list): New decalaration.
23802 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23804 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23805 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23806 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23807 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23808 (operator==, comp_dr_with_seg_len_pair): Ditto.
23809 (vect_prune_runtime_alias_test_list): Ditto.
23811 2017-05-26 Martin Liska <mliska@suse.cz>
23814 * params.def: Bound partial-inlining-entry-probability param.
23816 2017-05-26 Marek Polacek <polacek@redhat.com>
23819 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23822 2017-05-26 Richard Biener <rguenther@suse.de>
23824 PR tree-optimization/80842
23825 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23828 2017-05-26 Richard Biener <rguenther@suse.de>
23830 PR tree-optimization/80844
23831 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23833 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23835 * doc/md.texi (Machine Constraints): Update x86 family
23836 machine constraints section to match 'config/i386/constraints.md'.
23838 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23840 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23842 2017-05-25 Nathan Sidwell <nathan@acm.org>
23844 * doc/invoke.texi (--enable-languages): Update documentation.
23846 2017-05-25 Martin Liska <mliska@suse.cz>
23848 * dumpfile.c: Add TDF_FOLDING.
23849 * dumpfile.h (enum dump_kind): Likewise.
23850 * genmatch.c (dt_simplify::gen_1): Use it.
23852 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23854 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23856 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23858 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23859 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23861 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23863 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23864 * match.pd (X == C): Rewrite it here.
23865 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23866 with_certain_nonzero_bits2): New predicates.
23867 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23869 2017-05-24 Nathan Sidwell <nathan@acm.org>
23871 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23874 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23877 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23879 * config/powerpcspe: New port. Files are copied from the rs6000
23880 port, with "rs6000" in filenames replaced by "powerpcspe".
23882 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23884 PR rtl-optimization/80754
23885 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23887 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23889 * config/sparc/sparc.md (length): Return the correct value for -mflat
23890 sibcalls to match output_sibcall.
23892 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23896 * config/rs6000/rs6000.c (struct machine_function): Add new field
23898 (rs6000_get_separate_components): Init that field, use it.
23899 (rs6000_components_for_bb): Use the field.
23901 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23903 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23905 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23907 PR middle-end/80823
23908 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23910 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23913 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23914 address against address_operand predicate.
23915 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23917 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23919 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23920 parameters passed indirectly.
23922 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23924 * config/i386/i386.md (*movdi_internal): Remove SSE4
23925 alternative 18 (?r, *v). Update insn attributes.
23926 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23927 Update insn attributes.
23928 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23929 Update insn attributes.
23930 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23931 alternative 1 (r, v). Remove isa attribute.
23932 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23933 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23934 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23936 2017-05-23 Tom de Vries <tom@codesourcery.com>
23938 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23941 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23943 * cgraphunit.c (symbol_table::process_new_functions): Update.
23944 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23945 (inline_generate_summary): Rename to ...
23946 (ipa_fn_summary_generate): ... this one.
23947 (inline_read_summary): Rename to ...
23948 (ipa_fn_summary_read): ... this one.
23949 (inline_write_summary): Rename to ...
23950 (ipa_fn_summary_write): ... this one.
23951 (inline_free_summary): Rename to ...
23952 (ipa_free_fn_summary): ... this one.
23953 (pass_data_local_fn_summary, pass_local_fn_summary,
23954 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23955 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23956 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23957 make_pass_ipa_fn_summary): New.
23958 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23959 inline_write_summary, inline_free_summary): Remove.
23960 (ipa_free_fn_summary) : New.
23961 * ipa-inline.c (ipa_inline): Update.
23962 (pass_ipa_inline): Do not generate summaries.
23963 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23965 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23966 and add pass_ipa_fn_summary.
23967 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23969 (make_pass_inline_parameters): Remove.
23971 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23973 * omp-low.c (struct omp_context): Remove "default_kind" member.
23976 * omp-offload.c (execute_oacc_device_lower): Remove the
23977 parallelism dimensions function attributes for unparallelized
23978 OpenACC kernels constructs.
23980 2017-05-23 Martin Liska <mliska@suse.cz>
23982 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23984 (cgraph_edge::make_speculative): Likewise.
23985 (cgraph_edge::resolve_speculation): Likewise.
23986 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23987 (cgraph_node::dump): Likewise.
23988 * cgraph.h: Likewise.
23989 * cgraphunit.c (analyze_functions): Likewise.
23990 (symbol_table::compile): Likewise.
23991 * ipa-cp.c (print_all_lattices): Likewise.
23992 (determine_versionability): Likewise.
23993 (initialize_node_lattices): Likewise.
23994 (ipcp_verify_propagated_values): Likewise.
23995 (estimate_local_effects): Likewise.
23996 (update_profiling_info): Likewise.
23997 (create_specialized_node): Likewise.
23998 (perhaps_add_new_callers): Likewise.
23999 (decide_about_value): Likewise.
24000 (decide_whether_version_node): Likewise.
24001 (identify_dead_nodes): Likewise.
24002 (ipcp_store_bits_results): Likewise.
24003 * ipa-devirt.c (dump_targets): Likewise.
24004 (ipa_devirt): Likewise.
24005 * ipa-icf.c (sem_item::dump): Likewise.
24006 (sem_function::equals): Likewise.
24007 (sem_variable::equals): Likewise.
24008 (sem_item_optimizer::read_section): Likewise.
24009 (sem_item_optimizer::execute): Likewise.
24010 (congruence_class::dump): Likewise.
24011 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
24012 (dump_inline_summary): Likewise.
24013 (estimate_node_size_and_time): Likewise.
24014 (inline_analyze_function): Likewise.
24015 * ipa-inline-transform.c (inline_call): Likewise.
24016 * ipa-inline.c (report_inline_failed_reason): Likewise.
24017 (want_early_inline_function_p): Likewise.
24018 (edge_badness): Likewise.
24019 (update_edge_key): Likewise.
24020 (inline_small_functions): Likewise.
24021 * ipa-profile.c (ipa_profile): Likewise.
24022 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
24023 (ipa_make_edge_direct_to_target): Likewise.
24024 (remove_described_reference): Likewise.
24025 (ipa_impossible_devirt_target): Likewise.
24026 (propagate_controlled_uses): Likewise.
24027 (ipa_print_node_params): Likewise.
24028 (ipcp_transform_function): Likewise.
24029 * ipa-pure-const.c (pure_const_read_summary): Likewise.
24030 (propagate_pure_const): Likewise.
24031 * ipa-reference.c (generate_summary): Likewise.
24032 (read_write_all_from_decl): Likewise.
24033 (propagate): Likewise.
24034 (ipa_reference_read_optimization_summary): Likewise.
24035 * ipa-utils.c (ipa_merge_profiles): Likewise.
24036 * ipa.c (walk_polymorphic_call_targets): Likewise.
24037 (symbol_table::remove_unreachable_nodes): Likewise.
24038 (ipa_single_use): Likewise.
24039 * passes.c (execute_todo): Likewise.
24040 * predict.c (drop_profile): Likewise.
24041 * symtab.c (symtab_node::get_dump_name): New function.
24042 (symtab_node::dump_name): Likewise.
24043 (symtab_node::dump_asm_name): Likewise.
24044 (symtab_node::dump_references): Likewise.
24045 (symtab_node::dump_referring): Likewise.
24046 (symtab_node::dump_base): Likewise.
24047 (symtab_node::debug_symtab): Likewise.
24048 * tree-sra.c (convert_callers_for_node): Likewise.
24049 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24050 * value-prof.c (init_node_map): Likewise.
24052 2017-05-23 Martin Liska <mliska@suse.cz>
24054 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24055 and symtab_node::debug_symtab to symbol_table::debug.
24056 * cgraphunit.c (analyze_functions): Use the renamed function.
24057 (symbol_table::compile): Likewise.
24058 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24059 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24060 * passes.c (execute_todo): Likewise.
24061 * symtab.c (symbol_table::dump): New function.
24062 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24064 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24066 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24067 that nonconst implies exec.
24069 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24071 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24072 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24073 (inline_edge_summary_vec): Turn into ...
24074 (ipa_call_summaries): ... this one.
24075 (redirect_to_unreachable, edge_set_predicate,
24076 evaluate_properties_for_edge, inline_summary_alloc,
24077 reset_ipa_call_summary, reset_inline_summary,
24078 inline_summary_t::duplicate): Update.
24079 (inline_edge_duplication_hook): Turn to ...
24080 (ipa_call_summary_t::duplicate): ... this one.
24081 (inline_edge_removal_hook): Turn to ...
24082 (ipa_call_summary_t::remove): ... this one.
24083 (dump_inline_edge_summary): Turn to ...
24084 (dump_ipa_call_summary): ... this one.
24085 (estimate_function_body_sizes): Update.
24086 (inline_update_callee_summaries): Update.
24087 (remap_edge_change_prob): Update.
24088 (remap_edge_summaries): Update.
24089 (inline_merge_summary): Update.
24090 (do_estimate_edge_time): Update.
24091 (inline_generate_summary): Update.
24092 (inline_read_section): Update.
24093 (inline_read_summary): Update.
24094 (inline_free_summary): Update.
24095 * ipa-inline.c (can_inline_edge_p): Update.
24096 (compute_inlined_call_time): Update.
24097 (want_inline_small_function_p): Update.
24098 (edge_badness): Update.
24099 (early_inliner): Update.
24100 * ipa-inline.h (inline_edge_summary): Turn to ...
24101 (ipa_call_summary): ... this one.
24102 (ipa_call_summary_t): New class.
24103 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24104 (ipa_call_summaries): New.
24105 (inline_edge_summary): Remove.
24106 (estimate_edge_growth): Update.
24107 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24108 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24109 * ipa-split.c (execute_split_functions): Update.
24110 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24112 2017-05-23 Tom de Vries <tom@codesourcery.com>
24114 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24115 attributes): Document rdrand effective target.
24117 2017-05-23 Tom de Vries <tom@codesourcery.com>
24119 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24120 attributes): Sort alphabetically.
24122 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24124 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24126 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24129 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24130 V2DF/V2DI splat into two separate patterns, one that handles
24131 registers, and the other that only handles memory. Drop support
24132 for splatting from a GPR on ISA 2.07 and then splitting the
24133 splat into direct move and splat.
24134 (vsx_splat_<mode>_reg): Likewise.
24135 (vsx_splat_<mode>_mem): Likewise.
24137 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24139 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24141 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24143 PR middle-end/80809
24144 * omp-low.c (finish_taskreg_remap): New function.
24145 (finish_taskreg_scan): If unit size of ctx->record_type
24146 is non-constant, unshare the size expression and replace
24147 decls in it with possible outer var refs.
24149 PR middle-end/80809
24150 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24151 GOVD_SHARED rather than GOVD_PRIVATE with it.
24152 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24153 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24155 PR middle-end/80853
24156 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24157 as last argument to build_outer_var_ref for pointer bases of array
24158 section reductions.
24160 2017-05-19 Martin Sebor <msebor@redhat.com>
24162 * print-tree.c (print_node): Print DECL_READ_P flag.
24164 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24166 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24167 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24168 * cgraph.c: Likewise.
24169 * cgraphunit.c: Likewise.
24170 * gengtype.c: Likewise.
24171 * ipa-cp.c: Likewise.
24172 * ipa-devirt.c: Likewise.
24173 * ipa-icf.c: Likewise.
24174 * ipa-predicate.c: Likewise.
24175 * ipa-profile.c: Likewise.
24176 * ipa-prop.c: Likewise.
24177 * ipa-split.c: Likewise.
24179 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24180 edge_predicate_pool, dump_inline_hints,
24181 inline_summary::account_size_time, redirect_to_unreachable,
24182 edge_set_predicate, set_hint_predicate,
24183 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24184 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24185 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24186 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24187 ipa_call_summary_t::remove, initialize_growth_caches,
24188 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24189 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24190 mark_modified, unmodified_parm_1, unmodified_parm,
24191 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24192 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24193 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24194 will_be_nonconstant_predicate, record_modified_bb_info,
24195 get_minimal_bb, record_modified, param_change_prob,
24196 phi_result_unknown_predicate, predicate_for_phi_result,
24197 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24198 estimate_function_body_sizes, compute_inline_parameters,
24199 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24200 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24201 inline_update_callee_summaries, remap_edge_change_prob,
24202 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24203 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24204 inline_analyze_function, inline_summary_t::insert,
24205 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24206 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24207 inline_free_summary): Move to ipa-fnsummary.h
24208 (predicate_t): Remove.
24209 * ipa-fnsummary.c: New file.
24210 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24211 (enum inline_hints_vals, inline_hints, agg_position_info,
24212 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24213 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24214 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24215 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24216 inline_read_summary, inline_write_summary, inline_free_summary,
24217 inline_analyze_function, initialize_inline_failed,
24218 inline_merge_summary, inline_update_overall_summary,
24219 compute_inline_parameters): Move to ipa-fnsummary.h
24220 * ipa-fnsummary.h: New file.
24221 * ipa-inline-transform.h: Include ipa-inline.h.
24222 * ipa-inline.c: LIkewise.
24224 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24226 * ipa-inline.c (edge_badness): Use inlined_time instead of
24227 inline_summaries->get.
24229 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24231 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24233 2017-05-22 Nathan Sidwell <nathan@acm.org>
24235 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24236 (fdump-lang): Document 'raw' option.
24237 * dumpfile.h (TDI_tu): Delete.
24238 * dumpfile.c (dump_files): Remove translation-unit.
24239 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24241 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24243 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24244 command option from $(AWK) call.
24245 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24246 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24247 [FORMAT]: Remove handling of variable.
24248 * config/avr/t-multilib: Regenerate.
24250 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24252 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24254 (dump_inline_summary): Do not print self_time.
24255 (estimate_function_body_sizes): Do not set self_time.
24256 (compute_inline_parameters): Likewise.
24257 (inline_read_section, inline_write_summary): Do not stream self_time.
24258 * ipa-inline.h (inline_summary): Drop self_time.
24260 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24262 * ipa-inline-analysis.c (account_size_time): Rename to ...
24263 (inline_summary::account_size_time): ... this one.
24264 (reset_ipa_call_summary): Turn to ...
24265 (ipa_call_summary::reset): ... this one.
24266 (reset_inline_summary): Turn to ...
24267 (inline_summary::reset): ... this one.
24268 (inline_summary_t::remove): Update.
24269 (inline_summary_t::duplicate): Update.
24270 (ipa_call_summary_t::remove): Update.
24271 (dump_inline_summary): Update.
24272 (estimate_function_body_sizes): Update.
24273 (compute_inline_parameters): Update.
24274 (estimate_node_size_and_time): Update.
24275 (inline_merge_summary): Update.
24276 (inline_update_overall_summary): Update.
24277 (inline_read_section): Update.
24278 (inline_write_summary): Update.
24279 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24280 add account_size_time and reset member functions.
24281 (ipa_call_summary): Add reset function.
24282 * ipa-predicate.h (predicate::operator &): Constify.
24284 2017-05-22 Richard Biener <rguenther@suse.de>
24286 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24288 2017-05-19 Jason Merrill <jason@redhat.com>
24290 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24292 2017-05-19 Marek Polacek <polacek@redhat.com>
24295 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24296 TYPE_OVERFLOW_WRAPS checks.
24298 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24300 * tree-core.h (enum omp_clause_default_kind): Add
24301 "OMP_CLAUSE_DEFAULT_PRESENT".
24302 * tree-pretty-print.c (dump_omp_clause): Handle it.
24303 * gimplify.c (enum gimplify_omp_var_data): Add
24304 "GOVD_MAP_FORCE_PRESENT".
24305 (gimplify_adjust_omp_clauses_1): Map it to
24306 "GOMP_MAP_FORCE_PRESENT".
24307 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24309 * gimplify.c (oacc_default_clause): Clarify.
24311 2017-05-19 Nathan Sidwell <nathan@acm.org>
24313 LANG_HOOK_REGISTER_DUMPS
24314 * toplev.c (general_init): Call register dump lang hook.
24315 * doc/invoke.texi: Document -fdump-lang option family.
24316 * dumpfile.c (dump_files): Remove class dump here.
24317 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24318 * dumpfile.h (tree_dump_index): Remove TDI_class.
24319 * langhooks-def.h (lhd_register_dumps): Declare.
24320 (LANG_HOOKS_REGISTER_DUMPS): Define.
24321 (LANG_HOOKS_INITIALIZER): Add it.
24322 * langhooks.c (lhd_register_dumps): Define.
24323 * langhooks.h (struct lang_hooks): Add register_dumps.
24325 2017-05-19 Nathan Sidwell <nathan@acm.org>
24327 * context.h (context::set_passes): New.
24328 * context.c (context::context): Do not create pass manager.
24329 * toplev.c (general_init): Create pass manager here.
24331 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24333 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24334 use this splitter if two add or or instructions would also work for
24335 the constant we want to generate.
24337 2017-05-19 Richard Biener <rguenther@suse.de>
24340 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24341 predicate evaluation.
24343 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24345 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24347 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24350 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24352 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24353 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24354 (inline_edge_summary_vec): Turn into ...
24355 (ipa_call_summaries): ... this one.
24356 (redirect_to_unreachable, edge_set_predicate,
24357 evaluate_properties_for_edge, inline_summary_alloc,
24358 reset_ipa_call_summary, reset_inline_summary,
24359 inline_summary_t::duplicate): Update.
24360 (inline_edge_duplication_hook): Turn to ...
24361 (ipa_call_summary_t::duplicate): ... this one.
24362 (inline_edge_removal_hook): Turn to ...
24363 (ipa_call_summary_t::remove): ... this one.
24364 (dump_inline_edge_summary): Turn to ...
24365 (dump_ipa_call_summary): ... this one.
24366 (estimate_function_body_sizes): Update.
24367 (inline_update_callee_summaries): Update.
24368 (remap_edge_change_prob): Update.
24369 (remap_edge_summaries): Update.
24370 (inline_merge_summary): Update.
24371 (do_estimate_edge_time): Update.
24372 (inline_generate_summary): Update.
24373 (inline_read_section): Update.
24374 (inline_read_summary): Update.
24375 (inline_free_summary): Update.
24376 * ipa-inline.c (can_inline_edge_p): Update.
24377 (compute_inlined_call_time): Update.
24378 (want_inline_small_function_p): Update.
24379 (edge_badness): Update.
24380 (early_inliner): Update.
24381 * ipa-inline.h (inline_edge_summary): Turn to ...
24382 (ipa_call_summary): ... this one.
24383 (ipa_call_summary_t): New class.
24384 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24385 (ipa_call_summaries): New.
24386 (inline_edge_summary): Remove.
24387 (estimate_edge_growth): Update.
24388 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24389 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24390 * ipa-split.c (execute_split_functions): Update.
24391 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24393 2017-05-19 Richard Biener <rguenther@suse.de>
24395 PR middle-end/80764
24396 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24398 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24400 * config/rs6000/rs6000.c (struct machine_function): Add field
24401 fpr_is_wrapped_separately.
24402 (rs6000_get_separate_components): Use 64 components. Handle the
24403 new FPR components.
24404 (rs6000_components_for_bb): Handle the FPR components.
24405 (rs6000_emit_prologue_components): Handle the FPR components.
24406 (rs6000_emit_epilogue_components): Handle the FPR components.
24407 (rs6000_set_handled_components): Handle the FPR components.
24408 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24409 that are already separately shrink-wrapped.
24410 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24411 that are already separately shrink-wrapped.
24413 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24416 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24419 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24420 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24421 cases where the register allocator uses a move and an offsettable
24422 memory operation to/from a FPR register on ISA 2.06/2.07.
24423 (define_peephole2 for Altivec d-form store): Likewise.
24425 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24428 * config/i386/mmx.md (*mov<mode>_internal): Enable
24429 alternatives 11, 12, 13 and 14 also for 32bit targets.
24430 Remove alternatives 15, 16, 17 and 18.
24431 * config/i386/sse.md (vec_concatv2di): Change
24432 alternative (!x, *y) to (x, ?!*Yn).
24434 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24436 * dumpfile.h (enum dump_kind): Remove stray comma.
24438 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24440 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24441 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24442 predicate::num_conditions
24443 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24444 (CHANGED): turn into predicate::changed.
24445 (agg_position_info): Move to ipa-predicate.h
24446 (add_condition, predicate::add_clause, predicate::operator &=,
24447 predicate::or_with, predicate::evaluate, predicate::probability,
24448 dump_condition, dump_clause, predicate::dump,
24449 predicate::remap_after_duplication, predicate::remap_after_inlining,
24450 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24451 (evaluate_conditions_for_known_args): Update.
24452 (set_cond_stmt_execution_predicate): Update.
24453 * ipa-inline.h: Include ipa-predicate.h
24454 (condition, inline_param_summary, conditions, agg_position_info,
24455 predicate): Move to ipa-predicate.h
24456 * ipa-predicate.c: New file.
24457 * ipa-predicate.h: New file.
24459 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24461 * final.c (leaf_function_p): Check we are not in a sequence.
24463 2017-05-18 Martin Liska <mliska@suse.cz>
24465 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24466 * dumpfile.c (dump_register): Use new enum dump_kind.
24467 (get_dump_file_name): Likewise.
24468 (dump_enable_all): Likewise.
24469 (dump_switch_p_1): Likewise.
24470 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24471 * dumpfile.h (enum dump_kind): New enum type.
24472 (struct dump_file_info): Create constructor and
24473 format fields and comments.
24474 * passes.c (pass_manager::register_one_dump_file):
24476 * statistics.c (statistics_early_init): Likewise.
24477 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24478 TDF_TREE with TDF_SLIM.
24479 (gather_memory_references_ref): Likewise.
24481 2017-05-18 Martin Liska <mliska@suse.cz>
24483 * vec.h (struct vnull): Use it.
24485 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24487 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24488 (true_predicate, false_predicate, true_predicate_p,
24489 false_predicate_p): Remove.
24490 (single_cond_predicate, not_inlined_predicate): Turn to member function
24492 (add_condition): Update.
24493 (add_clause): Turn to...
24494 (predicate::add_clause): ... this one; update; allow passing NULL
24496 (and_predicates): Turn to ...
24497 (predicate::operator &=): ... this one.
24498 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24499 (or_predicates): Turn to ...
24500 (predicate::or_with): ... this one.
24501 (evaluate_predicate): Turn to ...
24502 (predicate::evaluate): ... this one.
24503 (predicate_probability): Turn to ...
24504 (predicate::probability): ... this one.
24505 (dump_condition): Update.
24506 (dump_predicate): Turn to ...
24507 (predicate::dump): ... this one.
24508 (account_size_time): Update.
24509 (edge_set_predicate): Update.
24510 (set_hint_predicate): UPdate.
24511 (evaluate_conditions_for_known_args): Update.
24512 (evaluate_properties_for_edge): Update.
24513 (remap_predicate_after_duplication): Turn to...
24514 (predicate::remap_after_duplication): ... this one.
24515 (remap_hint_predicate_after_duplication): Update.
24516 (inline_summary_t::duplicate): UPdate.
24517 (dump_inline_edge_summary): Update.
24518 (dump_inline_summary): Update.
24519 (set_cond_stmt_execution_predicate): Update.
24520 (set_switch_stmt_execution_predicate): Update.
24521 (compute_bb_predicates): Update.
24522 (will_be_nonconstant_expr_predicate): Update.
24523 (will_be_nonconstant_predicate): Update.
24524 (phi_result_unknown_predicate): Update.
24525 (predicate_for_phi_result): Update.
24526 (array_index_predicate): Update.
24527 (estimate_function_body_sizes): Update.
24528 (estimate_node_size_and_time): Update.
24529 (estimate_ipcp_clone_size_and_time): Update.
24530 (remap_predicate): Rename to ...
24531 (predicate::remap_after_inlining): ... this one.
24532 (remap_hint_predicate): Update.
24533 (inline_merge_summary): Update.
24534 (inline_update_overall_summary): Update.
24535 (estimate_size_after_inlining): Update.
24536 (read_predicate): Rename to ...
24537 (predicate::stream_in): ... this one.
24538 (read_inline_edge_summary): Update.
24539 (write_predicate): Rename to ...
24540 (predicate::stream_out): ... this one.
24541 (write_inline_edge_summary): Update.
24542 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24543 (clause_t): Turn to uint32_t
24544 (predicate): Turn to class; implement constructor and operators
24546 (size_time_entry): Update.
24547 (inline_summary): Update.
24548 (inline_edge_summary): Update.
24550 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24552 * fold-const.c (fold_binary_loc): Move transformation...
24553 * match.pd (C - X CMP X): ... here.
24555 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24557 * config/sparc/sparc.c (sparc_option_override): Set function
24558 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24559 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24561 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24563 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24565 2017-05-18 Marek Polacek <polacek@redhat.com>
24568 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24569 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24570 gimple_assign_load_p.
24572 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24574 PR middle-end/80692
24575 * real.c (do_compare): Give decimal_do_compare preference over
24576 comparing just the signs.
24578 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24580 * doc/md.texi (Canonicalization of Instructions): Describe the
24581 canonical form of instructions that inherently set a condition
24584 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24586 PR middle-end/80775
24587 * tree-cfg.c: Move deletion of unreachable case statements to after
24588 the merging of consecutive case labels.
24590 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24592 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24593 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24594 restoring of callee-saved registers.
24596 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24598 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24599 * config/visium/visium.c (single_set_and_flags): Likewise.
24600 * config/visium/visium.md (Substitutions): Likewise.
24602 2017-05-17 Martin Liska <mliska@suse.cz>
24604 * cfg.c: Introduce dump_flags_t type and
24605 use it instead of int type.
24607 * cfghooks.c: Likewise.
24608 * cfghooks.h (struct cfg_hooks): Likewise.
24609 * cfgrtl.c: Likewise.
24610 * cfgrtl.h: Likewise.
24611 * cgraph.c (cgraph_node::get_body): Likewise.
24612 * coretypes.h: Likewise.
24613 * domwalk.c: Likewise.
24614 * domwalk.h: Likewise.
24615 * dumpfile.c (struct dump_option_value_info): Likewise.
24616 (dump_enable_all): Likewise.
24617 (dump_switch_p_1): Likewise.
24618 (opt_info_switch_p): Likewise.
24619 * dumpfile.h (enum tree_dump_index): Likewise.
24620 (struct dump_file_info): Likewise.
24621 * genemit.c: Likewise.
24622 * generic-match-head.c: Likewise.
24623 * gengtype.c (open_base_files): Likewise.
24624 * gimple-pretty-print.c: Likewise.
24625 * gimple-pretty-print.h: Likewise.
24626 * graph.c (print_graph_cfg): Likewise.
24627 * graphite-scop-detection.c (dot_all_sese): Likewise.
24628 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24629 * loop-unroll.c (report_unroll): Likewise.
24630 * passes.c (pass_manager::register_one_dump_file): Likewise.
24631 * print-tree.c: Likewise.
24632 * statistics.c: Likewise.
24633 * tree-cfg.c: Likewise.
24634 * tree-cfg.h: Likewise.
24635 * tree-dfa.c: Likewise.
24636 * tree-dfa.h: Likewise.
24637 * tree-dump.c (dump_function): Likewise.
24638 * tree-dump.h (struct dump_info): Likewise.
24639 * tree-pretty-print.c: Likewise.
24640 * tree-pretty-print.h: Likewise.
24641 * tree-ssa-live.c: Likewise.
24642 * tree-ssa-live.h: Likewise.
24643 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24644 * tree-vect-loop.c: Likewise.
24645 * tree-vect-slp.c: Likewise.
24647 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24648 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24650 PR tree-optimization/80457
24651 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24652 of all arguments to a statement as scalar_to_vec operations.
24653 (vectorizable_call): Adjust call to vect_model_simple_cost for
24655 (vectorizable_conversion): Likewise.
24656 (vectorizable_assignment): Likewise.
24657 (vectorizable_shift): Likewise.
24658 (vectorizable_operation): Likewise.
24659 (vectorizable_comparison): Likewise.
24660 (vect_is_simple_cond): Record the def types for operands.
24661 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24662 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24663 for statement argument count.
24665 2017-05-16 Carl Love <cel@us.ibm.com>
24667 * config/rs6000/rs6000-c: Add support for built-in functions
24668 vector unsigned long long vec_bperm (vector unsigned long long,
24669 vector unsigned char)
24670 vector signed long long vec_mule (vector signed int,
24672 vector unsigned long long vec_mule (vector unsigned int,
24673 vector unsigned int)
24674 vector signed long long vec_mulo (vector signed int,
24676 vector unsigned long long vec_mulo (vector unsigned int,
24677 vector unsigned int)
24678 vector signed char vec_sldw (vector signed char,
24679 vector signed char,
24681 vector unsigned char vec_sldw (vector unsigned char,
24682 vector unsigned char,
24684 vector signed short vec_sldw (vector signed short,
24685 vector signed short,
24687 vector unsigned short vec_sldw (vector unsigned short,
24688 vector unsigned short,
24690 vector signed int vec_sldw (vector signed int,
24693 vector unsigned int vec_sldw (vector unsigned int,
24694 vector unsigned int,
24696 vector signed long long vec_sldw (vector signed long long,
24697 vector signed long long,
24699 vector unsigned long long vec_sldw (vector unsigned long long,
24700 vector unsigned long long,
24702 * config/rs6000/rs6000-c: Add support for built-in functions
24703 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24704 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24705 * doc/extend.texi: Update the built-in documentation for the
24706 new built-in functions.
24708 2017-05-16 Marek Polacek <polacek@redhat.com>
24712 * tree.c (save_expr): Don't fold the expression.
24714 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24716 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24717 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24718 and (?*y,m). Update insn attributes.
24720 2017-05-16 Martin Liska <mliska@suse.cz>
24722 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24723 flags argument of print_gimple_stmt, print_gimple_expr,
24724 print_generic_stmt and print_generic_expr.
24725 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24726 * coretypes.h: Likewise.
24727 * except.c (dump_eh_tree): Likewise.
24728 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24729 * gimple-pretty-print.h: Likewise.
24730 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24731 (backprop::push_to_worklist): Likewise.
24732 (backprop::pop_from_worklist): Likewise.
24733 (backprop::process_use): Likewise.
24734 (backprop::intersect_uses): Likewise.
24735 (note_replacement): Likewise.
24736 * gimple-ssa-store-merging.c
24737 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24738 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24739 (pass_store_merging::execute): Likewise.
24740 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24741 (ssa_base_cand_dump_callback): Likewise.
24742 (dump_incr_vec): Likewise.
24743 (replace_refs): Likewise.
24744 (replace_mult_candidate): Likewise.
24745 (create_add_on_incoming_edge): Likewise.
24746 (create_phi_basis): Likewise.
24747 (insert_initializers): Likewise.
24748 (all_phi_incrs_profitable): Likewise.
24749 (introduce_cast_before_cand): Likewise.
24750 (replace_one_candidate): Likewise.
24751 * gimplify.c (gimplify_expr): Likewise.
24752 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24753 (set_rename): Likewise.
24754 (rename_uses): Likewise.
24755 (copy_loop_phi_nodes): Likewise.
24756 (add_close_phis_to_merge_points): Likewise.
24757 (copy_loop_close_phi_args): Likewise.
24758 (copy_cond_phi_args): Likewise.
24759 (graphite_copy_stmts_from_block): Likewise.
24760 (translate_pending_phi_nodes): Likewise.
24761 * graphite-poly.c (print_pdr): Likewise.
24762 (dump_gbb_cases): Likewise.
24763 (dump_gbb_conditions): Likewise.
24764 (print_scop_params): Likewise.
24765 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24766 (build_cross_bb_scalars_use): Likewise.
24767 (gather_bbs::before_dom_children): Likewise.
24768 * hsa-dump.c (dump_hsa_immed): Likewise.
24769 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24770 (get_replacement_map): Likewise.
24771 * ipa-inline-analysis.c (dump_condition): Likewise.
24772 (estimate_function_body_sizes): Likewise.
24773 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24774 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24775 * ipa-prop.c (ipa_dump_param): Likewise.
24776 (ipa_print_node_jump_functions_for_edge): Likewise.
24777 (ipa_modify_call_arguments): Likewise.
24778 (ipa_modify_expr): Likewise.
24779 (ipa_dump_param_adjustments): Likewise.
24780 (ipa_dump_agg_replacement_values): Likewise.
24781 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24782 * ipa-pure-const.c (check_stmt): Likewise.
24783 (pass_nothrow::execute): Likewise.
24784 * ipa-split.c (execute_split_functions): Likewise.
24785 * omp-offload.c (dump_oacc_loop_part): Likewise.
24786 (dump_oacc_loop): Likewise.
24787 * trans-mem.c (tm_log_emit): Likewise.
24788 (tm_memopt_accumulate_memops): Likewise.
24789 (dump_tm_memopt_set): Likewise.
24790 (dump_tm_memopt_transform): Likewise.
24791 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24792 (print_loop): Likewise.
24793 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24794 (chkp_gather_checks_info): Likewise.
24795 (chkp_get_check_result): Likewise.
24796 (chkp_remove_check_if_pass): Likewise.
24797 (chkp_use_outer_bounds_if_possible): Likewise.
24798 (chkp_reduce_bounds_lifetime): Likewise.
24799 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24800 (chkp_mark_completed_bounds): Likewise.
24801 (chkp_register_incomplete_bounds): Likewise.
24802 (chkp_mark_invalid_bounds): Likewise.
24803 (chkp_maybe_copy_and_register_bounds): Likewise.
24804 (chkp_build_returned_bound): Likewise.
24805 (chkp_get_bound_for_parm): Likewise.
24806 (chkp_build_bndldx): Likewise.
24807 (chkp_get_bounds_by_definition): Likewise.
24808 (chkp_generate_extern_var_bounds): Likewise.
24809 (chkp_get_bounds_for_decl_addr): Likewise.
24810 * tree-chrec.c (chrec_apply): Likewise.
24811 * tree-data-ref.c (dump_data_reference): Likewise.
24812 (dump_subscript): Likewise.
24813 (dump_data_dependence_relation): Likewise.
24814 (analyze_overlapping_iterations): Likewise.
24815 * tree-inline.c (expand_call_inline): Likewise.
24816 (tree_function_versioning): Likewise.
24817 * tree-into-ssa.c (dump_defs_stack): Likewise.
24818 (dump_currdefs): Likewise.
24819 (dump_names_replaced_by): Likewise.
24820 (dump_update_ssa): Likewise.
24821 (update_ssa): Likewise.
24822 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24823 * tree-parloops.c (build_new_reduction): Likewise.
24824 (try_create_reduction_list): Likewise.
24825 (ref_conflicts_with_region): Likewise.
24826 (oacc_entry_exit_ok_1): Likewise.
24827 (oacc_entry_exit_single_gang): Likewise.
24828 * tree-pretty-print.h: Likewise.
24829 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24830 (get_scalar_evolution): Likewise.
24831 (add_to_evolution): Likewise.
24832 (get_loop_exit_condition): Likewise.
24833 (analyze_evolution_in_loop): Likewise.
24834 (analyze_initial_condition): Likewise.
24835 (analyze_scalar_evolution): Likewise.
24836 (instantiate_scev): Likewise.
24837 (number_of_latch_executions): Likewise.
24838 (gather_chrec_stats): Likewise.
24839 (final_value_replacement_loop): Likewise.
24840 (scev_const_prop): Likewise.
24841 * tree-sra.c (dump_access): Likewise.
24842 (disqualify_candidate): Likewise.
24843 (create_access): Likewise.
24844 (reject): Likewise.
24845 (maybe_add_sra_candidate): Likewise.
24846 (create_access_replacement): Likewise.
24847 (analyze_access_subtree): Likewise.
24848 (analyze_all_variable_accesses): Likewise.
24849 (sra_modify_assign): Likewise.
24850 (initialize_constant_pool_replacements): Likewise.
24851 (find_param_candidates): Likewise.
24852 (decide_one_param_reduction): Likewise.
24853 (replace_removed_params_ssa_names): Likewise.
24854 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24855 * tree-ssa-copy.c (dump_copy_of): Likewise.
24856 (copy_prop_visit_cond_stmt): Likewise.
24857 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24858 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24859 (record_equivalences_from_stmt): Likewise.
24860 * tree-ssa-dse.c (compute_trims): Likewise.
24861 (delete_dead_call): Likewise.
24862 (delete_dead_assignment): Likewise.
24863 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24864 (forward_propagate_into_cond): Likewise.
24865 (pass_forwprop::execute): Likewise.
24866 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24867 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24869 (move_computations_worker): Likewise.
24870 (execute_sm): Likewise.
24871 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24872 (remove_exits_and_undefined_stmts): Likewise.
24873 (remove_redundant_iv_tests): Likewise.
24874 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24875 (adjust_iv_update_pos): Likewise.
24876 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24877 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24878 (value_replacement): Likewise.
24879 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24880 * tree-ssa-pre.c (print_pre_expr): Likewise.
24881 (get_representative_for): Likewise.
24882 (create_expression_by_pieces): Likewise.
24883 (insert_into_preds_of_block): Likewise.
24884 (eliminate_insert): Likewise.
24885 (eliminate_dom_walker::before_dom_children): Likewise.
24886 (eliminate): Likewise.
24887 (remove_dead_inserted_code): Likewise.
24888 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24889 * tree-ssa-reassoc.c (get_rank): Likewise.
24890 (eliminate_duplicate_pair): Likewise.
24891 (eliminate_plus_minus_pair): Likewise.
24892 (eliminate_not_pairs): Likewise.
24893 (undistribute_ops_list): Likewise.
24894 (eliminate_redundant_comparison): Likewise.
24895 (update_range_test): Likewise.
24896 (optimize_range_tests_var_bound): Likewise.
24897 (optimize_vec_cond_expr): Likewise.
24898 (rewrite_expr_tree): Likewise.
24899 (rewrite_expr_tree_parallel): Likewise.
24900 (linearize_expr): Likewise.
24901 (break_up_subtract): Likewise.
24902 (linearize_expr_tree): Likewise.
24903 (attempt_builtin_powi): Likewise.
24904 (attempt_builtin_copysign): Likewise.
24905 (transform_stmt_to_copy): Likewise.
24906 (transform_stmt_to_multiply): Likewise.
24907 (dump_ops_vector): Likewise.
24908 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24909 (print_scc): Likewise.
24910 (set_ssa_val_to): Likewise.
24911 (visit_reference_op_store): Likewise.
24912 (visit_use): Likewise.
24913 (sccvn_dom_walker::before_dom_children): Likewise.
24914 (run_scc_vn): Likewise.
24915 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24917 (expr_hash_elt::print): Likewise.
24918 (const_and_copies::pop_to_marker): Likewise.
24919 (const_and_copies::record_const_or_copy_raw): Likewise.
24920 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24921 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24922 (dump_predicates): Likewise.
24923 (find_uninit_use): Likewise.
24924 (warn_uninitialized_phi): Likewise.
24925 (pass_late_warn_uninitialized::execute): Likewise.
24926 * tree-ssa.c (verify_vssa): Likewise.
24927 (verify_ssa): Likewise.
24928 (maybe_optimize_var): Likewise.
24929 * tree-vrp.c (dump_value_range): Likewise.
24930 (dump_all_value_ranges): Likewise.
24931 (dump_asserts_for): Likewise.
24932 (register_edge_assert_for_2): Likewise.
24933 (vrp_visit_cond_stmt): Likewise.
24934 (vrp_visit_switch_stmt): Likewise.
24935 (vrp_visit_stmt): Likewise.
24936 (vrp_visit_phi_node): Likewise.
24937 (simplify_cond_using_ranges_1): Likewise.
24938 (fold_predicate_in): Likewise.
24939 (evrp_dom_walker::before_dom_children): Likewise.
24940 (evrp_dom_walker::push_value_range): Likewise.
24941 (evrp_dom_walker::pop_value_range): Likewise.
24942 (execute_early_vrp): Likewise.
24944 2017-05-16 Richard Biener <rguenther@suse.de>
24946 * dwarf2out.c (loc_list_from_tree_1): Do not create
24947 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24949 2017-05-16 Richard Biener <rguenther@suse.de>
24951 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24953 (note_variable_value_in_expr): If we resolved the decl ref
24954 do not push to the stack.
24956 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24958 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24959 operations in fast-math mode.
