1 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3 * tree-predcom.c: Add general comment on Store-Store chains.
4 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6 (get_chain_last_ref_at): Rename into...
7 (get_chain_last_write_at): ...this.
8 (get_chain_last_write_before_load): New function.
9 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
10 CT_STORE_STORE when write reference is added.
11 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
12 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
13 (initialize_root_vars_store_elim_1): Ditto.
14 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
15 definition is created.
16 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
17 chain by replacing it with dominant stored value.
19 2017-11-20 Bin Cheng <bin.cheng@arm.com>
21 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
23 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
25 * vr-values.c (extract_range_from_binary_expr): Use a full range
28 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30 * config/arm/arm.md (R4_REGNUM): Define constant.
31 (nonsecure_call_internal): Remove r4 clobber.
32 (nonsecure_call_value_internal): Likewise.
33 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
34 clobber and resequence match_operands.
35 (nonsecure_call_value_reg_thumb1_v5): Likewise.
36 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
37 (nonsecure_call_value_reg_thumb2): Likewise.
39 2017-11-20 Jakub Jelinek <jakub@redhat.com>
41 PR tree-optimization/78821
42 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
43 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
44 to the address of the base rather than the base itself.
45 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
46 (find_bswap_or_nop_finalize): New function.
47 (find_bswap_or_nop): Use it.
48 (bswap_replace): Return a tree rather than bool, change first
49 argument from gimple * to gimple_stmt_iterator, allow inserting
50 into an empty sequence, allow ins_stmt to be NULL - then emit
51 all stmts into gsi. Fix up MEM_REF address gimplification.
52 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
53 (struct store_immediate_info): Add N and INS_STMT non-static
55 (store_immediate_info::store_immediate_info): Initialize them
56 from newly added ctor args.
57 (merged_store_group::apply_stores): Formatting fixes. Sort by
59 (stmts_may_clobber_ref_p): For stores call also
60 refs_anti_dependent_p.
61 (gather_bswap_load_refs): New function.
62 (imm_store_chain_info::try_coalesce_bswap): New method.
63 (imm_store_chain_info::coalesce_immediate_stores): Use it.
64 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
65 (imm_store_chain_info::output_merged_store): Fail if number of
66 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
67 and NOP_EXPR rhs_code.
68 (pass_store_merging::process_store): Compute n and ins_stmt, if
69 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
70 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
72 (pass_store_merging::execute): Calculate dominators.
74 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
75 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
76 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
77 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
78 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
79 class pass_optimize_bswap, bswap_replace,
80 pass_optimize_bswap::execute): Moved to ...
81 * gimple-ssa-store-merging.c: ... this file.
82 Include optabs-tree.h.
83 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
84 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
85 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
86 anonymous namespace, remove static keywords.
87 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
88 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
90 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
93 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
95 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
97 * vec.h (debug_helper): New function.
98 (DEFINE_DEBUG_VEC): New macro.
99 * hash-set.h (debug_helper): New function.
100 (DEFINE_DEBUG_HASH_SET): New macro.
101 * cfg.c (debug_slim (edge)): New function.
102 Call DEFINE_DEBUG_VEC for edges.
103 Call DEFINE_DEBUG_HASH_SET for edges.
104 * cfghooks.c (debug_slim (basic_block)): New function.
105 Call DEFINE_DEBUG_VEC for basic blocks.
106 Call DEFINE_DEBUG_HASH_SET for basic blocks.
107 * print-tree.c (debug_slim): New function to handle trees.
108 Call DEFINE_DEBUG_VEC for trees.
109 Call DEFINE_DEBUG_HASH_SET for trees.
110 (debug (vec<tree, va_gc>) &): Remove.
111 (debug (<vec<tree, va_gc>) *): Remove.
112 * print-rtl.c (debug_slim): New function to handle const_rtx.
113 Call DEFINE_DEBUG_VEC for rtx_def.
114 Call DEFINE_DEBUG_VEC for rtx_insn.
115 Call DEFINE_DEBUG_HASH_SET for rtx_def.
116 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
117 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
118 (debug (vec<rtx_insn *> *ptr): Remove.
119 (debug_insn_vector): Remove.
120 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
122 2017-11-20 Tom de Vries <tom@codesourcery.com>
124 PR rtl-optimization/82020
125 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
126 IF_THEN_ELSE condition.
128 2017-11-19 Jeff Law <law@redhat.com>
130 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
131 of degenerates resulting from ignoring an edge.
133 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
136 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
138 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
141 * profile-count.c (profile_count::to_sreal_scale): Fix return value
142 for uninitialied counts.
144 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
147 * tree-inline.c (estimate_num_insns): Set to 1 at least.
149 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
152 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
155 2017-11-19 Tom de Vries <tom@codesourcery.com>
157 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
160 2017-11-19 Tom de Vries <tom@codesourcery.com>
162 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
165 2017-11-19 Tom de Vries <tom@codesourcery.com>
167 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
170 2017-11-19 Tom de Vries <tom@codesourcery.com>
172 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
174 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
176 2017-11-19 Tom de Vries <tom@codesourcery.com>
178 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
180 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
181 semicolon after MCORE_EXPORT_NAME call.
183 2017-11-19 Tom de Vries <tom@codesourcery.com>
186 * vmsdbgout.c (vmsdbgout_early_finish): New function.
187 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
189 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
191 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
192 (cgraph_edge::clone): Cleanup updating of profile.
193 * ipa-cp.c (update_profiling_info): Likewise.
194 * ipa-inline-transform.c (inline_transform): Likewise.
195 * ipa-inline.c (inline_small_functions): Add missing space to dump.
196 * ipa-split.c (execute_split_functions): Do not split when function
198 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
199 * profile-count.c (profile_count::dump): Add global0.
200 (profile_count::to_cgraph_frequency): Do not ICE when entry is
202 (profile_count::to_sreal_scale): Likewise.
203 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
204 (profile_count::combine_with_ipa_count): New function.
205 * profile-count.h (profile_guessed_global0adjusted): New.
206 (profile_count::adjusted_zero): New.
207 (profile_count::global0adjusted): New.
208 (profile_count::combine_with_ipa_count): New.
209 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
210 correct profile of return block of split functions.
211 (copy_cfg_body): Remove unused profile_count.
212 (copy_body): Likewise.
213 (expand_call_inline): Update.
214 (tree_function_versioning): Update.
216 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
218 * hash-set.h (hash_set::empty): New.
219 * tree-ssa-threadbackward.h: Delete.
220 * tree-ssa-threadbackward.c (class thread_jumps): New.
221 Move max_threaded_paths into class.
222 (fsm_find_thread_path): Remove arguments that are now in class.
223 (profitable_jump_thread_path): Rename to...
224 (thread_jumps::profitable_jump_thread_path): ...this.
225 (convert_and_register_jump_thread_path): Rename to...
226 (thread_jumps::convert_and_register_current_path): ...this.
227 (check_subpath_and_update_thread_path): Rename to...
228 (thread_jumps::check_subpath_and_update_thread_path): ...this.
229 (register_jump_thread_path_if_profitable): Rename to...
230 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
231 (handle_phi): Rename to...
232 (thread_jumps::handle_phi): ...this.
233 (handle_assignment): Rename to...
234 (thread_jumps::handle_assignment): ...this.
235 (fsm_find_control_statement_thread_paths): Rename to...
236 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
237 (find_jump_threads_backwards): Rename to...
238 (thread_jumps::find_jump_threads_backwards): ...this.
239 Initialize path local data.
240 (pass_thread_jumps::execute): Call find_jump_threads_backwards
241 from within thread_jumps class.
242 (pass_early_thread_jumps::execute): Same.
244 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
246 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
248 2017-11-17 Steve Ellcey <sellcey@cavium.com>
250 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
251 to canonical location.
253 2017-11-17 Steve Ellcey <sellcey@cavium.com>
256 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
258 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
260 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
262 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
263 rcrt1.o%s/grcrt1.o%s for -static-pie.
265 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
267 * i386.c (ix86_multiplication_cost, ix86_division_cost,
268 ix86_shift_rotate_cost): Break out from ...
269 (ix86_rtx_costs): ... here.
270 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
273 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
275 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
276 when entry block was promoted unlikely.
277 (estimate_bb_frequencies): Increase frequency scale.
278 * profile-count.h (profile_count): Export precision info.
280 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
282 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
283 disturbin profile of entry block.
285 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
287 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
290 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
292 * ipa-cp.c (update_profiling_info): Handle conversion to local
294 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
296 2017-11-17 Jeff Law <law@redhat.com>
298 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
299 from evrp_dom_walker class. Various methods moved into new class.
300 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
301 (evrp_range_analyzer::enter): New method.
302 (evrp_range_analyzer::leave): New method.
303 (evrp_dom_walker): Remove delegators no longer needed by this class.
304 Replace vr_values data member with evrp_range_analyzer
306 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
307 method extracted from evrp_dom_walker::before_dom_children.
308 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
309 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
311 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
312 Add private copy constructor and move assignment operators.
313 Privatize methods and class data where trivially possible.
314 (evrp_dom_walker::cleanup): New function, extracted from
315 execute_early_vrp. Simplify access to class data.
317 * vr-values.h (get_output_for_vrp): Prototype.
318 * vr-values.c (get_output_for_vrp): New function extracted from
319 vrp_visit_assignment_or_call and extract_range_from_stmt.
320 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
322 2017-11-17 Luis Machado <luis.machado@linaro.org>
324 * config/aarch64/aarch64.c
325 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
326 (qdf24xx_tunings) <autoprefetcher_model>: Set to
327 tune_params::AUTOPREFETCHER_WEAK.
329 2017-11-17 Tamar Christina <tamar.christina@arm.com>
332 * config/arm/arm.c (arm_valid_target_attribute_rec):
333 Parse "arch=" and "+<ext>".
334 (arm_valid_target_attribute_tree): Re-init global options.
335 (arm_option_override): Make non-static.
336 (arm_options_perform_arch_sanity_checks): Make errors fatal.
337 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
338 (__ARM_FEATURE_CRC32): Support undef.