24960 (vaddq_f16): Likewise.
24961 (vmul_f16): Likewise.
24962 (vmulq_f16): Likewise.
24963 (vsub_f16): Likewise.
24964 (vsubq_f16): Likewise.
24965 * config/arm/neon.md (add<mode>3): New.
24967 (fma:<VH:mode>3): New. Also remove outdated comment.
24970 2017-05-16 Martin Liska <mliska@suse.cz>
24974 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24975 (odr_types_equivalent_p): Likewise.
24977 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24979 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24981 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24984 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24985 non-interunit SSE move alternatives with '?'.
24986 (zero-extendsidi peephole2): New peephole to skip intermediate
24987 general register in SSE zero-extend sequence.
24989 2017-05-15 Jeff Law <law@redhat.com>
24991 * reorg.c (relax_delay_slots): Create a new variable to hold
24992 the temporary target rather than clobbering TARGET_LABEL.
24994 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24995 missing argument to extract_bit_field call.
24996 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24998 2017-05-15 Martin Liska <mliska@suse.cz>
25001 * gcc.c (process_command): Do not allow empty argument of -o option.
25003 2017-05-15 Renlin Li <renlin.li@arm.com>
25005 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
25006 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
25007 * config/aarch64/constraints.md (Usf): Add long call check.
25008 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
25009 (call_value): Likewise.
25010 (sibcall): Likewise.
25011 (sibcall_value): Likewise.
25013 (call_value_insn): New.
25014 (sibcall_insn): Update rtx pattern.
25015 (sibcall_value_insn): Likewise.
25016 (call_internal): Remove.
25017 (call_value_internal): Likewise.
25018 (sibcall_internal): Likewise.
25019 (sibcall_value_internal): Likewise.
25020 (call_reg): Likewise.
25021 (call_symbol): Likewise.
25022 (call_value_reg): Likewise.
25023 (call_value_symbol): Likewise.
25025 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
25028 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
25030 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
25032 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
25033 compatible with CCGOCmode and with CCZmode.
25035 2017-05-14 Martin Sebor <msebor@redhat.com>
25037 PR middle-end/77671
25038 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25039 (gimple_fold_builtin_snprintf): Same.
25040 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25041 (gimple_fold_builtin_snprintf): Same.
25042 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25043 of character types.
25044 (is_call_safe): New function.
25045 (try_substitute_return_value): Call it.
25046 (try_simplify_call): New function.
25047 (pass_sprintf_length::handle_gimple_call): Call it.
25049 2017-05-14 Martin Sebor <msebor@redhat.com>
25051 PR middle-end/80669
25052 * builtins.c (expand_builtin_stpncpy): Simplify.
25054 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25056 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25057 * config/i386/i386.h
25058 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25059 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25060 (struct machine_function): Add new members call_ms2sysv,
25061 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25062 (struct machine_frame_state): New fields sp_realigned and
25063 sp_realigned_offset.
25064 * config/i386/i386.c
25065 (enum xlogue_stub): New enum.
25066 (enum xlogue_stub_sets): New enum.
25067 (class xlogue_layout): New class.
25068 (struct ix86_frame): New fields stack_realign_allocate_offset,
25069 stack_realign_offset and outlined_save_offset. Modify comments to
25070 detail stack layout when using out-of-line stubs.
25071 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25072 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25073 -mcall-ms2sysv-xlogues.
25074 (stub_managed_regs): New static variable.
25075 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25076 registers managed by out-of-line stub.
25077 (disable_call_ms2sysv_xlogues): New function.
25078 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25079 m->call_ms2sysv when appropriate and compute frame layout for
25081 (sp_valid_at, fp_valid_at): New inline functions.
25082 (choose_basereg): New function.
25083 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25085 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25086 Use align parameter of choose_baseaddr to generated aligned SSE movs
25088 (pro_epilogue_adjust_stack): Modify to track
25089 machine_frame_state::sp_realigned.
25090 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25091 (ix86_nsaved_sseregs): Likewise.
25092 (ix86_emit_save_regs): Likewise.
25093 (ix86_emit_save_regs_using_mov): Likewise.
25094 (ix86_emit_save_sse_regs_using_mov): Likewise.
25095 (get_scratch_register_on_entry): Likewise.
25096 (gen_frame_set): New function.
25097 (gen_frame_load): Likewise.
25098 (gen_frame_store): Likewise.
25099 (emit_outlined_ms2sysv_save): Likewise.
25100 (emit_outlined_ms2sysv_restore): Likewise.
25101 (ix86_expand_prologue): Modify stack re-alignment code and call
25102 emit_outlined_ms2sysv_save when appropriate.
25103 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25104 parameter rtx_insn *insn, which allows the function to be used to only
25105 generate the notes.
25106 (ix86_expand_epilogue): Modify validity checks of frame and stack
25107 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25108 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25109 * config/i386/predicates.md
25110 (save_multiple): New predicate.
25111 (restore_multiple): Likewise.
25112 * config/i386/sse.md
25113 (save_multiple<mode>): New pattern.
25114 (save_multiple_realign<mode>): Likewise.
25115 (restore_multiple<mode>): Likewise.
25116 (restore_multiple_and_return<mode>): Likewise.
25117 (restore_multiple_leave_return<mode>): Likewise.
25118 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25120 2017-05-14 Julia Koval <julia.koval@intel.com>
25122 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25123 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25124 (__builtin_ia32_xsetbv): New builtins.
25125 * config/i386/i386.c (ix86_expand_special_args_builtin):
25127 (ix86_expand_builtin): Special expand for new intrinsics.
25128 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25129 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25130 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25132 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25134 * cfganal.c (inverted_post_order_compute): Change argument type
25136 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25137 * df-core.c (rest_of_handle_df_initialize): Adjust.
25138 (rest_of_handle_df_finish): Likewise.
25139 (df_analyze_1): Likewise.
25140 (df_analyze): Likewise.
25141 (loop_inverted_post_order_compute): Change argument to be a vec *.
25142 (df_analyze_loop): Adjust.
25143 (df_get_n_blocks): Likewise.
25144 (df_get_postorder): Likewise.
25145 * df.h (struct df_d): Change field to be a vec.
25146 * lcm.c (compute_laterin): Adjust.
25147 (compute_available): Likewise.
25148 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25149 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25150 * tree-ssa-pre.c (compute_antic): Likewise.
25152 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25154 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25155 (depth_first_search::depth_first_search): Change structure init
25156 function to this constructor.
25157 (depth_first_search::add_bb): Rename function to this member.
25158 (depth_first_search::execute): Likewise.
25159 (flow_dfs_compute_reverse_finish): Adjust.
25161 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25163 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25164 (longest_simple_path): Likewise.
25165 * shrink-wrap.c (spread_components): Likewise.
25166 (disqualify_problematic_components): Likewise.
25167 (emit_common_heads_for_components): Likewise.
25168 (emit_common_tails_for_components): Likewise.
25169 (insert_prologue_epilogue_for_components): Likewise.
25171 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25173 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25176 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25178 * df-core.c (df_set_blocks): Start using auto_bitmap.
25179 (df_compact_blocks): Likewise.
25180 * df-problems.c (df_rd_confluence_n): Likewise.
25181 * df-scan.c (df_insn_rescan_all): Likewise.
25182 (df_process_deferred_rescans): Likewise.
25183 (df_update_entry_block_defs): Likewise.
25184 (df_update_exit_block_uses): Likewise.
25185 (df_entry_block_bitmap_verify): Likewise.
25186 (df_exit_block_bitmap_verify): Likewise.
25187 (df_scan_verify): Likewise.
25188 * lra-constraints.c (lra_constraints): Likewise.
25189 (undo_optional_reloads): Likewise.
25190 (lra_undo_inheritance): Likewise.
25191 * lra-remat.c (calculate_gen_cands): Likewise.
25192 (do_remat): Likewise.
25193 * lra-spills.c (assign_spill_hard_regs): Likewise.
25194 (spill_pseudos): Likewise.
25195 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25196 (bitmap_set_subtract_values): Likewise.
25198 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25200 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25201 management with auto_bitmap.
25202 (fix_inter_tick): Likewise.
25203 (fix_recovery_deps): Likewise.
25204 * ira.c (add_store_equivs): Likewise.
25205 (find_moveable_pseudos): Likewise.
25206 (split_live_ranges_for_shrink_wrap): Likewise.
25207 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25208 (rtx_reuse_manager::seen_def_p): Likewise.
25209 (rtx_reuse_manager::set_seen_def): Likewise.
25210 * print-rtl.h (class rtx_reuse_manager): Likewise.
25212 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25214 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25216 (migrate_btr_def): Likewise.
25217 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25218 * df-core.c (loop_post_order_compute): Likewise.
25219 (loop_inverted_post_order_compute): Likewise.
25220 * hsa-common.h: Likewise.
25221 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25222 * init-regs.c (initialize_uninitialized_regs): Likewise.
25223 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25224 (inline_small_functions): Likewise.
25225 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25226 * ira.c (combine_and_move_insns): Likewise.
25227 (build_insn_chain): Likewise.
25228 * loop-invariant.c (find_invariants): Likewise.
25229 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25230 * predict.c (tree_predict_by_opcode): Likewise.
25231 (predict_paths_leading_to): Likewise.
25232 (predict_paths_leading_to_edge): Likewise.
25233 (estimate_loops_at_level): Likewise.
25234 (estimate_loops): Likewise.
25235 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25236 (spread_components): Likewise.
25237 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25238 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25239 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25240 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25241 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25242 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25243 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25244 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25245 (mark_threaded_blocks): Likewise.
25246 (thread_through_all_blocks): Likewise.
25247 * tree-ssa.c (verify_ssa): Likewise.
25248 (execute_update_addresses_taken): Likewise.
25249 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25251 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25253 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25255 (post_order_compute): Likewise.
25256 (inverted_post_order_compute): Likewise.
25257 (pre_and_rev_post_order_compute_fn): Likewise.
25259 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25261 * genrecog.c (int_set::int_set): Explicitly construct our
25262 auto_vec base class.
25263 * vec.h (auto_vec::auto_vec): New constructor.
25265 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25267 * bitmap.h (class auto_bitmap): New constructor taking
25268 bitmap_obstack * argument.
25270 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25272 * bitmap.h (class auto_bitmap): Change type of m_bits to
25273 bitmap_head, and adjust ctor / dtor and member operators.
25275 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25277 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25278 when returned register mode doesn't match original mode.
25280 2017-05-12 Jeff Law <law@redhat.com>
25281 Jakub Jelinek <jakub@redhat.com>
25283 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25284 we look for cc setter after the compare-elim changes.
25285 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25286 within the vector to match what compare-elim now expects.
25287 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25288 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25290 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25291 after the compare-elim changes.
25292 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25293 the vector to match what compare-elim now expects.
25294 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25295 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25296 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25297 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25298 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25300 * config/visium/visium.c (single_set_and_flags): Fix where
25301 we look for cc setter after the compare-elim changes.
25302 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25303 with the vector to match what compare-elim now expects.
25304 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25305 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25306 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25307 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25308 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25309 (neg<mode>2_insn_set_overflow): Likewise.
25311 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25313 PR middle-end/79794
25314 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25315 maybe_expand_insn call, set ops[0].target. If still set after call,
25316 set alt_rtl. Add extra arg to recursive calls.
25317 (extract_bit_field): Add alt_rtl argument. Pass to
25319 * expmed.h (extract_bit_field): Fix prototype.
25320 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25321 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25322 to extract_bit_field_calls.
25323 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25324 Pass alt_rtl to extract_bit_field calls.
25325 * calls.c (store_unaligned_arguments_into_psuedos)
25326 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25327 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25329 * optabs.h (struct expand_operand): Add target bitfield.
25331 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25333 * compare-elim.c (try_eliminate_compare): Canonicalize
25334 operation with embedded compare to
25335 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25336 (set (reg) (operation)].
25338 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25340 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25343 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25344 cost of adding a carry flag for ADC instruction.
25345 [case MINUS]: Ignore the cost of subtracting a carry flag
25346 for SBB instruction.
25348 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25350 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25352 * config/rs6000/bmiintrin.h: New file.
25353 * config/rs6000/bmi2intrin.h: New file.
25354 * config/rs6000/x86intrin.h: New file.
25356 2017-05-12 Jeff Law <law@redhat.com>
25358 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25361 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25363 PR middle-end/80707
25364 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25366 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25369 early expansion of vector divide builtins.
25370 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25371 builtins identified as having unsigned arguments.
25373 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25375 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25376 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25377 expansion of vector logical operations (and, andc, or, xor,
25380 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25382 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25383 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25385 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25387 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25388 early GIMPLE expansion of vector multiplies.
25390 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25392 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25393 TARGET_HAVE_MOVT conditional.
25394 (movt splitter): Likewise.
25396 2017-05-12 Richard Biener <rguenther@suse.de>
25398 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25399 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25400 Fold all stmts not inplace.
25402 2017-05-12 Richard Biener <rguenther@suse.de>
25404 PR tree-optimization/80713
25405 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25406 inserted_exprs bit for not removed stmts.
25408 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25410 PR middle-end/69921
25411 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25412 parallelized" attribute for parallelized OpenACC kernels.
25413 * omp-offload.c (execute_oacc_device_lower): Use it.
25415 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25416 Set "oacc kernels" attribute.
25417 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25418 parameter. Adjust all users.
25419 (oacc_fn_attrib_kernels_p): Remove function.
25420 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25421 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25422 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25423 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25424 assert "oacc kernels" attribute is set.
25426 2017-05-11 Carl Love <cel@us.ibm.com>
25428 * config/rs6000/rs6000-c: Add support for built-in functions
25429 vector unsigned char vec_popcnt (vector signed char)
25430 vector unsigned char vec_popcnt (vector unsigned char)
25431 vector unsigned short vec_popcnt (vector signed short)
25432 vector unsigned short vec_popcnt (vector unsigned short)
25433 vector unsigned int vec_popcnt (vector signed int)
25434 vector unsigned int vec_popcnt (vector unsigned int)
25435 vector unsigned long long vec_popcnt (vector signed long long)
25436 vector unsigned long long vec_popcnt (vector unsigned long long)
25437 vector signed long long vec_slo (vector signed long long,
25438 vector signed char)
25439 vector signed long long vec_slo (vector signed long long,
25440 vector unsigned char)
25441 vector unsigned long long vec_slo (vector unsigned long long,
25442 vector signed char)
25443 vector unsigned long long vec_slo (vector unsigned long long,
25444 vector unsigned char)
25445 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25446 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25447 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25448 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25449 * doc/extend.texi: Update the built-in documentation file for the
25450 new built-in functions.
25452 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25454 * attribs.h (sorted_attr_string): Move machine independent
25455 functions for target clone support from the i386 port to common
25456 code. Rename ix86_function_versions to common_function_versions.
25457 Rename make_name to make_unique_name.
25458 (common_function_versions): Likewise.
25459 (make_unique_name): Likewise.
25460 (make_dispatcher_decl): Likewise.
25461 (is_function_default_version): Likewise.
25462 * attribs.c (attr_strcmp): Likewise.
25463 (sorted_attr_string): Likewise.
25464 (common_function_versions): Likewise.
25465 (make_unique_name): Likewise.
25466 (make_dispatcher_decl): Likewise.
25467 (is_function_default_version): Likewise.
25468 * config/i386/i386.c (attr_strcmp): Likewise.
25469 (sorted_attr_string): Likewise.
25470 (ix86_function_versions): Likewise.
25471 (make_name): Likewise.
25472 (make_dispatcher_decl): Likewise.
25473 (is_function_default_version): Likewise.
25474 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25476 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25479 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25480 Account for direct move costs for vec_construct of integer
25483 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25486 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25487 (UNSPEC_STX_ATOMIC): Ditto.
25488 (loaddi_via_sse): New insn.
25489 (storedi_via_sse): Ditto.
25490 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25491 Update corresponding peephole2 patterns.
25492 (atomic_storedi_fpu): Ditto.
25494 2017-05-11 Julia Koval <julia.koval@intel.com>
25496 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25497 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25499 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25500 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25501 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25503 2017-05-11 Nathan Sidwell <nathan@acm.org>
25505 * graphite-poly.c: Include dumpfile.h.
25507 * dumpfle.h (dump_function): Declare here ...
25508 * tree-dump.h (dump_function): ... not here.
25509 * dumpfile.c: #include tree-cfg.h.
25510 (dump_function): Move here from ...
25511 * tree-dump.c (dump_function): ... here.
25512 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25513 * graphite-poly.c: Don't include tree-dump.h.
25514 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25515 * print-tree.c: Likewise.
25516 * stor-layout.c: Likewise.
25517 * tree-nested.c: Likewise.
25519 * dumpfile.c (dump_start): Use TDF_FLAGS.
25520 (dump_enable_all): Fix TDF_KIND check thinko.
25522 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25524 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25525 array entries to represent two legal parameterizations of the
25526 overloaded __builtin_cmpb function, as represented by the
25527 P6_OV_BUILTIN_CMPB constant.
25528 (altivec_resolve_overloaded_builtin): Add special case handling
25529 for the __builtin_cmpb function, as represented by the
25530 P6_OV_BUILTIN_CMPB constant.
25531 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25532 (BU_P6_64BIT_2): New macro.
25533 (BU_P6_OVERLOAD_2): New macro
25534 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25535 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25536 (CMPB): Add overload support to represent both 32-bit and 64-bit
25537 compare-bytes function.
25538 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25539 support for TARGET_CMPB.
25540 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25541 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25542 documentation of the __builtin_cmpb overloaded built-in function.
25544 2017-05-11 Richard Biener <rguenther@suse.de>
25546 PR tree-optimization/80705
25547 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25548 bases are not vectorizable.
25550 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25552 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25553 when counting register pressure.
25555 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25557 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25558 (iv_ca_more_deps): Renamed to ...
25559 (iv_ca_compare_deps): ... this.
25560 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25564 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25566 (determine_group_iv_costs): ... here.
25567 (find_inv_vars_cb): Record inv var if it's not recorded before.
25569 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25571 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25572 (get_shiftadd_cost): Ditto.
25574 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25576 * tree-ssa-address.c: Include header file.
25577 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25579 (add_to_parts): Refactor.
25580 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25581 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25584 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25586 PR tree-optimization/53090
25587 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25589 (extract_cond_operands): Detect condition with IV on both sides
25590 and return COMP_IV_EXPR_2.
25591 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25592 (rewrite_use_compare): Simplify by removing call to function
25593 extract_cond_operands.
25595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25597 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25598 (extract_cond_operands): Detect condition comparing against non-
25599 invariant bound and return appropriate enum value.
25600 (find_interesting_uses_cond): Update use of extract_cond_operands.
25601 Handle its return value accordingly.
25602 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25604 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25606 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25607 nonlinear iv_use computation in loop invariant sensitive way.
25609 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25611 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25612 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25614 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25616 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25617 (dump_cand): Support iv_cand.inv_exprs.
25618 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25620 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25623 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25625 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25627 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25628 as local function. Include necessary header files.
25629 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25633 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25635 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25637 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25638 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25639 RSHIFT_EXPR and BIT_NOT_EXPR.
25641 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25643 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25644 (adjust_setup_cost): New parameter supporting round up adjustment.
25645 (struct address_cost_data): Delete.
25646 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25647 (split_address_cost, ptr_difference_cost): Delete.
25648 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25649 (struct ainc_cost_data): New struct.
25650 (get_address_cost_ainc): New function.
25651 (get_address_cost, get_computation_cost): Reimplement.
25652 (determine_group_iv_cost_address): Record inv_expr for all uses of
25654 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25655 (iv_ca_has_deps): Reimplemented to ...
25656 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25658 (iv_ca_extend): Call iv_ca_more_deps.
25660 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25662 * tree-ssa-address.c (struct mem_address): Move to header file.
25663 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25664 * tree-ssa-address.h (struct mem_address): Move from C file.
25665 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25669 * tree-affine.h (aff_combination_type): New interface.
25670 (aff_combination_zero_p): Remove static.
25671 (aff_combination_const_p): New interface.
25672 (aff_combination_singleton_var_p): New interfaces.
25674 2017-05-11 Richard Biener <rguenther@suse.de>
25676 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25677 Skip unreachable blocks and destinations.
25678 (eliminate): Move stmt removal and fixup ...
25679 (fini_eliminate): ... here. Skip inserted exprs.
25680 (pass_pre::execute): Move fini_pre after fini_eliminate.
25681 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25682 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25683 PRE to get rid of dead code that has invalid SSA form and
25684 split critical edges again.
25686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25688 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25690 2017-05-11 Richard Biener <rguenther@suse.de>
25692 * passes.c (execute_function_todo): Verify loops if they are
25693 said to be up-to-date.
25694 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25695 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25697 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25700 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25701 handle calling assemble_external ourself.
25704 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25705 modes with zero size. Enhance comment.
25707 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25709 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25710 built-ins for vec_xl and vec_xst with short and char pointer
25713 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25715 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25716 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25717 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25718 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25719 (_mm_maskz_min_round_ss): New intrinsics.
25720 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25721 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25722 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25723 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25724 (__builtin_ia32_minss_mask_round): New builtins.
25725 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25726 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25727 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25729 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25730 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25732 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25735 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25737 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25738 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25739 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25740 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25741 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25742 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25743 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25744 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25745 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25746 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25747 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25748 (__builtin_ia32_mulss_mask_round): New builtins.
25749 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25750 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25751 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25753 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25754 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25756 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25759 2017-05-10 Julia Koval <julia.koval@intel.com>
25761 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25762 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25763 (_mm256_setr_m128i): New intrinsics.
25765 2017-05-10 Julia Koval <julia.koval@intel.com>
25767 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25768 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25769 (_mm_maskz_rcp14_ss): New intrinsics.
25770 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25771 (__builtin_ia32_rcp14ss_mask): New builtins.
25772 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25774 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25776 PR tree-optimization/51513
25777 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25778 (assert_unreachable_fallthru_edge_p): Use it.
25779 (group_case_labels_stmt): Likewise.
25780 * tree-cfg.h: Prototype it.
25781 * stmt.c: Include cfghooks.h and tree-cfg.h.
25782 (emit_case_dispatch_table) <gap_label>: New local variable.
25783 Use it to fill dispatch table gaps.
25784 Test for default_label before updating probabilities.
25785 (expand_case) <default_label>: Remove unneeded initialization.
25786 Test for unreachable default case statement and remove its edge.
25787 Set default_label accordingly.
25788 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25790 2017-05-10 Carl Love <cel@us.ibm.com>
25792 * config/rs6000/rs6000-c: Add support for built-in functions
25793 vector signed char vec_neg (vector signed char)
25794 vector signed short int vec_neg (vector short int)
25795 vector signed int vec_neg (vector signed int)
25796 vector signed long long vec_neg (vector signed long long)
25797 vector float vec_neg (vector float)
25798 vector double vec_neg (vector double)
25799 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25801 * config/rs6000/altivec.h: Add define for vec_neg
25802 * doc/extend.texi: Update the built-in documentation for the
25803 new built-in functions.
25805 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25807 PR tree-optimization/77644
25808 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25810 2017-05-10 Nathan Sidwell <nathan@acm.org>
25812 * dumpfile.h (TDI_lang_all): New.
25813 (TDF_KIND): New. Renumber others
25814 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25816 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25818 (get_dump_file_name): Adjust suffix generation.
25819 (dump_enable_all): Use TDF_KIND.
25820 * doc/invoke.texi (-fdump-lang-all): Document.
25822 * dumpfile.h: Tabify.
25824 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25827 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25828 Move member access before delete.
25830 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25832 * tree-inline.c (expand_call_inline): Split block at stmt
25835 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25838 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25839 are now unused after splitting mov{sf,sd}_hardfloat.
25840 (f32_lr2): Likewise.
25841 (f32_lm): Likewise.
25842 (f32_lm2): Likewise.
25843 (f32_li): Likewise.
25844 (f32_li2): Likewise.
25845 (f32_lv): Likewise.
25846 (f32_sr): Likewise.
25847 (f32_sr2): Likewise.
25848 (f32_sm): Likewise.
25849 (f32_sm2): Likewise.
25850 (f32_si): Likewise.
25851 (f32_si2): Likewise.
25852 (f32_sv): Likewise.
25853 (f32_dm): Likewise.
25854 (f32_vsx): Likewise.
25855 (f32_av): Likewise.
25856 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25857 For movsf, order stores so the VSX stores occur before the GPR
25858 store which encourages the register allocator to use a traditional
25859 FPR instead of a GPR. For movsd, order the stores so that the GPR
25860 store comes before the VSX stores to allow the power6 to work.
25861 This is due to the power6 not having a 32-bit integer store
25862 instruction from a FPR.
25863 (movsf_hardfloat): Likewise.
25864 (movsd_hardfloat): Likewise.
25866 2017-05-09 Martin Sebor <msebor@redhat.com>
25868 PR translation/80280
25869 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25872 PR translation/80280
25873 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25874 data member added in r247778.
25875 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25877 2017-05-09 Nathan Sidwell <nathan@acm.org>
25879 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25881 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25884 2017-05-09 Marek Polacek <polacek@redhat.com>
25886 * doc/invoke.texi: Fix typo.
25888 2017-05-09 Richard Biener <rguenther@suse.de>
25890 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25891 (vrp_val_is_min): Likewise.
25892 (set_value_range_to_value): Likewise.
25893 (set_value_range_to_nonnegative): Likewise.
25894 (gimple_assign_nonzero_p): Likewise.
25895 (gimple_stmt_nonzero_p): Likewise.
25896 (vrp_int_const_binop): Likewise. Remove unreachable case.
25897 (adjust_range_with_scev): Adjust comments.
25898 (compare_range_with_value): Likewise.
25899 (extract_range_from_phi_node): Likewise.
25900 (test_for_singularity): Likewise.
25902 2017-05-09 Richard Biener <rguenther@suse.de>
25904 * tree-vrp.c (get_single_symbol): Add assert that we don't
25905 get overflowed constants as invariant part.
25906 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25907 checks. Use wi::cmp instead of recursing for integer constants.
25908 (compare_values): Just ignore whether we assumed undefined
25909 overflow instead of failing the compare.
25910 (extract_range_for_var_from_comparison_expr): Add comment before the
25911 TREE_NO_WARNING sets.
25912 (test_for_singularity): Likewise.
25913 (extract_range_from_comparison): Do not disable optimization
25914 when we assumed undefined overflow.
25915 (extract_range_basic): Remove init of unused var.
25917 2017-05-09 Richard Biener <rguenther@suse.de>
25919 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25920 (extract_range_from_multiplicative_op_1): Adjust.
25921 (extract_range_from_binary_expr_1): Use int_const_binop.
25923 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25926 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25927 rs6000_store_data_bypass_p in seven define_bypass directives and
25928 in several comments.
25929 * config/rs6000/rs6000-protos.h: Add prototype for
25930 rs6000_store_data_bypass_p function.
25931 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25932 function implements slightly different (rs6000-specific) semantics
25933 than store_data_bypass_p, returning false rather than aborting
25934 with assertion error when arguments do not satisfy the
25935 requirements of store data bypass.
25936 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25937 rs6000_store_data_bypass_p.
25939 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25941 * config/xtensa/xtensa-protos.h
25942 (xtensa_initial_elimination_offset): New declaration.
25943 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25944 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25945 macro definition, add case for FRAME_POINTER_REGNUM when
25946 FRAME_GROWS_DOWNWARD.
25947 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25948 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25949 xtensa_initial_elimination_offset.
25951 2017-05-08 Nathan Sidwell <nathan@acm.org>
25953 * doc/invoke.texi: Alphabetize -fdump options.
25955 2017-05-08 Martin Sebor <msebor@redhat.com>
25957 PR translation/80280
25958 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25960 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25962 * target.def (compute_frame_layout): New optional target hook.
25963 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25964 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25965 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25967 * reload1.c (verify_initial_elim_offsets): Likewise.
25968 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25969 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25970 (arm_get_frame_offsets): Split up into this ...
25971 (arm_compute_frame_layout): ... and this function.
25973 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25975 * config/aarch64/constraints.md (Usa): New constraint.
25976 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25978 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25980 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25981 with_multilib_list after it has been checked.
25983 2017-05-08 Richard Biener <rguenther@suse.de>
25985 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25986 (bitmap_set_subtract_values): Likewise.
25988 2017-05-08 Richard Biener <rguenther@suse.de>
25990 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25991 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25993 (gimple_stmt_nonzero_warnv_p): Rename to ...
25994 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25996 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25997 (extract_range_basic): Adjust, do not disable propagation on
25998 strict overflow sensitive simplification.
25999 (vrp_visit_cond_stmt): Likewise.
26001 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
26003 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
26004 body size unconditionally.
26006 2017-05-07 Jeff Law <law@redhat.com>
26009 2017-05-06 Jeff Law <law@redhat.com>
26010 PR tree-optimization/78496
26011 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26014 PR tree-optimization/78496
26015 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26016 (simplify_stmt_using_ranges): Call it.
26017 (vrp_dom_walker::before_dom_children): Extract equivalences
26018 from an ASSERT_EXPR with an equality comparison against a
26021 2017-05-06 Jeff Law <law@redhat.com>
26023 PR tree-optimization/78496
26024 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
26027 PR tree-optimization/78496
26028 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
26029 (simplify_stmt_using_ranges): Call it.
26030 (vrp_dom_walker::before_dom_children): Extract equivalences
26031 from an ASSERT_EXPR with an equality comparison against a
26034 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26036 * lra-constraints.c (lra_copy_reg_equiv): New function.
26037 (split_reg): Use it to copy equivalence information from the
26038 original register to the spill register.
26040 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26042 PR rtl-optimization/75964
26043 * simplify-rtx.c (simplify_const_relational_operation): Remove
26044 invalid handling of comparisons of integer ABS.
26046 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26048 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26049 initialize to zero.
26050 (init_regs): Remove declaration.
26051 (function_arg_advance_32): Initialize error_p as boolean variable.
26053 2017-05-05 Nathan Sidwell <nathan@acm.org>
26055 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26056 lines. Use for (;;).
26058 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26060 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26061 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26062 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26063 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26064 VF=2 that require versioning.
26066 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26068 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26071 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26073 * diagnostic.h (diagnostic_override_option_index): Convert from
26074 macro to inline function.
26076 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26078 * diagnostic.c (last_module_changed_p): New function.
26079 (set_last_module): New function.
26080 (diagnostic_report_current_module): Convert macro usage to
26081 the above functions.
26082 * diagnostic.h (diagnostic_context::last_module): Strengthen
26083 from const line_map * to const line_map_ordinary *.
26084 (diagnostic_last_module_changed): Delete macro.
26085 (diagnostic_set_last_module): Delete macro.
26087 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26089 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26090 with diagnostic_report_diagnostic.
26091 (diagnostic_n_impl_richloc): Likewise.
26092 * diagnostic.h (report_diagnostic): Delete macro.
26093 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26094 with diagnostic_report_diagnostic.
26095 * substring-locations.c (format_warning_va): Likewise.
26097 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26099 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26100 save/restor of format_spec. Move option-printing code to...
26101 (print_option_information): ...this new function, and
26102 reimplement by simply printing to the pretty_printer,
26103 rather than appending to the format string.
26105 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26107 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26108 handling logic into...
26109 (update_effective_level_from_pragmas): ...this new function.
26111 2017-05-04 Andrew Waterman <andrew@sifive.com>
26113 * config/riscv/riscv.opt (mstrict-align): New option.
26114 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26115 (SLOW_UNALIGNED_ACCESS): Define.
26116 (riscv_slow_unaligned_access): Declare.
26117 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26119 (riscv_slow_unaligned_access): New variable.
26120 (rocket_tune_info): Set slow_unaligned_access to true.
26121 (optimize_size_tune_info): Set slow_unaligned_access to false.
26122 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26123 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26124 (riscv_option_override): Set riscv_slow_unaligned_access.
26125 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26127 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26129 * config/riscv/riscv.md: Unify indentation.
26131 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26136 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26138 (extendsi<mode>2): Add support for doing sign extension via
26139 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26140 don't have ISA 3.0 instructions.
26141 (extendsi<mode>2 splitter): Likewise.
26142 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26143 generate the normal insns since SImode can now go in vector
26144 registers. Disallow the special UNSPECs needed for previous
26145 machines to hide SImode being used. Add new insns
26146 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26147 (fix_trunc<mode>si2_stfiwx): Likewise.
26148 (fix_trunc<mode>si2_internal): Likewise.
26149 (fixuns_trunc<mode>si2): Likewise.
26150 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26151 (fctiw<u>z_<mode>_smallint): Likewise.
26152 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26153 of floating point to 32-bit integer from doing a direct move to
26154 the GPR registers to do a store.
26155 (fctiwz_<mode>): Break long line.
26157 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26159 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26160 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26161 (addr_list, addr_offset_valid_p): New.
26162 (split_address_groups): Check offset validity with above function.
26163 (gt-tree-ssa-loop-ivopts.h): Include header file.
26165 2017-05-05 Nathan Sidwell <nathan@acm.org>
26167 * config.gcc (arm*-*-*): Add missing 'fi'.
26169 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26171 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26172 included in -fopt-info does not matter.
26173 * doc/optinfo.texi (-fopt-info): Fix description of default
26174 behavour. Explicitly say order of options included in -fopt-info
26177 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26179 * config.gcc: Allow combinations of aprofile and rmprofile values for
26180 --with-multilib-list.
26181 * config/arm/t-multilib: New file.
26182 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26183 variables. Remove setting of ISA and floating-point ABI in
26184 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26185 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26186 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26187 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26189 * config/arm/t-rmprofile: Likewise except for the matches changes.
26190 * doc/install.texi (--with-multilib-list): Document the combination of
26191 aprofile and rmprofile values and warn about pitfalls in doing that.
26193 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26195 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26196 (movdi_aarch64): Likewise.
26198 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26200 PR tree-optimization/80632
26201 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26203 (build_arrays): Initialize it for virtual phis.
26204 (fix_phi_nodes): Use it for virtual phis.
26206 PR tree-optimization/80558
26207 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26208 [x, y] op z into [x op, y op z] for op & or | if conditions
26211 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26212 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26215 * config/arm/arm.md (use_literal_pool): Remove.
26216 (64-bit immediate split): No longer takes cost into consideration
26217 if arm_disable_literal_pool is enabled.
26218 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26219 used when arm_disable_literal_pool is enabled.
26220 (arm_max_const_double_inline_cost): Remove use of
26221 arm_disable_literal_pool.
26222 (push_minipool_fix): Add assert.
26223 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26224 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26225 (no_literal_pool_sf_immediate): New.
26227 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26229 PR tree-optimization/80613
26230 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26231 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26233 2017-05-05 Richard Biener <rguenther@suse.de>
26235 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26237 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26239 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26240 of this flag from insn conditions due to removal from r247495.
26242 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26244 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26246 (arm_early_store_addr_dep_ptr): Likewise.
26247 * config/arm/aarch-common-protos.h
26248 (arm_early_load_addr_dep_ptr): Add prototype.
26249 (arm_early_store_addr_dep_ptr): Likewise.
26250 * config/arm/cortex-a53.md: Add new bypasses.
26252 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26254 * tree.c (next_type_uid): Change type to unsigned.
26255 (type_hash_canon): Decrement back next_type_uid if
26256 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26257 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26260 2017-05-04 Martin Sebor <msebor@redhat.com>
26262 * builtins.c: Fix a trivial typo in a comment.
26264 PR middle-end/79234
26265 * builtins.c (check_sizes): Adjust to handle reading past the end.
26266 Avoid printing excessive upper bound of ranges. Use %E to print
26267 tree nodes instead of converting them to %wu.
26268 (expand_builtin_memchr): New function.
26269 (compute_dest_size): Rename...
26270 (compute_objsize): ...to this.
26271 (expand_builtin_memcpy): Adjust.
26272 (expand_builtin_mempcpy): Adjust.
26273 (expand_builtin_strcat): Adjust.
26274 (expand_builtin_strcpy): Adjust.
26275 (check_strncat_sizes): Adjust.
26276 (expand_builtin_strncat): Adjust.
26277 (expand_builtin_strncpy): Adjust and simplify.
26278 (expand_builtin_memset): Adjust.
26279 (expand_builtin_bzero): Adjust.
26280 (expand_builtin_memcmp): Adjust.
26281 (expand_builtin): Handle memcmp.
26282 (maybe_emit_chk_warning): Check strncat just once.
26284 2017-05-04 Martin Sebor <msebor@redhat.com>
26286 PR preprocessor/79214
26287 PR middle-end/79222
26288 PR middle-end/79223
26289 * builtins.c (check_sizes): Add inlining context and issue
26290 warnings even when -Wno-system-headers is set.
26291 (check_strncat_sizes): Same.
26292 (expand_builtin_strncat): Same.
26293 (expand_builtin_memmove): New function.
26294 (expand_builtin_stpncpy): Same.
26295 (expand_builtin): Handle memmove and stpncpy.
26297 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26299 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26300 which is not used any more.
26302 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26304 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26306 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26308 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26309 (cortexa53_tunings): Likewise.
26310 (cortexa57_tunings): Likewise.
26311 (cortexa72_tunings): Likewise.
26312 (cortexa73_tunings): Likewise.
26314 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26316 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26317 Set loop alignment to 8.
26319 2017-05-04 Martin Sebor <msebor@redhat.com>
26321 PR translation/80280
26322 * builtins.c (expand_builtin_object_size): Add missing quoting to
26323 %D and like directives.
26324 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26325 (hsa_type_for_tree_type): Same.
26326 (verify_function_arguments): Same.
26327 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26328 * varasm.c (get_section): Same.
26331 2017-05-04 Martin Sebor <msebor@redhat.com>
26333 PR translation/80280
26334 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26336 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26338 * config/aarch64/aarch64.c (generic_addrcost_table):
26339 Change HI/TI mode setting.
26341 2017-05-04 Martin Jambor <mjambor@suse.cz>
26343 PR tree-optimization/80622
26344 * tree-sra.c (comes_initialized_p): New function.
26345 (build_accesses_from_assign): Only set write lazily when
26346 comes_initialized_p is false.
26347 (analyze_access_subtree): Use comes_initialized_p.
26348 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26349 instead of testing for PARM_DECL.
26351 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26353 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26354 constraint on operand 0 to allow more general addressing modes.
26355 Adjust output template.
26356 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26358 * config/aarch64/aarch64-protos.h
26359 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26360 * config/aarch64/constraints.md (Dp): New address constraint.
26361 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26364 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26366 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26367 update use of estimate_ipcp_clone_size_and_time.
26368 (estimate_local_effects): Update use of
26369 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26370 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26371 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26372 Return nonspecialized time.
26374 2017-05-04 Richard Biener <rguenther@suse.de>
26376 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26377 for the last VUSE which def dominates the PHI. Directly call
26379 (get_continuation_for_phi_1): Remove.
26381 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26383 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26384 to explain the use of truncating division. Cap the number of
26385 iterations to the maximum given by nb_iterations_upper_bound,
26388 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26390 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26391 * configure: Regenerate.
26392 * config.in: Regenerate.
26393 * config/i386/driver-mingw32.c: new file.
26394 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26395 * config.host: Link driver-mingw32.o on MinGW host.
26396 * doc/install.texi: Document new --enable-mingw-wildcard configure
26399 2017-05-04 Marek Polacek <polacek@redhat.com>
26401 PR tree-optimization/80612
26402 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26404 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26405 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26407 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26408 (movt splitter): Likewise.
26409 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26410 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26411 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26412 block for Thumb-1 with MOVT.
26413 (thumb2_legitimate_address_p): Move code block ...
26414 (can_avoid_literal_pool_for_label_p): ... into this new function.
26415 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26417 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26418 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26419 "M-profile targets with the MOVT instruction".
26421 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26423 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26424 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26425 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26427 2017-05-04 Martin Liska <mliska@suse.cz>
26429 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26430 variable cond_code.
26432 2017-05-04 Richard Biener <rguenther@suse.de>
26434 * tree.c (array_at_struct_end_p): Handle arrays at struct
26435 end with flexarrays more conservatively. Refactor and treat
26436 arrays of arrays or aggregates more strict. Fix
26437 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26438 * tree.c (array_at_struct_end_p): Adjust prototype.