339 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
340 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
342 2017-11-17 David Malcolm <dmalcolm@redhat.com>
344 * gdbinit.in (break-on-diagnostic): New command.
346 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
348 * config/i386/i386.c (ix86_expand_epilogue): Change simple
349 return to indirect jump for EH return if control-flow
350 protection is enabled. Change explicit 'false' argument in
351 pro_epilogue_adjust_stack with a value of
353 * config/i386/i386.md (simple_return_indirect_internal):
354 Remove SImode restriction to support 64-bit.
356 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
358 * combine.c (added_notes_insn): New.
359 (try_combine): Handle added_notes_insn like added_links_insn.
360 Rewrite return value code.
361 (distribute_notes): Set added_notes_insn to the earliest insn we added
364 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
366 PR rtl-optimization/82621
367 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
368 dest of one of those SETs is unused.
370 2017-11-17 Richard Biener <rguenther@suse.de>
373 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
374 * tree-pretty-print.c (dump_generic_node): Handle
375 annot_expr_parallel_kind.
376 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
377 * gimplify.c (gimple_boolify): Likewise.
379 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
381 * config.gcc (extra_headers): Add cet.h for x86 targets.
382 * config/i386/cet.h: New file.
383 * doc/install.texi: Add --enable-cet/--disable-cet.
385 2017-11-17 Richard Biener <rguenther@suse.de>
387 PR tree-optimization/83017
388 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
389 (gen_parallel_loop): Properly count iterations.
390 (parallelize_loops): Handle loop->can_be_parallel independent
391 of flag_loop_parallelize_all. Make static profitability test match
393 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
394 * invoke.texi (parloops-min-per-thread): Document.
396 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
398 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
399 upstreaming review comments.
401 2017-11-17 Tamar Christina <tamar.christina@arm.com>
403 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
404 with fast unaligned access.
405 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
407 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
409 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
410 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
412 2017-11-17 Richard Biener <rguenther@suse.de>
414 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
415 folding of references.
417 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
420 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
421 of replacing call to strncmp with corresponding call to strcmp when
424 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
426 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
427 option prefer-avx256 for skylake-avx512 configuration.
428 * config/i386/i386.c (ix86_option_override_internal): Ditto.
429 (get_builtin_code_for_version): Ditto.
431 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
432 Monk Chiang <sh.chiang04@gmail.com>
434 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
435 (FIXED_REGISTERS): Reserve more register numbers.
436 (CALL_USED_REGISTERS): Likewise.
437 (REG_ALLOC_ORDER): Likewise.
438 (REG_CLASS_CONTENTS): Likewise.
439 (REGISTER_NAMES): Likewise.
441 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
442 Kito Cheng <kito.cheng@gmail.com>
444 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
446 * config/nds32/iterators.md: Add vector mode iterators and attributes.
448 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
450 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
451 parameter list for vec_splats.
453 2017-11-16 Joseph Myers <joseph@codesourcery.com>
455 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
457 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
460 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
463 * doc/install.texi: Document incompatibility of Perl >=5.6.26
464 with the required version of automake 1.11.6.
466 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
468 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
470 (DU_C2_3_power9): Correct reservation combinations.
471 (FP_DIV_power9, VEC_DIV_power9): New.
472 (power9-alu): Split out rotate/shift...
473 (power9-rot): ...to here, correct dispatch resource.
474 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
476 (power9-fp): Correct latency.
477 (power9-sdiv): Add div/sqrt resource.
478 (power9-ddiv): Correct latency, add div/sqrt resource.
479 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
480 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
482 (power9-qpdiv, power9-qpmul): Adjust resource usage.
484 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
486 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
487 switch statement mapping KF built-ins to TF built-ins if we don't
488 have the proper ISA 3.0 assembler support.
490 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
492 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
493 (lower_emutls_function_body): Do not compute it.
495 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
497 * ipa-split.c (split_bb_info): Turn time to sreal.
498 (split_point): Likewise.
499 (dump_split_point): Likewise.
500 (fine_split_points): Likewise.
501 (execute_split_functions): Only zero split_bbs; turn time to sreals.
503 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
505 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
508 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
510 * predict.c (combine_predictions_for_bb): Preserve zero predicted
512 (expensive_function_p): Remove useless assert.
513 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
515 2017-11-16 Martin Sebor <msebor@redhat.com>
517 PR tree-optimization/82588
518 PR tree-optimization/82583
519 * tree-vrp.c (check_array_ref): Handle flexible array members,
520 string literals, and inner indices.
521 (search_for_addr_array): Add detail to diagnostics.
523 2017-11-16 Nathan Sidwell <nathan@acm.org>
527 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
528 (SET_DECL_ASSEMBLER_NAME): Forward to
529 overwrite_decl_assembler_name.
530 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
531 (overwrite_decl_assembler_name): Declare.
532 * tree.c (overwrite_decl_assembler_name): New.
533 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
534 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
535 (LANG_HOOKS_INITIALIZER): Add it.
536 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
537 * langhooks.c (lhd_set_decl_assembler_name): Use
538 SET_DECL_ASSEMBLER_NAME.
539 (lhd_overwrite_decl_assembler_name): Default implementation.
541 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
542 Jackson Woodruff <jackson.woodruff@arm.com>
544 PR tree-optimization/71026
545 * match.pd: Canonicalize constant multiplies in division.
547 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
549 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
550 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
551 Remove OPT_fomit_frame_pointer entry.
552 * common/config/arc/arc-common.c: Likewise.
553 * common/config/arm/arm-common.c: Likewise.
554 * common/config/avr/avr-common.c: Likewise.
555 * common/config/c6x/c6x-common.c: Likewise.
556 * common/config/cr16/cr16-common.c: Likewise.
557 * common/config/cris/cris-common.c: Likewise.
558 * common/config/epiphany/epiphany-common.c: Likewise.
559 * common/config/fr30/fr30-common.c: Likewise.
560 * common/config/frv/frv-common.c: Likewise.
561 * common/config/ia64/ia64-common.c: Likewise.
562 * common/config/iq2000/iq2000-common.c: Likewise.
563 * common/config/lm32/lm32-common.c: Likewise.
564 * common/config/m32r/m32r-common.c: Likewise.
565 * common/config/mcore/mcore-common.c: Likewise.
566 * common/config/microblaze/microblaze-common.c: Likewise.
567 * common/config/mips/mips-common.c: Likewise.
568 * common/config/mmix/mmix-common.c: Likewise.
569 * common/config/mn10300/mn10300-common.c: Likewise.
570 * common/config/nios2/nios2-common.c: Likewise.
571 * common/config/pa/pa-common.c: Likewise.
572 * common/config/pdp11/pdp11-common.c: Likewise.
573 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
574 * common/config/riscv/riscv-common.c: Likewise.
575 * common/config/rs6000/rs6000-common.c: Likewise.
576 * common/config/rx/rx-common.c: Likewise.
577 * common/config/s390/s390-common.c: Likewise.
578 * common/config/sh/sh-common.c: Likewise.
579 * common/config/sparc/sparc-common.c: Likewise.
580 * common/config/tilegx/tilegx-common.c: Likewise.
581 * common/config/tilepro/tilepro-common.c: Likewise.
582 * common/config/v850/v850-common.c: Likewise.
583 * common/config/visium/visium-common.c: Likewise.
584 * common/config/xstormy16/xstormy16-common.c: Likewise.
585 * common/config/xtensa/xtensa-common.c: Likewise.
586 * invoke.texi (-fomit-frame-pointer): Update documentation.
588 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
590 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
592 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
594 * cfg.c (scale_bbs_frequencies_int,
595 cale_bbs_frequencies_gcov_type): Remove.
596 * cfg.h (scale_bbs_frequencies_int,
597 cale_bbs_frequencies_gcov_type): Remove.
599 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
601 * tree-ssa-loop-manip.c
602 (scale_dominated_blocks_in_loop): Update to profile counts.
603 (tree_transform_and_unroll_loop): Likewise.
605 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
607 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
608 scale_bbs_frequencies_int.
610 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
612 * final.c (compute_alignments): Use counts rather than frequencies.
614 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
616 * cfgloopanal.c: Include sreal.h
617 (average_num_loop_insns): Use counts and sreal for accounting.
619 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
621 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
624 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
625 Kito Cheng <kito.cheng@gmail.com>
627 * config/nds32/constraints.md: Provide more constraints.
628 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
629 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
630 support constraints usage.
632 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
634 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
636 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
637 Kito Cheng <kito.cheng@gmail.com>
639 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
640 * config/nds32/nds32.opt: Refine the layout.
641 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
642 TARGET_EXT_STRING): Support new options.
643 * config/nds32/nds32.h: Likewise.
644 * config/nds32/nds32.md: Likewise.
645 * config/nds32/nds32-predicates.c: Likewise.
646 * config/nds32/constraints.md: Likewise.
647 * common/config/nds32/nds32-common.c: Likewise.
649 2017-11-16 Julia Koval <julia.koval@intel.com>
652 * config/i386/gfniintrin.h: Add sse check.
653 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
655 2017-11-16 Julia Koval <julia.koval@intel.com>
657 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
658 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
659 (ix86_handle_option): Handle -mavx512vbmi2.
660 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
661 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
662 * config/i386/i386-c.c (__AVX512VBMI2__): New.
663 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
664 (ix86_valid_target_attribute_inner_p): Ditto.
665 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
666 * config/i386/i386.opt (mavx512vbmi2): New option.
667 * doc/invoke.texi: Add new option.
669 [2017-11-16 Julia Koval <julia.koval@intel.com>
671 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
672 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
673 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
674 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
675 _mm512_gf2p8mul_epi8): New intrinsics.
676 * config/i386/i386-builtin-types.def
677 (V64QI_FTYPE_V64QI_V64QI): New type.
678 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
679 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
680 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
681 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
682 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
683 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
685 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
687 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
688 explicitly as a stream of bytes.
690 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
692 * config/rs6000/altivec.h (vec_xst_be): New #define.