26439 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26440 * gimple-fold.c (get_range_strlen): Likewise.
26441 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26443 2017-05-04 Richard Biener <rguenther@suse.de>
26445 PR tree-optimization/31130
26446 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26448 (supports_overflow_infinity): Likewise.
26449 (is_negative_overflow_infinity): Likewise.
26450 (is_positive_overflow_infinity): Likewise.
26451 (is_overflow_infinity): Likewise.
26452 (stmt_overflow_infinity): Likewise.
26453 (overflow_infinity_range_p): Likewise.
26454 (usable_range_p): Remove as always returning true.
26455 (make_overflow_infinity): Remove.
26456 (negative_overflow_infinity): Likewise.
26457 (positive_overflow_infinity): Likewise.
26458 (avoid_overflow_infinity): Likewise.
26459 (set_value_range): Adjust accordingly.
26460 (set_value_range_to_nonnegative): Likewise, remove now unused
26461 overflow_infinity arg.
26462 (vrp_operand_equal_p): Adjust.
26463 (update_value_range): Likewise.
26464 (range_int_cst_singleton_p): Likewise.
26465 (operand_less_p): Likewise.
26466 (compare_values_warnv): Likewise.
26467 (extract_range_for_var_from_comparison_expr): Likewise.
26468 (vrp_int_const_binop): Likewise.
26469 (zero_nonzero_bits_from_vr): Likewise.
26470 (extract_range_from_multiplicative_op_1): Likewise.
26471 (extract_range_from_binary_expr_1): Likewise.
26472 (extract_range_from_unary_expr): Likewise.
26473 (extract_range_from_comparison): Likewise.
26474 (extract_range_basic): Likewise.
26475 (adjust_range_with_scev): Likewise.
26476 (compare_ranges): Likewise.
26477 (compare_range_with_value): Likewise.
26478 (dump_value_range): Likewise.
26479 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26481 (simplify_cond_using_ranges): Adjust.
26483 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26485 * brig-builtins.def: Added a builtin for class_f64.
26486 * builtin-types.def: Added a builtin type needed by class_f64.
26488 2017-05-03 Jason Merrill <jason@redhat.com>
26490 * timevar.def: Add TV_CONSTEXPR.
26492 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26494 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26496 2017-05-03 Martin Jambor <mjambor@suse.cz>
26498 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26499 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26500 * ipa-cp.c (ipcp_propagate_stage): Do not call
26501 ipa_update_after_lto_read.
26502 * ipa-inline.c (ipa_inline): Likewise.
26504 2017-05-03 Martin Jambor <mjambor@suse.cz>
26506 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26507 tag. Added a default constructor and a destructor.
26508 (ipa_edge_args_sum_t): New class;
26509 (ipa_edge_args_sum): Declare.
26510 (ipa_edge_args_vector): Remove declaration.
26511 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26512 (ipa_free_edge_args_substructures): Remove declaration.
26513 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26514 (ipa_edge_args_info_available_for_edge_p): Likewise.
26515 * ipa-prop.c (ipa_edge_args_vector): Removed.
26516 (edge_removal_hook_holder): Likewise.
26517 (edge_duplication_hook_holder): Likewise.
26518 (ipa_edge_args_sum): New variable.
26519 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26520 ipa_edge_args_vector.
26521 (ipa_free_edge_args_substructures): Likewise.
26522 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26523 ipa_edge_args_vector.
26524 (ipa_edge_removal_hook): Turned into method
26525 ipa_edge_args_sum_t::remove.
26526 (ipa_edge_duplication_hook): Turned into method
26527 ipa_edge_args_sum_t::duplicate.
26528 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26529 registering edge hooks.
26530 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26531 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26532 ipa_edge_args_sum instead of ipa_edge_args_vector.
26533 * ipa-profile.c (ipa_profile): Likewise.
26535 2017-05-03 Martin Jambor <mjambor@suse.cz>
26537 * symbol-summary.h (function_summary): New method exists.
26538 (function_summary::symtab_removal): Deallocate through release.
26539 (call_summary): New class.
26540 (gt_ggc_mx): New overload.
26541 (gt_pch_nx): Likewise.
26542 (gt_pch_nx): Likewise.
26544 2017-05-03 Jeff Law <law@redhat.com>
26546 PR tree-optimization/78496
26547 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26548 from simplify_cond_using_ranges. Split off code to walk
26549 backwards through casts into ...
26550 (simplify_cond_using_ranges_2): New function.
26551 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26552 (execute_vrp): After identifying jump threads, call
26553 simplify_cond_using_ranges_2.
26555 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26558 * ipa-inline.h (inline_summary): Add ctor.
26559 (create_ggc): Do not use ggc_cleared_alloc.
26561 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26562 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26564 * gcc.c (handle_braces): Support escaping in switch matching
26566 * doc/invoke.texi (Spec Files): Document it.
26567 Remove superfluous @code markup in items.
26569 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26571 * diagnostic-show-locus.c (struct column_range): New struct.
26572 (get_affected_columns): New function.
26573 (get_printed_columns): New function.
26574 (struct correction): New struct.
26575 (correction::ensure_capacity): New function.
26576 (correction::ensure_terminated): New function.
26577 (struct line_corrections): New struct.
26578 (line_corrections::~line_corrections): New dtor.
26579 (line_corrections::add_hint): New function.
26580 (layout::print_trailing_fixits): Reimplement in terms of the new
26582 (selftest::test_overlapped_fixit_printing): New function.
26583 (selftest::diagnostic_show_locus_c_tests): Call it.
26585 2017-05-03 Nathan Sidwell <nathan@acm.org>
26587 Canonicalize canonical type hashing
26588 * tree.h (type_hash_canon_hash): Declare.
26589 * tree.c (type_hash_list, attribute_hash_list): Move into
26590 type_hash_canon_hash.
26591 (build_type_attribute_qual_variant): Break out hash code calc into
26592 type_hash_canon_hash.
26593 (type_hash_canon_hash): New. Generic type hash computation.
26594 (build_range_type_1, build_array_type_1, build_function_type,
26595 build_method_type_directly, build_offset_type, build_complex_type,
26596 make_vector_type): Call it.
26598 2017-05-03 Richard Biener <rguenther@suse.de>
26600 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26601 When all DRs have unknown misaligned do not always peel
26602 when there is a store but apply the same costing model as if
26603 there were only loads.
26605 2017-05-03 Richard Biener <rguenther@suse.de>
26608 PR tree-optimization/80492
26609 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26610 compare_base_decls returning dont-know properly.
26612 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26614 * config/arm/iterators.md (CCSI): New mode iterator.
26615 (arch): New mode attribute.
26616 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26617 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26618 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26619 code iterator for success result mode.
26620 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26621 the corresponding new insn generators.
26623 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26626 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26627 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26629 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26631 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26632 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26633 (DDR_A): Wrap DDR argument in brackets.
26634 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26635 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26636 (DDR_REVERSED_P): Likewise.
26638 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26640 PR tree-optimization/79472
26641 * tree-switch-conversion.c (struct switch_conv_info): Add
26642 contiguous_range and default_case_nonstandard fields.
26643 (collect_switch_conv_info): Compute contiguous_range and
26644 default_case_nonstandard fields, don't clear final_bb if
26645 contiguous_range and only the default case doesn't have the required
26647 (check_all_empty_except_final): Set default_case_nonstandard instead
26648 of failing if contiguous_range and the default case doesn't have empty
26650 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26651 and only the default case doesn't have the required constants. Skip
26653 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26654 if default_case_nonstandard.
26655 (build_constructors): Build constant 1 just once. Assert that default
26656 values aren't inserted in between cases if contiguous_range. Skip
26658 (build_arrays): Skip virtual phis.
26659 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26660 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26661 Handle virtual phis.
26662 (gen_inbound_check): Handle default_case_nonstandard case.
26663 (process_switch): Adjust check_final_bb caller. Call
26664 gather_default_values with the first non-default case instead of
26665 default case if default_case_nonstandard.
26667 2017-05-02 Nathan Sidwell <nathan@acm.org>
26669 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26670 check. Fix formatting.
26672 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26674 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26675 errors when comparing specialized and unspecialized times.
26677 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26679 * diagnostic-show-locus.c
26680 (layout::should_print_annotation_line_p): Make private.
26681 (layout::print_annotation_line): Make private.
26682 (layout::annotation_line_showed_range_p): Make private.
26683 (layout::show_ruler): Make private.
26684 (layout::print_source_line): Make private. Pass in line and
26685 line_width, rather than calling location_get_source_line. Drop
26687 (layout::print_leading_fixits): New method.
26688 (layout::print_any_fixits): Rename to...
26689 (layout::print_trailing_fixits): ...this, and make private.
26690 Don't print newline fixits.
26691 (diagnostic_show_locus): Move logic for printing one row into...
26692 (layout::print_line): ...this new function. Move the
26693 location_get_source_line call and error-handling from
26694 print_source_line to here. Call print_leading_fixits, and rename
26695 print_any_fixits to print_trailing_fixits.
26696 (selftest::test_fixit_insert_containing_newline): Update now that
26697 newlines are partially supported.
26698 (selftest::test_fixit_insert_containing_newline_2): New test.
26699 (selftest::test_fixit_replace_containing_newline): Update comments.
26700 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26701 * edit-context.c (class added_line): New class.
26702 (class edited_line): Describe newline handling in comment.
26703 (edited_line::actually_edited_p): New method.
26704 (edited_line::print_content): Delete redundant decl.
26705 (edited_line::m_predecessors): New field.
26706 (edited_file::print_content): Call edited_line::print_content.
26707 (edited_file::print_diff): Update to support newlines.
26708 (edited_file::print_diff_hunk): Likewise.
26709 (edited_file::print_run_of_changed_lines): New function.
26710 (edited_file::print_diff_line): Convert to...
26711 (print_diff_line): ...this.
26712 (edited_file::get_effective_line_count): New function.
26713 (edited_line::edited_line): Initialize new field m_predecessors.
26714 (edited_line::~edited_line): Clean up m_predecessors.
26715 (edited_line::apply_fixit): Handle newlines.
26716 (edited_line::get_effective_line_count): New function.
26717 (edited_line::print_content): New function.
26718 (edited_line::print_diff_lines): New function.
26719 (selftest::test_applying_fixits_insert_containing_newline): New
26721 (selftest::test_applying_fixits_replace_containing_newline): New
26723 (selftest::insert_line): New function.
26724 (selftest::test_applying_fixits_multiple_lines): Add example of
26726 (selftest::edit_context_c_tests): Call the new tests.
26728 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26730 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26731 parameter cand. Update dump information.
26732 (get_computation_cost): Update uses.
26734 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26736 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26737 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26738 (get_computation_at, rewrite_use_address): Update use of
26739 get_computation_aff.
26741 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26743 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26744 (get_computation): Delete.
26745 (get_computation_cost): Implement like get_computation_cost_at.
26746 Use get_computation_at.
26747 (get_computation_cost_at): Delete.
26748 (rewrite_use_nonlinear_expr): Use get_computation_at.
26749 (rewrite_use_compare, remove_unused_ivs): Ditto.
26751 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26753 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26755 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26757 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26758 (ivopts_global_cost_for_size): Rename parameter and update uses.
26759 (iv_ca_recount_cost): Update uses.
26760 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26761 candidates seperately in n_invs and n_cands.
26762 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26764 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26766 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26767 (find_inv_vars_cb): New.
26768 (find_depends): Renamed to ...
26769 (find_inv_vars): ... this.
26770 (add_candidate_1, force_var_cost): Call find_inv_vars.
26771 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26775 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26776 inv_vars. Add inv_exprs.
26777 (struct iv_cand): Rename depends_on to inv_vars.
26778 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26779 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26780 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26781 (dump_cand): Dump inv_vars.
26782 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26783 (record_invariant, find_depends, add_candidate_1): Ditto.
26784 (set_group_iv_cost, force_var_cost): Ditto.
26785 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26786 (get_computation_cost_at, get_computation_cost): Ditto.
26787 (determine_group_iv_cost_generic): Ditto.
26788 (determine_group_iv_cost_address): Ditto.
26789 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26790 (determine_group_iv_costs): Ditto.
26791 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26792 (iv_ca_set_remove_invariants): Renamed to ...
26793 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26794 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26795 (iv_ca_set_add_invariants): Renamed to ...
26796 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26797 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26798 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26799 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26800 (create_new_ivs): Remove useless dump.
26802 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26804 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26806 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26807 (iv_ca_set_no_cp, create_new_iv): Ditto.
26809 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26811 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26813 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26815 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26816 function tree_check2.
26818 2017-05-02 Martin Liska <mliska@suse.cz>
26820 * doc/gcov.texi: Add missing preposition.
26821 * gcov.c (function_info::function_info): Properly fill up
26822 all member variables.
26824 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26826 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26828 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26830 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26832 2017-05-02 Martin Liska <mliska@suse.cz>
26835 * lto-streamer-in.c (lto_read_tree_1): Remove
26836 LTO_STREAMER_DEBUG.
26837 * lto-streamer.c (struct tree_hash_entry): Likewise.
26838 (struct tree_entry_hasher): Likewise.
26839 (tree_entry_hasher::hash): Likewise.
26840 (tree_entry_hasher::equal): Likewise.
26841 (lto_streamer_init): Likewise.
26842 (lto_orig_address_map): Likewise.
26843 (lto_orig_address_get): Likewise.
26844 (lto_orig_address_remove): Likewise.
26845 * lto-streamer.h: Likewise.
26846 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26847 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26849 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26851 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26852 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26853 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26854 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26855 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26856 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26857 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26858 (mm_maskz_sub_ss): New intrinsics.
26859 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26860 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26861 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26862 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26863 (__builtin_ia32_subss_mask_round): New builtins.
26864 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26865 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26866 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26868 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26869 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26871 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26874 2017-05-02 Martin Jambor <mjambor@suse.cz>
26876 PR tree-optimization/78687
26877 * tree-sra.c (access): New field parent.
26878 (process_subtree_disqualification): New function.
26879 (disqualify_candidate): Call it.
26880 (build_accesses_from_assign): Reset write flag if creating an
26882 (build_access_subtree): Fill in parent field and also prpagate
26883 down grp_write flag.
26884 (create_artificial_child_access): New parameter set_grp_write, set
26885 grp_write to its value.
26886 (propagate_subaccesses_across_link): Also propagate grp_write flag
26888 (propagate_all_subaccesses): Push the closest parent back to work
26889 queue if add_access_to_work_queue returned true.
26891 2017-05-02 Richard Biener <rguenther@suse.de>
26893 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26894 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26895 -fstrict-overflow documentation.
26896 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26897 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26898 flag_strict_overflow.
26899 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26900 * lto-opts.c (lto_write_options): Do not stream it.
26901 * lto-wrapper.c (merge_and_complain): Do not handle it.
26902 * opts.c (default_options_table): Do not set -fstrict-overflow.
26903 (finish_options): Likewise do not clear it when sanitizing.
26904 * simplify-rtx.c (simplify_const_relational_operation): Do not
26905 test flag_strict_overflow.
26907 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26909 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26910 using enabled attribute.
26911 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26912 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26913 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26914 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26915 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26916 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26917 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26918 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26919 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26920 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26922 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26924 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26926 2017-05-02 Richard Biener <rguenther@suse.de>
26928 PR tree-optimization/80591
26930 2017-04-10 Richard Biener <rguenther@suse.de>
26932 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26935 2017-05-02 Richard Biener <rguenther@suse.de>
26937 PR tree-optimization/80549
26938 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26939 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26940 headers if they do not have a preheader.
26942 2017-05-02 Martin Liska <mliska@suse.cz>
26945 * common.opt: Fix typo.
26946 * doc/invoke.texi: Likewise.
26948 2017-05-01 Jan Beulich <jbeulich@suse.com>
26950 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26951 swapping, add (x,x,m,x,n) alternative.
26953 2017-05-01 Nathan Sidwell <nathan@acm.org>
26955 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26956 unnecessary unadjusted_alignment check.
26958 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26961 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26963 * gimplify.c (gimplify_cilk_detach): New function.
26964 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26965 * tree-core.h: Document EXPR_CILK_SPAWN.
26966 * tree.h (EXPR_CILK_SPAWN): Define.
26968 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26970 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26971 to use new fixit_hint representation, using the "replace" logic.
26972 (get_line_span_for_fixit_hint): Likewise.
26973 (layout::print_any_fixits): Likewise.
26974 (selftest::test_one_liner_many_fixits): Rename to...
26975 (selftest::test_one_liner_many_fixits_1): ...this, and update
26976 comment and expected output to reflect that the multiple fix-it
26977 hints are now consolidated into one insertion.
26978 (selftest::test_one_liner_many_fixits_2): New test.
26979 (selftest::test_diagnostic_show_locus_one_liner): Update for
26981 (selftest::test_fixit_consolidation): Update for fix-it API
26983 * diagnostic.c (print_parseable_fixits): Likewise.
26984 * edit-context.c (edited_line::m_line_events): Convert from
26985 auto_vec <line_event *> to auto_vec <line_event>.
26986 (class line_event): Convert from abstract base class to a concrete
26987 class, taking over the role of replace_event.
26988 (class insert_event): Delete.
26989 (class replace_event): Rename to class line_event. Convert to
26991 (edit_context::add_fixits): Reimplement.
26992 (edit_context::apply_insert): Delete.
26993 (edit_context::apply_replace): Rename to...
26994 (edit_context::apply_fixit): ...this. Convert to half-open range.
26995 (edited_file::apply_insert): Delete.
26996 (edited_file::apply_replace): Rename to...
26997 (edited_file::apply_fixit): ...this.
26998 (edited_line::~edited_line): Drop deletion of events.
26999 (edited_line::apply_insert): Delete.
27000 (edited_line::apply_replace): Rename to...
27001 (edited_line::apply_fixit): ...this. Convert to half-open range.
27002 Update for change to type of m_line_events.
27003 * edit-context.h (edit_context::apply_insert): Delete.
27004 (edit_context::apply_replace): Rename to...
27005 (edit_context::apply_fixit): ...this.
27007 2017-05-01 Martin Sebor <msebor@redhat.com>
27009 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
27012 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
27015 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
27016 __get_cpuid_max returns 0.
27017 (__get_cpuid_count): Ditto.
27019 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
27021 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
27022 replacement expression is another instance of one of its arguments.
27024 2017-05-01 Jakub Jelinek <jakub@redhat.com>
27027 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
27028 check for stack push/pop autoinc.
27029 * config/i386/i386.c (ix86_agi_dependent): Return false
27030 if the only reason why modified_in_p returned true is that
27031 addr is SP based and set_insn is a push or pop.
27033 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27035 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27038 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27041 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27042 (account_size_time): Use two predicates - exec_pred and
27044 (evaluate_conditions_for_known_args): Compute both clause and
27046 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27047 (inline_summary_t::duplicate): Update.
27048 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27050 (compute_inline_parameters): Likewise.
27051 (estimate_edge_size_and_time): Update caluclation of time.
27052 (estimate_node_size_and_time): Compute both time and nonspecialized
27054 (estimate_ipcp_clone_size_and_time): Update.
27055 (inline_merge_summary): Update.
27056 (do_estimate_edge_time): Update.
27057 (do_estimate_edge_size): Update.
27058 (do_estimate_edge_hints): Update.
27059 (inline_read_section, inline_write_summary): Stream both new predicates.
27060 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27062 (compute_inlined_call_time): Cleanup.
27063 (big_speedup_p): Update.
27064 (edge_badness): Update.
27065 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27066 (size_time_entry): Replace predicate by exec_predicate and
27067 nonconst_predicate.
27068 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27069 (estimate_edge_time): Return also nonspec_time.
27070 (reset_edge_growth_cache): Update.
27072 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27074 PR rtl-optimization/80491
27075 * ifcvt.c (noce_process_if_block): When looking for x setter
27076 with missing else_bb, don't check only the insn right before
27077 cond_earliest, but look for the last insn that x is modified in
27078 within the same bb.
27080 PR rtl-optimization/80491
27081 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27083 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27085 PR tree-optimization/80487
27086 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27088 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27090 PR tree-optimization/79697
27091 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27092 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27093 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27095 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27096 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27098 2017-04-28 Martin Sebor <msebor@redhat.com>
27100 PR tree-optimization/80523
27101 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27102 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27104 (maybe_warn, format_directive, parse_directive): Use new functions.
27105 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27107 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27109 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27111 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27113 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27114 target_header_dir): Set correctly.
27115 * configure: Regenerated.
27116 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27117 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27118 instead of SYSTEM_HEADER_DIR.
27120 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27122 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27123 (estimate_local_effects): Likewise.
27124 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27125 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27126 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27127 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27128 do_estimate_edge_time, estimate_edge_time): Likewise.
27129 * ipa-inline-analysis.c (estimate_node_size_and_time,
27130 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27131 (estimate_time_after_inlining): Remove.
27133 2017-04-28 Martin Liska <mliska@suse.cz>
27135 * doc/gcov.texi: Enhance documentation of gcov.
27137 2017-04-28 Martin Liska <mliska@suse.cz>
27139 * doc/gcov.texi: Sort options in alphabetic order.
27140 * doc/gcov-dump.texi: Likewise.
27141 * doc/gcov-tool.texi: Likewise.
27142 * gcov.c (print_usage): Likewise.
27143 * gcov-dump.c (print_usage): Likewise.
27144 * gcov-tool.c (print_merge_usage_message): Likewise.
27145 (print_rewrite_usage_message): Likewise.
27146 (print_overlap_usage_message): Likewise.
27148 2017-04-28 Martin Liska <mliska@suse.cz>
27150 PR gcov-profile/53915
27151 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27153 2017-04-28 Martin Liska <mliska@suse.cz>
27155 PR gcov-profile/79891
27156 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27157 is marked by compiler as living on a line.
27158 (get_cycles_count): Remove usage of the union.
27159 (output_intermediate_file): Likewise.
27160 (find_source): Fix GNU coding style.
27161 (accumulate_line_counts): Remove old non-all block mode.
27162 (output_lines): Remove usage of the union.
27163 * profile.c (output_location): Include all BBs, even if
27164 belonging to a same line (and file) as a previous BB.
27166 2017-04-28 Martin Liska <mliska@suse.cz>
27168 * gcov.c (process_args): Handle new argument 'w'.
27169 (read_graph_file): Assign ID to BBs.
27170 (output_branch_count): Display BB # if verbose flag is set.
27171 (output_lines): Likewise for arcs.
27172 (print_usage): Add '--verbose' option help.
27173 * doc/gcov.texi: Document --verbose (-w) option.
27175 2017-04-28 Martin Liska <mliska@suse.cz>
27177 * gcov.c (struct block_location_info): New struct.
27178 (process_file): Fill up the new structure.
27179 (read_graph_file): Replace usage of encoding by the newly added
27181 (add_line_counts): Likewise.
27182 (accumulate_line_counts): Remove usage of the union.
27183 (function_info::function_info): New function.
27184 (function_info::~function_info): Likewise.
27185 (process_file): Call delete instead of release_function.
27186 (release_function): Release the function.
27187 (release_structures): Call delete instead of release_function.
27188 (solve_flow_graph): Replace usage of num_blocks.
27189 (find_exception_blocks): Likewise.
27190 (output_lines): Fix GNU coding style.
27192 2017-04-28 Martin Liska <mliska@suse.cz>
27195 * coverage.c (coverage_remove_note_file): New function.
27196 * coverage.h: Declare the function.
27197 * toplev.c (finalize): Clean if an error has been seen.
27199 2017-04-28 Martin Liska <mliska@suse.cz>
27201 PR gcov-profile/80031
27202 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27203 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27204 * gcov.c (read_graph_file): Read just number of blocks.
27205 * profile.c (branch_prob): Do not stream 0 flags per a basic
27208 2017-04-28 Martin Liska <mliska@suse.cz>
27210 * gcov-dump.c (tag_*): Add new argument to declarations.
27211 (dump_gcov_file): Likewise.
27212 (tag_blocks): Add and use new argument depth.
27213 (tag_arcs): Likewise.
27214 (tag_lines): Likewise.
27215 (tag_counters): Likewise.
27216 (tag_summary): Likewise.
27217 (dump_working_sets): Use depth to do a proper indentation.
27219 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27222 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27223 * symtab.c (symtab_node::debug_symtab): Move definition here.
27225 2017-04-28 Richard Biener <rguenther@suse.de>
27227 * lto-streamer.h (LTO_major_version): Bump to 7.
27229 2017-04-28 Richard Biener <rguenther@suse.de>
27231 * tree-vrp.c (assert_info): New struct.
27232 (add_assert_info): New helper.
27233 (register_edge_assert_for_2): Refactor to add asserts to a vector
27235 (register_edge_assert_for_1): Likewise.
27236 (register_edge_assert_for): Likewise.
27237 (finish_register_edge_assert_for): New helper actually registering
27238 asserts where live on edge.
27239 (find_conditional_asserts): Adjust.
27240 (find_switch_asserts): Likewise.
27241 (evrp_dom_walker::try_find_new_range): Generalize.
27242 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27244 2017-04-27 Marek Polacek <polacek@redhat.com>
27247 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27248 arg10 and arg11 to itype.
27250 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27252 * doc/extend.texi (Object Size Checking): Improve grammar.
27254 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27257 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27258 that the logic for permitting reciprocal estimates matches that
27261 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27264 * tree.c (type_cache_hasher::equal): Only compare
27265 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27266 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27267 non-aggregate element types.
27268 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27269 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27271 2017-04-27 Richard Biener <rguenther@suse.de>
27273 PR middle-end/80533
27274 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27275 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27276 keeping a reference to a trailing array.
27278 2017-04-27 Richard Biener <rguenther@suse.de>
27280 PR middle-end/80539
27281 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27282 being in loop-closed SSA form conservatively.
27283 (chrec_fold_multiply_poly_poly): Likewise.
27285 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27287 PR middle-end/79665
27288 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27289 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27291 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27294 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27295 (aarch64_function_arg_alignment): Return unsigned int again, but still
27296 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27297 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27298 Don't emit -Wpsabi note.
27299 (aarch64_function_arg_boundary): Likewise.
27300 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27303 2017-04-26 Nathan Sidwell <nathan@acm.org>
27305 * tree.h (crc32_unsigned_n): Declare.
27306 (crc32_unsigned, crc32_unsigned): Make inline.
27307 * tree.c (crc32_unsigned_bits): Replace with ...
27308 (crc32_unsigned_n): ... this.
27309 (crc32_unsigned, crc32_byte): Remove.
27310 (crc32_string): Remove unnecessary braces.
27312 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27314 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27315 * ipa-inline-analysis.c (MAX_TIME): Remove.
27316 (account_size_time): Use sreal for time.
27317 (dump_inline_summary): Update.
27318 (estimate_function_body_sizes): Update.
27319 (estimate_edge_size_and_time): Update.
27320 (estimate_calls_size_and_time): Update.
27321 (estimate_node_size_and_time): Update.
27322 (inline_merge_summary): Update.
27323 (inline_update_overall_summary): Update.
27324 (estimate_time_after_inlining): Update.
27325 (inline_read_section): Update.
27326 (inline_write_summary): Update.
27327 * ipa-inline.c (compute_uninlined_call_time): Update.
27328 (compute_inlined_call_time): Update.
27329 (recursive_inlining): Update.
27330 (inline_small_functions): Update.
27331 (dump_overall_stats): Update.
27332 * ipa-inline.h: Include sreal.h.
27333 (size_time_entry): Turn time to sreal.
27334 (inline_summary): Turn self_time nad time to sreal.
27336 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27338 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27340 (sreal::stream_out, sreal::stream_in): New.
27341 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27343 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27345 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27348 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27351 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27353 (dimode_scalar_chain::compute_convert_gain): Ditto.
27354 (dimode_scalar_chain::make_vector_copies): Ditto.
27355 (dimode_scalar_chain::convert_reg): Ditto.
27356 (dimode_scalar_chain::convert_insn): Ditto.
27357 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27358 (VI248_AVX512BW_1): New mode iterator.
27359 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27360 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27363 2017-04-25 Martin Sebor <msebor@redhat.com>
27365 PR tree-optimization/80497
27366 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27367 constants are representable in HOST_WIDE_INT.
27368 (parse_directive): Ditto.
27370 2017-04-25 Martin Sebor <msebor@redhat.com>
27373 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27374 (new_zero_array): Adjust signature.
27375 (dom_info::dom_init): Used unsigned rather that size_t.
27376 (dom_info::dom_info): Same.
27378 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27379 Jakub Jelinek <jakub@redhat.com>
27382 * config/arm/arm.c: Include gimple.h.
27383 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27384 returns negative, increment ncrn only if it returned positive.
27385 (arm_needs_doubleword_align): Return int instead of bool,
27386 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27387 members, but if there is any such non-FIELD_DECL
27388 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27389 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27390 returns negative, increment nregs only if it returned positive.
27391 (arm_setup_incoming_varargs): Likewise.
27392 (arm_function_arg_boundary): Emit -Wpsabi note if
27393 arm_needs_doubleword_align returns negative, return
27394 DOUBLEWORD_ALIGNMENT only if it returned positive.
27396 2017-04-25 Marek Polacek <polacek@redhat.com>
27399 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27400 first argument to type.
27402 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27405 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27406 type checks to test for compatibility instead of equality.
27408 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27409 Jakub Jelinek <jakub@redhat.com>
27412 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27414 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27415 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27416 the alignment computation, but return their maximum in warn_alignment.
27417 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27418 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27420 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27421 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27424 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27426 * config/arc/simdext.md (dmpyh): Fix typo.
27428 2017-04-25 Richard Biener <rguenther@suse.de>
27430 PR tree-optimization/80492
27431 * alias.c (compare_base_decls): Handle registers with asm
27432 specification conservatively.
27433 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27434 compare_base_decls returning dont-know properly.
27436 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27438 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27439 (legitimate_offset_address_p): New function.
27440 (arc_legitimate_address_p): Use above function.
27442 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27444 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27446 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27448 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27449 ACCH registers whenever they are available.
27451 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27453 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27454 double regs fix when not used.
27456 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27458 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27460 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27461 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27463 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27465 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27466 when emitting short ADD instructions.
27468 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27470 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27472 (cmpsi_cc_c_insn): Likewise.
27473 (cbranchsi4_scratch): Compute proper instruction length using
27474 compact_hreg_operand.
27475 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27477 2017-04-25 Richard Biener <rguenther@suse.de>
27479 PR middle-end/80509
27480 * passes.c (pass_manager::pass_manager): Initialize
27481 m_name_to_pass_map.
27483 2017-04-25 Richard Biener <rguenther@suse.de>
27485 PR tree-optimization/79201
27486 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27488 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27491 * config/s390/vector.md: Split MEM->GPR vector moves for
27492 non-s_operand addresses.
27494 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27497 * config/s390/predicates.md (reload_const_wide_int_operand): New
27499 * config/s390/s390.md ("movti"): Remove d/P alternative.
27500 ("movti_bigconst"): New pattern definition.
27502 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27505 * s390-protos.h (s390_expand_cs_hqi): Removed.
27506 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27507 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27508 modes as well as CCZ1mode and CCZmode.
27509 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27510 signature of s390_emit_compare_and_swap.
27511 (s390_expand_cs_hqi): Likewise, make static.
27512 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27513 compare-and-swap, in some cases.
27514 (s390_expand_cs): Wrapper function.
27515 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27517 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27518 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27519 patterns for small and large integers. Forbid symref memory operands.
27520 Move expander to s390.c. Require cc register.
27521 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27522 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27523 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27524 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27525 symref memory operands. Remove CC mode and call s390_match_ccmode
27527 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27529 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27531 * config/s390/s390.md (define_peephole2): New peephole to help
27532 combining the load-and-test pattern with volatile memory.
27534 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27536 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27537 with CCZmode for TARGET_Z196.
27539 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27541 PR rtl-optimization/80501
27542 * combine.c (make_compound_operation_int): Set subreg_code to SET
27543 even for AND with mask of the sign bit of mode.
27545 PR rtl-optimization/80500
27546 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27547 sum's initial value.
27549 2017-04-25 Julian Brown <julian@codesourcery.com>
27550 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27552 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27554 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27556 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27558 2017-04-25 Julian Brown <julian@codesourcery.com>
27559 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27561 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27562 (thunderx2t99_sha): New Reservation.
27564 2017-04-25 Julian Brown <julian@codesourcery.com>
27565 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27567 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27568 type for 1-element load.
27570 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27572 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27574 2017-04-24 Martin Jambor <mjambor@suse.cz>
27576 PR tree-optimization/80293
27577 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27578 char arrays not totally scalarizable if it is false.
27579 (analyze_all_variable_accesses): Pass correct value in the new
27580 parameter. Add a statistics counter.
27582 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27584 PR middle-end/79931
27585 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27587 2017-04-24 Richard Biener <rguenther@suse.de>
27589 PR tree-optimization/80494
27590 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27591 out for complex types.
27593 2017-04-24 Richard Biener <rguenther@suse.de>
27595 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27596 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27597 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27598 (DFS): Adjust and never fail.
27599 (sccvn_dom_walker::fail): Remove.
27600 (sccvn_dom_walker::before_dom_children): Adjust.
27601 (run_scc_vn): Likewise and never fail.
27602 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27603 (pass_fre::execute): Likewise.
27605 2017-04-24 Richard Biener <rguenther@suse.de>
27607 PR tree-optimization/79725
27608 * tree-ssa-sink.c (statement_sink_location): Return whether
27609 failure reason was zero uses. Move that check later.
27610 (sink_code_in_bb): Deal with zero uses by removing the stmt
27613 2017-04-24 Richard Biener <rguenther@suse.de>
27616 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27617 pointer-based references.
27619 2017-04-24 Richard Biener <rguenther@suse.de>
27622 * pass_manager.h (pass_manager::operator new): Remove.
27623 (pass_manager::operator delete): Likewise.
27624 * passes.c (pass_manager::operator new): Remove.
27625 (pass_manager::operator delete): Likewise.
27626 (pass_manager::pass_manager): Zero individual pass members.
27628 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27631 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27632 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27633 Check "XEXP (src, 1)" operand here.
27634 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27635 Check "XEXP (src, 1)" operand here.
27636 (dimode_scalar_chain::make_vector_copies): Detect count register
27637 of a shift instruction. Zero extend count register from QImode
27638 to DImode to satisfy vector shift pattern count operand predicate.
27639 Substitute vector shift count operand with a DImode copy.
27640 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27643 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27645 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27646 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27647 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27648 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27649 (UNSPEC_NOREX_MEM): Remove definition.
27651 2017-04-21 Richard Biener <rguenther@suse.de>
27653 PR tree-optimization/79547
27654 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27655 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27656 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27657 without any constraints.
27659 2017-04-21 Richard Biener <rguenther@suse.de>
27661 PR tree-optimization/78847
27662 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27664 2017-04-21 Richard Biener <rguenther@suse.de>
27666 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27667 (build_distinct_type_copy): Likewise.
27668 (build_variant_type_copy): Likewise.
27669 * tree.c (build_qualified_type): Pass down mem-stat info.
27670 (build_distinct_type_copy): Likewise.
27671 (build_variant_type_copy): Likewise.
27673 2017-04-21 Richard Biener <rguenther@suse.de>
27675 PR tree-optimization/80237
27676 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27678 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27679 for a simplified result.
27681 2016-04-21 Richard Biener <rguenther@suse.de>
27683 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27684 sth as strict as a simple_iv but a chrec without symbols and an
27685 operand defined in the loop we are peeling (and not some subloop).
27686 (propagate_constants_for_unrolling): Propagate all constants.
27688 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27691 * config/i386/i386.c (print_reg): Remove assert for disalowed
27692 regno values, call output_operand_lossage instead.
27694 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27697 * config/i386/constraints.md (Yc): New register constraint.
27698 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27699 Use Yc constraint for alternative 2 of operand 0. Remove
27700 preferred_for_speed attribute.
27702 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27704 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27705 lastprivate clauses in SIMT case.
27707 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27709 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27711 2017-04-20 Richard Biener <rguenther@suse.de>
27713 PR tree-optimization/57796
27714 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27715 as N scalar stores.
27716 (vect_model_load_cost): Cost gathers as N scalar loads.
27718 2017-04-20 Richard Biener <rguenther@suse.de>
27720 * ggc-page.c (ggc_allocated_p): Rename to ...
27721 (safe_lookup_page_table_entry): ... this and return the lookup
27723 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27725 2017-04-20 Richard Biener <rguenther@suse.de>
27727 PR tree-optimization/80453
27728 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27729 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27730 from the conditions.
27731 (vn_phi_eq): Pass them down.
27732 (vn_phi_lookup): Record them.
27733 (vn_phi_insert): Likewise.
27735 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27737 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27738 uninitialized variable warning to avoid buffer overrun.
27740 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27743 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27744 is suppressed for '{ 0 }' in C.
27746 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27748 * BASE-VER: Set to 8.0.0.
27750 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27752 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27753 priority .init_array and .fini_array section with SECTION_NOTYPE
27756 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27758 PR middle-end/80423
27759 * tree.h (build_array_type): Add typeless_storage default argument.
27760 * tree.c (type_cache_hasher::equal): Also compare
27761 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27762 (build_array_type): Add typeless_storage argument, set
27763 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27765 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27766 (build_array_type): Likewise. Add typeless_storage argument.
27768 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27769 Jakub Jelinek <jakub@redhat.com>
27771 PR tree-optimization/80426
27772 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27773 operation on symbolic operands, also compute the overflow for the
27774 invariant part when the operation degenerates into a negation.
27776 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27779 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27780 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27783 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27785 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27788 * config/avr/avr.c (tree.h): Include it.
27789 (cgraph.h): Include it.
27790 (avr_encode_section_info): Don't warn for uninitialized progmem
27791 variable if it's just an alias.
27793 2017-04-19 Richard Biener <rguenther@suse.de>
27796 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27797 when needed by AutoPGO.
27799 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27802 * doc/lto.texi: Remove an extra 'that'.
27804 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27806 PR rtl-optimization/80429
27807 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27808 are only used in debug insns.
27810 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27811 Vladimir Makarov <vmakarov@redhat.com>
27813 * config/sparc/predicates.md (input_operand): Add comment. Return
27814 true for any memory operand when LRA is in progress.
27815 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27817 2017-04-18 Jeff Law <law@redhat.com>
27820 * mips.md ({return,simple_return}_internal): Do not overwrite
27823 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27825 PR tree-optimization/80443
27826 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27827 instead of adding 1, subtract -1 and similarly instead of subtracting
27830 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27832 PR rtl-optimization/80357
27833 * haifa-sched.c (tmp_bitmap): New variable.
27834 (model_recompute): Handle duplicate use records.
27835 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27836 (free_global_sched_pressure_data): Free it.
27838 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27841 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27842 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27843 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27844 instead of SYSTEM_HEADER_DIR.
27846 2017-04-18 Jeff Law <law@redhat.com>
27848 PR middle-end/80422
27849 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27850 predecessors after walking up the insn chain.
27852 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27855 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27856 sizetype type into debug info.
27858 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27861 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27862 unneeded test for TARGET_UPPER_REGS_SF.
27863 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27865 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27868 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27869 instead of gsi_after_labels.
27871 2017-04-18 Jeff Law <law@redhat.com>
27873 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27877 2017-04-13 Jeff Law <law@redhat.com>
27878 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27879 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27881 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27884 * config/avr/avr.c (intl.h): Include it.
27885 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27887 2017-04-18 Martin Liska <mliska@suse.cz>
27889 PR gcov-profile/78783
27890 * gcov-tool.c (gcov_output_files): Validate that destination
27891 file is either removed by the tool or by a user.
27893 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27894 Guy Benyei <guybe@mellanox.com>
27896 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27897 block, and do not negate it, the stored id is already negative.
27899 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27901 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27903 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27906 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27907 masks of options that should be turned off if the VSX vector
27908 options are turned off.
27909 (OTHER_P8_VECTOR_MASKS): Likewise.
27910 (OTHER_VSX_VECTOR_MASKS): Likewise.
27911 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27912 rs6000_disable_incompatible_switches to validate no type switches
27914 (rs6000_incompatible_switch): New function to disallow turning on
27915 other vector options if -mno-vsx, -mno-power8-vector, or
27916 -mno-power9-vector are specified.
27918 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27920 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27922 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27924 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27925 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27926 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27927 (ARG_POINTER_CFA_OFFSET): Likewise.
27929 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27931 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27932 conditions to take advantage of various optimizations.
27934 2017-04-13 Jeff Law <law@redhat.com>
27936 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27937 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27938 (zero_extendsidi2_dext): Likewise.
27940 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27943 * fold-const.c (fold_ternary_loc): Revert
27944 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27947 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27949 PR rtl-optimization/80343
27950 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27951 new scratch pseudo.
27953 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27956 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27957 to ubsan_encode_value.
27959 2017-04-13 Jeff Law <law@redhat.com>
27961 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27962 appearing in DEBUG_INSNs.
27964 2017-04-13 Martin Liska <mliska@suse.cz>
27966 PR gcov-profile/80413
27967 * gcov-io.c (gcov_write_string): Copy to buffer just when
27968 allocated size is greater than zero.
27970 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27973 * dwarf2out.c (decls_for_scope): Ignore declarations of
27974 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27976 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27979 * ipa-visibility.c (non_local_p): Fix typos.
27980 (localize_node): When localizing symbol in same comdat group,
27981 dissolve the group only when we know external symbols are going
27983 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27985 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27987 PR tree-optimization/79390
27988 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27989 order does not result in usable sequence, retry with reversed operand
27995 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27996 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27997 op0 instead of fold_convert_loc (loc, type, arg0).
27999 2017-04-12 Jeff Law <law@redhat.com>
28001 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
28002 has a delay slot in the generated code.
28004 * config/cris/cris.md (cris_preferred_reload_class): Return
28005 GENNONACR_REGS rather than GENERAL_REGS.
28007 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28010 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
28011 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
28012 signedness of the result type.
28014 2017-04-12 Richard Biener <rguenther@suse.de>
28015 Jeff Law <law@redhat.com>
28017 PR tree-optimization/80359
28018 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
28019 trim stores to TARGET_MEM_REFs.
28021 2017-04-12 Richard Biener <rguenther@suse.de>
28023 PR tree-optimization/79390
28024 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
28025 threading case even more.
28027 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
28030 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
28031 for quad_address_p for TImode, instead of just not indexed_address.
28033 2017-04-12 Richard Biener <rguenther@suse.de>
28034 Bernd Edlinger <bernd.edlinger@hotmail.de>
28036 PR middle-end/79671
28037 * alias.c (component_uses_parent_alias_set_from): Handle
28038 TYPE_TYPELESS_STORAGE.
28039 (get_alias_set): Likewise.
28040 * tree-core.h (tree_type_common): Add typeless_storage flag.
28041 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28042 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28043 for types containing members with TYPE_TYPELESS_STORAGE.
28044 (place_field): Likewise.
28045 (layout_type): Likewise for ARRAY_TYPE.
28046 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28047 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28048 TYPE_TYPELESS_STORAGE.
28049 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28051 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28054 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28055 first argument to type.
28057 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28061 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28062 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28063 (rs6000_expand_binop_builtin): Likewise.
28064 (rs6000_expand_ternop_builtin): Likewise; also add missing
28065 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28066 vshasigma built-ins.
28067 * doc/extend.texi: Document that vec_xxpermdi's third argument
28068 must be a constant.
28070 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28072 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28073 Use shift_const cost parameter when calculating gain of STV shifts.
28075 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28077 PR rtl-optimization/70478
28078 * lra-constraints.c (process_alt_operands): Check memory for
28079 disfavoring memory insn operand.
28081 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28083 PR middle-end/80100
28084 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28085 left shift in unsigned HOST_WIDE_INT type.
28087 PR rtl-optimization/80385
28088 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28089 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28092 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28093 if they have any depend clauses.
28095 2017-04-11 Martin Liska <mliska@suse.cz>
28098 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28099 * ipa-split.c (split_function): Create a local comdat symbol
28100 if caller is in a comdat group.
28102 2017-04-11 Martin Liska <mliska@suse.cz>
28105 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28108 2017-04-11 Martin Sebor <msebor@redhat.com>
28110 PR middle-end/80364
28111 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28112 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28114 (directive::set_width, directive::set_precision, format_character):
28116 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28119 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28122 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28123 conflict, set target->arch_name instead of target->cpu_name.
28125 2017-04-11 Richard Biener <rguenther@suse.de>
28127 PR tree-optimization/80374
28128 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28129 build_zero_cst, remove fold_convertible_p check again.
28131 2017-04-11 Martin Liska <mliska@suse.cz>
28134 * ubsan.c (instrument_object_size): Do not instrument register
28137 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28140 * config/i386/i386-builtin-types.def
28141 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28142 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28143 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28144 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28145 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28146 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28147 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28148 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28149 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28150 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28151 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28152 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28153 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28154 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28155 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28156 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28157 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28158 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28159 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28160 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28161 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28162 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28163 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28164 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28165 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28166 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28167 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28168 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28169 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28170 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28171 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28172 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28173 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28174 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28175 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28176 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28177 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28178 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28179 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28180 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28181 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28182 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28183 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28184 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28185 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28186 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28188 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28189 flag to second_arg_count, handle 4 argument function type _COUNT
28190 aliases, handle second_arg_count on second argument rather than last.
28192 2017-04-10 Jeff Law <law@redhat.com>
28194 PR tree-optimization/80374
28195 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28196 record anything if we can not convert integer_zero_node to the
28199 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28202 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28203 Enhance special handling given to the TARGET_P9_MINMAX option in
28204 relation to certain other options.
28206 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28208 PR tree-optimization/80153
28209 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28210 remove POINTER_PLUS_EXPR's base part directly, rather than through
28213 2017-04-10 Richard Biener <rguenther@suse.de>
28214 Bin Cheng <bin.cheng@arm.com>
28216 PR tree-optimization/80153
28217 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28218 the first element of pointer type aff_tree. Build result expr in
28220 (add_elt_to_tree): Convert to type unconditionally. Remove other
28221 fold_convert calls.
28222 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28223 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28225 2017-04-10 Richard Biener <rguenther@suse.de>
28227 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28230 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28232 PR rtl-optimization/70478
28233 * lra-constraints.c (curr_small_class_check): New.
28234 (update_and_check_small_class_inputs): New.
28235 (process_alt_operands): Update curr_small_class_check. Disfavor
28236 alternative insn memory operands. Check available regs for small
28239 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28242 * config/mips/mips.opt (-mvirt): Update description.
28243 * doc/invoke.texi (-mvirt): Likewise.
28245 2017-04-10 Richard Biener <rguenther@suse.de>
28247 PR middle-end/80362
28248 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28249 looking for NEGATE_EXPR in -A / -B to A / B folding.
28251 2017-04-10 Martin Liska <mliska@suse.cz>
28253 PR gcov-profile/80224
28254 * gcov.c (print_usage): Fix usage string.
28255 (get_gcov_intermediate_filename): Remove.
28256 (output_gcov_file): Use both for normal and intermediate format.
28257 (generate_results): Do not initialize special file for
28258 intermediate format.
28260 2017-04-10 Richard Biener <rguenther@suse.de>
28262 PR tree-optimization/80304
28263 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28266 2017-04-10 Nathan Sidwell <nathan@acm.org>
28269 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28270 (rs6000_init_builtins): Use it.
28272 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28274 * config/arm/arm.md (<mrc>): Add mode to SET source.
28275 (<mrrc>): Likewise.
28277 2017-04-10 Richard Biener <rguenther@suse.de>
28279 PR middle-end/80344
28280 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28282 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28285 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28286 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28287 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28288 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28289 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28290 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28291 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28292 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28293 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28294 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28295 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28296 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28297 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28298 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28299 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28300 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28301 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28302 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28303 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28304 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28305 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28306 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28307 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28309 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28311 PR rtl-optimization/70478
28312 * lra-constraints.c: Reverse the last patch.
28314 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28316 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28317 Add comment for WCHAR_T.
28319 2017-04-08 Martin Liska <mliska@suse.cz>
28322 2017-04-07 Martin Liska <mliska@suse.cz>
28325 * ipa-split.c (split_function): Add function part to a same comdat
28328 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28331 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28333 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28335 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28336 (vec_pairing): ...to this.
28337 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28338 (rs6000_sched_init): Adjust for name change.
28339 (struct rs6000_sched_context): Likewise.
28340 (rs6000_init_sched_context): Likewise.
28341 (rs6000_set_sched_context): Likewise.
28343 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28349 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28351 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28352 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28353 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28355 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28357 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28359 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28361 PR rtl-optimization/70703
28362 * ira-color.c (update_conflict_hard_regno_costs): Use
28363 int64_t instead of HOST_WIDE_INT.
28365 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28367 PR rtl-optimization/70478
28368 * lra-constraints.c (process_alt_operands): Disfavor alternative
28369 insn memory operands.
28371 2017-04-07 Jeff Law <law@redhat.com>
28373 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28374 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28376 2017-04-07 Martin Liska <mliska@suse.cz>
28379 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28380 Show error message instead of an ICE.
28382 2017-04-07 Martin Liska <mliska@suse.cz>
28385 * ipa-split.c (split_function): Add function part to a same comdat
28388 2017-04-07 Richard Biener <rguenther@suse.de>
28390 PR middle-end/80341
28391 * tree.c (get_unwidened): Also handle ! for_type case for
28393 * convert.c (do_narrow): Split out from ...
28394 (convert_to_integer_1): ... here. Do not pass final truncation
28395 type to get_unwidened for TRUNC_DIV_EXPR.
28397 2017-04-07 Richard Biener <rguenther@suse.de>
28399 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28401 (aff_combination_const): Adjust.
28402 (aff_combination_scale): Likewise.
28403 (aff_combination_add_elt): Likewise.
28404 (aff_combination_add_cst): Likewise.
28405 (aff_combination_convert): Likewise.
28406 (add_elt_to_tree): Likewise. Remove unused argument.
28407 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28409 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28411 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28413 * config/arm/arm.c (arm_default_short_enums): Use
28414 ARM_DEFAULT_SHORT_ENUMS.
28415 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28417 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28420 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28421 members with redundant out-of-class redeclaration.
28423 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28426 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28427 * config/i386/i386.md (*zero_extendsidi2):
28428 Add (?*x,*x) and (?*v,*v) alternatives.
28430 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28433 * config/i386/i386.c (ix86_expand_builtin)
28434 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28435 mode from insn data. Convert operands to insn operand mode.
28436 Copy operands that don't satisfy insn predicate to a register.
28438 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28440 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28443 2017-04-06 Richard Biener <rguenther@suse.de>
28445 PR tree-optimization/80334
28446 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28447 preserve alignment of accesses.
28449 2017-04-06 Richard Biener <rguenther@suse.de>
28451 PR tree-optimization/80262
28452 * tree-sra.c (build_ref_for_offset): Preserve address-space
28454 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28455 Drop useless address-space information on MEM_REF offsets.
28457 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28459 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28461 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28463 PR rtl-optimization/70703
28464 * ira-color.c (update_conflict_hard_regno_costs): Use
28465 HOST_WIDE_INT instead of long.
28467 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28470 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28471 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28473 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28474 for x86_64 target. Handle -m3dnowa option.
28476 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28478 PR rtl-optimization/70703
28479 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28480 (update_conflict_hard_regno_costs): Use long instead of unsigned
28481 arithmetic for cost calculation.
28483 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28484 Bernd Edlinger <bernd.edlinger@hotmail.de>
28487 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28490 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28493 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28494 ptr_mode with Pmode throughout.
28495 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28496 into probe_stack_range and use DImode.
28498 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28501 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28502 call_eh_return is true.
28504 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28506 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28507 Initialize last_match_fntype_index.
28509 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28512 * tree-nvr.c: Include internal-fn.h.
28513 (pass_return_slot::execute): Ignore internal calls without
28516 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28517 Richard Biener <rguenther@suse.de>
28520 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28521 captures used multiple times, except for the last use.
28522 * generic-match-head.c: Include gimplify.h.
28524 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28526 PR tree-optimization/79390
28527 * target.h (struct noce_if_info): Declare.
28528 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28529 * target.def (noce_conversion_profitable_p): New target hook.
28530 * ifcvt.h (struct noce_if_info): New type, moved from ...
28531 * ifcvt.c (struct noce_if_info): ... here.
28532 (noce_conversion_profitable_p): Renamed to ...
28533 (default_noce_conversion_profitable_p): ... this. No longer
28535 (noce_try_store_flag_constants, noce_try_addcc,
28536 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28537 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28538 instead of noce_conversion_profitable_p.
28539 * config/i386/i386.c: Include ifcvt.h.
28540 (ix86_option_override_internal): Don't override
28541 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28542 (ix86_noce_conversion_profitable_p): New function.
28543 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28544 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28545 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28546 * doc/tm.texi: Regenerated.
28548 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28550 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28553 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28556 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28557 instructions for small multiply cores.
28559 2017-04-04 Jeff Law <law@redhat.com>
28561 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28563 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28564 that are not set by the loop over the elements.
28566 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28569 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28570 int mode, convert_modes it to mode as unsigned, otherwise use
28571 lowpart_subreg to mode rather than SImode.
28572 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28573 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28574 Use DImode instead of SImode for the shift count operand.
28575 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28578 2017-04-04 Richard Biener <rguenther@suse.de>
28580 PR middle-end/80281
28581 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28582 arithmetic done for the negate or the plus. Simplify.
28583 (A - (-B) -> A + B): Likewise.
28584 * fold-const.c (split_tree): Make sure to not negate pointers.
28586 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28588 PR rtl-optimization/60818
28589 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28590 a compare of comparisons with the thing compared if this results
28591 in a different machine mode.
28593 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28595 * alias.c (base_alias_check): Fix typo in comment.
28596 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28597 * cgraphunit.c (symbol_table::compile): Likewise.
28598 * collect2.c (maybe_run_lto_and_relink): Likewise.
28599 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28600 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28601 * config/avr/avr.c (avr_map_op_t): Likewise.
28602 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28603 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28604 * config/epiphany/epiphany.md (movcc): Likewise.
28605 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28606 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28608 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28609 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28610 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28611 * config/sh/sh.c (sh_rtx_costs): Likewise.
28612 * fold-const.c (fold_truth_andor): Likewise.
28613 * genautomata.c (collapse_flag): Likewise.
28614 * gengtype.h (struct type::u::s): Likewise.
28615 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28616 * input.c (FORMAT_AMOUNT): Likewise.
28617 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28618 (known_aggs_to_agg_replacement_list): Likewise.
28619 * ipa-inline-analysis.c: Likewise.
28620 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28621 * ipa-polymorphic-call.c
28622 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28623 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28624 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28626 * modulo-sched.c (apply_reg_moves): Likewise.
28627 * omp-expand.c (build_omp_regions_1): Likewise.
28628 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28629 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28630 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28631 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28632 * value-prof.c: Likewise.
28633 * var-tracking.c (val_reset): Likewise.
28635 2017-04-03 Richard Biener <rguenther@suse.de>
28637 PR tree-optimization/80275
28638 * fold-const.c (split_address_to_core_and_offset): Handle
28641 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28643 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28644 descriptors is at least equal to that of functions.
28646 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28648 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28650 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28653 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28654 (mov<IMOD4:mode>): New expander.
28655 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28657 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28659 PR rtl-optimization/79405
28660 * fwprop.c (propagations_left): New variable.
28661 (forward_propagate_into): Decrement it.
28662 (fwprop_init): Initialize it.
28663 (fw_prop): If the variable has reached zero, stop propagating.
28664 (fwprop_addr): Ditto.
28666 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28669 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28670 a FUNCTION_DECL, pass it as decl instead of origin to
28673 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28675 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28678 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28681 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28682 TARGET_VSX_SMALL_INTEGER.
28684 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28686 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28687 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28689 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28691 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28692 extraction from odd-numbered MSA register.
28694 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28696 PR middle-end/80173
28697 * expmed.c (store_bit_field_1): Don't attempt to create
28698 a word subreg out of hard registers wider than word if they
28699 have HARD_REGNO_NREGS of 1 for their mode.
28701 PR middle-end/80163
28702 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28703 conversions to integer types wider than word and pointer.
28706 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28707 (rtx_equal_for_cselib_p): Pass 0 to it.
28708 * cselib.c (cselib_hasher::equal): Likewise.
28709 (rtx_equal_for_cselib_1): Add depth argument. If depth
28710 is 128, don't look up VALUE locs and punt. Increment
28711 depth in recursive calls when walking VALUE locs.
28713 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28715 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28716 (make_gcov_file_name): Use the canonical path name for generating
28718 (read_line): Fix handling of files with ascii null bytes.
28720 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28722 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28723 to initialise a vector register instead
28724 of using a const_int.
28726 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28728 PR translation/80189
28729 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28730 diagnostic messages.
28732 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28735 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28736 (dfp_diex_<mode>): Update mode of operand 1.
28737 * doc/extend.texi (dxex, dxexq): Document change to return type.
28738 (diex, diexq): Document change to argument type.
28740 2017-03-30 Martin Jambor <mjambor@suse.cz>
28743 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28744 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28745 it reflects the signature changes performed at the callee side.
28746 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28747 to cgraph_build_function_type_skip_args.
28748 (build_function_decl_skip_args): Adjust call to the above function.
28750 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28753 * config/i386/sse.md
28754 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28755 register as dest whenever it is a MEM not rtx_equal_p to the
28756 corresponding dup operand, and when forcing into reg move the
28757 reg into the memory afterwards.
28758 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28759 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28760 for the force_reg mode.
28761 (avx512vl_vextractf128<mode>): Use register as dest either
28762 always when a MEM, or when it is a MEM not rtx_equal_p to the
28763 corresponding dup operand, or even not when it is a CONST_VECTOR
28764 depending on the mode and lo vs. hi.
28765 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28767 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28768 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28769 Likewise. Require that operands[2] is even.
28770 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28771 Remove extraneous parens. Require that operands[2] is a multiple
28773 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28774 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28775 disallow memory then.
28777 2017-03-30 Richard Biener <rguenther@suse.de>
28779 PR tree-optimization/77498
28780 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28781 to non-constants over backedges.
28783 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28785 PR rtl-optimization/80233
28786 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28787 as last_combined_insn. Do not test for BARRIER_P separately.
28789 2017-03-29 Andreas Schwab <schwab@suse.de>
28792 * calls.c (prepare_call_address): Convert funexp to Pmode before
28793 copying to temp reg.
28795 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28797 PR tree-optimization/80158
28798 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28799 Handle possible future case of more than one alternate
28801 (replace_rhs_if_not_dup): Likewise.
28802 (replace_one_candidate): Likewise.
28804 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28806 PR rtl-optimization/80193
28807 * ira.c (ira): Do not check allocation for LRA.
28809 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28811 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28812 (nvptx_output_simt_exit): Declare.
28813 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28814 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28815 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28816 (nvptx_declare_function_name): ...here. Emit declaration of local
28817 memory space buffer for omp_simt_enter insn.
28818 (nvptx_output_unisimt_switch): New.
28819 (nvptx_output_softstack_switch): New.
28820 (nvptx_output_simt_enter): New.
28821 (nvptx_output_simt_exit): New.
28822 * config/nvptx/nvptx.h (struct machine_function): New fields
28823 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28824 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28825 (UNSPECV_SIMT_EXIT): Ditto.
28826 (omp_simt_enter_insn): New insn.
28827 (omp_simt_enter): New expansion.
28828 (omp_simt_exit): New insn.
28829 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28831 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28832 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28833 (expand_GOMP_SIMT_EXIT): New.
28834 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28835 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28836 (GOMP_SIMT_EXIT): Ditto.
28837 * target-insns.def (omp_simt_enter): New insn.
28838 (omp_simt_exit): Ditto.
28839 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28841 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28842 (lower_rec_input_clauses): Likewise.
28843 (lower_lastprivate_clauses): Handle SIMT privatization.
28845 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28846 (ompdevlow_adjust_simt_enter): New.
28847 (find_simtpriv_var_op): New.
28848 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28849 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28851 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28852 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28853 (copy_decl_for_dup_finish): Ditto.
28855 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28857 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28860 * config/i386/i386.c (ix86_option_override_internal): Always
28861 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28863 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28865 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28867 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28869 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28870 mark new edge's irreducible flag accordign to it.
28871 (vect_do_peeling): Check loop preheader edge's irreducible flag
28872 and pass it to function slpeel_add_loop_guard.
28874 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28876 PR tree-optimization/80218
28877 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28878 Update block frequencies and counts.
28880 2017-03-28 Richard Biener <rguenther@suse.de>
28882 PR tree-optimization/78644
28883 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28884 of a simplification result we may not use it at all.
28886 2017-03-28 Richard Biener <rguenther@suse.de>
28889 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28890 without arguments, generate default definition of a SSA name.
28892 2017-03-28 Richard Biener <rguenther@suse.de>
28894 PR middle-end/80222
28895 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28896 TYPE_REF_CAN_ALIAS_ALL references.
28897 * fold-const.c (fold_indirect_ref_1): Likewise.
28899 2017-03-28 Martin Liska <mliska@suse.cz>
28902 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28903 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28905 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28906 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28908 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28909 (EXTRA_SPECS): Define.
28910 (SUBTARGET_EXTRA_SPECS): Likewise.
28911 (SUBTARGET_CPP_SPEC): Likewise.
28912 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28913 SUBTARGET_EXTRA_SPECS.
28914 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28916 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28918 * config/arc/simdext.md (vst64_insn): Update pattern.
28919 (vld32wh_insn): Likewise.
28920 (vld32wl_insn): Likewise.
28921 (vld64_insn): Likewise.
28922 (vld32_insn): Likewise.
28924 2017-03-28 Marek Polacek <polacek@redhat.com>
28927 * fold-const.c (fold_comparison): Use protected_set_expr_location
28928 instead of SET_EXPR_LOCATION.
28930 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28932 * tree.c (add_expr): Avoid name lookup warning.
28934 2017-03-27 Jeff Law <law@redhat.com>
28936 PR tree-optimization/80216
28937 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28938 function name. Limit recursion depth.
28939 (record_temporary_equivalences): Corresponding changes.
28941 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28943 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28946 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28949 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28951 * cfgcleanup.c (reg_note_cfa_p): New array.
28952 (insns_have_identical_cfa_notes): New function.
28953 (old_insns_match_p): Don't cross-jump in between /f
28954 and non-/f instructions. If both i1 and i2 are frame related,
28955 verify all CFA notes, their order and content.
28957 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28960 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28961 HImode and SImode with zero extend to DImode to one insn.
28962 (bswap<mode>2_extenddi): Likewise.
28963 (bswapsi2_extenddi): Likewise.
28964 (bswaphi2_extendsi): Likewise.
28965 (bswaphi2): Combine bswap HImode and SImode into one insn.
28966 Separate memory insns from swapping register.
28967 (bswapsi2): Likewise.
28968 (bswap<mode>2): Likewise.
28969 (bswaphi2_internal): Delete, no longer used.
28970 (bswapsi2_internal): Likewise.
28971 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28972 store, and gpr<-gpr swap insns.
28973 (bswap<mode>2_store): Likewise.
28974 (bswaphi2_reg): Register only splitter, combine with the splitter.
28975 (bswaphi2 splitter): Likewise.
28976 (bswapsi2_reg): Likewise.
28977 (bswapsi2 splitter): Likewise.
28978 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28979 the insns into load, store, and register/register insns.
28980 (bswapdi2_ldbrx): Likewise.
28981 (bswapdi2_load): Likewise.
28982 (bswapdi2_store): Likewise.
28983 (bswapdi2_reg): Likewise.
28985 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28987 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28988 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28990 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28993 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28996 special handling for target option conflicts between dform
28997 options (-mpower9-dform, -mpower9-dform-vector,
28998 -mpower9-dform-scalar) and -mno-direct-move.
29000 2017-03-27 Richard Biener <rguenther@suse.de>
29002 PR tree-optimization/80181
29003 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
29005 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29007 * config/arc/predicates.md (move_double_src_operand): Replace the
29008 call to move_double_src_operand with a call to address_operand.
29010 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29012 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
29013 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
29014 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
29016 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29018 * config/arc/predicates.md (long_immediate_loadstore_operand):
29019 Consider scaled addresses cases.
29021 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
29023 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
29024 restored when in interrupt.
29025 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
29026 doesn't have delay slot.
29028 2017-03-27 Richard Biener <rguenther@suse.de>
29031 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
29032 inlined thunk clones.
29034 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29037 * asan.c (instrument_derefs): Copy over last operand from
29038 original COMPONENT_REF to the new COMPONENT_REF with
29039 DECL_BIT_FIELD_REPRESENTATIVE.
29040 * ubsan.c (instrument_object_size): Likewise.
29042 2017-03-27 Richard Biener <rguenther@suse.de>
29044 PR tree-optimization/80170
29045 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29046 sure DR/SCEV didnt fold in constants we do not see when looking
29047 at the reference base alignment.
29049 2017-03-27 Richard Biener <rguenther@suse.de>
29051 PR middle-end/80171
29052 * gimple-fold.c (fold_ctor_reference): Properly guard against
29053 NULL return value from canonicalize_constructor_val.
29055 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29058 * config/i386/i386.c (ix86_expand_builtin)
29059 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29060 flags reg setting and flags reg using instructions.
29061 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29062 clobbering instructions to zero extend op2.
29064 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29066 * doc/install.texi (Configuration) <--with-aix-soname>:
29067 Update link to AIX ld.
29069 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29071 PR rtl-optimization/80160
29072 PR rtl-optimization/80159
29073 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29074 reg_alternate_class into account.
29076 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29079 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29080 to consider in curr_insn_transform.
29082 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29084 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29085 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29086 and emit_mode_inner.
29088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29090 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29091 argument to the overloaded builtin variants. Use the new flag to
29092 deprecate certain builtin variants.
29093 * config/s390/s390-builtin-types.def: Add new builtin types.
29094 * config/s390/s390-builtins.h: Support new flags field for
29095 overloaded builtins.
29096 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29097 (s390_macro_to_expand): Enable vector float data type.
29098 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29099 builtins by incrementing the __VEC__ version number.
29100 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29102 (s390_resolve_overloaded_builtin): Emit error messages depending
29103 on the builtin flags.
29104 * config/s390/s390.c (s390_expand_builtin): Support additional
29105 flags argument. Change error message to match the messages
29106 emitted in s390-c.c.
29107 * config/s390/s390.md: New UNSPEC_* constants.
29108 (op_type): Add new instruction types.
29109 * config/s390/vecintrin.h: Add new builtins and test data class
29111 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29112 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29113 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29114 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29115 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29116 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29118 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29119 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29120 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29121 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29123 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29124 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29125 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29126 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29128 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29129 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29130 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29131 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29132 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29133 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29134 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29136 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29137 ("vec_scatter_element<V_HW_4:mode>_DI")
29138 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29139 ("vec_fpint<mode>", "vflls")
29140 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29141 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29142 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29143 ("*vec_cmphe<mode>_cc"): ... these.
29145 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29146 mode constant instead of magic value.
29148 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29150 * config/s390/s390.c (s390_expand_vec_compare): Support other
29151 vector floating point modes than just V2DF.
29152 (s390_expand_vcond): Likewise.
29153 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29154 (s390_cannot_change_mode_class): Prevent mode changes between TF
29155 and V1TF in vector registers.
29156 * config/s390/s390.md (DF, SF): New mode attributes.
29157 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29158 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29159 SFmode support for VRs.
29160 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29162 (VFT, VF_HW): New mode iterators.
29163 (vw, sdx): New mode attributes.
29164 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29165 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29166 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29167 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29168 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29169 also the new vector floating point modes. Renaming to ...
29171 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29172 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29173 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29174 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29175 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29176 ("vec_unordered<mode>"): ... these.
29178 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29179 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29180 ("*vec_extendv2df"): New insn definitions.
29182 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29184 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29185 ("mulditi3_2", "*muldi3_sign"): New patterns.
29186 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29187 rename the pattern definition.
29189 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29191 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29193 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29195 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29197 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29198 instruction if possible.
29199 * config/s390/vector.md (vec_halfnumelts): New mode
29201 ("*vec_vllezlf<mode>"): New pattern.
29203 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29205 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29206 ("popcountv4si2", "popcountv2di2"): Rename to ...
29207 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29208 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29210 ("popcount<mode>2_vxe"): New pattern.
29212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29214 * common/config/s390/s390-common.c (processor_flags_table): Add
29216 * config.gcc: Add arch12.
29217 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29218 Default to arch12 for unknown CPU model numbers.
29219 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29220 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29221 PROCESSOR_max sanity check.
29222 * config/s390/s390-opts.h (enum processor_type): Add
29224 * config/s390/s390.c (processor_table): Add arch12.
29225 (s390_expand_builtin): Add check for B_VXE flag.
29226 (s390_issue_rate): Add PROCESSOR_ARCH12.
29227 (s390_get_sched_attrmask): Likewise.
29228 (s390_get_unit_mask): Likewise.
29229 (s390_sched_score): Enable z13 scheduling for arch12.
29230 (s390_sched_reorder): Likewise.
29231 (s390_sched_variable_issue): Likewise.
29232 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29234 (s390_tune_attr): Use z13 scheduling also for arch12.
29235 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29236 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29237 (TARGET_VXE_P): New macros.
29238 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29239 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29240 * config/s390/s390.opt: Add arch12 as processor_type.
29242 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29244 * config/s390/s390.md
29245 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29246 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29247 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29249 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29250 Rename expanders to ...
29252 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29253 ("fixuns_truncdddi2_emu"): ... these.
29255 ("fixuns_trunc<mode>si2_emu"): New expander.
29257 ("*fixuns_truncdfdi2_z13"): Rename to ...
29258 ("*fixuns_truncdfdi2_vx"): ... this.
29260 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29262 * config/s390/2964.md: Remove the single element vector compare
29263 instructions which are no longer used.
29264 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29266 (s390_canonicalize_comparison): Remove handling of DFmode
29268 (s390_expand_vec_compare_scalar): Remove function.
29269 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29270 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29272 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29274 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29276 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29278 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29279 will anyway by matched by mov<mode>_64dfp.
29281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29283 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29284 vlef/vstef. Add missing operand to vleif.
29286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29288 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29289 pair for all vector types with 64 bit elements.
29290 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29291 * config/s390/vector.md (V_HW_64): ... here.
29292 (V_128_NOSINGLE): New mode iterator.
29293 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29294 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29295 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29296 ("*vec_load_pairv2di"): Change to ...
29297 ("*vec_load_pair<mode>"): ... this one.
29299 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29301 * config/s390/constraints.md: Add comments.
29302 (jKK): Reject element sizes > 8 bytes.
29303 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29305 * config/s390/s390.md: Add the s_operand checks formerly in
29306 s390_split_ok_p to various splitters where they are still
29308 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29309 for 128 bit vectors. Plus two splitters.
29311 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29313 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29319 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29320 error if the boundary argument is not constant.
29322 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29324 PR rtl-optimization/80112
29325 * loop-doloop.c (doloop_condition_get): Don't check condition
29326 if cmp isn't SET with IF_THEN_ELSE src.
29328 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29330 PR tree-optimization/80158
29331 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29332 replacing a candidate statement, also replace it for the
29333 candidate's alternate interpretation.
29334 (replace_rhs_if_not_dup): Likewise.
29335 (replace_one_candidate): Likewise.
29337 2017-03-24 Richard Biener <rguenther@suse.de>
29339 PR tree-optimization/80167
29340 * graphite-isl-ast-to-gimple.c
29341 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29343 (translate_isl_ast_to_gimple::get_rename): Likewise.
29345 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29348 handling of certain combinations of target options, including the
29349 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29350 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29352 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29355 * config/arm/arm.md (*load_multiple): Add reload_completed to
29356 matching condition.
29358 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29359 Richard Biener <rguenth@suse.de>
29361 PR tree-optimization/79908
29362 PR tree-optimization/80136
29363 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29364 been cast away, gimplify_and_add suffices.
29366 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29368 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29370 2017-03-23 Richard Biener <rguenther@suse.de>
29372 PR tree-optimization/80032
29373 * gimplify.c (gimple_push_cleanup): Forced unconditional
29374 cleanups still have to go to the conditional_cleanups
29377 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29379 PR tree-optimization/80072
29380 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29382 (next_operand_entry_id): Change type to unsigned int.
29383 (sort_by_operand_rank): Make sure to return the right return value
29384 even if unsigned fields are bigger than INT_MAX.
29385 (struct oecount): Change cnt and id type to unsigned int.
29386 (oecount_hasher::equal): Formatting fix.
29387 (oecount_cmp): Make sure to return the right return value
29388 even if unsigned fields are bigger than INT_MAX.
29389 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29392 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29393 TREE_READONLY on result if writing it more than once.
29396 * doc/invoke.texi (-fsanitize=thread): Document that with
29397 -fnon-call-exceptions atomics are not able to throw
29401 * tsan.c: Include tree-eh.h.
29402 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29403 maybe_clean_or_replace_eh_stmt where needed.
29404 (instrument_memory_accesses): Add cfg_changed argument.
29405 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29406 if it returned true.
29407 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29409 PR rtl-optimization/63191
29410 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29411 wrapper function, moved the whole old content into ...
29412 (ix86_delegitimize_address_1): ... this. New inline function.
29413 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29414 true as last argument instead of ix86_delegitimize_address.
29416 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29418 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29419 cortexa57_branch_cost.
29421 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29423 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29425 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29428 * doc/md.texi (Constraints): Document wA constraint.
29429 * config/rs6000/constraints.md (wA): New.
29430 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29431 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29432 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29433 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29435 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29438 * gimplify.c (is_oacc_declared): New function.
29439 (oacc_default_clause): Use it to set default flags for acc declared
29440 variables inside parallel regions.
29441 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29442 declared variables.
29443 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29444 declare attribute to any decl as necessary.
29446 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29449 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29450 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29451 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29452 (arm_arch_lpae): This.
29453 * config/arm/arm.c (arm_arch7ve): Rename into ...
29454 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29455 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29458 2017-03-22 Martin Liska <mliska@suse.cz>
29461 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29462 error message instead of an ICE.
29464 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29466 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29468 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29470 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29472 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29475 2017-03-21 Martin Sebor <msebor@redhat.com>
29477 * doc/extend.texi: Use "cannot" instead of "can't."
29478 * doc/hostconfig.texi: Same.
29479 * doc/install.texi: Same.
29480 * doc/invoke.texi: Same.
29481 * doc/loop.texi: Same.
29482 * doc/md.texi: Same.
29483 * doc/objc.texi: Same.
29484 * doc/rtl.texi: Same.
29485 * doc/tm.texi: Same.
29486 * doc/tm.texi.in: Same.
29487 * doc/trouble.texi: Same.
29489 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29492 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29493 (collect_checksum_attributes): Set it.
29494 (die_checksum_ordered): Use it.
29496 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29498 PR tree-optimization/79908
29499 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29500 change: For a VA_ARG whose LHS has been cast away, use
29501 force_gimple_operand to construct the side effects.
29503 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29505 PR translation/80001
29506 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29507 more amenable to translation.
29508 (oacc_loop_auto_partitions): Likewise.
29510 2017-03-21 Marek Polacek <polacek@redhat.com>
29511 Martin Sebor <msebor@redhat.com>
29513 PR tree-optimization/80109
29514 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29515 on INTEGRAL_TYPE_P.
29517 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29518 Segher Boessenkool <segher@kernel.crashing.org>
29521 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29522 check reg_used_between_p between insn and one of succ or succ2
29523 depending on if succ is artificial insn not inserted into insn
29526 2017-03-21 Martin Liska <mliska@suse.cz>
29528 PR gcov-profile/80081
29529 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29530 * doc/gcc.texi: Include gcov-dump stuff.
29531 * doc/gcov-dump.texi: New file.
29533 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29535 PR rtl-optimization/79150
29536 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29537 conditional jump, if the jump is the last insn of the loop.
29539 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29540 Richard Biener <rguenth@suse.de>
29542 PR tree-optimization/79908
29543 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29544 been cast away, use force_gimple_operand to construct the side
29547 2017-03-21 Martin Liska <mliska@suse.cz>
29549 PR libfortran/79956
29550 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29553 2017-03-21 Brad Spengler <spender@grsecurity.net>
29556 * plugin.c (htab_hash_plugin): New function.
29557 (add_new_plugin): Use it and adjust.
29558 (parse_plugin_arg_opt): Adjust.
29559 (init_one_plugin): Likewise.
29561 2017-03-21 Richard Biener <rguenther@suse.de>
29563 PR tree-optimization/80032
29564 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29565 if set force the cleanup to happen unconditionally.
29566 (gimplify_target_expr): Push inserted clobbers with force_uncond
29567 to avoid them being removed by control-dependent DCE.
29569 2017-03-21 Richard Biener <rguenther@suse.de>
29571 PR tree-optimization/80122
29572 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29573 va_arg_pack_len when the inlined call stmt requires pack
29575 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29577 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29580 * tsan.c (instrument_builtin_call): If the memory model argument
29581 is not a constant, assume it is valid.
29584 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29587 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29589 PR rtl-optimization/79910
29590 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29591 if its dest is used by an insn before I2 (other than the combined
29592 insns themselves, which are properly handled already).
29594 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29597 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29599 * combine.c (record_used_regs): New static function.
29600 (try_combine): Handle situations where there is an additional
29601 instruction between I2 and I3 which needs to have a LOG_LINK
29605 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29607 * combine.c (try_combine): Delete redundant i1 test. Call
29608 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29610 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29613 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29614 alternatives 13/14.
29616 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29618 PR tree-optimization/80054
29619 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29620 the optimization if a PHI or any of its arguments is not dominated
29621 by the candidate's basis. Use gphi* rather than gimple* as
29623 (replace_profitable_candidates): Clean up a gimple* variable that
29624 should be a gphi* variable.
29626 2017-03-20 Martin Sebor <msebor@redhat.com>
29629 * doc/extend.texi (attribute constructor): Document present limitation.
29631 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29634 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29635 __POWER9_VECTOR__ #ifdef control, change template definition to
29636 use Power9-specific built-in function.
29637 (vec_any_eq): Likewise.
29638 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29639 to control outcomes from this test.
29640 (vector_ae_<mode>p): For VEC_F modes, likewise.
29642 2017-03-20 Ian Lance Taylor <iant@google.com>
29644 * config/i386/i386.c (ix86_function_regparm): Save an extra
29645 register for -fsplit-stack with DECL_STATIC_CHAIN.
29647 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29650 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29651 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29653 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29655 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29657 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29660 2017-03-20 Marek Polacek <polacek@redhat.com>
29663 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29665 2017-03-20 Richard Biener <rguenther@suse.de>
29667 PR tree-optimization/80113
29668 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29669 allocate extra SSA name for PHI def.
29670 (add_close_phis_to_outer_loops): Likewise.
29671 (add_close_phis_to_merge_points): Likewise.
29672 (copy_loop_close_phi_args): Likewise.
29673 (copy_cond_phi_nodes): Likewise.
29675 2017-03-20 Martin Liska <mliska@suse.cz>
29677 PR middle-end/79753
29678 * tree-chkp.c (chkp_build_returned_bound): Do not build
29679 returned bounds for a LHS that's not a BOUNDED_P type.
29681 2017-03-20 Martin Liska <mliska@suse.cz>
29685 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29686 COMPLEX_CST and VECTOR_CST.
29688 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29691 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29692 target operand. A new splitter adds the clobber statement in case
29693 the target operand is dead anyway.
29695 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29697 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29698 to age-old versions of binutils and glibc.
29700 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29702 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29704 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29706 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29708 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29710 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29711 requirement for binutils 2.13.
29713 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29715 * combine.c (try_combine): Delete redundant i1 test. Call
29716 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29718 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29720 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29721 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29723 <riscv64-*-elf>: Re-arrange section
29724 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29725 <riscv32-*-linux>: Likewise.
29726 <riscv64-*-elf>: Likewise
29727 <riscv64-*-linux>: Likewise.
29729 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29732 * aarch64.opt(verbose-cost-dump): Fix typo.
29734 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29737 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29738 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29740 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29742 * reload.c (find_reloads): When reloading a nonoffsettable address,
29743 use RELOAD_OTHER for it and its address reloads.