693 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
694 and externalize from *altivec_vperm_<mode>_internal.
695 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
697 (XL_BE_V8HI): Likewise.
698 (XL_BE_V4SI): Likewise.
699 (XL_BE_V4SI): Likewise.
700 (XL_BE_V2DI): Likewise.
701 (XL_BE_V4SF): Likewise.
702 (XL_BE_V2DF): Likewise.
703 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
704 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
705 all array entries with these keys: VSX_BUILTIN_VEC_XL,
706 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
707 VSX_BUILTIN_VEC_XST_BE.
708 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
709 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
711 (altivec_init_builtins): Replace conditional calls to def_builtin
712 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
713 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
714 with unconditional calls. Remove calls to def_builtin for
715 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
716 __builtin_vec_xst_be.
717 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
718 to define_expand, and add alternate RTL generation for P8.
719 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
721 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
722 add alternate RTL generation for P8.
723 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
724 vsx_ld_elemrev_v16qi.
725 (vsx_st_elemrev_v8hi): Convert define_insn
726 to define_expand, and add alternate RTL generation for P8.
727 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
729 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
730 add alternate RTL generation for P8.
731 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
732 vsx_st_elemrev_v16qi.
734 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
737 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
738 TARGET_AVX512ER check.
739 (ix86_option_override_internal): Set MASK_VZEROUPPER if
740 neither -mzeroupper nor -mno-zeroupper is used and
741 TARGET_EMIT_VZEROUPPER is set.
742 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
743 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
745 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
747 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
748 folding of vector compares.
749 (fold_build_vec_cmp): New helper function.
750 (fold_compare_helper): New helper function.
751 (builtin_function_type): Add compare builtins to the list of functions
752 having unsigned arguments. Cosmetic updates to comment indentation.
753 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
754 the not+eq combination.
756 2017-11-15 Bin Cheng <bin.cheng@arm.com>
758 PR tree-optimization/82726
759 PR tree-optimization/70754
760 * tree-predcom.c (order_drefs_by_pos): New function.
761 (combine_chains): Move code setting has_max_use_after to...
762 (try_combine_chains): ...here. New parameter. Sort combined chains
763 according to position information.
764 (tree_predictive_commoning_loop): Update call to above function.
765 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
767 2017-11-15 Bin Cheng <bin.cheng@arm.com>
769 PR tree-optimization/82726
771 2017-01-23 Bin Cheng <bin.cheng@arm.com>
773 PR tree-optimization/70754
774 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
775 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
776 combined stmt before it if not NULL.
777 (combine_chains): Process refs reversely and compute dominance point
781 2017-02-23 Bin Cheng <bin.cheng@arm.com>
783 PR tree-optimization/79663
784 * tree-predcom.c (combine_chains): Process refs in reverse order
785 only for ZERO length chains, and add explaining comment.
787 2017-11-15 Tamar Christina <tamar.christina@arm.com>
789 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
790 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
791 * doc/invoke.texi (ARM Options): Add armv8.3-a.
793 2017-11-15 Tamar Christina <tamar.christina@arm.com>
795 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
797 2017-11-15 Martin Liska <mliska@suse.cz>
799 * tree-cfg.c (pass_warn_function_return::execute):
800 Compare warn_return_type for greater than zero.
802 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
806 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
807 to return true on Xeon and not on Xeon Phi.
808 (ix86_check_avx256_register): Changed to ...
809 (ix86_check_avx_upper_register): ... this. Add extra check for
810 VALID_AVX512F_REG_OR_XI_MODE.
811 (ix86_avx_u128_mode_needed): Changed
812 ix86_check_avx256_register to ix86_check_avx_upper_register.
813 (ix86_check_avx256_stores): Changed to ...
814 (ix86_check_avx_upper_stores): ... this. Changed
815 ix86_check_avx256_register to ix86_check_avx_upper_register.
816 (ix86_avx_u128_mode_after): Changed
817 avx_reg256_found to avx_upper_reg_found. Changed
818 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
819 (ix86_avx_u128_mode_entry): Changed
820 ix86_check_avx256_register to ix86_check_avx_upper_register.
821 (ix86_avx_u128_mode_exit): Ditto.
822 * config/i386/i386.h: (host_detect_local_cpu): New define.
824 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
826 * config/arm/xgene1.md (xgene1): Split into automatons
827 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
828 (xgene1_f_load): Adjust reservations and/or types.
829 (xgene1_f_store): Likewise.
830 (xgene1_load_pair): Likewise.
831 (xgene1_store_pair): Likewise.
832 (xgene1_fp_load1): Likewise.
833 (xgene1_load1): Likewise.
834 (xgene1_store1): Likewise.
835 (xgene1_move): Likewise.
836 (xgene1_alu): Likewise.
837 (xgene1_simd): Likewise.
838 (xgene1_bfm): Likewise.
839 (xgene1_neon_load1): Likewise.
840 (xgene1_neon_store1): Likewise.
841 (xgene1_neon_logic): Likewise.
842 (xgene1_neon_st1): Likewise.
843 (xgene1_neon_ld1r): Likewise.
844 (xgene1_alu_cond): Added.
845 (xgene1_shift_reg): Likwise.
846 (xgene1_bfx): Likewise.
847 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
849 2017-11-15 Jakub Jelinek <jakub@redhat.com>
852 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
854 (can_widen_mult_without_libcall): New function.
855 (expand_mul_overflow): If only checking unsigned mul overflow,
856 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
857 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
858 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
859 (expand_DIVMOD): Formatting fix.
860 * expmed.h (expand_mult): Add NO_LIBCALL argument.
861 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
862 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
864 PR tree-optimization/82977
865 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
866 constructed temporary to strlen_to_stridx.put.
868 2017-11-15 Martin Liska <mliska@suse.cz>
870 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
871 * configure: Regenerate.
873 2017-11-15 Martin Liska <mliska@suse.cz>
876 * config/sh/sh-mem.cc: Use proper probability for
879 2017-11-14 Jeff Law <law@redhat.com>
881 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
882 the red zone for stack_clash_protection_final_dynamic_probe targets
883 when the total dynamic stack size is zero bytes.
885 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
886 blocks is post order.
888 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
890 * dumpfile.h (TDF_COMPARE_DEBUG): New.
891 * final.c (rest_of_clean_state): Set it for the
892 -fcompare-debug dump.
893 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
894 class when TDF_COMPARE_DEBUG is set.
896 * dwarf2out.c (gen_producer_string): Discard
899 2017-11-15 Joseph Myers <joseph@codesourcery.com>
902 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
903 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
904 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
905 (__TGMATH_REAL_2_3): Remove macros.
906 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
907 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
909 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
911 (remquo): Define using __TGMATH_REAL_3.
913 2017-11-14 Jeff Law <law@redhat.com>
915 * vr-values.c: New file with contents extracted from tree-vrp.c.
916 * Makefile.in (OBJS): Add vr-values.o
917 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
918 (set_value_range, set_and_canonicalize_value_range): Likewise.
919 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
920 (value_range_constant_singleton, symbolic_range_p): Likewise.
921 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
922 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
923 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
924 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
925 (find_case_label_range, find_case_label_index): Likewise.
926 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
927 (range_int_cst_singleton_p, value_inside_range): Likewise.
928 (get_single_symbol): Likewise.
929 (switch_update): Move structure definition here.
930 (to_remove_edges, to_update_switch_stmts): Provide externs.
931 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
932 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
933 (vrp_val_is_min, set_value_range): Likewise.
934 (set_and_canonicalize_value_range, copy_value_range): Likewise.
935 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
936 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
937 (range_is_nonnull, range_int_cst_p): Likewwise.
938 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
939 (get_single_symbol, operand_less_p): Likewise
940 (compare_values_warnv, compare_values): Likewise.
941 (value_inside_range, value_range_constant_singleton): Likewise.
942 (zero_nonzero_bitgs_from_vr): Likewise.
943 (extract_range_from_binary_expr_1): Likewise.
944 (overflow_comparison_p): Likewise.
945 (to_remove_edges, to_update_switch_stmts): Likewise.
946 (find_case_label-index, find_case_label_range): Likewise.
947 (switch_update, set_value_range_to_nonnegative): Remove.
948 (set_value_range_to_truthvalue): Likewise.
949 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
950 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
951 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
952 (find_case_label_ranges, test_for_singularity): Likewise.
953 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
954 (x_vr_values): Move to its remaining use site.
956 2017-11-10 Jeff Law <law@redhat.com>
958 * vr-values.h (VR_INITIALIZER): Move #define here.
959 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
960 * Makefile.in (OBJS): Add tree-evrp.o
961 * tree-vrp.h (assert_info): Move structure definition here.
962 (set_value_range_to_varying): Prototype.
963 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
964 (infer_value_range, register_edge_assert_for): Likewise.
965 (stmt_interesting_for_vrp): Likewise.
966 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
967 (set_value_range_to_varying): No longer static.
968 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
969 (infer_value_range, register_edge_assert_for): Likewise.
971 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
973 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
974 generate the XXBRD instruction.
976 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
977 (rs6000_builtin_type_compatible): Treat _Float128 and long double
978 as being compatible if -mabi=ieeelongdouble.
979 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
980 to setup float128 built-ins with hardware support.
981 (BU_FLOAT128_HW_2): Likewise.
982 (BU_FLOAT128_HW_3): Likewise.
983 (BU_FLOAT128_HW_VSX_1): Likewise.
984 (BU_FLOAT128_HW_VSX_2): Likewise.
985 (scalar_extract_expq): Change float128 built-in functions to
986 accommodate having both KFmode and TFmode functions. Use the
987 KFmode variant as the default.
988 (scalar_extract_sigq): Likewise.
989 (scalar_test_neg_qp): Likewise.
990 (scalar_insert_exp_q): Likewise.
991 (scalar_insert_exp_qp): Likewise.
992 (scalar_test_data_class_qp): Likewise.
993 (sqrtf128_round_to_odd): Delete processing the round to odd
994 built-in functions as special built-in functions, and define them
995 as float128 built-ins. Use the KFmode variant as the default.