29745 PR rtl-optimization/79910
29746 * combine.c (record_used_regs): New static function.
29747 (try_combine): Handle situations where there is an additional
29748 instruction between I2 and I3 which needs to have a LOG_LINK
29751 2017-03-17 Jeff Law <law@redhat.com>
29753 PR tree-optimization/71437
29754 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29755 conditional in the hash table first.
29756 (vrp_dom_walker::before_dom_children): Extract condition from
29757 ASSERT_EXPR. Record condition, its inverion and any implied
29758 conditions as well.
29760 2017-03-17 Marek Polacek <polacek@redhat.com>
29761 Markus Trippelsdorf <markus@trippelsdorf.de>
29763 PR tree-optimization/80079
29764 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29767 2017-03-17 Richard Biener <rguenther@suse.de>
29769 PR middle-end/80075
29770 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29771 Properly verify the LHS before the RHS possibly claims to be
29773 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29776 2017-03-17 Martin Jambor <mjambor@suse.cz>
29778 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29779 (List of -O2 options): Likewise.
29780 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29783 2017-03-17 Tom de Vries <tom@codesourcery.com>
29785 * gcov-dump.c (print_usage): Print bug_report_url.
29787 2017-03-17 Richard Biener <rguenther@suse.de>
29789 PR middle-end/80050
29790 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29791 (parser::peek): Likewise.
29793 2017-03-17 Richard Biener <rguenther@suse.de>
29795 PR tree-optimization/80048
29796 * sese.c (free_sese_info): Properly release rename_map and
29797 copied_bb_map elements.
29799 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29801 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29802 Add linked-list forward and backlinks. Insert on
29803 construction, remove on destruction.
29804 (class pass_store_merging): Add m_stores_head field.
29805 (pass_store_merging::terminate_and_process_all_chains):
29806 Iterate over m_stores_head list.
29807 (pass_store_merging::terminate_all_aliasing_chains):
29809 (pass_store_merging::execute): Check for debug stmts first.
29810 Push new chains onto the m_stores_head stack.
29812 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29815 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29816 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29817 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29819 2017-03-16 Jeff Law <law@redhat.com>
29821 PR tree-optimization/71437
29822 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29823 member function. Implementation moved into after_dom_children
29824 member function and into the threader's thread_outgoing_edges
29826 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29827 some code into new thread_outgoing_edges.
29828 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29829 definition. Simplify marker handling (do it here). Assume we always
29830 have the available expression and the const/copies tables.
29831 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29833 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29834 * tree-vrp.c (equiv_stack): No longer file scoped.
29835 (vrp_dom_walker): New class.
29836 (vrp_dom_walker::before_dom_children): New member function.
29837 (vrp_dom_walker::after_dom_children): Likewise.
29838 (identify_jump_threads): Setup domwalker. Use it rather than
29839 walking edges in a random order by hand. Simplify setup/finalization.
29840 (finalize_jump_threads): Remove.
29841 (vrp_finalize): Do not call identify_jump_threads here.
29842 (execute_vrp): Do it here instead and call thread_through_all_blocks
29845 PR tree-optimization/71437
29846 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29848 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29850 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29851 (dom_opt_dom_walker::thread_across_edge): Remove
29852 handle_dominating_asserts argument. All callers changed.
29853 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29854 changes. Remove calls to lhs_of_dominating_assert. Other
29855 uses of handle_dominating_asserts turn into unconditional code
29856 (simplify_control_stmt_condition_1): Likewise.
29857 (simplify_control_stmt_condition): Likewise.
29858 (thread_through_normal_block, thread_across_edge): Likewise.
29859 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29860 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29861 object if it is not an SSA_NAME.
29862 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29863 before calling into the VRP specific simplifiers.
29864 (identify_jump_threads): Remove handle_dominating_asserts
29867 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29870 * tree-diagnostic.c (default_tree_printer): No longer static.
29871 * tree-diagnostic.h (default_tree_printer): New prototype.
29873 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29875 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29876 Change ins into fmov.
29878 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29880 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29881 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29882 Use h_con constraint for operand 1.
29883 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29884 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29886 2017-03-15 Jeff Law <law@redhat.com>
29888 PR tree-optimization/71437
29889 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29890 (record_temporary_equivalences): Use it.
29892 PR tree-optimization/71437
29893 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29894 tree-ssa-scopedtables.
29895 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29896 (record_conditions, record_cond, vuse_eq): Likewise.
29897 (record_edge_info): Adjust to API tweak of record_conditions.
29898 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29899 (record_temporary_equivalences, optimize_stmt): Likewise.
29900 (eliminate_redundant_computations): Likewise.
29901 (record_equivalences_from_stmt): Likewise.
29902 * tree-ssa-scopedtables.c: Include options.h and params.h.
29903 (vuse_eq): New function, moved from tree-ssa-dom.c
29904 (build_and_record_new_cond): Likewise.
29905 (record_conditions): Likewise. Accept vector of conditions rather
29906 than edge_equivalence structure for first argument.
29907 for the first argument.
29908 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29909 from tree-ssa-dom.c.
29910 (avail_exprs_stack::record_cond): Likewise.
29911 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29912 from tree-ssa-dom.c.
29913 (avail_exprs_stack): Add new member functions lookup_avail_expr
29915 (record_conditions): Declare.
29917 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29920 * lra-constraints.c (process_alt_operands): Increase reject for
29921 reloading an input/output operand.
29923 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29926 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29927 insns to convert from signed/unsigned char/short to IEEE 128-bit
29929 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29931 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29934 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29935 subreg of inner mode for values already in registers.
29937 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29939 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29940 iteration reg is used after the loop.
29942 2017-03-14 Martin Sebor <msebor@redhat.com>
29944 PR tree-optimization/79800
29945 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29946 precision in negative-positive range.
29947 (format_floating): Call non-const overload with adjusted precision.
29949 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29952 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29955 2017-03-14 Martin Sebor <msebor@redhat.com>
29957 PR middle-end/80020
29958 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29959 * builtins.def (aligned_alloc): Use it.
29962 * Makefile.in (GTFILES): Add calls.c.
29963 * calls.c: Include "gt-calls.h".
29965 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29967 PR rtl-optimization/79728
29968 * regs.h (struct target_regs): New field
29969 x_contains_allocatable_regs_of_mode.
29970 (contains_allocatable_regs_of_mode): New macro.
29971 * reginfo.c (init_reg_sets_1): Initialize it, and change
29972 contains_reg_of_mode so it includes global regs as well.
29973 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29974 rather than contains_regs_of_mode.
29976 2017-03-14 Martin Liska <mliska@suse.cz>
29978 * doc/invoke.texi: Document options that can't be combined with
29979 -fcheck-pointer-bounds.
29981 2017-03-14 Martin Liska <mliska@suse.cz>
29983 PR middle-end/79831
29984 * doc/invoke.texi (-Wchkp): Document the option.
29986 2017-03-14 Martin Liska <mliska@suse.cz>
29988 * Makefile.in: Install gcov-dump.
29990 2017-03-14 Martin Liska <mliska@suse.cz>
29992 * multiple_target.c (expand_target_clones): Bail out for
29993 an invalid attribute.
29995 2017-03-14 Richard Biener <rguenther@suse.de>
29997 * alias.c (struct alias_set_entry): Pack properly.
29998 * cfgloop.h (struct loop): Likewise.
29999 * cse.c (struct set): Likewise.
30000 * ipa-utils.c (struct searchc_env): Likewise.
30001 * loop-invariant.c (struct invariant): Likewise.
30002 * lra-remat.c (struct cand): Likewise.
30003 * recog.c (struct change_t): Likewise.
30004 * rtl.h (struct address_info): Likewise.
30005 * symbol-summary.h (function_summary): Likewise.
30006 * tree-loop-distribution.c (struct partition): Likewise.
30007 * tree-object-size.c (struct object_size_info): Likewise.
30008 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
30009 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
30010 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
30011 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
30012 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
30013 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
30014 (struct _stmt_vec_info): Likewise.
30016 2017-03-14 Martin Liska <mliska@suse.cz>
30019 * multiple_target.c (create_dispatcher_calls): Check that
30020 a target can create a function dispatcher.
30022 2017-03-14 Martin Liska <mliska@suse.cz>
30025 * multiple_target.c (expand_target_clones): Drop local.local
30026 flag for default implementation.
30028 2017-03-14 Richard Biener <rguenther@suse.de>
30030 PR tree-optimization/80030
30031 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
30033 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
30035 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30036 gcc_fallthrough() instead of __attribute__((fallthrough));
30038 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30040 * doc/gcc.texi: Remove "up" link to (DIR).
30041 * doc/gccint.texi: Ditto.
30043 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30045 * doc/install.texi (Specific) <avr>: Remove reference to
30048 2017-03-13 Jeff Law <law@redhat.com>
30050 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30051 attribute rather than comments.
30053 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30054 match_scratch operand is highest.
30056 2017-03-13 Martin Liska <mliska@suse.cz>
30058 PR middle-end/78339
30059 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30060 is a CHKP clone, use original declaration.
30062 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30064 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30065 (arc_conditional_register_usage): Use a different allocation order
30066 when optimizing for size.
30067 * common/config/arc/arc-common.c (arc_option_optimization_table):
30068 Section anchors default on when optimizing for size.
30070 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30072 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30074 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30076 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30077 * config/arc/arc.md (cpu_facility): Add cd variant.
30078 (*movqi_insn): Add code density variant.
30079 (*movhi_insn): Likewise.
30080 (*movqi_insn): Likewise.
30081 (*addsi3_mixed): Likewise.
30082 (subsi3_insn): Likewise.
30084 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30086 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30088 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30090 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30091 expressions with MINUS and UNARY ops.
30093 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30096 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30098 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30099 between vec_select and vector argument.
30100 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30101 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30102 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30103 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30104 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30105 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30107 2017-03-13 Richard Biener <rguenther@suse.de>
30110 * params.def (vect-max-peeling-for-alignment): Fix typo.
30112 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30114 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30115 issue that only occurred with binutils below 2.18.
30117 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30119 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30120 refer to binutils 2.11/2.12 minimum.
30122 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30124 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30125 ftp.kernel.org and simplify binutils requirement.
30127 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30129 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30131 (Optimize Options): Ditto. Also remove redundancy.
30133 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30135 PR translation/79848
30136 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30138 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30139 to G_ to avoid double translation.
30141 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30143 PR translation/79923
30144 * auto-profile.c (get_combined_location): Convert leading
30145 character of diagnostics to lower case and remove trailing period.
30146 (read_profile): Likewise for various diagnostics.
30147 * config/arm/arm.c (arm_option_override): Remove trailing period
30148 from various diagnostics.
30149 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30150 (msp430_expand_delay_cycles): Likewise.
30152 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30155 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30156 full command-line argument, rather than just "str".
30157 (aarch64_validate_march): Likewise.
30158 (aarch64_validate_mtune): Likewise.
30160 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30162 PR rtl-optimization/78911
30163 * lra-assigns.c (must_not_spill_p): New function.
30164 (spill_for): Use it.
30166 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30168 PR tree-optimization/79981
30169 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30170 ATOMIC_COMPARE_EXCHANGE ifn result.
30171 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30172 IFN_ATOMIC_COMPARE_EXCHANGE.
30174 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30177 * opts.c (parse_sanitizer_options): Add missing question mark to
30178 "did you mean" message.
30180 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30182 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30184 (VMULEUH_UNS): Likewise.
30185 (VMULOUB_UNS): Likewise.
30186 (VMULOUH_UNS): Likewise.
30187 * config/rs6000/rs6000.c (builtin_function_type): Remove
30188 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30190 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30193 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30194 x with result of extra_parsing_for_operand_code_0.
30195 (function_reader::extra_parsing_for_operand_code_0): Convert
30196 return type from void to rtx, returning x. When reading
30197 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30198 larger size containing struct block_symbol.
30200 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30202 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30203 -mfloat128-hardware without -m64.
30205 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30208 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30209 entries to the case statement that marks unsigned arguments to
30210 overloaded functions.
30212 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30214 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30215 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30217 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30220 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30221 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30223 2017-03-10 Martin Liska <mliska@suse.cz>
30227 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30228 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30231 2017-03-10 Olivier Hainque <hainque@adacore.com>
30233 * tree-switch-conversion (array_value_type): Start by resetting
30234 candidate type to it's main variant.
30236 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30238 PR rtl-optimization/79909
30239 * combine.c (try_combine): Use simplify_replace_rtx on individual
30240 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30241 of the whole CALL_INSN_FUNCTION_USAGE.
30243 PR tree-optimization/79972
30244 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30245 get_range_info on SSA_NAMEs. Formatting fixes.
30247 2017-03-10 Richard Biener <rguenther@suse.de>
30248 Jakub Jelinek <jakub@redhat.com>
30250 PR tree-optimization/77975
30251 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30252 edge to be constant.
30253 (get_val_for): For constant x return it. Formatting fix.
30254 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30255 would use the same bases as the current one.
30257 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30259 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30260 instead of vec_select for V1TImode.
30261 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30263 (VSX_LE_128): Add V1TI to this mode iterator.
30264 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30265 (*vsx_le_perm_store_<mode>): Likewise.
30266 (pre-reload splitter for VSX stores): Likewise.
30267 (post-reload splitter for VSX stores): Likewise.
30268 (*vsx_xxpermdi2_le_<mode>): Likewise.
30269 (*vsx_lxvd2x2_le_<mode>): Likewise.
30270 (*vsx_stxvd2x2_le_<mode>): Likewise.
30272 2017-03-09 Michael Eager <eager@eagercon.com>
30274 Correct failures with --enable-checking=yes,rtl.
30276 * config/microblaze/microblaze.c (microblaze_expand_shift):
30277 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30278 test for const0_rtx.
30279 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30280 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30282 2017-03-09 Richard Biener <rguenther@suse.de>
30284 PR tree-optimization/79977
30285 * graphite-scop-detection.c (scop_detection::merge_sese):
30286 Handle the case of extra exits to blocks dominating the entry.
30288 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30290 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30293 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30295 PR rtl-optimization/79949
30296 * lra-constraints.c (process_alt_operands): Check memory when
30297 trying to predict a cycle. Print about the overall increase.
30299 2017-03-09 Richard Biener <rguenther@suse.de>
30301 PR middle-end/79971
30302 * gimple-expr.c (useless_type_conversion_p): Preserve
30303 TYPE_SATURATING for fixed-point types.
30305 2017-03-09 Richard Biener <rguenther@suse.de>
30308 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30309 alignment of BLKmode params.
30311 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30314 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30315 (VALL_NO_V2Q): Likewise.
30317 * config/aarch64/aarch64-simd.md
30318 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30320 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30321 VALL_NO_V2Q mode iterator.
30322 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30324 2017-03-09 Martin Liska <mliska@suse.cz>
30326 PR tree-optimization/79631
30327 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30328 tree_int_cst_sign_bit just for INTEGER constants.
30330 2017-03-09 Martin Liska <mliska@suse.cz>
30334 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30337 2017-03-09 Marek Polacek <polacek@redhat.com>
30340 * tree.c (inchash::add_expr): Handle TREE_VEC.
30342 2017-03-09 Martin Liska <mliska@suse.cz>
30345 (chkp_narrow_size_and_offset): New function.
30346 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30347 (void chkp_parse_bit_field_ref): New function.
30348 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30349 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30351 2017-03-09 Martin Liska <mliska@suse.cz>
30354 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30355 (chkp_find_bounds_1): Remove gcc_unreachable.
30357 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30360 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30361 BUILT_IN_SYNC*, determine the access type from the size suffix and
30362 always build a MEM_REF with that type. Handle forgotten
30363 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30366 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30367 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30368 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30369 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30370 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30371 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30372 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30373 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30374 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30375 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30376 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30377 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30378 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30379 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30380 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30381 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30382 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30383 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30384 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30385 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30386 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30387 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30388 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30389 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30390 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30391 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30392 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30393 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30394 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30395 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30396 definitions outside of __OPTIMIZE__ guarded section.
30399 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30400 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30401 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30402 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30405 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30407 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30408 ("vfenez<mode>"): Add missing constraints.
30410 2017-03-08 Martin Sebor <msebor@redhat.com>
30413 * config/nds32/nds32.c (nds32_option_override):
30414 Fix misspelled diagnostic.
30416 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30419 * gimplify.c (gimplify_omp_for): Replace index var in outer
30420 taskloop statement with an artificial variable and add
30421 OMP_CLAUSE_PRIVATE clause for it.
30423 2017-03-08 Richard Biener <rguenther@suse.de>
30425 PR tree-optimization/79955
30426 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30427 for accesses that are completely outside of the variable.
30429 2017-03-08 Andrew Haley <aph@redhat.com>
30431 PR tree-optimization/79943
30432 * tree-ssa-loop-split.c (compute_new_first_bound): When
30433 calculating the new upper bound, (END-BEG) should be added, not
30436 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30438 * config/avr/avr.md (setmemhi): Make sure match_dup
30439 operand number comes before match_scratch.
30441 2017-03-08 Richard Biener <rguenther@suse.de>
30443 PR tree-optimization/79920
30444 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30445 with ncopies == 1 to ...
30446 (vect_transform_slp_perm_load): ... here. Properly compute
30447 all element loads by iterating VF times over the group. Do
30448 not handle ncopies (computed in a broken way) in
30449 vect_create_mask_and_perm.
30451 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30454 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30455 is a uniform vector, use uniform_vector_p return value instead of
30456 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30458 2017-03-07 Marek Polacek <polacek@redhat.com>
30460 PR middle-end/79809
30461 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30462 (alloca_call_type): Likewise.
30464 2017-03-07 Martin Liska <mliska@suse.cz>
30466 * gcov.c (process_args): Put comment to correct location.
30468 2017-03-07 Martin Liska <mliska@suse.cz>
30470 PR middle-end/68270
30471 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30472 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30473 (chkp_narrow_bounds_for_field): Likewise.
30474 (chkp_parse_array_and_component_ref): Pass one more argument to
30477 2017-03-07 Richard Biener <rguenther@suse.de>
30479 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30482 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30484 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30485 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30487 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30490 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30491 to end of description.
30492 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30494 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30496 PR rtl-optimization/79901
30497 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30499 (*avx512f_<code><mode>3<mask_name>): ... this.
30500 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30501 iterator instead of VI8_AVX2_AVX512BW.
30503 PR rtl-optimization/79901
30504 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30505 min/max expander, expand it using expand_vec_cond_expr.
30508 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30511 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30514 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30515 to void * for PCH reasons.
30516 * dwarf2out.c (output_loc_operands, output_die): Cast
30517 v.val_vec.array to unsigned char *.
30519 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30522 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30525 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30527 PR rtl-optimization/79571
30528 * lra-constraints.c (process_alt_operands): Calculate static
30529 reject and subtract it from overall when only addresses will be
30532 2017-03-06 Julia Koval <julia.koval@intel.com>
30535 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30536 incoming stack boundary to 128 for 64-bit targets.
30538 2017-03-06 Richard Biener <rguenther@suse.de>
30540 PR tree-optimization/79894
30541 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30542 to NULL after folding it.
30544 2017-03-06 Richard Biener <rguenther@suse.de>
30546 PR tree-optimization/79824
30547 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30548 check disabling peeling for gaps.
30550 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30552 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30553 attributes): Document gettimeofday.
30555 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30557 * config/s390/s390.c (s390_option_override_internal): Set
30558 PARAM_MIN_VECT_LOOP_BOUND
30560 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30562 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30563 * config/s390/s390.md: Likewise.
30565 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30568 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30569 (<avx2_avx512>_perm<mode>): Rename to ...
30570 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30572 (<avx512>_perm<mode>_mask): Rename to ...
30573 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30575 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30576 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30577 instead of VI8F_256_512.
30578 (avx512f_perm<mode>): New define_expand.
30579 (avx512f_perm<mode>_mask): Likewise.
30580 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30581 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30583 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30585 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30586 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30588 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30590 2017-03-06 Martin Liska <mliska@suse.cz>
30593 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30594 when having a SSA NAME w/o VAR_DECL assigned to it.
30596 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30598 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30599 msa_dpsub_<su>_d): Fix MODE for vec_select.
30601 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30603 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30605 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30607 2017-03-06 Richard Biener <rguenther@suse.de>
30609 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30610 * plugin.c (register_plugin_info): Likewise.
30611 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30613 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30615 * config/i386/sse.md (sse_storehps, sse_storelps,
30616 avx_<castmode><avxsizesuffix>_<castmode>,
30617 avx512f_<castmode><avxsizesuffix>_<castmode>,
30618 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30619 in condition that at least one operand is not a MEM.
30621 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30623 PR middle-end/79805
30624 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30625 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30627 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30628 gimple_call_nothrow_p flag based on whether original builtin can throw.
30629 If it can, emit following stmts on the fallthrough edge.
30630 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30631 don't create new bb if inserting just debug stmts on the edge, try to
30632 insert them on the fallthru bb or just reset debug stmts.
30634 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30637 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30638 restore recog_data (including the operand rtxes inside it) around
30639 the call to get_insn_template.
30641 2017-03-03 Martin Sebor <msebor@redhat.com>
30643 PR tree-optimization/79699
30644 * context.c (context::~context): Free MPFR caches to avoid
30645 a memory leak on program exit.
30647 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30649 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30650 Use wide_int::ulow () instead of .elt (0).
30652 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30654 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30655 (*pushxf): Limit oF constraint to 32bit targets and add oC
30656 constraint for 64bit targets.
30657 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30658 (*pushdf): Change rmF constraint to rmC.
30660 2017-03-03 Martin Liska <mliska@suse.cz>
30662 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30663 Remove unused variable.
30665 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30668 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30669 is a memory operand, increase num_memory.
30670 (ix86_expand_args_builtin): Likewise.
30672 2017-03-03 Jan Hubicka <jh@suse.cz>
30675 * ipa-devirt.c (maybe_record_node): Properly handle
30676 __cxa_pure_virtual visibility.
30678 2017-03-03 Martin Liska <mliska@suse.cz>
30680 PR tree-optimization/79803
30681 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30683 (pass_loop_prefetch::execute): Disabled optimization if an
30684 assumption about L1 cache size is not met.
30686 2017-03-03 Martin Liska <mliska@suse.cz>
30688 PR rtl-optimization/79574
30689 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30690 (hash_scan_set): Likewise.
30691 (dump_hash_table): Likewise.
30692 (hoist_code): Likewise.
30694 2017-03-03 Richard Biener <rguenther@suse.de>
30696 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30697 in place of uhigh ().
30698 (fixed_convert_from_real): Likewise.
30700 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30703 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30705 2017-03-03 Richard Biener <rguenther@suse.de>
30707 PR middle-end/79818
30708 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30709 TYPE_OVERFLOW_UNDEFINED check.
30711 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30713 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30715 (vector_ae_<mode>_p): Likewise.
30716 (vector_nez_<mode>_p): Likewise.
30717 (vector_ne_v2di_p): Likewise.
30718 (vector_ae_v2di_p): Likewise.
30719 (vector_ne_<mode>_p): Likewise.
30720 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30722 (vsx_tsqrt<mode>2_fe): Likewise.
30724 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30727 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30729 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30731 PR rtl-optimization/79780
30732 * cprop.c (one_cprop_pass): When second and further conditional trap
30733 in a single basic block is turned into an unconditional trap, turn it
30734 into a deleted note to avoid RTL verification failures.
30736 2017-03-02 Richard Biener <rguenther@suse.de>
30738 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30740 2017-03-02 Richard Biener <rguenther@suse.de>
30742 PR tree-optimization/79345
30744 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30745 param and abort the walk, returning -1 if it is hit.
30746 (walk_aliased_vdefs): Take a limit param and pass it on.
30747 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30748 defaulting to 0 and return a signed int.
30749 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30750 (check_defs): New helper.
30751 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30752 about uninitialized memory.
30753 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30754 bogus uninitialized warning.
30755 (fixed_convert_from_real): Likewise.
30757 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30759 PR tree-optimization/66768
30760 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30761 iv_use if base object can't be determined.
30763 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30765 PR tree-optimization/79345
30766 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30767 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30768 (get_pattern_stats): Initialize it.
30769 * genemit.c (gen_expand): Verify match_scratch numbers come after
30770 match_operand/match_dup numbers.
30771 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30772 match_scratch numbers.
30773 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30775 * config/s390/s390.md (trunctdsd2): Likewise.
30777 2017-03-02 Richard Biener <rguenther@suse.de>
30779 * wide-int.h (wide_int_storage::operator=): Implement in terms
30782 2017-03-02 Richard Biener <rguenther@suse.de>
30784 PR tree-optimization/79777
30785 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30786 the to insert expression to sth existing.
30788 2017-03-01 Martin Sebor <msebor@redhat.com>
30790 PR middle-end/79692
30791 * gimple-ssa-sprintf.c
30792 (directive::known_width_and_precision): New function.
30793 (format_integer): Use it.
30794 (get_mpfr_format_length): Consider the full range of precision
30795 when computing %g output with the # flag. Set the likely byte
30796 count to 3 rather than 1 when precision is indeterminate.
30797 (format_floating): Correct the lower bound of precision.
30799 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30801 * doc/invoke.texi: Document default code model for 64-bit Linux.
30803 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30806 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30807 udiv rather than div since input pattern is unsigned.
30809 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30811 * config/i386/i386.c (print_reg): Warn for values of
30812 unsupported size in integer register.
30814 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30817 * config/rs6000/predicates.md (current_file_function_operand): Do
30818 not allow self calls to be local if the function is replaceable.
30820 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30823 * config/rs6000/altivec.h (vec_ctz and others): Change the
30824 preprocessor macro that controls conditional compilation from
30825 _ARCH_PWR9 to __POWER9_VECTOR__.
30826 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30827 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30828 control (instead of _ARCH_PWR9 control) so that template
30829 definition uses power9-specific function.
30830 (vec_any_eq): Likewise.
30831 (vec_all_ne): Change macro definition to use a power9-specific
30832 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30833 _ARCH_PWR9 control).
30834 (vec_any_eq) Likewise.
30835 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30836 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30837 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30838 support for xvcmpnedp instruction.
30839 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30840 macro expansion so that Power9 implementation of vec_all_ne does
30841 not use the AltiVec predicate framework.
30842 (VCMPNEH_P): Likewise.
30843 (VCMPNEW_P): Likewise.
30844 (VCMPNED_P): Likewise.
30845 (VCMPNEFP_P): Likewise.
30846 (VCMPNEDP_P): Likewise.
30847 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30848 implementation of vec_any_eq to not use AltiVec predicate
30850 (VCMPAEH_P): Likewise.
30851 (VCMPAEW_P): Likewise.
30852 (VCMPAED_P): Likewise.
30853 (VCMPAEFP_P): Likewise.
30854 (VCMPAEDP_P): Likewise.
30855 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30856 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30857 not use the AltiVec predicate framework.
30858 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30859 of vec_any_eq to not use AltiVec predicate framework.
30860 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30861 support for predefined __POWER9_VECTOR__ macro to indicate that
30862 Power9 instruction selection is enabled.
30863 (altivec_overloaded_builtins): Remove extraneous
30864 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30865 function argument types RS6000_BTI_bool_V16QI and
30866 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30867 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30868 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30869 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30870 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30871 Power9 for implementations of vec_cmpne. Change the signature for
30872 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30873 (representing vec_all_ne) to remove the previously described first
30874 argument of type RS6000_BTI_INTSI, as this was an artifact of
30875 reliance on the AltiVec predicate framework, which is no longer
30876 used in the implementation of these functions. Add
30877 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30878 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30879 since, unlike the AltiVec predicate framework implementation, we
30880 do not share function descriptors between vec_alle and vec_anyeq.
30881 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30882 set of modes that receive special treatment even when
30883 TARGET_P9_VECTOR is true. The special treatment emits code that
30884 does not depend on Power9 instructions.
30885 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30886 define_expand to not rely on AltiVec predicate framework.
30887 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30889 (vector_ne_v2di_p): Change this define_expand to not rely on
30890 AltiVec predicate framework.
30891 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30893 (vector_ne_<mode>_p): Change this define_expand to not rely on
30894 AltiVec predicate framework.
30895 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30897 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30898 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30899 define_insn pattern.
30900 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30901 define_insn pattern because the xvcmpne<VSs>. instruction is not
30903 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30904 instruction is not supported.
30906 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30908 * config/nvptx/nvptx.c: Include intl.h.
30910 2017-03-01 Martin Jambor <mjambor@suse.cz>
30913 * ipa-prop.h (ipa_bits): Removed field known.
30914 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30915 to pointers. Adjusted their comments to warn about their sharing.
30916 (ipcp_transformation_summary): Change bits to a vector of pointers.
30917 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30918 (ipa_get_ipa_bits_for_value): Declare.
30919 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30920 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30921 (ipa_bits_hash_table): Likewise.
30922 (ipa_vr_ggc_hash_traits): Likewise.
30923 (ipa_vr_hash_table): Likewise.
30924 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30925 being pointers and vr_known being removed.
30926 (ipa_set_jf_unknown): Likewise.
30927 (ipa_get_ipa_bits_for_value): New function.
30928 (ipa_set_jfunc_bits): Likewise.
30929 (ipa_get_value_range): New overloaded functions.
30930 (ipa_set_jfunc_vr): Likewise.
30931 (ipa_compute_jump_functions_for_edge): Use the above functions to
30932 construct bits and vr parts of jump functions.
30933 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30934 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30936 (ipcp_grow_transformations_if_necessary): Also allocate
30937 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30939 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30940 them. Fix too long lines.
30941 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30942 vr_known being removed.
30943 (ipa_read_jump_function): Use new setter functions to construct bits
30944 and vr parts of jump functions or set them to NULL.
30945 (write_ipcp_transformation_info): Adjust for bits being pointers.
30946 (read_ipcp_transformation_info): Likewise.
30947 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30949 Include gt-ipa-prop.h.
30950 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30952 (ipcp_store_bits_results): Likewise.
30953 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30954 Do not write to existing jump functions but use a temporary instead.
30956 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30959 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30960 attempt to use its first operand as BIT_FIELD_REF base.
30962 2017-03-01 Richard Biener <rguenther@suse.de>
30964 PR middle-end/79721
30965 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30966 interpolating formula in wrapping arithmetic.
30967 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30969 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30971 PR tree-optimization/79734
30972 * tree-vect-generic.c (expand_vector_condition): Optimize
30973 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30974 Handle VEC_COND_EXPR where comparison has different inner width from
30975 type's inner width.
30977 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30979 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30980 markup, and similar issues. Remove @opindex entries for things
30981 that aren't options. Add missing -mmpy-option entries.
30983 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30985 PR tree-optimization/79737
30986 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30987 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30988 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30989 instead of byte_size. Formatting fix.
30990 (shift_bytes_in_array_right): Formatting fix.
30992 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30995 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30996 condition on optimize for the leaf function test.
30998 2017-02-28 Martin Liska <mliska@suse.cz>
31001 * read-rtl-function.c (function_reader::handle_unknown_directive):
31002 Bail out when one uses -flto.
31004 2017-02-28 Martin Liska <mliska@suse.cz>
31006 * common.opt: Replace space with tabular for options of <number>
31008 * config/i386/i386.opt: Show <number> value for
31009 -mlarge-data-threshold.
31010 * opts.c (print_filtered_help): Do not display number in hexadecimal
31013 2017-02-28 Martin Liska <mliska@suse.cz>
31015 * common.opt: Fix --help=option -Q for options which are of
31018 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
31020 * config/i386/i386.c (print_reg): Error out for values
31021 of 8-bit size in invalid integer register.
31023 2017-02-28 Martin Sebor <msebor@redhat.com>
31025 PR tree-optimization/79691
31026 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
31028 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31031 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
31032 gcc_unreachable with output_operand_lossage.
31034 2017-02-28 Richard Biener <rguenther@suse.de>
31036 PR tree-optimization/79740
31037 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31039 (visit_nary_op): Insert the nary into the hashtable if we
31040 pattern-matched sth.
31041 * tree-ssa-pre.c (eliminate_insert): Robustify.
31043 2017-02-28 Richard Biener <rguenther@suse.de>
31045 PR middle-end/79731
31046 * fold-const.c (decode_field_reference): Reject out-of-bound
31049 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31051 * config/i386/i386.c: Include intl.h.
31052 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31053 instead of just cond ? "..." : "...".
31054 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31055 * coverage.c (read_counts_file): Likewise.
31056 * omp-offload.c: Include intl.h.
31057 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31058 of just cond ? "..." : "...".
31059 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31060 of just cond ? "..." : "...".
31062 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31065 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31067 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31068 'tune for' CPU name.
31069 * config/arm/arm-cpu-data.h: Regenerated.
31071 2017-02-28 Richard Biener <rguenther@suse.de>
31073 PR tree-optimization/79732
31074 * tree-inline.c (expand_call_inline): Do not shadow var.
31076 2017-02-28 Richard Biener <rguenther@suse.de>
31078 PR tree-optimization/79723
31079 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31080 address-space properly.
31082 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31084 * doc/optinfo.texi (Optimization groups): Fix option used for
31086 * doc/invoke.texi (-fopt-info): Document "omp".
31087 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31088 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31089 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31090 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31091 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31093 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31095 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31096 OPTGROUP_OMP with "omp".
31098 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31101 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31102 for arithmetic shift of unsigned V2DI.
31104 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31106 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31107 arc/linux.h headers.
31108 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31109 (LINK_SPEC): Likewise.
31110 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31111 (EXTRA_SPECS): Likewise.
31112 (STARTFILE_SPEC): Likewise.
31113 (ENDFILE_SPEC): Likewise.
31114 (LIB_SPEC): Likewise.
31115 (TARGET_SDATA_DEFAULT): Likewise.
31116 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31117 (MULTILIB_DEFAULTS): Likewise.
31118 (DWARF2_UNWIND_INFO): Likewise.
31119 * config/arc/big.h: New file.
31120 * config/arc/elf.h: Likewise.
31121 * config/arc/linux.h: Likewise.
31122 * config/arc/t-uClibc: Remove.
31124 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31126 PR tree-optimization/77536
31127 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31128 (tree_transform_and_unroll_loop): Use above function to compute the
31129 estimated niter of unrolled loop and use it when scaling profile.
31130 Also use count info rather than frequency if it's non-zero.
31131 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31132 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31133 (vect_transform_loop): Call above function.
31135 2017-02-27 Richard Biener <rguenther@suse.de>
31137 PR tree-optimization/45397
31138 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31139 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31140 (visit_nary_op): Add pattern matching for CSEing sign-changed
31141 or truncated operations with wider ones.
31143 2017-02-27 Richard Biener <rguenther@suse.de>
31145 PR tree-optimization/79690
31146 * tree-vect-stmts.c (vectorizable_store): Use vector type
31147 built from the DR with address-space.
31149 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31151 * doc/invoke.texi (Optimize Options): Refine the description
31152 of asan-use-after-return.
31154 2017-02-25 Alan Modra <amodra@gmail.com>
31156 PR rtl-optimization/79584
31157 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31158 base, not ad->base_term, the reg within base. Remove assertion
31159 that ad->base == ad->base_term. Replace gen_int_mode using
31160 bogus mode with const0_rtx.
31162 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31164 PR middle-end/79396
31165 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31166 FMA_EXPR like tcc_binary or tcc_unary.
31168 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31171 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31173 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31174 (output_loc_operands): Handle DW_OP_call_ref and
31175 DW_OP_GNU_variable_value.
31176 (struct variable_value_struct): New type.
31177 (struct variable_value_hasher): Likewise.
31178 (variable_value_hash): New variable.
31179 (string_types): Remove.
31180 (copy_loc_descr): New function.
31181 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31182 (prepend_loc_descr_to_each): New function.
31183 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31184 instead of add_loc_descr_to_each if the first argument is single
31185 location list and the second has multiple.
31186 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31187 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31188 when looking for variable value which doesn't have other location info.
31189 (loc_list_from_tree): Formatting fix.
31190 (gen_array_type_die): Simplify DW_AT_string_length handling.
31191 (adjust_string_types): Remove.
31192 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31193 string_types. Call resolve_variable_values.
31194 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31195 (resolve_addr_in_expr): Likewise. Add A argument.
31196 (copy_deref_exprloc): Remove deref argument. Adjust for the
31197 original expression being DW_OP_GNU_variable_value with optionally
31198 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31199 optionally after it.
31200 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31201 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31202 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31203 (variable_value_hasher::hash, variable_value_hasher::equal): New
31205 (resolve_variable_value_in_expr, resolve_variable_value,
31206 resolve_variable_values, note_variable_value_in_expr,
31207 note_variable_value): New functions.
31208 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31211 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31214 * opts.h (handle_generated_option): Add GENERATED_P argument.
31215 * opts-common.c (handle_option): Adjust function comment.
31216 (handle_generated_option): Add GENERATED_P argument, pass it to
31218 (control_warning_option): Pass false to handle_generated_option
31220 * opts.c (maybe_default_option): Pass true to handle_generated_option
31222 * optc-gen.awk: Likewise.
31224 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31226 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31227 a REG, look at the REG it is a SUBREG of.
31228 (splitter for cmpeqsi_t): Ditto.
31230 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31232 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31233 the special USEs with the pattern of the insn, not the insn itself.
31235 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31238 * doc/invoke.texi: Document -mload-store-pairs.
31240 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31241 Sandra Loosemore <sandra@codesourcery.com>
31243 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31244 argument isn't a CONST_INT.
31245 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31246 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31247 (nios2_validate_compare): Bypass alternate compare logic if *op2
31248 is not a CONST_INT.
31249 (ldstwm_operation_p): Return false if first_base is not a REG or
31250 if first_offset is not a CONST_INT.
31252 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31254 * config/cris/cris.md: Use correct operand in a define_peephole2.
31256 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31258 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31260 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31262 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31263 this_insn if it is an INSN or JUMP_INSN.
31264 (force_offsettable): Look at base, not at addr.
31265 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31266 on things that aren't necessarily CONST_INTs.
31268 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31270 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31271 -mfpmath=sse is the default also for x86-32 targets with SSE2
31272 instruction set when @option{-ffast-math} is enabled
31274 2017-02-24 Jeff Law <law@redhat.com>
31276 PR rtl-optimizatoin/79286
31277 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31280 2017-02-24 Richard Biener <rguenther@suse.de>
31282 PR tree-optimization/79389
31283 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31286 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31288 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31289 function comment to reflect reality.
31290 (loop_exits_before_overflow): Fix typo in function description.
31292 2017-02-24 Richard Biener <rguenther@suse.de>
31294 PR tree-optimization/79389
31295 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31296 properly that a threading opportunity exists. Detect conditional
31297 copy/constant propagation opportunities.
31299 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31301 * config/visium/visium.md (type): Add trap.
31302 (b): New mode attribute.
31303 (*btst): Rename into...
31304 (*btst<mode>): ...this and adjust.
31305 (*cbranchsi4_btst_insn): Rename into...
31306 (*cbranch<mode>4_btst_insn): ...this and adjust.
31307 (trap): New define_insn.
31309 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31311 PR tree-optimization/79389
31312 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31313 (noce_reversed_cond_code): New function.
31314 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31315 reversed_comparison_code. Formatting fix.
31316 (noce_try_store_flag): Test rev_cond != NULL in addition to
31317 reversed_comparison_code.
31318 (noce_try_store_flag_constants): Likewise.
31319 (noce_try_store_flag_mask): Likewise.
31320 (noce_try_addcc): Use rev_cond if non-NULL instead of
31321 reversed_comparison_code.
31322 (noce_try_cmove_arith): Likewise. Formatting fixes.
31323 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31324 (noce_find_if_block): Initialize rev_cond.
31325 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31326 instead of false as last argument never attempt to reverse it
31329 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31331 PR tree-optimization/79663
31332 * tree-predcom.c (combine_chains): Process refs in reverse order
31333 only for ZERO length chains, and add explaining comment.
31335 2017-02-23 Jeff Law <law@redhat.com>
31337 PR tree-optimization/79578
31338 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31339 in call to operand_equal_p.
31341 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31344 * config/i386/cpuid.h: Fix another undefined behavior.