996 (truncf128_round_to_odd): Likewise.
997 (addf128_round_to_odd): Likewise.
998 (subf128_round_to_odd): Likewise.
999 (mulf128_round_to_odd): Likewise.
1000 (divf128_round_to_odd): Likewise.
1001 (fmaf128_round_to_odd): Likewise.
1002 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1003 support for KFmode and TFmode xststdcqp calls.
1004 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1005 point, switch the built-in handlers for the get/set float128
1006 exponent, get float128 mantissa, float128 test built-ins, and the
1007 float128 round to odd built-in functions. Eliminate creating the
1008 float128 round to odd built-in functions as special built-ins.
1009 (rs6000_init_builtins): Eliminate special creation of the float128
1010 round to odd built-in functions.
1011 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1012 function insns to support both TFmode and KFmode variants.
1013 (xsxsigqp_<mode>): Likewise.
1014 (xsiexpqpf_<mode>): Likewise.
1015 (xsiexpqp_<mode>): Likewise.
1016 (xststdcqp_<mode>): Likewise.
1017 (xststdcnegqp_<mode>): Likewise.
1018 (xststdcqp_<mode>): Likewise.
1020 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1022 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1023 unused path_in_freq_ptr parameter.
1024 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1026 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1028 * ipa-inline.c (edge_badness): Dump sreal frequency.
1029 (compute_inlined_call_time): Match natural implementaiton ...
1030 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1031 forgotten division by CGRAPH_FREQ_BASE.
1033 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1035 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1036 Solaris 11. Update comment.
1037 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1039 * config/sol2.h (STARTFILE_SPEC): Likewise.
1040 * configure: Regenerate.
1042 2017-11-14 Carl Love <cel@us.ibm.com>
1044 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1045 le_ and be_ prefixes to swap* variables. Remove
1046 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1049 2017-11-14 Jason Merrill <jason@redhat.com>
1051 Support GTY((cache)) on hash_map.
1052 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1053 (ggc_cache_remove): Override it instead of ggc_mx.
1054 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1055 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1056 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1057 (simple_cache_map_traits): Override maybe_mx.
1058 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1059 (hash_map): Friend gt_cleare_cache.
1060 (gt_cleare_cache): New.
1061 * tree.h (tree_cache_traits): New hash_map traits class.
1062 (tree_cache_map): New typedef.
1064 2017-11-14 Richard Biener <rguenther@suse.de>
1066 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1067 paramter and handling.
1068 (cleanup_control_flow_bb): Likewise.
1069 (cleanup_control_flow_pre): New helper performing a DFS walk
1070 to call cleanup_control_flow_bb in PRE order.
1071 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1072 via cleanup_control_flow_pre.
1074 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1076 * config/aarch64/aarch64-simd.md
1077 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1078 (*aarch64_simd_bsl<mode>_alt): Likewise.
1079 (aarch64_simd_bsldi_internal): New.
1080 (aarch64_simd_bsldi_alt): Likewise.
1082 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1084 * tracer.c (better_p): Do not compare frequencies.
1085 * reg-stack.c (better_edge): Likewise.
1086 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1089 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1091 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1092 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1093 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1094 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1095 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1096 * predict.c (maybe_hot_frequency_p): Inline to ...
1097 (maybe_hot_count_p): ... here; rewrite to counts.
1098 (counts_to_freqs): Rename to ...
1099 (update_max_bb_count): ... this one.
1100 (expensive_function_p): Use counts.
1101 (estimate_bb_frequencies): Update.
1102 (rebuild_frequencies): Update.
1103 * predict.h (counts_to_freqs): Rename to ...
1104 (update_max_bb_count): ... this one.
1105 * profile.c (compute_branch_probabilities): Add debug info
1106 * tree-inline.c (expand_call_inline): Update debug info.
1107 (optimize_inline_calls): Use update_max_bb_count..
1108 (tree_function_versioning): Use update_max_bb_count..
1109 * value-prof.c (gimple_value_profile_transformations):
1110 Do not use update_max_bb_count.
1112 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1114 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1115 always use frequencies.
1117 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1119 * bb-reorder.c: Remove frequencies from comments.
1120 (better_edge_p): Use profile counts.
1121 (find_traces): Dump profile counts.
1122 (rotate_loop): Use profile counts.
1123 (find_traces_1_round): Likewise.
1124 (connect_better_edge_p): Use counts instead of probabilities for
1126 (copy_bb_p): Drop early check for non-0 frequency.
1127 (sanitize_hot_paths): Update comments.
1129 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1131 * ipa-split.c (struct split_point): Add count.
1132 (consider_split): Do not compute incoming frequency; compute incoming
1133 count and store it to split_point.
1134 (split_function): Set count of the call to split part correctly.
1136 2017-11-13 Carl Love <cel@us.ibm.com>
1138 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1140 2017-11-13 Tom Tromey <tom@tromey.com>
1142 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1144 2017-11-13 Carl Love <cel@us.ibm.com>
1146 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1147 Add support for builtins:
1148 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1149 vector {un,}signed {char,int,short},
1150 vector {un,}signed {char,int,short}) arguments.
1151 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1152 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1153 Add BU_P9V_AV_2 expansions for the builtins.
1154 * config/rs6000/altivec.h (vec_first_match_index,
1155 vec_first_mismatch_index, vec_first_match_or_eos_index,
1156 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1157 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1158 new extern declaration.
1159 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1160 * config/rs6000/vsx.md (first_match_index_<mode>,
1161 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1162 first_mismatch_or_eos_index_<mode>): Add define expand.
1163 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1164 * doc/extend.texi: Update the built-in documenation file for the new
1167 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1169 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1170 into the min/max operations for _Float<N> and _Float<N>X types.
1172 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1175 * dwarf2out.c (do_eh_frame): New static variable.
1176 (dwarf2out_begin_prologue): Set it.
1177 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1179 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1181 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1183 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1185 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1188 * debug.h (dwarf2out_do_eh_frame): Declare.
1189 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1190 (dwarf2out_do_frame): Use it.
1191 (dwarf2out_do_cfi_asm): Likewise.
1192 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1193 (dwarf2out_assembly_start): Likewise.
1194 (dwarf2out_begin_prologue): Fix comment.
1195 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1196 if the target needs either debug or unwind DWARF2 info.
1197 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1198 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1199 -ftrapping-math, -ftrapv and -fwrapv.
1201 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1203 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1204 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1205 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1206 (estimate_edge_size_and_time): Likewise.
1207 (ipa_merge_fn_summary_after_inlining): Likewise.
1208 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1209 (compute_uninlined_call_time): Use sreal_frequency.
1210 (compute_inlined_call_time): Likewise.
1211 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1212 * profile-count.c: Include sreal.h.
1213 (profile_count::to_sreal_scale): New.
1214 * profile-count.h: Forward declare sreal.
1215 (profile_count::to_sreal_scale): Declare.
1217 2017-11-13 Nathan Sidwell <nathan@acm.org>
1219 * diagnostic.c (maybe_line_and_column): New.
1220 (diagnostic_get_location_text): Use it.
1221 (diagnostic_report_current_module): Likewise.
1222 (test_diagnostic_get_location_text): Add tests.
1224 2017-11-13 Luis Machado <luis.machado@linaro.org>
1226 * doc/md.texi (Specifying processor pipeline description): Fix
1227 incorrect latency for the div instruction example.
1229 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1231 PR tree-optimization/78821
1232 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1233 that bit_not_p is the same.
1234 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1235 (split_group): Count precisely bit_not_p bits in each statement.
1236 (invert_op): New function.
1237 (imm_store_chain_info::output_merged_store): Use invert_op to
1238 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1239 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1241 2017-11-13 Martin Liska <mliska@suse.cz>
1243 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1244 (struct source_info): Likewise.
1245 (add_branch_counts): Likewise.
1246 (add_line_counts): Likewise.
1247 (function_summary): Likewise.
1248 (output_intermediate_line): Likewise.
1249 (generate_results): Likewise.
1251 2017-11-13 Martin Liska <mliska@suse.cz>
1253 * gcov.c (struct block_info): Remove typedef for block_t.
1254 (struct line_info): Likewise.
1255 (line_info::has_block): Likewise.
1256 (EXIT_BLOCK): Likewise.
1257 (unblock): Likewise.
1258 (circuit): Likewise.
1259 (get_cycles_count): Likewise.
1260 (process_file): Likewise.
1261 (read_graph_file): Likewise.
1262 (solve_flow_graph): Likewise.
1263 (find_exception_blocks): Likewise.
1264 (add_line_counts): Likewise.
1265 (accumulate_line_info): Likewise.
1266 (output_line_details): Likewise.
1268 2017-11-13 Martin Liska <mliska@suse.cz>
1270 * gcov.c (struct arc_info): Remove typedef for arc_t.
1271 (struct line_info): Likewise.
1272 (add_branch_counts): Likewise.
1273 (output_branch_count): Likewise.
1274 (function_info::~function_info): Likewise.
1275 (circuit): Likewise.
1276 (output_intermediate_line): Likewise.
1277 (read_graph_file): Likewise.
1278 (solve_flow_graph): Likewise.
1279 (find_exception_blocks): Likewise.
1280 (add_line_counts): Likewise.
1281 (accumulate_line_info): Likewise.
1282 (output_line_details): Likewise.
1283 (output_function_details): Likewise.
1285 2017-11-13 Martin Liska <mliska@suse.cz>
1287 * gcov.c (struct function_info): Remove typedef for function_t.
1288 (struct source_info): Likewise.
1289 (source_info::get_functions_at_location): Likewise.
1290 (solve_flow_graph): Likewise.
1291 (find_exception_blocks): Likewise.
1292 (add_line_counts): Likewise.
1293 (output_intermediate_file): Likewise.
1294 (process_file): Likewise.
1295 (generate_results): Likewise.
1296 (release_structures): Likewise.
1297 (read_graph_file): Likewise.
1298 (read_count_file): Likewise.
1299 (accumulate_line_counts): Likewise.
1300 (output_lines): Likewise.