31346 2017-02-23 Richard Biener <rguenther@suse.de>
31348 PR tree-optimization/79683
31349 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31350 vector types for data-refs.
31352 2017-02-23 Martin Liska <mliska@suse.cz>
31354 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31356 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31358 PR middle-end/79665
31359 * internal-fn.c (get_range_pos_neg): Moved to ...
31360 * tree.c (get_range_pos_neg): ... here. No longer static.
31361 * tree.h (get_range_pos_neg): New prototype.
31362 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31363 are known to be in between 0 and signed maximum inclusive, try to
31364 expand both unsigned and signed divmod and use the cheaper one from
31367 2017-02-22 Jeff Law <law@redhat.com>
31369 PR tree-optimization/79578
31370 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31371 to compare base operands.
31373 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31376 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31377 gpc_reg_operand instead of fpr_reg_operand.
31379 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31381 * config/mips/mips.c (mips_return_in_memory): Force FP
31382 vector types to be returned in memory for o32 ABI.
31384 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31386 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31387 instead of DW_TAG_member for static data member declarations and don't
31388 set no_linkage_name for static inline data members.
31389 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31392 2017-02-22 Martin Liska <mliska@suse.cz>
31394 * doc/invoke.texi: Replace inequality signs with square brackets
31397 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31400 * lra-constraints.c (simplify_operand_subreg): Handle
31401 WORD_REGISTER_OPERATIONS targets.
31403 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31406 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31407 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31408 elimination by swapping fld*.
31410 2017-02-22 Richard Biener <rguenther@suse.de>
31412 PR tree-optimization/79673
31413 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31414 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31415 irrelevant address-space qualifiers and avoiding a
31416 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31418 2017-02-22 Richard Biener <rguenther@suse.de>
31420 PR tree-optimization/79666
31421 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31422 to not symbolically negate if that may introduce undefined
31425 2017-02-22 Martin Liska <mliska@suse.cz>
31428 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31429 * data-streamer-out.c (streamer_write_gcov_count_stream):
31431 * value-prof.c (stream_out_histogram_value): Make assert more
31432 precise based on type of counter.
31434 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31437 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31438 Use nonimmediate_operand instead of memory_operand for operand 1.
31439 (float-extend standard_x87sse_constant_load splitter): Ditto.
31441 2017-02-21 Jeff Law <law@redhat.com>
31443 PR tree-optimization/79621
31444 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31445 blocks with edges to themselves.
31447 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31450 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31451 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31452 Use gimple_call_builtin_p.
31455 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31456 on temporarily removed DEBUG_INSNs.
31458 PR tree-optimization/79649
31459 * tree-loop-distribution.c (classify_partition): Give up on
31460 non-generic address space loads/stores.
31462 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31464 * doc/loop.texi (Loop manipulation): Remove nonexistent
31465 tree_ssa_loop_version from the documentation.
31466 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31468 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31471 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31472 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31473 * config/rs6000/rs6000.c: Include except.h.
31474 (rs6000_expand_split_stack_prologue): Call
31475 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31477 2017-02-21 Martin Jambor <mjambor@suse.cz>
31480 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31481 have been analyzed.
31483 2017-02-21 Martin Jambor <mjambor@suse.cz>
31485 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31486 for backward compatibility only.
31487 * doc/invoke.texi (Option Summary): Remove all references to
31488 -fipa-cp-alignment.
31490 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31494 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31496 * lra-constraints.c (curr_insn_transform): Handle
31497 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31499 2017-02-21 Martin Liska <mliska@suse.cz>
31501 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31503 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31505 PR translation/79638
31506 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31508 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31511 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31512 (arm_function_ok_for_sibcall): Return false for an indirect call by
31513 descriptor if all the argument registers are used.
31514 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31515 alignment of the function.
31517 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31519 PR tree-optimization/61441
31520 * simplify-rtx.c (simplify_const_unary_operation): For
31521 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31522 the sNaN unmodified.
31524 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31526 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31527 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31528 instead of SYSTEM_HEADER_DIR.
31530 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31531 Martin Liška <mliska@suse.cz>
31533 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31534 Fix typos and grammar, use active voice, and clarify.
31536 2017-02-20 Marek Polacek <polacek@redhat.com>
31538 PR middle-end/79537
31539 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31542 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31544 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31547 * config/i386/i386.c (ix86_expand_builtin): Handle
31548 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31549 ix86_builtins_isa[fcode].isa as a requirement of those
31550 flags and any other flag in the bitmask.
31551 (ix86_init_mmx_sse_builtins): Use 0 instead of
31552 ~OPTION_MASK_ISA_64BIT as mask.
31553 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31554 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31555 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31556 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31558 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31561 * lra-constraints.c (split_reg): Check requested split mode
31562 is supported by the register.
31564 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31566 * lra-constraints.c (simplify_operand_subreg): Remove early
31569 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31572 * lra-constraints.c (curr_insn_transform): Tighten condition
31573 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31575 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31578 * lra-constraints.c (curr_insn_transform): Handle
31579 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31581 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31584 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31586 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31588 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31591 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31594 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31596 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31599 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31601 * final.c (last_columnnum, override_columnnum): New variables.
31602 (final_start_function): Set last_columnnum, pass it to begin_prologue
31603 hook and pass 0 to dwarf2out_begin_prologue.
31604 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31605 to source_line debug hook.
31606 (notice_source_line): Compute last_columnnum and for debug_column_info
31607 return true on column changes.
31608 * debug.h (struct gcc_debug_hooks): Add column argument to
31609 source_line and begin_prologue hooks.
31610 (debug_nothing_int_charstar_int_bool): Remove prototype.
31611 (debug_nothing_int_int_charstar,
31612 debug_nothing_int_int_charstar_int_bool): New prototypes.
31613 (dwarf2out_begin_prologue): Add column argument.
31614 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31615 begin_prologue hooks.
31616 (debug_nothing_int_charstar_int_bool): Remove.
31617 (debug_nothing_int_int_charstar,
31618 debug_nothing_int_int_charstar_int_bool): New functions.
31619 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31620 through to dwarf2out_source_line.
31621 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31622 (dwarf2out_source_line): Add column argument, emit it if requested.
31623 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31625 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31626 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31627 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31628 through to dwarf2out_begin_prologue.
31629 (vmsdbgout_source_line): Add column argument, pass it through to
31630 dwarf2out_source_line.
31631 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31632 dbxout_source_line caller.
31633 (dbxout_source_line): Add column argument.
31635 * common.opt (gno-column-info, gcolumn-info): New options.
31636 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31637 (check_die): Also test for multiple DW_AT_decl_column attributes.
31638 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31639 DW_AT_decl_column if requested.
31640 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31642 (gen_variable_die): Likewise.
31643 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31644 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31647 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31648 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31649 (ix86_handle_option): Handle OPT_m3dnowa.
31650 * doc/invoke.texi (-m3dnowa): Document.
31651 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31652 -m3dnowa instead of -m3dnow -march=athlon.
31655 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31656 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31658 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31661 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31662 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31663 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31664 generator for vsx_xxpermdi_<mode>_be.
31665 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31666 force big-endian semantics.
31667 (vsx_xxpermdi_<mode>_be): New define_expand with same
31668 implementation as previous version of vsx_xxpermdi_<mode>.
31670 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31672 PR tree-optimization/79327
31673 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31674 variable, its initialization and use.
31676 2017-02-17 Julia Koval <julia.koval@intel.com>
31678 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31679 (OPTION_MASK_ISA_PKU_UNSET): New.
31680 (ix86_handle_option): Handle -mrdpid.
31681 * config/i386/cpuid.h (bit_RDPID): New.
31682 * config/i386/driver-i386.c (host_detect_local_cpu):
31683 Detect RDPID feature.
31684 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31685 * config/i386/i386-c.c (ix86_target_macros_internal):
31687 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31688 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31689 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31690 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31691 * config/i386/i386.md (define_insn "rdpid"): New.
31692 * config/i386/i386.opt Add -mrdpid.
31693 * config/i386/immintrin.h (_rdpid_u32): New.
31695 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31697 PR rtl-optimization/79541
31698 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31699 instead of transforming it into USE.
31701 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31703 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31704 If HONOR_SNANS (SFmode) force the input to a register.
31705 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31706 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31707 an frsp or similar insn.
31709 2017-02-17 Martin Liska <mliska@suse.cz>
31711 PR rtl-optimization/79577
31712 * params.def (selsched-max-sched-times): Increase minimum to 1.
31714 2017-02-17 Martin Liska <mliska@suse.cz>
31716 PR rtl-optimization/79574
31717 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31719 2017-02-17 Martin Liska <mliska@suse.cz>
31721 PR tree-optimization/79529
31722 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31723 ssa_defined_default_def_p to handle cases which are implicitly
31725 * tree-ssa.c (ssa_defined_default_def_p): New function.
31726 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31727 which are implicitly defined.
31728 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31730 2017-02-17 Richard Biener <rguenther@suse.de>
31732 PR middle-end/79576
31733 * params.def (max-ssa-name-query-depth): Limit to 10.
31735 2017-02-17 Richard Biener <rguenther@suse.de>
31737 PR tree-optimization/79552
31738 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31741 2017-02-17 Richard Biener <rguenther@suse.de>
31744 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31746 2017-02-17 Marek Polacek <polacek@redhat.com>
31748 PR middle-end/79536
31749 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31750 (fold_negate_expr): New wrapper.
31752 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31754 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31755 Correct terminology and de-emphasize pre-standard behavior.
31757 2017-02-16 Alan Modra <amodra@gmail.com>
31759 PR rtl-optimization/79286
31760 * ira.c (def_dominates_uses): New function.
31761 (update_equiv_regs): Don't create an equivalence for insns that
31762 may trap where the register def does not dominate the use.
31764 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31766 PR rtl-optimization/78127
31767 * lra.c (lra): Call lra_eliminate before finish the loop after
31770 2017-02-16 Richard Biener <rguenther@suse.de>
31772 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31774 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31775 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31776 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31777 (isl_val_int_from_wi): New function.
31778 (extract_affine_gmp): Rename to ...
31779 (extract_affine_wi): ... this, take a widest_int.
31780 (extract_affine_int): Just wrap extract_affine_wi.
31781 (add_param_constraints): Use isl_val_int_from_wi.
31782 (add_loop_constraints): Likewise, and extract_affine_wi.
31784 2017-02-15 Jeff Law <law@redhat.com>
31786 PR middle-end/79521
31787 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31788 ira_init_register_move_cost_if_necessary.
31790 2017-02-15 Martin Sebor <msebor@redhat.com>
31792 PR middle-end/32003
31793 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31794 removed in a prior commit.
31796 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31798 PR tree-optimization/79347
31799 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31800 counters during peeling.
31802 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31804 * Makefile.in (site.exp): Remove "set ISLVER".
31806 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31809 * real.c (real_from_integer): Call real_convert even for decimal.
31811 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31814 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31816 2017-02-14 Andrew Pinski <apinski@cavium.com>
31818 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31819 cores and change the partno/implementer to be correct.
31820 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31821 the 'B" as the implementer.
31822 * config/aarch64/aarch64-tune.md: Regenerate.
31824 2017-02-14 Carl Love <cel@us.ibm.com>
31826 * config/rs6000/rs6000.c: Add case statement entry to make the
31827 xvcvuxdsp built-in argument unsigned.
31828 * config/rs6000/vsx.md: Fix the source and return operand types so they
31829 match the instruction definitions from the ISA document. Fix typo
31830 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31833 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31836 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31837 member early_clobber_alts.
31838 * lra-lives.c (reg_early_clobber_p): New.
31839 (process_bb_lives): Use it.
31840 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31841 (debug_operand_data): Initialize early_clobber_alts.
31842 (setup_operand_alternative): Set up early_clobber_alts.
31843 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31844 alternatives to new_insn_reg.
31845 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31847 (lra_update_insn_regno_info): Pass the new arg.
31849 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31851 PR middle-end/79505
31852 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31853 (new_oacc_loop_raw): Don't clear already cleared fields.
31856 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31857 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31858 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31860 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31863 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31865 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31868 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31869 the extra instruction to the right place to store 128-bit constant
31872 2017-02-14 Martin Sebor <msebor@redhat.com>
31874 PR middle-end/79448
31875 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31876 warning for strings of unknown length.
31878 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31880 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31882 2017-02-14 Jeff Law <law@redhat.com>
31885 * ira-costs.c (scan_one_insn): Initialize register move costs
31886 for pseudos seen in USE/CLOBBER insns.
31888 PR tree-optimization/79095
31889 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31890 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31891 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31892 if the operands are known to be not equal, then the resulting range
31894 (intersect_ranges): If the new range is ~[0,0] and the old range is
31895 wide, then prefer ~[0,0].
31896 * tree-vrp.c (overflow_comparison_p_1): New function.
31897 (overflow_comparison_p): New function.
31898 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31899 if NAME is used in an overflow test.
31900 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31901 overflow check that can be expressed as an equality test, then adjust
31902 ops to be that equality test.
31904 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31906 * config/s390/s390-builtin-types.def: Remove flags argument.
31907 * config/s390/s390.c (s390_init_builtins): Likewise.
31909 2017-02-14 Martin Liska <mliska@suse.cz>
31911 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31912 vector. Fix trailing white spaces.
31914 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31916 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31919 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31921 PR rtl-optimization/68664
31922 * config/arm/arm.c (arm_sched_can_speculate_insn):
31923 New function. Declare prototype.
31924 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31926 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31928 PR rtl-optimization/68664
31929 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31931 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31933 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31935 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31936 max skip bytes for function, loop and jump.
31938 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31940 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31941 ABS_EXPR for gimple dump.
31943 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31946 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31948 PR tree-optimization/79408
31949 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31950 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31951 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31952 also if rhs1 is INTEGER_CST.
31954 2017-02-14 Richard Biener <rguenther@suse.de>
31956 PR middle-end/79432
31957 * tree-into-ssa.c (insert_phi_nodes): When the function can
31958 have abnormal edges rewrite SSA names with broken use-def
31959 dominance out of SSA and register them for PHI insertion.
31961 2017-02-13 Martin Sebor <msebor@redhat.com>
31963 PR middle-end/79496
31964 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31965 clearing info.nowrite flag when snprintf size argument is a range.
31967 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31969 * cprop.c (cprop_jump): Add missing space in string literal.
31970 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31971 (get_constraint_for_component_ref): Likewise.
31972 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31973 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31974 * lra-constraints.c (process_alt_operands): Likewise.
31975 * ipa-inline.c (inline_small_functions): Likewise.
31976 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31977 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31978 * trans-mem.c (diagnose_tm_1_op): Likewise.
31979 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31980 (grid_parallel_clauses_gridifiable): Likewise.
31982 * config/nvptx/mkoffload.c (process): Add space in between
31985 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31986 "MOD4_SSE_REGS" and "ALL_REGS".
31988 * spellcheck.c (test_data): Add , in between "foo" and "food".
31990 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31993 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31994 boundary crossing check and subsequent code generation agree.
31996 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31998 * config/aarch64/aarch64.c (has_memory_op): Delete.
31999 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
32002 2017-02-13 Jakub Jelinek <jakub@redhat.com>
32004 PR rtl-optimization/79388
32005 PR rtl-optimization/79450
32006 * combine.c (distribute_notes): When removing TEM_INSN for which
32007 corresponding dest has last value recorded, invalidate that last
32010 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32012 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
32013 of explicit '@'. Add missing assembly comment marker on branch costs
32016 2017-02-13 Nathan Sidwell <nathan@acm.org>
32018 * gengtype-lex.l (<in_struct>): Add '/'.
32020 2017-02-13 Martin Liska <mliska@suse.cz>
32023 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
32025 2017-02-13 Richard Biener <rguenther@suse.de>
32027 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
32029 * configure: Re-generate.
32030 * config.in: Likewise.
32031 * graphite-dependences.c: Simplify as if
32032 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
32033 * graphite-isl-ast-to-gimple.c: Likewise.
32034 * graphite-optimize-isl.c: Likewise.
32035 * graphite-poly.c: Likewise.
32036 * graphite-sese-to-poly.c: Likewise.
32037 * graphite.h: Likewise.
32038 * toplev.c: Include isl/version.h and use isl_version () for
32039 printing the ISL version.
32040 * doc/install.texi: Update ISL requirement.
32042 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32044 * doc/standards.texi (Standards): Update reference to
32047 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32049 * doc/extend.texi (Named Address Spaces): sourceware.org now
32051 * doc/install.texi (Binaries): Ditto.
32054 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32056 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32057 standard terminology "stringize"/"stringizing" throughout.
32058 * doc/cppinternals.texi: Likewise.
32060 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32062 * doc/extend.texi: Fix some spelling mistakes and typos.
32063 * doc/invoke.texi: Likewise.
32065 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32068 * params.def (inline-min-speedup) Change from 10 to 8.
32070 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32072 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32075 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32078 * ipa-inline-analysis.c (get_minimal_bb): New function.
32079 (record_modified): Use it.
32080 (remap_edge_change_prob): Handle also ancestor functions.
32082 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32084 * doc/contrib.texi (Contributors): Remove broken link into
32085 the Mauve CVS repository.
32087 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32089 PR middle-end/79454
32090 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32091 result computation whenever lhs doesn't have vector mode, not
32092 just when it has BLKmode.
32094 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32096 * doc/makefile.texi (profiledbootstrap): Refer to the
32097 installation instructions only in textual form.
32099 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32102 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32104 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32106 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32107 (Specific): Update mingw-w64 reference.
32109 (Specific): Remove broken link to Renesas RX processor.
32111 2017-02-10 Richard Biener <rguenther@suse.de>
32113 * toplev.c (process_options): Do not mention obsolete graphite
32114 options when printing sorry message about missing graphite support.
32115 Mention -floop-nest-optimize.
32117 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32119 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32120 (vtst_p16): Likewise.
32121 (vtstq_p8): Likewise.
32122 (vtstq_p16): Likewise.
32124 (vtstq_p64): Likewise.
32125 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32126 (vset_lane_p64): New.
32127 (vsetq_lane_p64): New.
32129 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32131 PR tree-optimization/79411
32132 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32133 stmt operands are SSA_NAMEs used in abnormal phis.
32134 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32137 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32140 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32143 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32145 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32147 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32149 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32153 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32154 not arbitrary TREE_CONSTANT.
32157 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32158 "omp declare target link" attribute unless is_global_var.
32159 * omp-offload.c (find_link_var_op): Likewise.
32161 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32162 Chung-Lin Tang <cltang@codesourcery.com>
32164 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32166 (gimplify_adjust_omp_clauses): Don't delete TILE.
32167 (gimplify_omp_for): Deal with TILE.
32168 * internal-fn.c (expand_GOACC_TILE): New function.
32169 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32171 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32172 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32174 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32175 avoid DIV for outermost collapse var.
32176 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32177 Remove out of date comments, fix whitespace.
32178 * omp-general.c (omp_extract_for_data): Deal with tiling.
32179 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32180 adjust OLF_DIM_BASE value.
32181 (struct omp_for_data): Add tiling field.
32182 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32183 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32184 for auto loops. Remove default auto determining, moved to
32185 oacc_loop_fixed_partitions.
32186 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32187 stmts, add e_mask field.
32188 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32189 (oacc_thread_numbers): Use oacc_dim_call.
32190 (oacc_xform_tile): New.
32191 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32192 (finish_oacc_loop): Adjust for ifns vector.
32193 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32194 add case for GOACC_TILE fns.
32195 (oacc_loop_xform_loop): Delete.
32196 (oacc_loop_process): Iterate over call list directly, and add
32197 handling for GOACC_TILE fns.
32198 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32200 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32201 vector partitioning to outer loops. Assign 2 partitions to loops
32202 when available. Add TILE handling.
32203 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32204 (execite_oacc_device_lower): Process GOACC_TILE fns,
32205 ignore unknown specs.
32206 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32207 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32208 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32210 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32212 * configure.ac (ACX_BUGURL): Update.
32213 * configure: Regenerate.
32215 2017-02-09 Richard Biener <rguenther@suse.de>
32217 PR tree-optimization/69823
32218 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32219 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32221 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32223 * config/arc/arc-c.def: Add __NPS400__ definition.
32224 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32225 (TARGET_NPS400): Define.
32227 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32229 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32231 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32232 pointer, arch_info.
32233 (arc_cpu_types): Fill the arch_info field with a pointer into the
32234 arc_arch_types table.
32235 (arc_selected_cpu): Declare.
32236 * config/arc/arc.c (arc_selected_cpu): Make global.
32237 (arc_selected_arch): Delete.
32238 (arc_base_cpu): Delete.
32239 (arc_override_options): Remove references to deleted variables,
32240 update access to arch information.
32241 (ARC_OPT): Update access to arch information.
32242 (ARC_OPTX): Likewise.
32243 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32244 (TARGET_ARC600): Update access to arch information.
32245 (TARGET_ARC601): Likewise.
32246 (TARGET_ARC700): Likewise.
32247 (TARGET_EM): Likewise.
32248 (TARGET_HS): Likewise.
32249 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32252 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32255 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32256 condition/operands for integer GE/LE/GEU/LEU operations.
32258 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32260 PR translation/79397
32261 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32264 2017-02-08 Martin Jambor <mjambor@suse.cz>
32267 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32268 whether allocation happened.
32269 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32270 nothing was allocated.
32272 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32274 PR tree-optimization/79408
32275 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32276 constant, but SSA_NAME with a known integer range, use the minimum
32277 of that range instead of op1 to determine if modulo can be replaced
32278 with its first operand.
32280 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32282 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32284 2017-02-08 Richard Biener <rguenther@suse.de>
32286 PR tree-optimization/71824
32287 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32288 Check all loops contained in the merged region.
32290 2017-02-07 Andrew Pinski <apinski@cavium.com>
32292 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32294 2017-02-07 Andrew Pinski <apinski@cavium.com>
32296 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32297 (thunderxt88): Likewise.
32298 (thunderxt81): Disable LSE and change v8.1 to v8.
32299 (thunderxt83): Likewise.
32301 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32302 Richard Biener <rguenther@suse.de>
32304 PR middle-end/79399
32305 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32306 type from int to size_t.
32307 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32309 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32311 PR rtl-optimization/79386
32312 * cprop.c (bypass_conditional_jumps): Initialize
32313 bypass_last_basic_block already before splitting bbs after
32314 unconditional traps...
32315 (bypass_conditional_jumps): ... rather than here.
32318 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32319 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32320 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32321 fix -masm=intel patterns.
32323 2017-02-07 Richard Biener <rguenther@suse.de>
32325 PR tree-optimization/79256
32326 PR middle-end/79278
32327 * builtins.c (get_object_alignment_2): Use min_align_of_type
32328 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32329 and ADJUST_FIELD_ALIGN.
32331 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32333 * doc/tm.texi: Regenerate.
32334 * stor-layout.c (layout_decl): Adjust.
32335 (update_alignment_for_field): Likewise.
32336 (place_field): Likewise.
32337 (min_align_of_type): Likewise.
32338 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32339 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32340 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32341 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32342 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32343 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32344 * config/i386/i386.c (x86_field_alignment): Likewise.
32345 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32346 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32347 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32348 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32349 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32350 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32354 2017-01-30 Richard Biener <rguenther@suse.de>
32356 PR tree-optimization/79256
32357 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32358 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32361 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32363 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32364 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32365 builtins to SImode and emit a zero-extend, if necessary.
32367 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32369 * docs/invoke.texi (RISC-V Options): Alphabetize.
32371 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32373 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32376 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32378 * config/riscv/riscv.c: New file.
32379 * common/config/riscv/riscv-common.c: Likewise.
32380 * config.gcc: Likewise.
32381 * config/riscv/constraints.md: Likewise.
32382 * config/riscv/elf.h: Likewise.
32383 * config/riscv/generic.md: Likewise.
32384 * config/riscv/linux.h: Likewise.
32385 * config/riscv/multilib-generator: Likewise.
32386 * config/riscv/peephole.md: Likewise.
32387 * config/riscv/pic.md: Likewise.
32388 * config/riscv/predicates.md: Likewise.
32389 * config/riscv/riscv-builtins.c: Likewise.
32390 * config/riscv/riscv-c.c: Likewise.
32391 * config/riscv/riscv-ftypes.def: Likewise.
32392 * config/riscv/riscv-modes.def: Likewise.
32393 * config/riscv/riscv-opts.h: Likewise.
32394 * config/riscv/riscv-protos.h: Likewise.
32395 * config/riscv/riscv.h: Likewise.
32396 * config/riscv/riscv.md: Likewise.
32397 * config/riscv/riscv.opt: Likewise.
32398 * config/riscv/sync.md: Likewise.
32399 * config/riscv/t-elf-multilib: Likewise.
32400 * config/riscv/t-linux: Likewise.
32401 * config/riscv/t-linux-multilib: Likewise.
32402 * config/riscv/t-riscv: Likewise.
32403 * configure.ac: Likewise.
32404 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32405 Waterman as RISC-V maintainers.
32406 * doc/install.texi: Add RISC-V entries.
32407 * doc/invoke.texi: Add RISC-V options section.
32408 * doc/md.texi: Add RISC-V constraints section.
32409 * configure: Regenerated.
32411 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32414 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32415 false values to be constant vectors with all 0 or all 1 bits set.
32416 (vcondu<mode><mode>): Likewise.
32417 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32419 (fpmask_comparison_operator): Update comment.
32420 (vecint_comparison_operator): New predicate.
32421 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32422 vector conditionals when the true and false values are constant
32423 vectors with all 0 bits or all 1 bits set.
32425 2017-02-06 Martin Sebor <msebor@redhat.com>
32427 PR tree-optimization/79376
32428 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32430 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32432 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32433 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32434 to simplify split condition.
32436 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32438 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32439 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32442 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32444 PR rtl-optimization/68664
32445 * target.def (can_speculate_insn): New hook.
32446 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32447 * doc/tm.texi: Regenerate.
32448 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32449 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32450 (rs6000_sched_can_speculate_insn): New function.
32452 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32454 PR tree-optimization/79284
32455 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32456 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32457 vectorizable_mask_load_store, vectorizable_operation,
32458 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32459 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32460 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32461 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32462 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32463 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32464 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32465 is_gimple_assign (stmt). Replace another such test with
32466 is_gimple_assign (stmt).
32468 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32471 * config/avr/avr.c (rtl-iter.h): Include it.
32472 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32473 (avr_legitimate_combined_insn): ...and implementation.
32475 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32477 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32478 * config/s390/s390.c (s390_const_operand_ok)
32479 (s390_canonicalize_comparison, s390_extract_part)
32480 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32481 (s390_contiguous_bitmask_p, s390_rtx_costs)
32482 (legitimize_pic_address): Likewise.
32483 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32484 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32485 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32486 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32487 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32489 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32491 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32492 REGNO($0) == REGNO($1).
32494 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32496 * config/s390/linux.h(SIZE_TYPE): Add comment.
32498 2017-02-06 Julian Brown <julian@codesourcery.com>
32499 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32500 Virendra Pathak <virendra.pathak@broadcom.com>
32502 * config/aarch64/aarch64-cores.def: Change the scheduler
32504 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32505 * config/aarch64/thunderx2t99.md: New file.
32507 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32509 * doc/standards.texi (Go Language): Update link to language
32512 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32514 * tree-eh.c (lower_resx): Sanitize profile.
32515 (cleanup_empty_eh_move_lp): Likewise.
32517 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32520 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32522 * cfgloopmanip.h (loop_version): Update prototype.
32523 * modulo-sched.c (sms_schedule): Update call of loop_version.
32524 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32525 * tree-parloops.c (gen_parallel_loop): Likewise.
32526 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32527 * tree-ssa-loop-split.c (split_loop): Likewise.
32528 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32529 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32531 2017-02-05 Martin Liska <mliska@suse.cz>
32534 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32536 (print_operand_address): Initialize a struct to zero.
32538 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32540 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32541 garbage collector only in textual form.
32543 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32545 * doc/extend.texi (x86 specific memory model extensions for
32546 transactional memory): Simplify a phrase.
32548 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32551 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32552 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32553 (atomic_storedi_1): Likewise.
32555 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32557 PR tree-optimization/79338
32558 * tree-parloops.c (gather_scalar_reductions): Don't call
32559 vect_analyze_loop_form for loop->inner before destroying loop's
32562 2017-02-03 Martin Sebor <msebor@redhat.com>
32564 PR tree-optimization/79327
32565 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32566 when precision has resulted in leading zeros.
32567 (format_integer): Adjust the likely counter to assume an unknown
32568 argument that may be zero is non-zero.
32570 2017-02-03 Jason Merrill <jason@redhat.com>
32573 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32574 avoid copying non-taken branch.
32576 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32578 PR tree-optimization/79340
32579 * tree-vect-loop.c (vectorizable_reduction): Release
32580 vec_defs elements after safe_splicing them into other vectors.
32583 PR tree-optimization/79327
32584 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32585 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32587 (format_integer): Use wide_int_to_tree instead of build_int_cst
32588 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32589 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32590 of shortest and longest sequence.
32592 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32594 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32595 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32597 2017-02-03 Walter Lee <walt@tilera.com>
32600 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32601 after initial stackframe link reg save.
32602 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32604 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32607 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32608 wu for stxssp alternative.
32610 2017-02-03 Martin Sebor <msebor@redhat.com>
32612 PR tree-optimization/79352
32613 * gimple-fold.c (get_range_strlen): Add argument.
32614 (get_range_strlen): Change return type to bool.
32615 (get_maxval_strlen): Pass in a dummy argument.
32616 * gimple-fold.h (get_range_strlen): Change return type to bool.
32617 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32618 * tree.h (array_at_struct_end_p): Add argument.
32619 * tree.c (array_at_struct_end_p): Handle it.
32621 2017-02-03 Martin Liska <mliska@suse.cz>
32624 * multiple_target.c (create_dispatcher_calls): Redirect edge
32625 from a caller of a dispatcher.
32626 (expand_target_clones): Make the clones local.
32627 (ipa_target_clone): Do both target clones and resolvers.
32628 (ipa_dispatcher_calls): Remove the pass.
32629 (pass_dispatcher_calls::gate): Likewise.
32630 (make_pass_dispatcher_calls): Likewise.
32631 * passes.def (pass_target_clone): Put as very first IPA early
32634 2017-02-03 Martin Liska <mliska@suse.cz>
32636 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32637 in case of a function with ifunc attribute.
32639 2017-02-03 Martin Liska <mliska@suse.cz>
32641 * cgraph.c (cgraph_node::dump): Dump function version info.
32642 * symtab.c (symtab_node::dump_base): Add missing new line.
32644 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32646 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32647 (ifcombine_ifandif): Use it.
32649 2017-02-03 Martin Liska <mliska@suse.cz>
32651 * doc/invoke.texi: Document default value for
32652 use-after-scope-direct-emission-threshold.
32654 2017-02-03 Martin Liska <mliska@suse.cz>
32656 PR tree-optimization/79339
32657 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32658 (format_floating): Likewise.
32660 2017-02-03 Martin Liska <mliska@suse.cz>
32663 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32665 (ipa_node_params_t::remove): Likewise.
32666 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32667 initialization from removed ipa_node_params_t::insert.
32668 (ipa_node_params::~ipa_node_params): Move from removed
32669 ipa_node_params_t::release.
32670 * symbol-summary.h (symbol_summary::m_released): New member.
32671 Do not release a summary twice. Do not allow to call finalizer
32672 for types of a summary that live in GGC memory.
32674 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32676 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32679 2017-02-02 Martin Sebor <msebor@redhat.com>
32681 PR middle-end/79275
32682 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32683 (format_string): Tighten up the range of output for non-constant
32684 strings and correct the expected range for wide non-constant strings.
32686 2017-02-02 Martin Sebor <msebor@redhat.com>
32688 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32690 PR middle-end/32003
32691 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32693 (-fdump-tree-@var): Add to index and document how to come up
32694 with pass-specific option and dump file names.
32695 (-fdump-passes): Clarify where to look for output.
32697 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32699 PR middle-end/77445
32700 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32701 statistics of the analyzed path; allow threading for speed when
32702 any of BBs along the path are optimized for speed.
32704 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32706 PR middle-end/78468
32707 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32708 settings of the virtual registers.
32711 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32713 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32714 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32717 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32719 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32720 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32722 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32724 * config/s390/s390.md: Add missing comments with the expanded
32726 * config/s390/vector.md: Likewise.
32727 * config/s390/vx-builtins.md: Likewise.
32729 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32732 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32733 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32734 conditions on a single line.
32736 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32738 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32739 __S390_VX__ to __VX__.
32741 2017-02-01 Andrew Pinski <apinski@cavium.com>
32743 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32744 stmt_info to record_stmt_cost.
32745 (vect_get_known_peeling_cost): Pass stmt_info if known to
32747 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32748 cpu_vector_cost field into
32749 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32750 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32751 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32752 splitting of scalar_stmt_cost and vec_stmt_cost.
32753 (thunderx_vector_cost): Likewise.
32754 (cortexa57_vector_cost): LIkewise.
32755 (exynosm1_vector_cost): Likewise.
32756 (xgene1_vector_cost): Likewise.
32757 (thunderx2t99_vector_cost): Improve after the splitting of the two
32759 (aarch64_builtin_vectorization_cost): Update for the splitting of
32760 scalar_stmt_cost and vec_stmt_cost.
32762 2017-02-01 Torvald Riegel <triegel@redhat.com>
32763 Richard Henderson <rth@redhat.com>
32765 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32766 conditional on existance of a fast atomic load.
32767 * optabs-query.c (can_atomic_load_p): New function.
32768 * optabs-query.h (can_atomic_load_p): Declare it.
32769 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32770 no fast atomic load is available for the particular size of access.
32771 (expand_atomic_compare_and_swap): Likewise.
32772 (expand_atomic_load): Likewise.
32773 (expand_atomic_store): Likewise.
32774 (expand_atomic_fetch_op): Likewise.
32775 * testsuite/lib/target-supports.exp
32776 (check_effective_target_sync_int_128): Remove x86 because it provides
32777 no fast atomic load.
32778 (check_effective_target_sync_int_128_runtime): Likewise.
32780 2017-02-01 Richard Biener <rguenther@suse.de>
32782 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32783 (graphite_transform_loops): Provide opt-info for optimized nests.
32784 * tree-parloop.c (parallelize_loops): Provide opt-info for
32785 parallelized loops.
32787 2017-02-01 Richard Biener <rguenther@suse.de>
32789 PR middle-end/79315
32790 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32791 was not set before.
32793 2017-02-01 Richard Biener <rguenther@suse.de>
32795 PR tree-optimization/71824
32796 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32797 Verify the loops are valid in the merged SESE region.
32798 (scop_detection::can_represent_loop_1): Check analyzing the
32799 evolution of the number of iterations in the region succeeds.
32801 2017-01-31 Ian Lance Taylor <iant@golang.org>
32803 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32804 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32806 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32808 PR preprocessor/79210
32809 * input.c (get_substring_ranges_for_loc): Replace line_width
32810 assertion with error-handling.
32812 2017-01-31 Richard Biener <rguenther@suse.de>
32814 PR tree-optimization/77318
32815 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32816 (create_pw_aff_from_tree): Take loop parameter.
32817 (add_condition_to_pbb): Pass loop of the condition to
32818 create_pw_aff_from_tree.
32820 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32822 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32823 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32825 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32829 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32831 (convert_int_to_float128): Likewise.
32832 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32833 (convert_int_to_float128): Likewise.
32834 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32835 (UNSPEC_IEEE128_CONVERT): Likewise.
32836 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32837 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32838 Use local variables for IBM extended format.
32839 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32840 (fix_trunc<mode>si2_fprs): Likewise.
32841 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32842 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32843 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32844 to know that we can now have integers of all sizes in vector
32846 (fix<uns>_<mode>di2_hw): Likewise.
32847 (float<uns>_<mode>si2_hw): Likewise.
32848 (fix_<mode>si2_hw): Likewise.
32849 (fixuns_<mode>si2_hw): Likewise.
32850 (float<uns>_<mode>di2_hw): Likewise.
32851 (float_<mode>di2_hw): Likewise.
32852 (float_<mode>si2_hw): Likewise.
32853 (floatuns_<mode>di2_hw): Likewise.
32854 (floatuns_<mode>si2_hw): Likewise.
32855 (xscvqp<su>wz_<mode>): Delete, no longer used.
32856 (xscvqp<su>dz_<mode>): Likewise.
32857 (xscv<su>dqp_<mode>): Likewise.
32858 (ieee128_mfvsrd_64bit): Likewise.
32859 (ieee128_mfvsrd_32bit): Likewise.
32860 (ieee128_mfvsrwz): Likewise.
32861 (ieee128_mtvsrw): Likewise.
32862 (ieee128_mtvsrd_64bit): Likewise.
32863 (ieee128_mtvsrd_32bit): Likewise.
32865 2017-01-31 Martin Liska <mliska@suse.cz>
32868 * ipa-prop.c (ipa_free_all_node_params): Call release method
32869 instead of ~sumbol_summary to not to trigger double times
32872 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32874 PR tree-optimization/71691
32875 * bitmap.h (class auto_bitmap): New.
32876 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32877 is_maybe_undefined instead of ssa_undefined_value_p.
32879 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32881 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32882 __S390_ARCH_LEVEL__ to __ARCH__.
32884 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32886 PR tree-optimization/79267
32887 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32888 if should_remove_lhs_p is true.
32890 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32893 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32894 (add_alignment_attribute): New.
32895 (base_type_die): Add alignment attribute.
32896 (subrange_type_die): Likewise.
32897 (modified_type_die): Likewise.
32898 (gen_array_type_die): Likewise.
32899 (gen_descr_array_type_die: Likewise.
32900 (gen_enumeration_type_die): Likewise.
32901 (gen_subprogram_die): Likewise.
32902 (gen_variable_die): Likewise.
32903 (gen_field_die): Likewise.
32904 (gen_ptr_to_mbr_type_die): Likewise.
32905 (gen_struct_or_union_type_die): Likewise.
32906 (gen_subroutine_type_die): Likewise.
32907 (gen_typedef_die): Likewise.
32908 (base_type_cmp): Compare alignment attribute.
32910 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32913 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32914 (setb_unsigned) New pattern for setb with CCUNS.
32915 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32916 subfc./subfe sequence to avoid overflow problems. Generate a
32917 shorter sequence with cmpld/setb for power9.
32918 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32919 for generating subfc. instruction.
32920 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32921 now uses this instruction.
32923 2017-01-30 Ian Lance Taylor <iant@google.com>
32926 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32927 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32929 2017-01-30 Martin Sebor <msebor@redhat.com>
32931 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32932 Move constant to the right of a relational operator.
32933 (get_mpfr_format_length, format_character, format_string): Ditto.
32934 (should_warn_p, maybe_warn): Same.
32936 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32938 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32941 * asan.c (get_translation_unit_decl): Remove function.
32942 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32944 2017-01-30 Martin Liska <mliska@suse.cz>
32946 PR gcov-profile/79259
32947 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32948 -fprofile-generate.
32950 2017-01-30 Martin Liska <mliska@suse.cz>
32953 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32954 Initialize variables with NULL value.
32956 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32959 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32961 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32963 2017-01-30 Richard Biener <rguenther@suse.de>
32965 PR tree-optimization/79276
32966 * tree-vrp.c (process_assert_insertions): Properly adjust common
32967 when removing a duplicate.
32969 2017-01-30 Richard Biener <rguenther@suse.de>
32971 PR tree-optimization/79256
32972 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32973 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32975 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32977 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32980 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32981 ("*r<noxa>sbg_<mode>_sll_bitmask")
32982 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32983 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32984 Use contiguous_bitmask_nowrap_operand.
32986 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32989 * config/rs6000/altivec.h (vec_xl): Revise #define.
32990 (vec_xst): Likewise.
32992 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32994 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32996 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32998 PR rtl-optimization/79194
32999 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
33000 traps before call to bypass_conditional_jumps.
33002 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33004 PR tree-optimization/71374
33005 * lra-constraints.c (check_conflict_input_operands): New.
33006 (match_reload): Use it.
33008 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
33011 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
33012 account to calculate conflict_set.
33014 2017-01-27 Bin Cheng <bin.cheng@arm.com>
33016 PR rtl-optimization/78559
33017 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
33018 other_insn in combine.
33020 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33022 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
33023 uint16_type_node for BT_UINT16.
33025 2017-01-27 David Malcolm <dmalcolm@redhat.com>
33027 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
33028 "RTL Tests" to menu.
33029 (GIMPLE Tests): New node.
33030 (RTL Tests): New node.
33032 2017-01-27 Richard Biener <rguenther@suse.de>
33034 PR tree-optimization/79245
33035 * tree-loop-distribution.c (distribute_loop): Apply cost
33036 modeling also to detected patterns.
33038 2017-01-27 Richard Biener <rguenther@suse.de>
33040 PR tree-optimization/71433
33041 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33042 (compare_assert_loc): New function.
33043 (process_assert_insertions): Sort and optimize assert locations
33044 to remove duplicates and push down identical assertions on
33045 edges to their destination block.
33047 2017-01-27 Richard Biener <rguenther@suse.de>
33049 PR tree-optimization/79244
33050 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33051 out SSA names even if abnormal.
33053 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33055 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33056 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33057 instead of MPFR_RNDN.
33059 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33062 * arm.c (arm_option_override): Don't call build_target_option_node
33063 until after doing all option overrides.
33064 (arm_valid_target_attribute_tree): Likewise.
33066 2017-01-27 Martin Liska <mliska@suse.cz>
33068 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33069 for {cd}tors and C++ {cd}tors.
33071 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33073 * config/s390/s390.md ("*setmem_long_and")
33074 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33076 2017-01-26 Martin Sebor <msebor@redhat.com>
33078 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33081 2017-01-26 Martin Sebor <msebor@redhat.com>
33083 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33084 HAVE_DFmode before using XFmode or DFmode.
33085 (parse_directive): Avoid using the z length modifier to avoid
33086 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33088 PR middle-end/78703
33089 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33090 to accept adjustment as an array.
33091 (get_int_range): New function.
33092 (struct directive): Make width and prec arrays.
33093 (directive::set_width, directive::set_precision): Call get_int_range.
33094 (format_integer, format_floating): Handle width and precision ranges.
33095 (format_string, parse_directive): Same.
33097 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33100 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33101 comdat_type_p set, just clone them, but keep the children in the
33105 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33106 which have direct callers with -fvar-tracking-assignments enabled
33108 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33109 inside of type units.
33111 2017-01-26 Martin Sebor <msebor@redhat.com>
33113 PR middle-end/78703
33114 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33116 (struct format_result): Replace number_chars, number_chars_min,
33117 and number_chars_max with a single member of struct result_range.
33119 (format_result::operator+=): Adjust.
33120 (struct fmtresult): Remove bounded. Handle likely and unlikely
33122 (fmtresult::adjust_for_width_or_precision): New function.
33123 (fmtresult:type_max_digits): New function.
33124 (bytes_remaining): Handle likely and unlikely counters.
33125 (min_bytes_remaining): Remove.
33126 (format_percent): Simplify.
33127 (format_integer, format_floating): Set likely and unlikely counters.
33128 (get_string_length, format_character, format_string): Same.
33129 (format_plain, should_warn_p): New function.
33130 (maybe_warn): Call should_warn_p. Update diagnostic messages
33131 and handle those for all directives, including plain strings.
33132 (format_directive): Handle likely and unlikely counters.
33133 Remove unnecessary quoting from diagnostics. Add an informational
33135 (add_bytes): Remove.
33136 (pass_sprintf_length::compute_format_length): Simplify.
33137 (try_substitute_return_value): Handle likely and unlikely counters.
33139 2017-01-26 Carl Love <cel@us.ibm.com>
33141 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33142 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33144 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33147 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33148 endianess for subregs into account.
33149 * lra-constraints.c (lra_constraints): Do risky transformations
33150 always on the first iteration.
33151 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33152 last_call_used_reg_set.
33153 (process_bb_lives): Define and use last_call_used_reg_set.
33154 * lra.c (lra): Always continue after lra_constraints on the first
33157 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33159 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33161 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33163 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33165 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33166 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33167 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33168 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33169 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33170 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33171 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33172 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33173 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33175 2017-01-26 Marek Polacek <polacek@redhat.com>
33178 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33179 for the third operand.
33181 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33183 PR middle-end/79236
33184 * omp-low.c (struct omp_context): Add simt_stmt field.
33185 (scan_omp_for): Return omp_context *.
33186 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33187 context to the _simt_ SIMD stmt.
33188 (lower_omp_for): For combined SIMD with sibling _simt_
33189 SIMD, make sure to use the same decls in _looptemp_
33190 clauses as in the sibling.
33192 2017-01-26 David Sherwood <david.sherwood@arm.com>
33194 PR middle-end/79212
33195 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33198 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33201 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33202 emit fld b; fld a; if possible.
33204 * brig-builtins.def: Update copyright years.
33205 * config/arm/arm_acle_builtins.def: Update copyright years.
33207 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33210 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33211 constraint instead of o for the stxsd instruction.
33213 2017-01-25 Carl Love <cel@us.ibm.com>
33215 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33216 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33218 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33220 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33222 2017-01-25 Richard Biener <rguenther@suse.de>
33224 PR tree-optimization/69264
33225 * target.def (vector_alignment_reachable): Improve documentation.
33226 * doc/tm.texi: Regenerate.
33227 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33229 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33230 earlier changes with respect to TYPE_USER_ALIGN.
33231 (vector_alignment_reachable_p): Likewise. Improve dumping.
33233 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33236 * config/arm/arm.md (xordi3): Force constant operand into a register
33239 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33241 * doc/invoke.texi (-fstore-merging): Correct default optimization
33242 levels at which it is enabled.
33243 (-O): Move -fstore-merging from list to...
33246 2017-01-25 Richard Biener <rguenther@suse.de>
33249 * omp-expand.c: Include debug.h.
33250 (expand_omp_taskreg): Make sure to generate early debug before
33251 outlining anything from a function.
33252 (expand_omp_target): Likewise.
33253 (grid_expand_target_grid_body): Likewise.
33255 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33258 * asan.c (get_translation_unit_decl): New function.
33259 (asan_add_global): Extract modules file name from globals
33260 TRANSLATION_UNIT_DECL name.
33262 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33265 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33266 for long calls with APCS frame and VFP.
33268 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33270 * cfg.c (original_copy_tables_initialized_p): New function.
33271 * cfg.h (original_copy_tables_initialized_p): New decl.
33272 * cfgrtl.c (relink_block_chain): Guard the call to
33273 free_original_copy_tables with a call to
33274 original_copy_tables_initialized_p.
33275 * cgraph.h (symtab_node::native_rtl_p): New decl.
33276 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33277 (symtab_node::needed_p): Don't assert for early assembly output
33278 for __RTL functions.
33279 (cgraph_node::finalize_function): Set "force_output" for __RTL
33281 (cgraph_node::analyze): Bail out early for __RTL functions.
33282 (analyze_functions): Update assertion to support __RTL functions.
33283 (cgraph_node::expand): Bail out early for __RTL functions.
33284 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33286 * function.h (struct function): Update comment for field
33288 * gimple-expr.c: Include "tree-pass.h".
33289 (gimple_has_body_p): Return false for __RTL functions.
33290 * Makefile.in (OBJS): Add run-rtl-passes.o.
33291 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33293 (gcc::pass_manager::get_clean_slate): New accessor.
33294 * passes.c: Include "insn-addr.h".
33295 (should_skip_pass_p): Add logging. Update logic for running
33296 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33297 property-provider override so it is only done for gimple passes.
33299 (skip_pass): New function.
33300 (execute_one_pass): Call skip_pass when skipping passes.
33301 * read-md.c (md_reader::read_char): Support filtering
33302 the input to a subset of line numbers.
33303 (md_reader::md_reader): Initialize fields
33304 m_first_line and m_last_line.
33305 (md_reader::read_file_fragment): New function.
33306 * read-md.h (md_reader::read_file_fragment): New decl.
33307 (md_reader::m_first_line): New field.
33308 (md_reader::m_last_line): New field.
33309 * read-rtl-function.c (function_reader::create_function): Only
33310 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33311 curr_properties. Set DECL_INITIAL to a dummy block.
33312 (read_rtl_function_body_from_file_range): New function.
33313 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33315 * run-rtl-passes.c: New file.
33316 * run-rtl-passes.h: New file.
33318 2017-01-24 Jeff Law <law@redhat.com>
33320 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33323 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33325 PR tree-optimization/79159
33326 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33327 (record_nonwrapping_iv): Improve boundary using above function if no
33328 value range information.
33330 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33331 Martin Jambor <mjambor@suse.cz>
33333 * brig-builtins.def: New file.
33334 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33335 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33336 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33337 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33338 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33339 * builtin-types.def (BT_INT8): New.
33340 (BT_INT16): Likewise.
33341 (BT_UINT8): Likewise.
33342 (BT_UINT16): Likewise.
33343 (BT_FN_ULONG): Likewise.
33344 (BT_FN_UINT_INT): Likewise.
33345 (BT_FN_UINT_ULONG): Likewise.
33346 (BT_FN_UINT_LONG): Likewise.
33347 (BT_FN_UINT_PTR): Likewise.
33348 (BT_FN_ULONG_PTR): Likewise.
33349 (BT_FN_INT8_FLOAT): Likewise.
33350 (BT_FN_INT16_FLOAT): Likewise.
33351 (BT_FN_UINT32_FLOAT): Likewise.
33352 (BT_FN_UINT16_FLOAT): Likewise.
33353 (BT_FN_UINT8_FLOAT): Likewise.
33354 (BT_FN_UINT64_FLOAT): Likewise.
33355 (BT_FN_UINT16_UINT32): Likewise.
33356 (BT_FN_UINT32_UINT16): Likewise.
33357 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33358 (BT_FN_INT_PTR_INT): Likewise.
33359 (BT_FN_UINT_PTR_UINT): Likewise.
33360 (BT_FN_LONG_PTR_LONG): Likewise.
33361 (BT_FN_ULONG_PTR_ULONG): Likewise.
33362 (BT_FN_VOID_UINT64_UINT64): Likewise.
33363 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33364 (BT_FN_INT8_INT8_INT8): Likewise.
33365 (BT_FN_INT16_INT16_INT16): Likewise.
33366 (BT_FN_INT_INT_INT): Likewise.
33367 (BT_FN_UINT_FLOAT_UINT): Likewise.
33368 (BT_FN_FLOAT_UINT_UINT): Likewise.
33369 (BT_FN_ULONG_UINT_UINT): Likewise.
33370 (BT_FN_ULONG_UINT_PTR): Likewise.
33371 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33372 (BT_FN_UINT_UINT_UINT): Likewise.
33373 (BT_FN_VOID_UINT_PTR): Likewise.
33374 (BT_FN_UINT_UINT_PTR: Likewise.
33375 (BT_FN_UINT32_UINT64_PTR): Likewise.
33376 (BT_FN_INT_INT_UINT_UINT): Likewise.
33377 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33378 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33379 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33380 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33381 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33382 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33383 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33384 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33385 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33386 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33387 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33388 * doc/frontends.texi: List BRIG FE.
33389 * doc/install.texi (Testing): Add BRIG tesring requirements.
33390 * doc/invoke.texi (Overall Options): Mention BRIG.
33391 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33393 2017-01-24 Richard Biener <rguenther@suse.de>
33395 PR translation/79208
33396 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33398 2017-01-24 Martin Jambor <mjambor@suse.cz>
33401 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33402 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33403 and known_contexts.
33405 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33407 PR middle-end/79123
33408 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33409 casts from signed to unsigned really don't have a range.
33411 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33413 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33414 GMP_RNDx for compatiblity.
33416 2017-01-24 Martin Liska <mliska@suse.cz>
33419 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33420 that would prevent us to call alloca with -1 as argument.
33422 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33424 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33425 Avoid -Wformat-security warning.
33427 2017-01-23 Andrew Pinski <apinski@cavium.com>
33429 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33432 2017-01-23 Martin Sebor <msebor@redhat.com>
33434 PR middle-end/78703
33435 * gimple-ssa-sprintf.c (warn_level): New global.
33436 (format_integer): Use it here and throughout the rest of the file.
33437 Use the same switch to compute sign as base.
33438 (maybe_warn): New function.
33439 (format_directive): Factor out warnings into maybe_warn.
33440 Add debugging output. Use warn_level.
33441 (add_bytes): Use warn_level.
33442 (pass_sprintf_length::compute_format_length): Add debugging output.
33443 (try_substitute_return_value): Same.
33444 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33446 PR middle-end/78703
33447 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33448 (struct fmtresult, format_integer, format_floating): Adjust.
33449 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33450 (get_string_length, format_string,format_directive): Same.
33451 (pass_sprintf_length::compute_format_length): Same.
33452 (try_substitute_return_value): Simplify slightly.
33454 PR middle-end/78703
33455 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33456 (fmtresult::operator+=): Outlined.
33457 (struct fmtresult): Add ctors.
33458 (struct conversion_spec): Rename...
33459 (struct directive): ...to this. Add and remove data members.
33460 (directive::set_width, directive::set_precision): New functions.
33461 (format_percent): Use fmtresult ctor.
33462 (get_width_and_precision): Remove.
33463 (format_integer): Make naming changes. Avoid computing width and
33465 (format_floating): Same. Adjust indentation.
33466 (format_character, format_none): New functions.
33467 (format_string): Moved character handling to format_character.
33468 (format_directive): Remove arguments, change return type.
33469 (parse_directive): New function.
33470 (pass_sprintf_length::compute_format_length): Move directive
33471 parsing to parse_directive.
33473 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33475 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33476 (assign_assembler_name_if_needed): ... this.
33477 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33478 (assign_assembler_name_if_needed): ... this.
33479 (free_lang_data_in_cgraph): Adjust callers.
33480 * cgraphunit.c (cgraph_node::analyze): Likewise.
33481 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33484 2017-01-23 Richard Biener <rguenther@suse.de>
33486 PR tree-optimization/79088
33487 PR tree-optimization/79188
33488 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33489 resetting loop bounds after last path deletion. Reset loop
33490 bounds of the target loop, make code match the comments.
33491 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33492 Make sure loops need no fixups.
33494 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33496 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33497 exponent support with double type for first argument.
33498 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33499 type returned by __builtin_vec_extract_sig,
33500 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33501 functions from "vector int" to "vector unsigned int" or from
33502 "vector long long int" to "vector unsigned long long int".
33503 Changed type returned by __builtin_vec_extract_exp,
33504 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33505 functions from "vector int" to "vector unsigned int" or from
33506 "vector long long int" to "vector unsigned long long int".
33507 Changed return type of __builtin_vec_test_data_class,
33508 __builtin_vec_test_data_class_sp, and
33509 __builtin_vec_test_data_class_dp from "vector int" to
33510 "vector bool int" or from "vector long long int" to "vector bool
33511 long long int" and changed second argument type from "unsigned
33512 int" to "int". Added new overloaded function forms "vector float
33513 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33514 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33515 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33516 double, vector unsigned long long int)" and "vector double
33517 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33518 long int)". Changed return type of
33519 __builtin_scalar_test_data_class and
33520 __builtin_scalar_test_data_class_sp and
33521 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33522 int" and changed second argument from "unsigned int" to "int".
33523 Changed type returned by __builtin_scalar_test_neg,
33524 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33525 from "int" to "bool int". Added new overloaded function form
33526 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33527 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33528 exponent double-precision with floating point first argument.
33529 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33530 documentation of scalar_test_data_class, scalar_test_neg,
33531 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33532 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33533 vec_test_data_class built-in functions to reflect refinements in
33534 their type signatures.
33536 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33538 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33540 (aarch64_elf_asm_destructor): Likewise.
33542 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33544 PR rtl-optimization/78634
33545 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33546 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33547 * ifcvt.c (noce_try_cmove): Add missing cost check.
33549 PR rtl-optimization/71724
33550 * combine.c (if_then_else_cond): Look for situations where it is
33551 beneficial to undo the work of one of the recursive calls.
33553 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33555 PR tree-optimization/70754
33556 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33557 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33558 combined stmt before it if not NULL.
33559 (combine_chains): Process refs reversely and compute dominance point
33562 2017-01-23 Martin Liska <mliska@suse.cz>
33564 PR tree-optimization/79196
33565 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33566 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33568 (strlen_optimize_stmt): Call the renamed function.
33570 2017-01-23 Michael Matz <matz@suse.de>
33572 PR tree-optimization/78384
33573 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33575 2017-01-23 Richard Biener <rguenther@suse.de>
33577 PR tree-optimization/79186
33578 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33579 both incoming edges before moving an assert.
33581 2017-01-23 Martin Jambor <mjambor@suse.cz>
33583 * ipa-prop.c (load_from_param_1): Removed.
33584 (load_from_unmodified_param): Bits from load_from_param_1 put back
33586 (load_from_param): Removed.
33587 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33588 with stmt. Reverted back to use of load_from_unmodified_param.
33590 2017-01-23 Martin Jambor <mjambor@suse.cz>
33593 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33594 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33595 field a pointer to garbage collected vector, mark lattices and
33596 ipcp_orig_node with GTY((skip)).
33597 (ipa_get_param_count): Adjust to descriptors being a pointer.
33598 (ipa_get_param): Likewise.
33599 (ipa_get_type): Likewise.
33600 (ipa_get_param_move_cost): Likewise.
33601 (ipa_set_param_used): Likewise.
33602 (ipa_get_controlled_uses): Likewise.
33603 (ipa_set_controlled_uses): Likewise.
33604 (ipa_is_param_used): Likewise.
33605 (ipa_node_params_t): Move into garbage collector. New methods insert
33607 (ipa_node_params_sum): Annotate wth GTY(()).
33608 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33610 (ipa_load_from_parm_agg): Adjust declaration.
33611 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33612 * ipa-profile.c (ipa_profile): Likewise.
33613 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33614 (ipa_populate_param_decls): Make descriptors parameter garbage
33616 (ipa_dump_param): Adjust to descriptors being a pointer.
33617 (ipa_alloc_node_params): Likewise.
33618 (ipa_initialize_node_params): Likewise.
33619 (load_from_param_1): Make descriptors parameter garbage collected.
33620 (load_from_unmodified_param): Likewise.
33621 (load_from_param): Likewise.
33622 (ipa_load_from_parm_agg): Likewise.
33623 (ipa_node_params::~ipa_node_params): Removed.
33624 (ipa_free_all_node_params): Remove call to delete operator.
33625 (ipa_node_params_t::insert): New.
33626 (ipa_node_params_t::remove): Likewise.
33627 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33628 copy known_csts and known_contexts vectors.
33629 (ipa_read_node_info): Adjust to descriptors being a pointer.
33630 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33632 (ipcp_transform_function): Make descriptors variable garbage
33635 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33637 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33638 * config/i386/avx512dqintrin.h: Ditto.
33639 * config/i386/avx512fintrin.h: Ditto.
33640 * config/i386/i386.c: Handle new builtins.
33641 * config/i386/i386-builtin.def: Add new builtins.
33642 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33643 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33645 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33646 Martin Liska <mliska@suse.cz>
33648 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33649 * asan.c (asan_expand_poison_ifn): Support stores and use
33650 appropriate ASAN report function.
33651 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33652 * internal-fn.def (ASAN_POISON_USE): Declare.
33653 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33654 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33655 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33656 ASAN_POISON calls w/o LHS.
33657 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33658 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33659 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33660 * gimplify.c (asan_poison_variables): Add attribute
33661 use_after_scope_memory to variables that really needs to live
33663 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33664 having the attribute.
33666 2017-01-23 Martin Liska <mliska@suse.cz>
33668 * asan.c (create_asan_shadow_var): New function.
33669 (asan_expand_poison_ifn): Likewise.
33670 * asan.h (asan_expand_poison_ifn): New declaration.
33671 * internal-fn.c (expand_ASAN_POISON): Likewise.
33672 * internal-fn.def (ASAN_POISON): New builtin.
33673 * sanopt.c (pass_sanopt::execute): Expand
33674 asan_expand_poison_ifn.
33675 * tree-inline.c (copy_decl_for_dup_finish): Make function
33677 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33678 * tree-ssa.c (is_asan_mark_p): New function.
33679 (execute_update_addresses_taken): Rewrite local variables
33680 (identified just by use-after-scope as addressable) into SSA.
33682 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33684 * doc/install.texi (Specific): opensource.apple.com uses https
33685 now. Remove trailing slash.
33687 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33689 * README.Portability: Remove note on an Irix compatibility issue.
33691 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33693 * gcov.c (INCLUDE_ALGORITHM): Define.
33694 (INCLUDE_VECTOR): Define.
33695 No longer include <vector> and <algorithm> directly.
33697 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33699 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33701 * doc/invoke.texi (Code Gen Options): Ditto.
33703 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33706 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33708 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33710 rtl-optimization/79125
33711 * cprop.c (local_cprop_pass): Handle cases where we make an
33712 unconditional trap.
33714 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33718 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33719 read from, for big endian.
33721 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33723 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33724 register pauth builtins for LP64 only.
33726 2017-01-20 Marek Polacek <polacek@redhat.com>
33729 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33732 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33734 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33736 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33737 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33738 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33740 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33743 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33744 in big-endian mode when they are not a single duplicated value.
33746 2017-01-20 Richard Biener <rguenther@suse.de>
33748 * BASE-VER: Bump to 7.0.1.
33750 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33752 * omp-low.c (omplow_simd_context): New struct. Use it...
33753 (lower_rec_simd_input_clauses): ...here and...
33754 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33755 references to idx, lane, max_vf, is_simt.
33757 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33759 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33762 2017-01-20 Martin Jambor <mjambor@suse.cz>
33764 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33765 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33767 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33768 (GTFILES): Rename hsa.c to hsa-common.c.
33769 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33770 * hsa-dump.c: Likewise.
33771 * hsa-gen.c: Likewise.
33772 * hsa-regalloc.c: Likewise.
33773 * ipa-hsa.c: Likewise.
33774 * omp-expand.c: Likewise.
33775 * omp-low.c: Likewise.
33776 * toplev.c: Likewise.
33778 2017-01-20 Marek Polacek <polacek@redhat.com>
33781 * doc/invoke.texi: Document -Wduplicated-branches.
33782 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33783 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33784 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33785 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33786 return 0 only when not OEP_LEXICOGRAPHIC.
33787 (fold_build_cleanup_point_expr): Use the expression
33788 location when building CLEANUP_POINT_EXPR.
33789 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33790 * tree.c (add_expr): Handle error_mark_node.
33792 2017-01-20 Martin Liska <mliska@suse.cz>
33795 * tree-profile.c (init_ic_make_global_vars): Do not call
33797 (gimple_init_gcov_profiler): Likewise.
33799 2017-01-20 Martin Liska <mliska@suse.cz>
33802 * cgraph.h (maybe_create_reference): Remove argument and
33804 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33806 * ipa-cp.c (create_specialized_node): Likewise.
33807 * symtab.c (symtab_node::maybe_create_reference): Handle
33808 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33810 2017-01-20 Martin Liska <mliska@suse.cz>
33812 * read-rtl-function.c (function_reader::create_function): Use
33813 build_decl instread of build_decl_stat.
33815 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33817 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33818 * config/i386/avx512dqintrin.h: Ditto.
33819 * config/i386/avx512fintrin.h: Ditto.
33820 * config/i386/i386-builtin-types.def: Add new types.
33821 * config/i386/i386.c: Handle new types.
33822 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33823 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33824 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33825 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33826 (__builtin_ia32_kshiftridi): New.
33827 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33829 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33833 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33834 define to rs6000_init_stack_protect_guard.
33835 (rs6000_init_stack_protect_guard): New function.
33837 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33838 Yunqiang Su <yunqiang.su@imgtec.com>
33840 * config.gcc (supported_defaults): Add madd4.
33841 (with_madd4): Add validation.
33842 (all_defaults): Add madd4.
33843 * config/mips/mips.opt (mmadd4): New option.
33844 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33846 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33848 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33849 (ISA_HAS_FUSED_MADD4): Likewise.
33850 * doc/invoke.texi (-mmadd4): Document the new option.
33851 * doc/install.texi (--with-madd4): Document the new option.
33853 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33855 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33856 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33857 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33858 (aarch64_init_pauth_hint_builtins): New.
33859 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33860 (aarch64_expand_builtin): Expand new builtins.
33862 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33864 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33865 * combine-stack-adj.c (no_unhandled_cfa): Handle
33866 REG_CFA_TOGGLE_RA_MANGLE.
33867 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33868 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33869 info for return address signing.
33870 (aarch64_expand_epilogue): Likewise.
33872 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33874 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33875 * config/aarch64/aarch64-protos.h
33876 (aarch64_return_address_signing_enabled): New declaration.
33877 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33879 (aarch64_expand_prologue): Sign return address before it's pushed onto
33881 (aarch64_expand_epilogue): Authenticate return address fetched from
33883 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33884 (aarch64_attributes): New function attributes for "sign-return-address".
33885 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33886 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33887 ("*do_return"): Generate combined instructions according to key index.
33888 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33889 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33891 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33892 * config/aarch64/aarch64.opt (msign-return-address=): New.
33893 * doc/extend.texi (AArch64 Function Attributes): Documents
33894 "sign-return-address=".
33895 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33897 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33899 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33900 overall option summary.
33902 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33904 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33905 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33906 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33907 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33909 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33911 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33912 -mpower9-minmax by default for -mcpu=power9.
33913 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33914 128-bit floating point.
33916 2017-01-20 Alan Modra <amodra@gmail.com>
33918 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33919 optimizing for size.
33921 2017-01-20 Alan Modra <amodra@gmail.com>
33924 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33925 for strcmp and strncmp from corresponding builtin decl.
33927 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33929 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33930 instead of i386/rtems-64.h.
33931 * config/i386/rtems-64.h: Remove.
33933 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33937 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33939 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33941 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33943 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33944 Change int to HOST_WIDE_INT.
33945 * config/aarch64/aarch64-protos.h
33946 (aarch64_simd_gen_const_vector_dup): Likewise.
33947 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33949 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33951 * langhooks-def.h (lhd_type_for_size): New decl.
33952 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33953 * langhooks.c (lhd_type_for_size): New function, taken from
33956 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33958 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33959 define_bypass for CR latency.
33960 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33961 (power9-alu2): Add define_bypass for CR latency.
33963 (power9-mul): Update insn latency.
33964 (power9-mul-compare): Update insn latency, bypass latency and remove
33967 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33969 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33971 * config/aarch64/aarch64.md
33972 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33973 aarch64_nopcrelative_literal_loads.
33974 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33976 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33978 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33979 TARGET_LOONGSON_3A.
33980 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33982 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33985 * config.gcc (supported_defaults): Add lxc1-sxc1.
33986 (with_lxc1_sxc1): Add validation.
33987 (all_defaults): Add lxc1-sxc1.
33988 * config/mips/mips.opt (mlxc1-sxc1): New option.
33989 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33991 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33992 __mips_no_lxc1_sxc1.
33993 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33994 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33995 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33997 2017-01-19 Richard Biener <rguenther@suse.de>
33999 PR tree-optimization/72488
34000 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
34001 sure to restore SSA info.
34002 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
34004 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
34006 PR rtl-optimization/79121
34007 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
34008 of the inner type when shifting an extended value.
34010 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34013 * symtab.c (symtab_node::equal_address_to): Fix comparing of
34014 interposable aliases.
34016 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
34019 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
34020 Use the evmergelohi instruction.
34021 (mov_si<mode>_e500_subreg4_2_le): Likewise.
34022 (mov_sitf_e500_subreg8_2_be): Likewise.
34023 (mov_sitf_e500_subreg12_2_le): Likewise.
34024 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
34025 (mov_si<mode>_e500_subreg4_2_be): Likewise.
34026 (mov_sitf_e500_subreg8_2_le): Likewise.
34027 (mov_sitf_e500_subreg12_2_be): Likewise.
34029 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34031 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
34032 attribute from vecsimple to vecperm.
34033 (altivec_vbpermq2): Likewise.
34035 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34038 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34040 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34041 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34042 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34043 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34044 case where N arg is SIZE_MAX.
34045 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34046 (cmpstrsi): Add pattern.
34048 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34051 __builtin_vec_revb builtins.
34052 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34053 built-in functions to support generation of the ISA 3.0 XXBR<x>
34054 vector byte reverse instructions.
34055 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34056 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34057 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34058 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34059 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34060 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34061 (P9V_BUILTIN_VEC_REVB): Likewise.
34062 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34063 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34064 (p9_xxbrq_v16qi): Likewise.
34065 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34066 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34067 (p9_xxbrh_v8hi): Likewise.
34068 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34069 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34070 vec_revb built-in functions.
34072 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34074 PR rtl-optimization/78952
34075 * config/i386/i386.md (any_extract): New code iterator.
34076 (*insvqi_2): Use any_extract for source operand.
34077 (*insvqi_3): Use any_shiftrt for source operand.
34079 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34081 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34083 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34085 2017-01-18 Matthias Klose <doko@ubuntu.com>
34087 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34089 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34091 * config/rs6000/altivec.h (vec_bperm): Change #define.
34092 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34093 (altivec_vbpermq2): New define_insn.
34094 (altivec_vbpermd): Likewise.
34095 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34096 function interface.
34097 (VBPERMD): Likewise.
34098 (VBPERM): New polymorphic function interface.
34099 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34100 Add entries for P9V_BUILTIN_VEC_VBPERM.
34101 * doc/extend.texi: Add interfaces for vec_bperm.
34103 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34105 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34106 first letter of error messages.
34107 (s390_resolve_overloaded_builtin): Likewise.
34108 * config/s390/s390.c (s390_expand_builtin): Likewise.
34109 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34110 (s390_valid_target_attribute_inner_p): Likewise.
34111 * config/s390/s390.md ("tabort"): Likewise.
34113 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34115 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34116 (ISA_AVOID_DIV_HILO): New macro.
34117 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34118 (ISA_HAS_DDIV): Likewise.
34120 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34122 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34124 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34126 * doc/invoke.texi (fabi-version): Spelling fix.
34128 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34131 * doc/invoke.texi (fabi-version): Mention mangling fix for
34134 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34137 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34139 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34142 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34143 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34145 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34147 (stack_protect_test): Ditto.
34148 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34149 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34151 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34152 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34153 -mstack-protector-guard-offset=.
34154 (RS/6000 and PowerPC Options): Ditto.
34156 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34158 * config/i386/i386.h (MASK_CLASS_P): New define.
34159 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34160 there are no registers from different register sets also when
34161 mask registers are used. Update function comment.
34162 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34163 to (*k/*r) and (*k/*km) alternatives.
34165 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34167 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34168 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34169 (EH_RETURN_HANDLER_RTX): New define.
34170 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34171 Force frame pointer in EH return functions.
34172 (aarch64_expand_epilogue): Add barrier for eh_return.
34173 (aarch64_final_eh_return_addr): Remove.
34174 (aarch64_eh_return_handler_rtx): New function.
34175 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34177 (aarch64_eh_return_handler_rtx): New prototype.
34179 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34181 * config/rs6000/altivec.h (vec_rlmi): New #define.
34182 (vec_vrlnm): Likewise.
34183 (vec_rlnm): Likewise.
34184 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34185 (UNSPEC_VRLNM): Likewise.
34186 (VIlong): New mode iterator.
34187 (altivec_vrl<VI_char>mi): New define_insn.
34188 (altivec_vrl<VI_char>nm): Likewise.
34189 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34191 (VRLDNM): Likewise.
34192 (RLNM): New polymorphic function entry.
34193 (VRLWMI): New monomorphic function entry.
34194 (VRLDMI): Likewise.
34195 (RLMI): New polymorphic function entry.
34196 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34197 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34198 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34201 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34204 * dwarf2out.c (field_byte_offset): Restore the
34205 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34206 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34207 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34210 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34213 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34215 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34218 * dwarf2out.c (add_data_member_location_attribute): For constant
34219 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34220 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34221 DW_AT_byte_size attributes.
34223 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34225 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34226 after forcing to constant memory when the code model is medium.
34228 2017-01-17 Julia Koval <julia.koval@intel.com>
34231 * config/i386/avx512fintrin.h
34232 (_mm512_i32gather_ps): Change __addr type to void const*.
34233 (_mm512_mask_i32gather_ps): Ditto.
34234 (_mm512_i32gather_pd): Ditto.
34235 (_mm512_mask_i32gather_pd): Ditto.
34236 (_mm512_i64gather_ps): Ditto.
34237 (_mm512_mask_i64gather_ps): Ditto.
34238 (_mm512_i64gather_pd): Ditto.
34239 (_mm512_mask_i64gather_pd): Ditto.
34240 (_mm512_i32gather_epi32): Ditto.
34241 (_mm512_mask_i32gather_epi32): Ditto.
34242 (_mm512_i32gather_epi64): Ditto.
34243 (_mm512_mask_i32gather_epi64): Ditto.
34244 (_mm512_i64gather_epi32): Ditto.
34245 (_mm512_mask_i64gather_epi32): Ditto.
34246 (_mm512_i64gather_epi64): Ditto.
34247 (_mm512_mask_i64gather_epi64): Ditto.
34248 (_mm512_i32scatter_ps): Change __addr type to void*.
34249 (_mm512_mask_i32scatter_ps): Ditto.
34250 (_mm512_i32scatter_pd): Ditto.
34251 (_mm512_mask_i32scatter_pd): Ditto.
34252 (_mm512_i64scatter_ps): Ditto.
34253 (_mm512_mask_i64scatter_ps): Ditto.
34254 (_mm512_i64scatter_pd): Ditto.
34255 (_mm512_mask_i64scatter_pd): Ditto.
34256 (_mm512_i32scatter_epi32): Ditto.
34257 (_mm512_mask_i32scatter_epi32): Ditto.
34258 (_mm512_i32scatter_epi64): Ditto.
34259 (_mm512_mask_i32scatter_epi64): Ditto.
34260 (_mm512_i64scatter_epi32): Ditto.
34261 (_mm512_mask_i64scatter_epi32): Ditto.
34262 (_mm512_i64scatter_epi64): Ditto.
34263 (_mm512_mask_i64scatter_epi64): Ditto.
34264 * config/i386/avx512pfintrin.h
34265 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34266 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34267 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34268 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34269 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34270 (_mm512_prefetch_i32scatter_ps): Ditto.
34271 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34272 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34273 (_mm512_prefetch_i64scatter_pd): Ditto.
34274 (_mm512_prefetch_i64scatter_ps): Ditto.
34275 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34276 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34277 * config/i386/avx512vlintrin.h
34278 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34279 (_mm_mmask_i32gather_ps): Ditto.
34280 (_mm256_mmask_i32gather_pd): Ditto.
34281 (_mm_mmask_i32gather_pd): Ditto.
34282 (_mm256_mmask_i64gather_ps): Ditto.
34283 (_mm_mmask_i64gather_ps): Ditto.
34284 (_mm256_mmask_i64gather_pd): Ditto.
34285 (_mm_mmask_i64gather_pd): Ditto.
34286 (_mm256_mmask_i32gather_epi32): Ditto.
34287 (_mm_mmask_i32gather_epi32): Ditto.
34288 (_mm256_mmask_i32gather_epi64): Ditto.
34289 (_mm_mmask_i32gather_epi64): Ditto.
34290 (_mm256_mmask_i64gather_epi32): Ditto.
34291 (_mm_mmask_i64gather_epi32): Ditto.
34292 (_mm256_mmask_i64gather_epi64): Ditto.
34293 (_mm_mmask_i64gather_epi64): Ditto.
34294 (_mm256_i32scatter_ps): Change __addr type to void*.
34295 (_mm256_mask_i32scatter_ps): Ditto.
34296 (_mm_i32scatter_ps): Ditto.
34297 (_mm_mask_i32scatter_ps): Ditto.
34298 (_mm256_i32scatter_pd): Ditto.
34299 (_mm256_mask_i32scatter_pd): Ditto.
34300 (_mm_i32scatter_pd): Ditto.
34301 (_mm_mask_i32scatter_pd): Ditto.
34302 (_mm256_i64scatter_ps): Ditto.
34303 (_mm256_mask_i64scatter_ps): Ditto.
34304 (_mm_i64scatter_ps): Ditto.
34305 (_mm_mask_i64scatter_ps): Ditto.
34306 (_mm256_i64scatter_pd): Ditto.
34307 (_mm256_mask_i64scatter_pd): Ditto.
34308 (_mm_i64scatter_pd): Ditto.
34309 (_mm_mask_i64scatter_pd): Ditto.
34310 (_mm256_i32scatter_epi32): Ditto.
34311 (_mm256_mask_i32scatter_epi32): Ditto.
34312 (_mm_i32scatter_epi32): Ditto.
34313 (_mm_mask_i32scatter_epi32): Ditto.
34314 (_mm256_i32scatter_epi64): Ditto.
34315 (_mm256_mask_i32scatter_epi64): Ditto.
34316 (_mm_i32scatter_epi64): Ditto.
34317 (_mm_mask_i32scatter_epi64): Ditto.
34318 (_mm256_i64scatter_epi32): Ditto.
34319 (_mm256_mask_i64scatter_epi32): Ditto.
34320 (_mm_i64scatter_epi32): Ditto.
34321 (_mm_mask_i64scatter_epi32): Ditto.
34322 (_mm256_i64scatter_epi64): Ditto.
34323 (_mm256_mask_i64scatter_epi64): Ditto.
34324 (_mm_i64scatter_epi64): Ditto.
34325 (_mm_mask_i64scatter_epi64): Ditto.
34326 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34327 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34328 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34329 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34330 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34331 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34332 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34333 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34334 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34335 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34336 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34337 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34338 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34339 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34340 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34341 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34342 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34343 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34344 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34345 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34346 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34347 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34348 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34349 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34350 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34351 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34352 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34353 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34354 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34355 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34356 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34357 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34358 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34359 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34360 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34361 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34362 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34363 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34364 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34365 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34366 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34367 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34368 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34369 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34370 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34371 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34372 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34373 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34374 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34375 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34376 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34377 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34378 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34379 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34380 definitions accordingly.
34382 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34383 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34386 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34389 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34392 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34393 hard regs for inner regno.
34395 2017-01-17 Martin Liska <mliska@suse.cz>
34398 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34399 assumption and add comment.
34401 2017-01-17 Nathan Sidwell <nathan@acm.org>
34403 * ipa-visibility.c (localize_node): New function, broken out of ...
34404 (function_and_variable_visibility): ... here. Call it.
34406 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34408 PR middle-end/77445
34409 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34410 correctly set frequency of oudgoing edge.
34411 (duplicate_thread_path): Fix profile updating.
34413 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34416 * configure.ac: Add GCC_BASE_VER.
34417 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34418 version from BASE-VER file.
34419 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34420 (gcc.o): Depend on $(BASEVER).
34421 * common.opt (dumpfullversion): New option.
34422 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34423 * doc/invoke.texi: Document -dumpfullversion.
34424 * doc/install.texi: Document --with-gcc-major-version-only.
34425 * configure: Regenerated.
34427 2017-01-17 Richard Biener <rguenther@suse.de>
34429 PR tree-optimization/71433
34430 * tree-vrp.c (register_new_assert_for): Merge same asserts
34431 on all incoming edges.
34432 (process_assert_insertions_for): Handle insertions at the
34435 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34437 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34438 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34440 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34443 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34446 2017-01-17 Alan Modra <amodra@gmail.com>
34449 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34450 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34451 symbolic stack limit when pic.
34453 2017-01-16 Martin Sebor <msebor@redhat.com>
34455 PR tree-optimization/78608
34456 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34458 2017-01-16 Jeff Law <law@redhat.com>
34461 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34462 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34463 for several include directories that may be relative to sysroot.
34464 * config/i386/x-mingw32 (gplus_includedir): Define.
34465 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34466 (native_system_includedir): Likewise.
34467 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34468 override if TARGET_SYSTEM_ROOT is defined.
34469 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34471 PR tree-optimization/79090
34472 PR tree-optimization/33562
34473 PR tree-optimization/61912
34474 PR tree-optimization/77485
34475 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34476 and computed trims into the dump file.
34478 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34480 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34482 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34485 * gimplify.c (gimplify_init_constructor): If want_value and
34486 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34490 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34491 sequence. Formatting fixes.