1302 2017-11-13 Martin Liska <mliska@suse.cz>
1304 * gcov.c (function_info::function_info): Remove num_counts
1305 and add vector<gcov_type>.
1306 (function_info::~function_info): Use the vector.
1307 (process_file): Likewise.
1308 (read_graph_file): Likewise.
1309 (read_count_file): Likewise.
1310 (solve_flow_graph): Likewise.
1312 2017-11-13 Martin Liska <mliska@suse.cz>
1314 * gcov.c (function_info::is_artificial): New function.
1315 (process_file): Erase all artificial early.
1316 (generate_results): Skip as all artificial are already
1319 2017-11-13 Martin Liska <mliska@suse.cz>
1321 * gcov.c (read_graph_file): Store to global vector of functions.
1322 (read_count_file): Iterate the vector.
1323 (process_file): Likewise.
1324 (generate_results): Likewise.
1325 (release_structures): Likewise.
1327 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1329 PR tree-optimization/82954
1330 * gimple-ssa-store-merging.c
1331 (imm_store_chain_info::coalesce_immediate_stores): If
1332 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1334 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1336 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1337 Upddate call to ENDIAN_LANE_N.
1338 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1339 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1340 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1341 and use aarch64_endian_lane_rtx.
1342 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1344 2017-11-12 Tom de Vries <tom@codesourcery.com>
1346 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1348 2017-11-12 Tom de Vries <tom@codesourcery.com>
1350 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1352 2017-11-12 Tom de Vries <tom@codesourcery.com>
1354 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1356 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1357 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1358 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1360 2017-11-11 Martin Sebor <msebor@redhat.com>
1363 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1366 * prefic.c (translate_name): Replace strncpy with memcpy to
1367 avoid -Wstringop-truncation.
1369 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1371 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1373 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1375 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1377 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1379 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1382 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1384 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1385 enable generating XXBRH if the value is in a vector register.
1386 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1387 value is in a vector register.
1388 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1389 register to register bswap64's instead of doing the GPR sequence
1390 used on previous machines.
1391 (bswapdi2_xxbrd): New insn.
1392 (bswapdi2_reg): Disallow on ISA 3.0.
1393 (register to register bswap64 splitter): Do not split the insn on
1394 ISA 3.0 systems that use XXBRD.
1396 2017-11-10 Martin Sebor <msebor@redhat.com>
1399 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1401 (find_subframework_file): Same.
1403 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1405 * auto-profile.c (afdo_indirect_call): Drop frequency.
1406 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1407 (cgraph_node::create_edge): Drop frequency argument.
1408 (cgraph_node::create_indirect_edge): Drop frequency argument.
1409 (cgraph_edge::make_speculative): Drop frequency arguments.
1410 (cgraph_edge::resolve_speculation): Do not update frequencies
1411 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1412 (cgraph_node::dump): Check consistency in IPA mode.
1413 (cgraph_edge::maybe_hot_p): Use IPA counter.
1414 (cgraph_edge::verify_count_and_frequency): Rename to ...
1415 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1416 (cgraph_node::verify_node): Update.
1417 * cgraph.h (struct cgraph_edge): Drop frequency.
1418 (cgraph_edge::frequency): New function.
1419 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1421 (cgraph_edge::rebuild_edges): Likewise.
1422 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1423 (duplicate_thunk_for_node): Do not pass frequency.
1424 (cgraph_node::create_clone): Scale only counts.
1425 (cgraph_node::create_virtual_clone): Do not pass frequency.
1426 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1427 (cgraph_node::create_version_clone): Do not pass frequency.
1428 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1429 (cgraph_node::expand_thunk): Do not pass frequency.
1430 (cgraph_node::create_wrapper): Do not pass frequency.
1431 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1433 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1434 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1435 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1436 (gather_caller_stats): Use frequency function.
1437 (ipcp_cloning_candidate_p): Use frequency function.
1438 (ipcp_propagate_stage): Use frequency function.
1439 (get_info_about_necessary_edges): Use frequency function.
1440 (update_profiling_info): Update only IPA profile.
1441 (update_specialized_profile): Use frequency functoin.
1442 (perhaps_add_new_callers): Update only IPA profile.
1443 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1444 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1445 (dump_ipa_call_summary): Use frequency function.
1446 (estimate_edge_size_and_time): Use frequency function.
1447 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1448 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1449 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1450 (update_noncloned_counts): ... ths one; scale counts only.
1451 (clone_inlined_nodes): Do not scale frequency.
1452 (inline_call): Do not pass frequency.
1453 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1454 (compute_inlined_call_time): Use IPA profile.
1455 (want_inline_small_function_p): Use IPA profile.
1456 (want_inline_self_recursive_call_p): Use IPA profile.
1457 (edge_badness): Use IPA profile.
1458 (lookup_recursive_calls): Use IPA profile.
1459 (recursive_inlining): Do not pass frequency.
1460 (resolve_noninline_speculation): Do not update frequency.
1461 (inline_small_functions): Collect max of IPA profile.
1462 (dump_overall_stats): Dump IPA porfile.
1463 (dump_inline_stats): Dump IPA porfile.
1464 (ipa_inline): Collect IPA stats.
1465 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1466 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1467 (ipa_propagate_frequency): Use frequency function.
1468 (ipa_profile): Cleanup.
1469 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1470 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1471 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1472 (input_node): Do not stream frequency.
1473 (input_edge): Do not stream frequency.
1474 (merge_profile_summaries): Scale only IPA profiles.
1475 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1476 * predict.c (drop_profile): Do not recompute frequency.
1477 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1478 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1479 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1480 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1481 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1482 * tree-inline.c (copy_bb): Do not scale frequency.
1483 (expand_call_inline): Do not scale frequency.
1484 (tree_function_versioning): Do not scale frequency.
1485 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1487 2017-11-10 Julia Koval <julia.koval@intel.com>
1489 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1490 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1491 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1492 (_mm256_maskz_gf2p8affine_epi64_epi8)
1493 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1494 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1495 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1496 (__builtin_ia32_vgf2p8affineqb_v32qi)
1497 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1498 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1500 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1504 (arm_option_override): Refactor.
1505 (arm_option_reconfigure_globals): New.
1506 (arm_options_perform_arch_sanity_checks): New.
1507 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1509 (arm_options_perform_arch_sanity_checks): Likewise
1511 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1513 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1514 (power9-qpmul): New.
1515 * rs6000/rs6000.md ("type" attr): Add qmul.
1516 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1517 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1518 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1520 2017-11-10 Martin Sebor <msebor@redhat.com>
1523 * builtins.c (compute_objsize): Handle arrays that
1524 compute_builtin_object_size likes to fail for. Make extern.
1525 * builtins.h (compute_objsize): Declare.
1526 (check_strncpy_sizes): New function.
1527 (expand_builtin_strncpy): Call check_strncpy_sizes.
1528 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1529 -Wstringop-truncation.
1530 (gimple_fold_builtin_strncat): Same.
1531 * gimple.c (gimple_build_call_from_tree): Set call location.
1532 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1533 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1534 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1535 (handle_builtin_strlen): Use strlen_to_stridx.
1536 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1538 Use strlen_to_stridx.
1539 (pass_strlen::execute): Release strlen_to_stridx.
1540 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1541 (-Wstringop-truncation): Document new option.
1543 2017-11-10 Martin Liska <mliska@suse.cz>
1545 PR gcov-profile/82702
1546 * gcov.c (main): Handle intermediate files in a different
1548 (get_gcov_intermediate_filename): New function.
1549 (output_gcov_file): Remove support of intermediate files.
1550 (generate_results): Allocate intermediate file.
1551 (release_structures): Clean-up properly fn_end.
1552 (output_intermediate_file): Start iterating with line 1.
1554 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1556 PR tree-optimization/82929
1557 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1558 ops_swapped_p non-static data member.
1559 (store_immediate_info::store_immediate_info): Clear it.
1560 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1561 ops set ops_swapped_p.
1562 (count_multiple_uses): Handle ops_swapped_p.
1564 2017-11-10 Martin Liska <mliska@suse.cz>
1566 * coverage.c (coverage_init): Stream information about
1567 support of has_unexecuted_blocks.
1568 * doc/gcov.texi: Document that.
1569 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1570 * gcov.c (read_graph_file): Likewise.
1571 (output_line_beginning): Fix a small issue with
1574 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1576 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1577 reference of trivial component.
1579 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1582 * gimple-ssa-store-merging.c
1583 (pass_store_merging::terminate_all_aliasing_chains): For
1584 gimple_store_p stmts also call refs_output_dependent_p.
1586 PR rtl-optimization/82913
1587 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1590 2017-11-09 Jeff Law <law@redhat.com>
1592 * vr-values.h: New file with vr_values class.
1593 * tree-vrp.c: Include vr-values.h
1594 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1595 data objects into the vr_values class.
1596 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1597 (get_value_range): Make it a member function within vr_values class.
1598 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1599 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1600 (op_with_constant_singleton_value_range): Likewise.
1601 (extract_range_for_var_from_comparison_expr): Likewise.
1602 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1603 (extract_range_from_binary_expr): Likewise.
1604 (extract_range_from_unary_expr): Likewise.
1605 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1606 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1607 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1608 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1609 (compare_name_with_value, compare_names): Likewise.
1610 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1611 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1612 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1613 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1614 (extract_range_from_phi_node): Likewise.
1615 (simplify_truth_ops_using_ranges): Likewise.
1616 (simplify_div_or_mod_using_ranges): Likewise.
1617 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1618 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1619 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1620 (simplify_float_conversion_using_ranges): Likewise.
1621 (simplify_internal_call_using_ranges): Likewise.
1622 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1623 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1624 poitner via x_vr_values for calls into gimple folder.
1625 (vrp_initialize_lattice): Make this the vr_values ctor.
1626 (vrp_free_lattice): Make this the vr_values dtor.
1627 (set_vr_value): New function.
1628 (class vrp_prop): Add vr_values data member. Add various member
1629 functions as well as member functions that delegate to vr_values.
1630 (check_array_ref): Make a member function within vrp_prop class.
1631 (search_for_addr_array, vrp_initialize): Likewise.
1632 (vrp_finalize): Likewise. Revamp to avoid direct access to
1633 vr_value, values_propagated, etc.
1634 (check_array_bounds): Extract vrp_prop class instance pointer from
1635 walk info structure. Use it to call member functions.
1636 (check_all_array_refs): Make a member function within vrp_prop class.
1637 Smuggle class instance pointer via walk info structure.
1638 (x_vr_values): New local static.
1639 (vrp_valueize): Use x_vr_values to get class instance.
1640 (vr_valueize_1): Likewise.
1641 (class vrp_folder): Add vr_values data member. Add various member
1642 functions as well as member functions that delegate to vr_values.
1643 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1644 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1645 class instance from vr_values. Use it to call member functions.
1646 (vrp_dom_walker): Add vr_values data member.
1647 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1648 instance via x_vr_values.
1649 (identify_jump_threads): Accept vr_values as argument. Store
1650 it into the walker structure.
1651 (evrp_dom_walker): Add vr_values class data member. Add various
1653 (evrp_dom_walker::try_find_new_range): Use vr_values data
1654 member to access the memory allocator.
1655 (evrp_dom_walker::before_dom_children): Store vr_values class
1656 instance into the vrp_folder class.
1657 (evrp_dom_walker::push_value_range): Rework to avoid direct
1658 access to num_vr_values and vr_value.
1659 (evrp_dom_walker::pop_value_range): Likewise.
1660 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1661 Use vr_values to get to dump_all_value_ranges member function.
1662 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1663 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1664 Pass vr_values class instance down to identify_jump_threads.
1665 Remove call to vrp_free_lattice.
1666 (debug_all_value_ranges): Remove.
1668 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1669 (vrp_folder): Likewise.
1671 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1672 Get it from the existing bitmap instead.
1673 (vrp_intersect_ranges_1): Likewise.
1675 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1677 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1679 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1680 set bit_not_p to it.
1681 (imm_store_chain_info::coalesce_immediate_stores): Break group
1682 if bit_not_p is different.
1683 (count_multiple_uses, split_group,
1684 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1685 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1686 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1687 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1688 to store_immediate_info ctor.
1690 2017-11-09 Jim Wilson <jimw@sifive.com>
1692 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1693 (scan_prog_file): Likewise.
1695 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1697 * bb-reorder.c (max_entry_frequency): Remove.
1698 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1699 connect_traces, push_to_next_round_p): Remove prototypes.
1700 (find_traces_1_round): Use counts only.
1701 (push_to_next_round_p): Likewise.
1702 (find_traces): Likewise.
1703 (rotate_loop): Likewise.
1704 (find_traces_1_round): Likewise.
1705 (connect_traces): Likewise.
1706 (edge_order): Likewise.
1708 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1710 * config/arm/arm.c (output_return_instruction): Add comments to
1711 indicate requirement for cmse_nonsecure_entry return to account
1712 for the size of clearing instruction output here.
1713 (thumb_exit): Likewise.
1714 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1715 return in hardfloat mode.
1717 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1719 * config/rs6000/rs6000.c (machine_function): Add a bool,
1720 "toc_is_wrapped_separately".
1721 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1722 if it wasn't explicitly set or unset, we are optimizing for speed, and
1723 doing separate shrink-wrapping.
1724 (rs6000_get_separate_components): Enable the TOC component if
1725 saving the TOC register in the prologue.
1726 (rs6000_components_for_bb): Handle the TOC component.
1727 (rs6000_emit_prologue_components): Store the TOC register where needed.
1728 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1729 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1731 2017-11-09 Martin Jambor <mjambor@suse.cz>
1733 * ipa-param-manipulation.c: New file.
1734 * ipa-param-manipulation.h: Likewise.
1735 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1736 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1737 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1738 (ipa_parm_adjustment): Likewise.
1739 (ipa_parm_adjustment_vec): Likewise.
1740 (ipa_get_vector_of_formal_parms): Moved declaration to
1741 ipa-param-manipulation.h.
1742 (ipa_get_vector_of_formal_parm_types): Likewise.
1743 (ipa_modify_formal_parameters): Likewise.
1744 (ipa_modify_call_arguments): Likewise.
1745 (ipa_combine_adjustments): Likewise.
1746 (ipa_dump_param_adjustments): Likewise.
1747 (ipa_modify_expr): Likewise.
1748 (ipa_get_adjustment_candidate): Likewise.
1749 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1750 ipa-param-manipulation.c.
1751 (ipa_get_vector_of_formal_parm_types): Likewise.
1752 (ipa_modify_formal_parameters): Likewise.
1753 (ipa_modify_call_arguments): Likewise.
1754 (ipa_modify_expr): Likewise.
1755 (get_ssa_base_param): Likewise.
1756 (ipa_get_adjustment_candidate): Likewise.
1757 (index_in_adjustments_multiple_times_p): Likewise.
1758 (ipa_combine_adjustments): Likewise.
1759 (ipa_dump_param_adjustments): Likewise.
1760 * tree-sra.c: Also include ipa-param-manipulation.h
1761 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1764 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1765 Alan Hayward <alan.hayward@arm.com>
1766 David Sherwood <david.sherwood@arm.com>
1768 * doc/sourcebuild.texi (vect_masked_store): Document.
1770 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1771 Alan Hayward <alan.hayward@arm.com>
1772 David Sherwood <david.sherwood@arm.com>
1774 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1776 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1777 Alan Hayward <alan.hayward@arm.com>
1778 David Sherwood <david.sherwood@arm.com>
1780 * doc/sourcebuild.texi (vect_variable_length): Document.
1782 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1783 Alan Hayward <alan.hayward@arm.com>
1784 David Sherwood <david.sherwood@arm.com>
1786 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1788 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1794 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1795 Alan Hayward <alan.hayward@arm.com>
1796 David Sherwood <david.sherwood@arm.com>
1798 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1799 previously undocumented selectors.
1800 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1802 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1804 * doc/rtl.texi (const_vector): Say that elements can be
1805 const_wide_ints too.
1806 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1807 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1808 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1809 * optabs.c (expand_vector_broadcast): Likewise.
1811 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1812 Alan Hayward <alan.hayward@arm.com>
1813 David Sherwood <david.sherwood@arm.com>
1815 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1816 scaled index even if the unscaled address was invalid.
1817 Don't increase the complexity of using a scale in that case.
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/rtl.texi: Rewrite the subreg rules so that they partition
1824 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1825 UNITS_PER_WORD bytes.
1826 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1827 based on REGMODE_NATURAL_SIZE of the inner mode.
1828 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1829 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1830 * expmed.c (lowpart_bit_field_p): Divide the value up into
1831 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1832 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1833 whether something is likely to occupy more than one register.
1835 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1838 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1839 profile_count::adjust_for_ipa_scaling.
1840 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1841 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1843 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1845 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1847 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1848 (split_group): Add total_orig and total_new arguments, estimate the
1849 number of statements related to the store group without store merging
1850 and with store merging.
1851 (imm_store_chain_info::output_merged_store): Adjust split_group
1852 callers, punt if estimated number of statements with store merging
1853 is not smaller than estimated number of statements without it.
1855 (handled_load): Remove has_single_use checks.
1856 (pass_store_merging::process_store): Likewise.
1858 2017-11-09 Richard Biener <rguenther@suse.de>
1860 PR tree-optimization/82902
1861 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1863 2017-11-09 Martin Liska <mliska@suse.cz>
1866 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1869 2017-11-09 Martin Liska <mliska@suse.cz>
1871 PR tree-optimization/82669
1872 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1874 2017-11-09 Martin Liska <mliska@suse.cz>
1876 PR gcov-profile/48463
1877 * coverage.c (coverage_begin_function): Output also end locus
1878 of a function and information whether the function is
1880 * gcov-dump.c (tag_function): Parse and print the information.
1881 * gcov.c (INCLUDE_MAP): Add include.
1882 (INCLUDE_SET): Likewise.
1883 (struct line_info): Move earlier in the source file because
1884 of vector<line_info> in function_info structure.
1885 (line_info::line_info): Likewise.
1886 (line_info::has_block): Likewise.
1887 (struct source_info): Add new member index.
1888 (source_info::get_functions_at_location): New function.
1889 (function_info::group_line_p): New function.
1890 (output_intermediate_line): New function.
1891 (output_intermediate_file): Use the mentioned function.
1892 (struct function_start): New.
1893 (struct function_start_pair_hash): Likewise.
1894 (process_file): Add code that identifies group functions.
1895 Assign lines either to global or function scope.
1896 (generate_results): Skip artificial functions.
1897 (find_source): Assign index for each source file.
1898 (read_graph_file): Read new flag artificial and end_line.
1899 (add_line_counts): Assign it either to global of function scope.
1900 (accumulate_line_counts): Isolate core of the function to
1901 accumulate_line_info and call it for both function and global
1903 (accumulate_line_info): New function.
1904 (output_line_beginning): Fix GNU coding style.
1905 (print_source_line): New function.
1906 (output_line_details): Likewise.
1907 (output_function_details): Likewise.
1908 (output_lines): Iterate both source (global) scope and function
1910 (struct function_line_start_cmp): New class.
1911 * doc/gcov.texi: Reflect changes in documentation.
1913 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1916 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1917 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1918 and similarly for not instead of neg.
1920 2017-11-08 Andi Kleen <ak@linux.intel.com>
1922 * config/i386/i386.opt: Add -mforce-indirect-call.
1923 * config/i386/predicates.md: Check for flag_force_indirect_call.
1924 * doc/invoke.texi: Document -mforce-indirect-call
1926 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1928 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1930 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1931 config/riscv/riscv.c (predict.h): New include.
1932 (riscv_slow_unaligned_access_p): No longer static.
1933 (riscv_block_move_straight): Add require.
1934 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1936 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1939 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1940 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1941 nonimmediate_operand predicate for operand 1 instead of
1944 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1946 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1948 * config/aarch64/constraints.md (Uml): New constraint.