34492 (doloop_optimize): Formatting fixes.
34495 * gcc.c (debug_level_greater_than_spec_func): New function.
34496 (static_spec_functions): Add debug-level-gt spec function.
34497 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34499 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34500 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34501 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34502 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34503 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34504 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34506 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34508 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34509 QImode fixups to general and mask registers only.
34511 2017-01-16 Carl Love <cel@us.ibm.com>
34513 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34514 for built-in functions
34515 vector signed char vec_nabs (vector signed char)
34516 vector signed short vec_nabs (vector signed short)
34517 vector signed int vec_nabs (vector signed int)
34518 vector signed long long vec_nabs (vector signed long long)
34519 vector float vec_nabs (vector float)
34520 vector double vec_nabs (vector double)
34521 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34523 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34524 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34525 * doc/extend.texi: Update the documentation file for the new built-in
34528 2017-01-16 Martin Sebor <msebor@redhat.com>
34530 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34533 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34535 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34536 UNSPEC_VSX__XXSPLTD to require special splat handling.
34538 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34541 * system.h: Poison strndup.
34543 2017-01-16 Alan Modra <amodra@gmail.com>
34546 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34549 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34551 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34553 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34555 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34556 call recog here. Assert that INSN_CODE (insn) is non-negative.
34558 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34561 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34563 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34564 in the currently scheduled RTL fragment.
34566 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34568 PR rtl-optimization/78751
34569 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34572 2017-01-14 Jeff Law <law@redhat.com>
34574 PR tree-optimization/79090
34575 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34576 variable length stores.
34577 (compute_trims): Delete dead assignment to *trim_tail.
34578 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34581 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34583 PR rtl-optimization/78626
34584 PR rtl-optimization/78727
34585 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34586 of a block, and split such blocks after everything else is finished.
34588 2017-01-14 Alan Modra <amodra@gmail.com>
34591 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34592 target legitimate_combined_insn.
34593 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34594 (rs6000_legitimate_combined_insn): New function.
34595 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34597 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34598 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34599 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34601 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34603 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34605 2017-01-13 Jeff Law <law@redhat.com>
34607 PR tree-optimization/33562
34608 PR tree-optimization/61912
34609 PR tree-optimization/77485
34610 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34612 (delete_dead_assignment): Likewise.
34613 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34614 statement to delete_dead_call and delete_dead_assignment.
34616 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34619 * substring-locations.c (format_warning_va): Strengthen case 1 so
34620 that both endpoints of the substring must be within the format
34621 range for just the substring to be printed.
34623 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34625 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34626 * config/i386/i386.c (ix86_target_string): Add missing options
34627 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34628 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34629 flags_other and ix86_target_other to flags2_other. Display unknown
34631 (ix86_valid_target_attribute_inner_p): Add missing options and
34632 reorder options by implied ISAs, as in ix86_target_string.
34634 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34636 * hash-table.h (hash_table::too_empty_p): New function.
34637 (hash_table::expand): Use it.
34638 (hash_table::traverse): Likewise.
34639 (hash_table::empty_slot): Use sizeof (value_type) instead of
34640 sizeof (PTR) to convert bytes to elements. Shrink the table
34641 if the current size is excessive for the current number of
34644 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34646 * ira-costs.c (record_reg_classes): Break from the inner loop
34647 early once alt_fail is known to be true. Update outer loop
34648 handling accordingly.
34650 2017-01-13 Jeff Law <law@redhat.com>
34652 * tree-ssa-dse.c (decrement_count): New function.
34653 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34654 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34655 when we know the partially dead statement is a mem* function.
34657 PR tree-optimization/61912
34658 PR tree-optimization/77485
34659 * tree-ssa-dse.c: Include expr.h.
34660 (maybe_trim_constructor_store): New function.
34661 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34663 PR tree-optimization/33562
34664 PR tree-optimization/61912
34665 PR tree-optimization/77485
34666 * doc/invoke.texi: Document new dse-max-object-size param.
34667 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34668 * tree-ssa-dse.c: Include params.h.
34669 (dse_store_status): New enum.
34670 (initialize_ao_ref_for_dse): New, partially extracted from
34672 (valid_ao_ref_for_dse, normalize_ref): New.
34673 (setup_live_bytes_from_ref, compute_trims): Likewise.
34674 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34675 (maybe_trim_partially_dead_store): Likewise.
34676 (maybe_trim_complex_store): Likewise.
34677 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34678 Track what bytes live from the original store. Return tri-state
34679 for dead, partially dead or live.
34680 (dse_dom_walker): Add constructor, destructor and new private members.
34681 (delete_dead_call, delete_dead_assignment): New extracted from
34683 (dse_optimize_stmt): Make a member of dse_dom_walker.
34684 Use initialize_ao_ref_for_dse.
34686 PR tree-optimization/33562
34687 PR tree-optimization/61912
34688 PR tree-optimization/77485
34689 * sbitmap.h (bitmap_count_bits): Prototype.
34690 (bitmap_clear_range, bitmap_set_range): Likewise.
34691 * sbitmap.c (bitmap_clear_range): New function.
34692 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34694 2017-01-13 Martin Liska <mliska@suse.cz>
34697 * function.c (set_cfun): Add new argument force.
34698 * function.h (set_cfun): Likewise.
34699 * ipa-inline-transform.c (inline_call): Use the function when
34700 strict alising from is dropped for function we inline to.
34702 2017-01-13 Richard Biener <rguenther@suse.de>
34704 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34705 for dumping GIMPLE INTEGER_CSTs.
34707 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34709 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34710 to 201112L since C++17.
34712 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34715 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34716 if -fsanitize=kernel-address is present.
34718 2017-01-13 Richard Biener <rguenther@suse.de>
34720 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34721 as _Literal ( type ) number in case usual suffixes do not
34722 preserve all information.
34724 2017-01-13 Richard Biener <rguenther@suse.de>
34726 PR tree-optimization/77283
34727 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34728 and ssa-iterators.h.
34729 (is_feasible_trace): Implement a cost model based on joiner
34732 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34735 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34736 char or short to __float128/_Float128 directly.
34738 2017-01-12 Martin Sebor <msebor@redhat.com>
34740 to -Wformat-overflow.
34741 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34742 (min_bytes_remaining): Same.
34743 (get_string_length): Same.
34744 (format_string): Same.
34745 (format_directive): Same.
34747 (pass_sprintf_length::handle_gimple_call): Same.
34749 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34751 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34752 info.nowrite calls with no lhs that can't throw. Return bool
34753 whether gsi_remove has been called or not.
34754 (pass_sprintf_length::handle_gimple_call): Return bool whether
34755 try_substitute_return_value called gsi_remove. Formatting fix.
34756 (pass_sprintf_length::execute): Don't use gsi_remove if
34757 handle_gimple_call returned true.
34760 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34761 be removed due to side-effects, don't remove following barrier nor
34762 turn the successor edge into fallthru edge.
34764 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34767 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34768 element-reversing loads and stores as not swappable.
34770 2017-01-12 Nathan Sidwell <nathan@acm.org>
34771 Nicolai Stange <nicstange@gmail.com>
34773 * combine.c (try_combine): Don't ignore result of overlap checking
34774 loop. Combine overlap & asm check into single loop.
34776 2017-01-12 Richard Biener <rguenther@suse.de>
34778 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34779 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34781 2017-01-12 Richard Biener <rguenther@suse.de>
34783 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34784 and TS_TARGET_OPTION directly derive from TS_BASE.
34785 * tree-core.h (tree_optimization_option): Derive from tree_base.
34786 (tree_target_option): Likewise.
34788 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34790 * config/i386/i386.c (memory_address_length): Increase len
34791 only when rip_relative_addr_p returns false.
34793 2017-01-11 Julia Koval <julia.koval@intel.com>
34795 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34796 (OPTION_MASK_ISA_SGX_SET): New.
34797 (ix86_handle_option): Handle OPT_msgx.
34798 * config.gcc: Added sgxintrin.h.
34799 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34800 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34801 * config/i386/i386.c (ix86_target_string): Add -msgx.
34803 (ix86_option_override_internal): Handle new options.
34804 (ix86_valid_target_attribute_inner_p): Add sgx.
34805 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34806 * config/i386/i386.opt: Add msgx.
34807 * config/i386/sgxintrin.h: New file.
34808 * config/i386/x86intrin.h: Add sgxintrin.h.
34810 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34813 * fold-const.c (maybe_nonzero_address): Return 1 for function
34815 (tree_single_nonzero_warnv_p): Don't handle function local objects
34819 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34822 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34825 * opts.c: Include "spellcheck.h"
34826 (struct string_fragment): New struct.
34827 (struct edit_distance_traits<const string_fragment &>): New
34829 (get_closest_sanitizer_option): New function.
34830 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34832 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34834 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34836 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34837 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34838 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34839 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34840 for initial die_offset if dwarf_split_debug_info.
34841 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34842 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34844 (output_skeleton_debug_sections): Formatting fix. Use
34845 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34846 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34848 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34850 * config/arm/cortex-a53.md: Add bypasses for
34851 cortex_a53_r2f_cvt.
34852 (cortex_a53_r2f): Only use for transfers.
34853 (cortex_a53_f2r): Likewise.
34854 (cortex_a53_r2f_cvt): Add reservation for conversions.
34855 (cortex_a53_f2r_cvt): Likewise.
34857 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34859 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34860 to all inlined functions, change static to extern.
34862 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34865 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34867 (arm_assemble_integer): Likewise.
34869 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34871 * config.gcc: Use new awk script to check CPU, FPU and architecture
34872 parameters for --with-... options.
34873 * config/arm/parsecpu.awk: New file
34874 * config/arm/arm-cpus.in: New file.
34875 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34877 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34879 * config/arm/t-arm: Update dependency rules.
34880 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34881 of processing .def files.
34882 * config/arm/genopt.sh: Deleted.
34883 * config/arm/gentune.sh: Deleted.
34884 * config/arm/arm-cores.def: Deleted.
34885 * config/arm/arm-arches.def: Deleted.
34886 * config/arm/arm-fpus.def: Deleted.
34887 * config/arm/arm-tune.md: Regenerated.
34888 * config/arm/arm-tables.opt: Regenerated.
34889 * config/arm/arm-cpu.h: New generated file.
34890 * config/arm/arm-cpu-data.h: New generated file.
34891 * config/arm/arm-cpu-cdata.h: New generated file.
34893 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34896 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34898 (input_varpool_node): Unpack dynamically_initialized bit.
34900 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34902 PR rtl-optimization/79032
34903 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34904 the alignment of the adjusted memory reference against that of MODE,
34905 instead of the alignment of the original memory reference.
34907 2017-01-11 Martin Jambor <mjambor@suse.cz>
34909 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34911 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34912 decorated functions.
34914 2017-01-11 Richard Biener <rguenther@suse.de>
34916 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34917 set range/nonnull info for PHI results. Do not set it on
34918 stmts marked for removal.
34920 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34922 * expr.c (store_field): In the bitfield case, fetch the return value
34923 from the registers before applying a single big-endian adjustment.
34924 Always do a final load for a BLKmode value not larger than a word.
34926 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34929 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34930 that we correctly handle column numbers greater than
34931 LINE_MAP_MAX_COLUMN_NUMBER.
34933 2017-01-10 Martin Sebor <msebor@redhat.com>
34935 PR middle-end/78245
34936 * gimple-ssa-sprintf.c (get_destination_size): Call
34937 {init,fini}object_sizes.
34938 * tree-object-size.c (addr_object_size): Adjust.
34939 (pass_through_call): Adjust.
34940 (pass_object_sizes::execute): Adjust.
34941 * tree-object-size.h (fini_object_sizes): Declare.
34943 2017-01-10 Martin Sebor <msebor@redhat.com>
34945 PR tree-optimization/78775
34946 * builtins.c (get_size_range): Move...
34947 * calls.c: ...to here.
34948 (alloc_max_size): Accept zero argument.
34949 (operand_signed_p): Remove.
34950 (maybe_warn_alloc_args_overflow): Call get_size_range.
34951 * calls.h (get_size_range): Declare.
34953 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34955 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34956 from TI's devices.csv file as of September 2016.
34957 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34959 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34961 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34962 * doc/invoke.texi: Likewise.
34963 * doc/md.texi: Likewise.
34964 * doc/objc.texi: Likewise.
34966 2017-01-10 Joshua Conner <joshconner@google.com>
34968 * config/arm/fuchsia-elf.h: New file.
34969 * config/fuchsia.h: New file.
34970 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34971 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34973 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34975 2016-01-10 Richard Biener <rguenther@suse.de>
34977 PR tree-optimization/79034
34978 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34979 Propagate out degenerate PHIs in the joiner.
34981 2017-01-10 Martin Liska <mliska@suse.cz>
34983 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34984 (sort_congruence_classes_by_decl_uid): Likewise.
34985 (sort_congruence_class_groups_by_decl_uid): Likewise.
34986 (sem_item_optimizer::merge_classes): Sort class, groups in these
34987 classes and members in the groups by DECL_UID of declarations.
34988 This would make merge operations stable.
34990 2017-01-10 Martin Liska <mliska@suse.cz>
34992 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34993 usage of m_classes_vec.
34994 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34995 (sem_item_optimizer::get_group_by_hash): Likewise.
34996 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34997 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34998 (sem_item_optimizer::verify_classes): Likewise.
34999 (sem_item_optimizer::process_cong_reduction): Likewise.
35000 (sem_item_optimizer::dump_cong_classes): Likewise.
35001 (sem_item_optimizer::merge_classes): Likewise.
35002 * ipa-icf.h (congruence_class_hash): Rename from
35003 congruence_class_group_hash. Remove declaration of m_classes_vec.
35005 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
35007 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
35008 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
35009 * config.gcc: Add avx512vpopcntdqintrin.h.
35010 * config/i386/avx512vpopcntdqintrin.h: New.
35011 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
35012 * config/i386/i386-builtin-types.def: Add new types.
35013 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
35014 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
35015 __builtin_ia32_vpopcountq_v8di_mask): New.
35016 * config/i386/i386-c.c (ix86_target_macros_internal): Define
35017 __AVX512VPOPCNTDQ__.
35018 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
35019 (PTA_AVX512VPOPCNTDQ): Define.
35020 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
35021 TARGET_AVX512VPOPCNTDQ_P): Define.
35022 * config/i386/i386.opt: Add mavx512vpopcntdq.
35023 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
35024 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
35026 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35028 PR middle-end/77484
35029 * predict.def (PRED_CALL): Set to 67.
35031 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
35033 * expr.c (store_field): In the bitfield case, if the value comes from
35034 a function call and is of an aggregate type returned in registers, do
35035 not modify the field mode; extract the value in all cases if the mode
35036 is BLKmode and the size is not larger than a word.
35038 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35041 * config/i386/cpuid.h: Fix undefined behavior.
35043 2017-01-04 Jeff Law <law@redhat.com>
35045 PR tree-optimization/79007
35046 PR tree-optimization/67955
35047 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35048 conservative for pt.null when flag_non_call_exceptions is on.
35050 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35052 PR translation/79019
35053 PR translation/79020
35054 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35055 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35056 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35058 * config/avr/avr.opt (maccumulate-args): Likewise.
35059 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35060 * common.opt (freport-bug): Likewise.
35061 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35062 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35063 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35064 translatable string.
35065 * config/i386/i386.c (function_value_32): Likewise.
35066 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35067 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35069 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35070 * common/config/msp430/msp430-common.c (msp430_handle_option):
35072 * symtab.c (symtab_node::verify_base): Likewise.
35073 * opts.c (set_debug_level): Likewise.
35074 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35075 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35076 missing whitespace to translatable strings.
35077 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35078 * config/sh/superh.h: Likewise.
35079 * config/i386/xopintrin.h: Likewise.
35080 * config/i386/znver1.md: Likewise.
35081 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35082 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35083 * double-int.h (struct double_int): Likewise.
35084 * double-int.c (div_and_round_double): Likewise.
35085 * wide-int.cc: Likewise.
35086 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35087 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35088 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35089 (crossjumps_occurred): ... this.
35090 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35093 PR tree-optimization/78899
35094 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35095 returning bool return struct loop *, NULL for failure and the new
35097 (versionable_outer_loop_p): Don't version outer loop if it has
35098 dont_vectorized bit set.
35099 (tree_if_conversion): When versioning outer loop, ensure
35100 tree_if_conversion is performed also on the inner loop of the
35101 non-vectorizable outer loop copy.
35102 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35103 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35104 prevent vectorization of it.
35105 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35106 the outer loop vectorization of the non-scalar version is attempted
35107 before vectorization of the inner loop in scalar version. If
35108 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35109 vectorization of its inner loop.
35110 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35111 has 2 inner loops, rename also on edges from bb whose single pred
35112 is outer_loop->header. Fix typo in function comment.
35114 2017-01-09 Martin Sebor <msebor@redhat.com>
35117 * asan.c (asan_emit_stack_protection): Increase local buffer size
35118 to avoid snprintf truncation warning.
35120 2017-01-09 Andrew Pinski <apinski@cavium.com>
35122 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35123 to reference thunderx2t99 for the tuning structure
35124 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35126 (thunderx2t99_extra_costs): This.
35127 * config/aarch64/aarch64-tune.md: Regenerate.
35128 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35129 (vulcan_addrcost_table): This.
35130 (vulcan_regmove_cost): Rename to ...
35131 (thunderx2t99_regmove_cost): This.
35132 (vulcan_vector_cost): Rename to ...
35133 (thunderx2t99_vector_cost): this.
35134 (vulcan_branch_cost): Rename to ...
35135 (thunderx2t99_branch_cost): This.
35136 (vulcan_tunings): Rename to ...
35137 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35138 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35140 2017-01-09 Martin Jambor <mjambor@suse.cz>
35144 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35145 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35146 (propagate_vr_accross_jump_function): Use the above function for all
35147 value range computations for pass-through jump functions and type
35148 converasion from explicit value range values.
35149 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35150 parameters from TYPE_ARG_TYPES.
35151 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35152 (ipa_write_node_info): Stream type of the actual argument.
35153 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35155 2017-01-09 Martin Liska <mliska@suse.cz>
35158 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35159 (lookup_compiler): Do not show error message with have_E.
35161 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35163 PR tree-optimization/78938
35164 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35165 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35166 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35167 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35170 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35172 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35175 2017-01-09 Richard Biener <rguenther@suse.de>
35177 PR tree-optimization/78997
35178 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35179 name condition properly.
35181 2017-01-09 Richard Biener <rguenther@suse.de>
35184 * dwarf2out.c (is_cxx): New overload with context.
35185 (is_naming_typedef_decl): Use it.
35187 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35189 * invoke.texi (Option Summary): Correct spacing in option lists
35190 and add line breaks to fix over-long lines.
35192 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35194 PR middle-end/17660
35196 * extend.texi (Common Variable Attributes): Add xref to GCC
35197 Internals manual to explain mode attribute keywords.
35199 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35202 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35203 and Preprocessor Options.
35204 (Options for Linking): Document -pthread here....
35205 (RS/6000 and PowerPC Options): ...not here.
35206 (Solaris 2 Options): ...or here.
35207 * doc/cppopts.texi: Document -pthread.
35209 2017-01-08 Martin Sebor <msebor@redhat.com>
35211 PR middle-end/77708
35212 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35213 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35214 New member functions.
35215 (format_directive): Used them.
35217 (pass_sprintf_length::handle_gimple_call): Same.
35218 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35219 to avoid truncation for any argument.
35220 (extract_affine_mul): Same.
35221 * tree.c (get_file_function_name): Same.
35223 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35225 PR middle-end/77484
35226 * predict.def (PRED_INDIR_CALL): Set to 86.
35228 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35230 PR preprocessor/54124
35231 * doc/cppopts.texi: Reformat -d subtable to list the full name
35232 of the options. Add cross-reference to the docs for the general
35233 compiler -d options.
35234 * doc/invoke.texi (Developer Options): Add cross-reference to the
35235 preprocessor-specific -d option documentation.
35237 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35239 PR preprocessor/13498
35240 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35241 redudant material, and reflect new command-line options.
35242 (System Headers): Likewise.
35244 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35246 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35247 -isystem, and -idirafter. Copy-edit.
35248 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35249 default for -ftrack-macro-expansion. Delete obsolete and
35250 badly-formatted implementation details about -fdebug-cpp output.
35251 * doc/cppwarnopts.texi: Copy-edit.
35253 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35256 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35257 that the transition from a max line width >= 1<<10 to narrower
35258 lines works correctly.
35260 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35262 * doc/options.texi (PerFunction): New.
35263 * opt-functions.awk (switch_flags): Map both Optimization and
35264 PerFunction to CL_OPTIMIZATION.
35265 * opth-gen.awk: Test for PerFunction flag along with
35267 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35268 it only when the latter is present. Skip those that don't in
35269 the hash function generator.
35270 * common.opt (fvar-tracking): Mark as PerFunction instead of
35272 (fvar-tracking-assignments): Likewise.
35273 (fvar-tracking-assignments-toggle): Likewise.
35274 (fvar-tracking-uninit): Likewise.
35276 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35278 PR translation/79018
35279 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35282 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35285 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35286 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35287 TARGET_LONG_JUMP_TABLE_OFFSETS.
35288 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35289 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35290 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35291 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35292 * config/m68k/m68k.md (tablejump expander): Likewise.
35293 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35294 TARGET_LONG_JUMP_TABLE_OFFSETS.
35295 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35296 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35298 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35299 David Holsgrove <david.holsgrove@xilinx.com>
35301 * common/config/microblaze/microblaze-common.c
35302 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35303 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35305 * config/microblaze/microblaze.c (microblaze_must_save_register)
35306 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35308 (microblaze_eh_return): New function.
35309 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35310 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35311 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35312 * config/microblaze/microblaze.md (eh_return): New pattern.
35314 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35316 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35317 GCC_DIAGNOSTIC_STRINGIFY): Define.
35319 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35321 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35323 * config/arm/arm.md (<mcrr>): New.
35325 * config/arm/arm.c (arm_arch5te): New.
35326 (arm_option_override): Set arm_arch5te.
35327 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35329 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35330 (arm_mcrr_qualifiers): ... this. New.
35331 (MRRC_QUALIFIERS): Define to...
35332 (arm_mrrc_qualifiers): ... this. New.
35333 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35335 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35336 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35337 (MRRCI, mrrc, MRRC): New.
35338 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35339 VUNSPEC_MRRC2): New.
35341 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35343 * config/arm/arm.md (<mcr>): New.
35345 * config/arm/arm.c (arm_coproc_builtin_available): Add
35346 support for mcr, mrc, mcr2 and mrc2.
35347 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35348 (arm_mcr_qualifiers): ... this. New.
35349 (MRC_QUALIFIERS): Define to ...
35350 (arm_mrc_qualifiers): ... this. New.
35351 (MCR_QUALIFIERS): Define to ...
35352 (arm_mcr_qualifiers): ... this. New.
35353 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35355 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35356 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35357 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35358 VUNSPEC_MRC2): New.
35360 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35362 * config/arm/arm.md (*ldc): New.
35366 * config/arm/arm.c (arm_coproc_builtin_available): Add
35367 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35368 (arm_coproc_ldc_stc_legitimate_address): New.
35369 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35370 'qualifier_const_pointer'.
35371 (LDC_QUALIFIERS): Define to...
35372 (arm_ldc_qualifiers): ... this. New.
35373 (STC_QUALIFIERS): Define to...
35374 (arm_stc_qualifiers): ... this. New.
35375 * config/arm/arm-protos.h
35376 (arm_coproc_ldc_stc_legitimate_address): New.
35377 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35378 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35379 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35380 stc2, stcl, stc2l): New.
35381 * config/arm/constraints.md (Uz): New.
35382 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35383 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35384 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35385 VUNSPEC_STC2L): New.
35387 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35389 * config/arm/arm.md (<cdp>): New.
35390 * config/arm/arm.c (neon_const_bounds): Rename this ...
35391 (arm_const_bounds): ... this.
35392 (arm_coproc_builtin_available): New.
35393 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35394 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35395 (CDP_QUALIFIERS): Define to...
35396 (arm_cdp_qualifiers): ... this. New.
35398 (arm_expand_builtin_args): Add case for 6 arguments.
35399 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35400 (arm_const_bounds): ... this.
35401 (arm_coproc_builtin_available): New.
35402 * config/arm/arm_acle.h (__arm_cdp): New.
35404 * config/arm/arm_acle_builtins.def (cdp): New.
35406 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35407 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35408 'arm_const_bounds'.
35409 * config/arm/types.md (coproc): New.
35410 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35411 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35412 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35413 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35415 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35417 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35418 (UBINOP_QUALIFIERS): New.
35420 (acle_builtin_data): New. Change comment.
35421 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35422 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35423 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35424 arm_acle_builtins.def.
35425 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35426 (arm_init_acle_builtins): New.
35427 (CRC32_BUILTIN): Remove.
35428 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35429 crc32cb, crc32ch and crc32cw.
35430 (arm_init_crc32_builtins): Remove.
35431 (arm_init_builtins): Use arm_init_acle_builtins rather
35432 than arm_init_crc32_builtins.
35433 (arm_expand_acle_builtin): New.
35434 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35435 * config/arm/arm_acle_builtins.def: New.
35437 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35439 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35440 (arm_builtin_datum): ... this.
35441 (arm_init_neon_builtin): Rename to ...
35442 (arm_init_builtin): ... this. Add a new parameters PREFIX
35443 and USE_SIG_IN_NAME.
35444 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35445 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35446 'arm_builtin_datum'.
35447 (arm_init_vfp_builtins): Likewise.
35448 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35449 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35450 (arm_expand_neon_args): Rename to ...
35451 (arm_expand_builtin_args): ... this. Rename builtin_arg
35452 enum values and differentiate between ARG_BUILTIN_MEMORY
35453 and ARG_BUILTIN_NEON_MEMORY.
35454 (arm_expand_neon_builtin_1): Rename to ...
35455 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35456 values, arm_expand_builtin_args and add bool parameter NEON.
35457 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35458 (arm_expand_vfp_builtin): Likewise.
35459 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35461 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35463 PR middle-end/77484
35464 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35465 * predict.c (tree_estimate_probability_bb): Reverse direction of
35466 polymorphic call predictor.
35468 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35470 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35471 (determine_pass_name_match): ...this new function and...
35472 (should_skip_pass_p): ...this new function.
35474 2017-01-06 Nathan Sidwell <nathan@acm.org>
35476 * ipa-visibility.c (function_and_variable_visibility): Reformat
35477 comments and long lines. Remove extrneous if.
35478 * symtab.c (symtab_node::make_decl_local): Fix code format.
35479 (symtab_node::set_section_for_node): Fix comment typo.
35481 2017-01-06 Martin Liska <mliska@suse.cz>
35484 * lra-constraints.c: Rename invariant to lra_invariant.
35485 * predict.c (set_even_probabilities): Initialize e to NULL.
35487 2017-01-05 Martin Sebor <msebor@redhat.com>
35489 PR tree-optimization/78910
35490 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35491 (format_integer): Correct off-by-one error in the handling
35492 of precision with negative numbers in signed conversions..
35494 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35496 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35498 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35500 PR tree-optimization/71016
35501 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35502 factor_out_conditional_conversion. Formatting fix.
35503 (factor_out_conditional_conversion): Add cond_stmt argument.
35504 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35505 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35508 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35510 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35511 read-rtl-function.o, and selftest-rtl.o.
35512 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35513 (selftest::aarch64_test_loading_full_dump): New function.
35514 (selftest::aarch64_run_selftests): New function.
35515 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35516 selftest::aarch64_run_selftests.
35517 * config/i386/i386.c
35518 (selftest::ix86_test_loading_dump_fragment_1): New function.
35519 (selftest::ix86_test_loading_call_insn): New function.
35520 (selftest::ix86_test_loading_full_dump): New function.
35521 (selftest::ix86_test_loading_unspec): New function.
35522 (selftest::ix86_run_selftests): Call the new functions.
35523 * emit-rtl.c (maybe_set_max_label_num): New function.
35524 * emit-rtl.h (maybe_set_max_label_num): New decl.
35525 * function.c (instantiate_decls): Guard call to
35526 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35527 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35529 * gensupport.c (gen_reader::gen_reader): Pass "false"
35530 for new "compact" param of rtx_reader.
35531 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35532 rather than an empty string for NULL strings.
35533 * read-md.c: Potentially include config.h rather than bconfig.h.
35534 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35535 (have_error): New global, copied from errors.c.
35536 (md_reader::read_name): Rename to...
35537 (md_reader::read_name_1): ...this, adding "out_loc" param,
35538 and converting "missing name or number" to returning false, rather
35540 (md_reader::read_name): Reimplement in terms of read_name_1.
35541 (md_reader::read_name_or_nil): New function.
35542 (md_reader::read_string): Handle "(nil)" by returning NULL.
35543 (md_reader::md_reader): Add new param "compact".
35544 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35545 (md_reader::read_file): New method.
35546 * read-md.h (md_reader::md_reader): Add new param "compact".
35547 (md_reader::read_file): New method.
35548 (md_reader::is_compact): New accessor.
35549 (md_reader::read_name): Convert return type from void to file_location.
35550 (md_reader::read_name_or_nil): New decl.
35551 (md_reader::read_name_1): New decl.
35552 (md_reader::m_compact): New field.
35553 (noop_reader::noop_reader): Pass "false" for new "compact" param
35555 (rtx_reader::rtx_reader): Add new "compact" param.
35556 (rtx_reader::read_rtx_operand): Make virtual and convert return
35557 type from void to rtx.
35558 (rtx_reader::read_until): New decl.
35559 (rtx_reader::handle_any_trailing_information): New virtual function.
35560 (rtx_reader::postprocess): New virtual function.
35561 (rtx_reader::finalize_string): New virtual function.
35562 (rtx_reader::m_in_call_function_usage): New field.
35563 (rtx_reader::m_reuse_rtx_by_id): New field.
35564 * read-rtl-function.c: New file.
35565 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35566 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35567 (selftest::verify_three_block_rtl_cfg): New decl.
35568 * read-rtl-function.h: New file.
35569 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35570 For host, include function.h, memmodel.h, and emit-rtl.h.
35571 (one_time_initialization): New function.
35572 (struct compact_insn_name): New struct.
35573 (compact_insn_names): New array.
35574 (find_code): Handle insn codes in compact dumps.
35575 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35576 (bind_subst_iter_and_attr): Likewise.
35577 (add_condition_to_string): Likewise.
35578 (add_condition_to_rtx): Likewise.
35579 (apply_attribute_uses): Likewise.
35580 (add_current_iterators): Likewise.
35581 (apply_iterators): Likewise.
35582 (initialize_iterators): Guard usage of apply_subst_iterator with
35583 #ifdef GENERATOR_FILE.
35584 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35585 (md_reader::read_mapping): Likewise.
35586 (add_define_attr_for_define_subst): Likewise.
35587 (add_define_subst_attr): Likewise.
35588 (read_subst_mapping): Likewise.
35589 (check_code_iterator): Likewise.
35590 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35592 (one_time_initialization): New function.
35593 (rtx_reader::read_until): New method.
35594 (read_flags): New function.
35595 (parse_reg_note_name): New function.
35596 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35597 Handle reuse_rtx ids.
35598 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35599 Add parsing support for RTL dumps, mirroring the special-cases in
35600 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35601 values, and calling handle_any_trailing_information.
35602 (rtx_reader::read_rtx_operand): Convert return type from void
35603 to rtx, returning return_rtx. Handle case 'e'. Call
35604 finalize_string on XSTR and XTMPL fields.
35605 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35606 "(nil)" values were omitted. Call the postprocess vfunc on the
35608 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35609 class ctor. Initialize m_in_call_function_usage. Call
35610 one_time_initialization.
35611 * rtl-tests.c (selftest::test_uncond_jump): Call
35612 set_new_first_and_last_insn.
35613 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35614 * selftest-rtl.c: New file.
35615 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35616 (selftest::get_insn_by_uid): New decl.
35617 * selftest-run-tests.c (selftest::run_tests): Call
35618 read_rtl_function_c_tests.
35619 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35620 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35623 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35625 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35626 operands in a special way. Assert that pos+len <= mode precision.
35628 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35630 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35631 3 argument Alias with unlimited for the negative form.
35632 (fno-vect-cost-model): Removed.
35634 2017-01-05 Martin Liska <mliska@suse.cz>
35636 * hsa-gen.c (gen_hsa_divmod): New function.
35637 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35639 2017-01-05 Martin Liska <mliska@suse.cz>
35642 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35645 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35647 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35648 small constant length operands.
35650 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35652 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35653 between loop iterations.
35655 2017-01-05 Martin Liska <mliska@suse.cz>
35658 * gimplify.c (gimplify_decl_expr): Compare to
35659 asan_poisoned_variables instread of checking flags.
35660 (gimplify_target_expr): Likewise.
35661 (gimplify_expr): Likewise.
35662 (gimplify_function_tree): Conditionally initialize
35663 asan_poisoned_variables.
35665 2017-01-04 Jeff Law <law@redhat.com>
35667 PR tree-optimizatin/78812
35668 * rtl.h (contains_mem_rtx_p): Prototype.
35669 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35670 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35671 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35672 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35673 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35675 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35677 * input.c (assert_char_at_range): Default-initialize actual_range.
35679 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35681 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35682 to match the caller.
35684 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35686 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35687 insns after final jump in test to emit dummy move.
35689 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35691 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35692 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35694 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35696 * multiple_target.c (create_dispatcher_calls): Init e_next.
35697 * tree-ssa-loop-split.c (split_loop): Init border.
35698 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35701 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35706 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35707 (altivec_register_operand): Do not return true if the operand
35708 contains a SUBREG mixing SImode and SFmode.
35709 (vsx_register_operand): Likewise.
35710 (vsx_reg_sfsubreg_ok): New predicate.
35711 (vfloat_operand): Do not return true if the operand contains a
35712 SUBREG mixing SImode and SFmode.
35713 (vint_operand): Likewise.
35714 (vlogical_operand): Likewise.
35715 (gpc_reg_operand): Likewise.
35716 (int_reg_operand): Likewise.
35717 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35718 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35719 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35721 (rs6000_emit_move_si_sf_subreg): New helper function.
35722 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35723 fixup SUBREGs involving SImode and SFmode.
35724 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35725 numbers for the new peephole2 optimization.
35726 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35727 the GLIBC math library that do AND/IOR/XOR operations on single
35728 precision floating point.
35729 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35730 target macros to say whether we need to avoid SUBREGs mixing
35732 (TARGET_ALLOW_SF_SUBREG): Likewise.
35733 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35734 (UNSPEC_SI_FROM_SF): Likewise.
35735 (iorxor): Change spacing.
35736 (and_ior_xor): New iterator for AND, IOR, and XOR.
35737 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35738 (movdi_from_sf_zero_ext): Likewise.
35739 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35740 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35741 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35742 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35743 (fms<mode>4): Likewise.
35744 (fnma<mode>4): Likewise.
35745 (fnms<mode>4): Likewise.
35746 (nfma<mode>4): Likewise.
35747 (nfms<mode>4): Likewise.
35749 2017-01-04 Marek Polacek <polacek@redhat.com>
35752 * doc/invoke.texi: Document -Wpointer-compare.
35754 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35756 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35759 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35760 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35763 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35766 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35767 documentation of the powerpc_popcntb_ok attribute.
35768 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35769 code to issue warning messages if a requested CPU configuration is
35770 not supported by the binary (assembler and loader) toolchain.
35771 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35772 made to define a built-in function that has been disabled.
35773 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35774 made to define a built-in function that has been disabled.
35775 (altivec_init_builtins): Add comment explaining why definition
35776 of the DST built-in functions is not preceded by an assertion
35777 check. Add assertions to prevent ICE if attempts are made to
35778 define an altivec predicate or an abs* built-in function that has
35780 (htm_init_builtins): Add comment explaining why definition of the
35781 htm built-in functions is not preceded by an assertion check.
35783 2017-01-04 Jeff Law <law@redhat.com>
35785 PR tree-optimizatin/67955
35786 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35787 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35788 the points-to solution does not include pt_null. Use DECL_PT_UID
35791 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35793 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35794 Use gen_int_mode instead of gen_lopwart for const_int operands.
35796 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35798 PR tree-optimization/71563
35799 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35800 out of range value - has low bits known to be zero.
35802 2017-01-04 Alan Modra <amodra@gmail.com>
35804 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35805 * configure: Regenerate.
35806 * config.in: Regenerate.
35808 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35811 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35812 a substring of the message, but strcmp with the whole message. Ifdef
35813 ENABLE_NLS, translate the message first using dgettext.
35815 2017-01-03 Jeff Law <law@redhat.com>
35817 PR tree-optimizatin/78856
35818 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35819 (mark_threaded_blocks): Remove code to truncate thread paths that
35820 cross multiple loop headers. Instead invalidate the cached loop
35821 iteration information and handle case of a thread path walking
35822 into an irreducible region.
35824 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35827 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35828 assertions. Add support for doing the signbit if the IEEE 128-bit
35829 floating point value is in a GPR.
35830 * config/rs6000/rs6000.md (Fsignbit): Delete.
35831 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35832 Update the length attribute if the value is in a GPR.
35833 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35834 the sign or zero extension instruction, since the value is always 0/1.
35835 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35838 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35839 extracting SImode to a GPR register so that we can generate a
35840 store, limit the vector to be in a traditional Altivec register
35841 for the vextuwrx instruction.
35843 2017-01-03 Ian Lance Taylor <iant@google.com>
35845 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35847 2017-01-03 Martin Sebor <msebor@redhat.com>
35849 PR tree-optimization/78696
35850 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35851 precision. Use MPFR for %f for greater fidelity. Correct handling
35853 (pass_sprintf_length::compute_format_length): Set width and precision
35854 specified by asrerisk to void_node for vararg functions.
35855 (try_substitute_return_value): Adjust dump output.
35857 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35859 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35861 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35863 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35864 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35865 -mlra/-mno-lra was passed to the compiler.
35867 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35869 PR rtl-optimization/65618
35870 * emit-rtl.c (try_split): Move initialization of "before" and
35871 "after" to just before the call to emit_insn_after_setloc.
35873 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35875 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35877 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35879 * dwarf2out.c (gen_enumeration_type_die): When
35880 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35882 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35884 PR tree-optimization/78965
35885 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35886 Change first argument from const call_info & to call_info &. For %n
35887 set info.nowrite to false.
35889 PR middle-end/78901
35890 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35891 possibly throwing calls.
35893 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35894 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35895 and fns handling, rather than in a separate case SSA_NAME.
35897 2017-01-02 Jeff Law <law@redhat.com>
35899 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35900 fixes for first_period and second_period variables.
35902 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35905 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35906 (*insvqi_1): New insn pattern.
35907 (*insvqi_1_mem_rex64): Ditto.
35908 (*insvqi_2): Ditto.
35909 (*insvqi_3): Rename from *insvqi.
35911 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35913 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35915 * doc/cfg.texi (Edges): Remove reference to Java.
35916 (Maintaining the CFG): Ditto.
35918 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35920 PR middle-end/77674
35921 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35922 transparent aliases.
35924 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35926 PR middle-end/77484
35927 * predict.def (PRED_CALL): Update hitrate.
35928 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35929 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35930 into direct/indirect/polymorphic variants.
35932 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35934 Update copyright years.
35936 * gcc.c (process_command): Update copyright notice dates.
35937 * gcov-dump.c (print_version): Ditto.
35938 * gcov.c (print_version): Ditto.
35939 * gcov-tool.c (print_version): Ditto.
35940 * gengtype.c (create_file): Ditto.
35941 * doc/cpp.texi: Bump @copying's copyright year.
35942 * doc/cppinternals.texi: Ditto.
35943 * doc/gcc.texi: Ditto.
35944 * doc/gccint.texi: Ditto.
35945 * doc/gcov.texi: Ditto.
35946 * doc/install.texi: Ditto.
35947 * doc/invoke.texi: Ditto.
35949 Copyright (C) 2017 Free Software Foundation, Inc.
35951 Copying and distribution of this file, with or without modification,
35952 are permitted in any medium without royalty provided the copyright
35953 notice and this notice are preserved.