1949 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1952 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1954 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1955 of two vec_duplicates into a vec_concat.
1957 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1959 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1960 Simplify vec_merge of vec_duplicate and vec_concat.
1961 * config/aarch64/constraints.md (Utq): New constraint.
1962 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1965 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1967 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1968 Simplify vec_merge of vec_duplicate and const_vector.
1969 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1971 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1972 mode iterator. Update predicate on operand 1 to
1973 handle non-const_vec constants. Delete constraints.
1974 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1976 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1978 PR tree-optimization/78821
1979 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1981 (store_operand_info::store_operand_info): Initialize it to false.
1982 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1983 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1984 store in the group, and if chain_info is non-NULL, to ignore altogether
1986 (compatible_load_p): Fail if bit_not_p does not match.
1987 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1988 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1989 (pass_store_merging::process_store): Adjust
1990 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1991 call terminate_all_aliasing_chains newly when adding a store into
1992 a chain with non-NULL chain_info.
1994 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1996 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1998 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2000 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2002 (aarch64_layout_frame): Initialise emit_frame_chain.
2003 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2004 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2006 2017-11-08 Martin Liska <mliska@suse.cz>
2008 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2009 of gimple_call_internal_p.
2011 2017-11-07 Tom de Vries <tom@codesourcery.com>
2013 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2015 2017-11-07 Tom de Vries <tom@codesourcery.com>
2017 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2020 2017-11-08 Martin Liska <mliska@suse.cz>
2023 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2025 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2027 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2029 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2032 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2033 (*cmp<mode>_ccz_1): New insn with $k alternative.
2036 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2037 RTX_COMM_COMPARE as commutative as well.
2038 (ix86_binary_operator_ok): Formatting fix.
2039 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2040 *<code><mode>3<mask_name><round_saeonly_name>,
2041 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2042 *<s>mul<mode>3_highpart<mask_name>,
2043 *vec_widen_umult_even_v16si<mask_name>,
2044 *vec_widen_umult_even_v8si<mask_name>,
2045 *vec_widen_umult_even_v4si<mask_name>,
2046 *vec_widen_smult_even_v16si<mask_name>,
2047 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2048 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2049 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2050 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2051 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2052 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2053 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2054 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2055 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2056 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2057 ix86_binary_operator_ok. Formatting fixes.
2058 (*<plusminus_insn><mode>3<mask_name><round_name>,
2059 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2062 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2064 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2065 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2066 eq and ne if TARGET_ISEL.
2067 (cmp): New code_iterator.
2068 (UNS, UNSU_, UNSIK): New code_attrs.
2069 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2070 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2072 ("eq<mode>3"): ... this.
2073 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2075 ("ne<mode>3"): ... this.
2077 2017-11-07 Julia Koval <julia.koval@intel.com>
2080 * common/config/i386/i386-common.c
2081 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2082 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2083 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2084 * config/i386/i386.opt: Ditto.
2085 * config/i386/i386.c (ix86_target_string): Ditto.
2086 (ix86_option_override_internal): Ditto.
2087 (ix86_init_mpx_builtins): Move MPX to args2.
2088 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2089 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2090 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2091 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2092 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2093 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2094 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2096 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2099 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2100 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2101 (zero-extendsidi peephole2): Remove peephole.
2103 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2106 * stor-layout.c: Include attribs.h.
2107 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2108 explicit lookup of "aligned" attribute.
2110 2017-11-07 Andrew Waterman <andrew@sifive.com>
2112 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2113 (riscv_expand_block_move): Likewise.
2114 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2116 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2117 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2118 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2119 (riscv_adjust_block_mem): Likewise.
2120 (riscv_block_move_loop): Likewise.
2121 (riscv_expand_block_move): Likewise.
2122 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2124 2017-11-07 Michael Clark <michaeljclark@mac.com>
2126 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2127 (MUSL_DYNAMIC_LINKER): Likewise.
2129 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2131 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2132 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2135 2017-11-07 Richard Biener <rguenther@suse.de>
2137 * match.pd: Fix build.
2139 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2140 Jackson Woodruff <jackson.woodruff@arm.com>
2142 PR tree-optimization/71026
2143 * match.pd: Canonicalize negate in division.
2145 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2148 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2150 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2152 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2153 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2155 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2157 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2158 non-scalar integral types.
2159 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2160 (-(A-B), -(~A)): New transformations.
2162 2017-11-07 Tom de Vries <tom@codesourcery.com>
2164 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2165 semicolon after "do {} while (0)".
2166 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2167 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2168 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2169 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2170 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2172 2017-11-07 Tom de Vries <tom@codesourcery.com>
2174 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2175 after "do {} while (0)".
2176 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2177 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2178 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2179 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2180 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2182 2017-11-07 Tom de Vries <tom@codesourcery.com>
2185 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2187 (arm_rtx_costs_internal): Add missing semicolon after
2188 HANDLE_NARROW_SHIFT_ARITH call.
2190 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2192 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2193 disable isel if it was not set explicitly.
2195 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2197 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2199 (add_type_attribute) likewise.
2201 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2203 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2204 of struct ix86_frame.
2205 (ix86_initial_elimination_offset): Likewise.
2206 (ix86_expand_split_stack_prologue): Likewise.
2208 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2210 * tree-vrp.h (enum value_range_type): Update stale comment.
2212 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2213 Alan Hayward <alan.hayward@arm.com>
2214 David Sherwood <david.sherwood@arm.com>
2216 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2217 (aarch64_expand_vec_perm_const): Take the number of units too.
2218 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2219 (aarch64_expand_vec_perm_const): Likewise.
2220 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2221 (vec_perm<mode>): Update accordingly.
2223 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2224 Alan Hayward <alan.hayward@arm.com>
2225 David Sherwood <david.sherwood@arm.com>
2227 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2228 Take the number of units too.
2229 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2230 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2231 but check for a vector mode before rather than after the call.
2232 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2233 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2234 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2235 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2236 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2237 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2238 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2239 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2240 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2241 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2242 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2243 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2244 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2245 (aarch64_sqdmull2_n<mode>): Update accordingly.
2247 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2248 Alan Hayward <alan.hayward@arm.com>
2249 David Sherwood <david.sherwood@arm.com>
2251 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2252 the number of units too.
2253 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2254 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2255 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2256 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2257 (vec_store_lanesxi<mode>): Update accordingly.
2259 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2260 Alan Hayward <alan.hayward@arm.com>
2261 David Sherwood <david.sherwood@arm.com>
2263 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2264 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2265 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2266 of units rather than the mode.
2267 * config/aarch64/iterators.md (nunits): New mode attribute.
2268 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2269 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2270 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2271 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2272 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2273 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2274 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2275 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2276 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2277 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2278 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2279 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2280 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2281 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2282 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2283 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2284 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2285 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2286 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2287 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2288 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2289 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2290 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2291 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2292 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2293 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2294 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2295 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2296 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2297 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2298 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2299 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2300 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2301 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2302 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2303 (aarch64_simd_vec_setv2di): Likewise.
2305 2017-11-06 Carl Love <cel@us.ibm.com>
2307 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2309 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2310 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2312 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2314 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2315 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2316 Add power 8 macro expansions.
2317 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2318 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2319 power 8 instructions. (VSX_XXBR): Add iterator.
2321 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2323 * config/arm/arm.md (predicable_short_it): Change default to "no",
2324 improve documentation, remove uses that are identical to the default.
2325 (enabled_for_depr_it): Rename to enabled_for_short_it.
2326 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2327 Remove default uses.
2328 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2329 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2330 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2331 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2333 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2336 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2337 float128 helper macros, which are no longer used after deleting
2338 the old 'q' built-in functions, and moving the round to odd
2339 built-in functions to being special built-in functions.
2340 (BU_FLOAT128_2): Likewise.
2341 (BU_FLOAT128_1_HW): Likewise.
2342 (BU_FLOAT128_2_HW): Likewise.
2343 (BU_FLOAT128_3_HW): Likewise.
2344 (FABSQ): Delete old 'q' built-in functions.
2345 (COPYSIGNQ): Likewise.
2346 (SQRTF128_ODD): Move round to odd built-in functions to be
2347 special built-in functions, so that we can handle
2348 -mabi=ieeelongdouble.
2349 (TRUNCF128_ODD): Likewise.
2350 (ADDF128_ODD): Likewise.
2351 (SUBF128_ODD): Likewise.
2352 (MULF128_ODD): Likewise.
2353 (DIVF128_ODD): Likewise.
2354 (FMAF128_ODD): Likewise.
2355 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2356 built-in names to 'f128'.
2357 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2358 old 'q' built-in functions, as the machine independent code for
2359 'f128' built-in functions handles this.
2360 (rs6000_expand_builtin): Add expansion for float128 round to odd
2361 functions, keying off on -mabi=ieeelongdouble of whether to use
2362 the KFmode or TFmode variant.
2363 (rs6000_init_builtins): Initialize the _Float128 round to odd
2365 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2366 _Float128 'q' built-in functions are now mapped into the new
2367 'f128' built-in functions.
2369 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2371 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2372 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2374 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2377 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2378 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2379 * toplev.c: Include "ipa-fnsummary.h".
2380 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2382 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2384 PR tree-optimization/82838
2385 * gimple-ssa-store-merging.c
2386 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2387 on a separate gimple_seq which is then appended to seq.
2389 2017-11-06 Jeff Law <law@redhat.com>
2392 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2393 (get_probe_interval): New functions.
2394 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2395 (ix86_adjust_stack_and_probe): Likewise.
2396 (output_adjust_stack_and_probe): Likewise.
2397 (ix86_emit_probe_stack_range): Likewise.
2398 (ix86_expand_prologue): Likewise.
2400 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2402 PR tree-optimization/82816
2403 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2404 if the modes of the two types are the same.
2405 (convert_plusminus_to_widen): Likewise.
2407 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2409 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2411 (usadv16qi): New define_expand.
2412 (usadv8hi): New define_expand.
2414 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2417 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2419 (inline_transform): Likewise.
2420 * predict.c (counts_to_freqs): Remove useless conditional.
2421 * profile-count.h (profile_count::apply_scale): Move sanity check.
2422 * tree-inline.c (copy_bb): Always scale.
2423 (copy_cfg_body): Likewise.
2425 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2428 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2430 (*cmp_ite0): Add enabled_for_depr_it attribute.
2431 (*cmp_ite1): Likewise.
2433 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2435 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2438 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2440 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2441 return the value by pointer.
2442 (extract_range_from_multiplicative_op_1): Update accordingly.
2443 Return as soon as an operation fails.
2445 2017-11-05 Tom de Vries <tom@codesourcery.com>
2448 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2449 (DEF_SANITIZER_BUILTIN): ... here.
2450 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2451 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2453 2017-11-05 Tom de Vries <tom@codesourcery.com>
2456 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2458 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2459 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2460 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2462 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2463 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2464 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2466 2017-11-05 Tom de Vries <tom@codesourcery.com>
2469 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2472 2017-11-04 Michael Clark <michaeljclark@mac.com>
2474 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2475 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2477 (*addsi3_extended): Likewise.
2478 (*addsi3_extended2): Likewise.
2479 (<optab>si3): Likewise.
2480 (<optab>di3): Likewise.
2481 (<optab><mode>3): Likewise.
2482 (<*optabe>si3_internal): Likewise.
2483 (zero_extendqi<SUPERQI:mode>2): Likewise.
2484 (*add<mode>hi3): Likewise.
2485 (*xor<mode>hi3): Likewise.
2486 (<optab>di3): Likewise.
2487 (*<optab>si3_extend): Likewise.
2488 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2489 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2490 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2492 2017-11-04 Andrew Waterman <andrew@sifive.com>
2494 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2495 TARGET_STRICT_ALIGN based upon -mtune argument.
2497 2017-11-04 Andrew Waterman <andrew@sifive.com>
2499 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2501 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2503 config/i386/i386.c (choose_basereg): Use optional scratch
2504 register and add assertion.
2505 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2506 needed, and don't allocate stack.
2507 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2508 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2509 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2511 2017-11-03 Jeff Law <law@redhat.com>
2513 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2514 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2515 to probe at the start of a noreturn function.
2517 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2519 PR tree-optimization/78821
2520 * gimple-ssa-store-merging.c: Update the file comment.
2521 (MAX_STORE_ALIAS_CHECKS): Define.
2522 (struct store_operand_info): New type.
2523 (store_operand_info::store_operand_info): New constructor.
2524 (struct store_immediate_info): Add rhs_code and ops data members.
2525 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2526 and op1r arguments to the ctor, initialize corresponding data members.
2527 (struct merged_store_group): Add load_align_base and load_align
2529 (merged_store_group::merged_store_group): Initialize them.
2530 (merged_store_group::do_merge): Update them.
2531 (merged_store_group::apply_stores): Pick the constant for
2532 encode_tree_to_bitpos from one of the two operands, or skip
2533 encode_tree_to_bitpos if neither operand is a constant.
2534 (class pass_store_merging): Add process_store method decl. Remove
2535 bool argument from terminate_all_aliasing_chains method decl.
2536 (pass_store_merging::terminate_all_aliasing_chains): Remove
2537 var_offset_p argument and corresponding handling.
2538 (stmts_may_clobber_ref_p): New function.
2539 (compatible_load_p): New function.
2540 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2541 if there is overlap and rhs_code is not INTEGER_CST. For
2542 non-overlapping stores terminate group if rhs is not mergeable.
2543 (get_alias_type_for_stmts): Change first argument from
2544 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2545 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2546 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2548 (get_location_for_stmts): Change first argument from
2549 auto_vec<gimple *> & to vec<gimple *> &.
2550 (struct split_store): Remove orig_stmts data member, add orig_stores.
2551 (split_store::split_store): Create orig_stores rather than orig_stmts.
2552 (find_constituent_stmts): Renamed to ...
2553 (find_constituent_stores): ... this. Change second argument from
2554 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2555 to info structures rather than the statements.
2556 (split_group): Rename ALLOW_UNALIGNED argument to
2557 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2558 it. Adjust find_constituent_stores caller.
2559 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2560 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2561 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2562 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2563 (mem_valid_for_store_merging): New function.
2564 (handled_load): New function.
2565 (pass_store_merging::process_store): New method.
2566 (pass_store_merging::execute): Use process_store method. Adjust
2567 terminate_all_aliasing_chains caller.
2569 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2571 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2572 Return true for more constants, symbols and label references.
2573 (aarch64_valid_floating_const): Remove unused function.
2575 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2578 * config/aarch64/aarch64.c (aarch64_layout_frame):
2579 Undo forcing of LR at bottom of frame.
2581 2017-11-03 Jeff Law <law@redhat.com>
2584 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2585 for int_registers_saved.
2587 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2588 extracted from tree-ssa-dom.c.
2589 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2590 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2591 (record_equivalences_from_incoming_edge): Add additional argument
2592 to single_pred_edge_ignoring_loop_edges call.
2593 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2594 (uncprop_dom_walker::before_dom_children): Add additional argument
2595 to single_pred_edge_ignoring_loop_edges call.
2596 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2597 single_pred_edge_ignoring_loop_edges rather than open coding.
2598 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2600 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2602 * match.pd (-(-A)): Rewrite.
2604 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2606 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2607 (rs6000_emit_int_cmove): New declaration.
2608 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2609 (rs6000_emit_sISEL): Delete.
2610 (rs6000_emit_int_cmove): Make non-static.
2611 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2612 instead of rs6000_emit_sISEL.
2614 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2616 * asan.c (create_cond_insert_point): Maintain profile.
2617 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2619 * basic-block.h (struct basic_block_def): Remove frequency.
2620 (EDGE_FREQUENCY): Use to_frequency
2621 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2623 (find_traces): Update to use to_frequency.
2624 (find_traces_1_round): Likewise; use only IPA counts.
2625 (bb_to_key): Likewise.
2626 (connect_traces): Use IPA counts only.
2627 (copy_bb_p): Update to use to_frequency.
2628 (fix_up_crossing_landing_pad): Likewise.
2629 (sanitize_hot_paths): Likewise.
2630 * bt-load.c (basic_block_freq): Likewise.
2631 * cfg.c (init_flow): Set count_max to uninitialized.
2632 (check_bb_profile): Remove frequencies; check counts.
2633 (dump_bb_info): Do not dump frequencies.
2634 (update_bb_profile_for_threading): Update counts only.
2635 (scale_bbs_frequencies_int): Likewise.
2636 (MAX_SAFE_MULTIPLIER): Remove.
2637 (scale_bbs_frequencies_gcov_type): Update counts only.
2638 (scale_bbs_frequencies_profile_count): Update counts only.
2639 (scale_bbs_frequencies): Update counts only.
2640 * cfg.h (struct control_flow_graph): Add count-max.
2641 (update_bb_profile_for_threading): Update prototype.
2642 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2643 (find_many_sub_basic_blocks): Likewise.
2644 * cfgcleanup.c (try_forward_edges): Likewise.
2645 (try_crossjump_to_edge): Likewise.
2646 * cfgexpand.c (expand_gimple_cond): Likewise.
2647 (expand_gimple_tailcall): Likewise.
2648 (construct_init_block): Likewise.
2649 (construct_exit_block): Likewise.
2650 * cfghooks.c (verify_flow_info): Check consistency of counts.
2651 (dump_bb_for_graph): Do not dump frequencies.
2652 (split_block_1): Do not update frequencies.
2653 (split_edge): Do not update frequencies.
2654 (make_forwarder_block): Do not update frequencies.
2655 (duplicate_block): Do not update frequencies.
2656 (account_profile_record): Do not update frequencies.
2657 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2658 for global heuristics.
2659 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2660 (expected_loop_iterations_unbounded): Use counts only.
2661 * cfgloopmanip.c (scale_loop_profile): Simplify.
2662 (create_empty_loop_on_edge): Simplify
2664 (duplicate_loop_to_header_edge): Simplify
2665 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2666 (update_br_prob_note): Take care of removing note when profile
2668 (relink_block_chain): Do not dump frequency.
2669 (rtl_account_profile_record): Use to_frequency.
2670 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2671 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2672 (cgraph_update_edges_for_call_stmt_node): Likewise.
2673 (cgraph_edge::verify_count_and_frequency): Update.
2674 (cgraph_node::verify_node): Temporarily disable frequency verification.
2675 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2676 to_cgraph_frequency.
2677 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2678 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2680 (cgraph_node::expand_thunk): Update profile.
2681 * except.c (dw2_build_landing_pads): Do not update frequency.
2682 * final.c (compute_alignments): Use to_frequency.
2683 (dump_basic_block_info): Do not dump frequency.
2684 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2685 (dump_gimple_bb_header): Do not dump frequency.
2686 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2688 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2689 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2690 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2691 (init_before_recovery): Do not update frequency.
2692 (sched_create_recovery_edges): Do not update frequency.
2693 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2694 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2695 (ipa_cp_c_finalize): Set max_count to uninitialized.
2696 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2697 (param_change_prob): Use counts.
2698 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2700 * ipa-split.c (consider_split): Use to_frequency.
2701 (split_function): Use to_frequency.
2702 * ira-build.c (loop_compare_func): Likewise.
2703 (mark_loops_for_removal): Likewise.
2704 (mark_all_loops_for_removal): Likewise.
2705 * loop-doloop.c (doloop_modify): Do not update frequency.
2706 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2708 * lto-streamer-in.c (input_function): Update count_max.
2709 * omp-expand.c (expand_omp_taskreg): Update count_max.
2710 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2711 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2712 (maybe_hot_count_p): Use ipa counts only.
2713 (maybe_hot_bb_p): Simplify.
2714 (maybe_hot_edge_p): Simplify.
2715 (probably_never_executed): Do not take frequency argument.
2716 (probably_never_executed_bb_p): Do not pass frequency.
2717 (probably_never_executed_edge_p): Likewise.
2718 (combine_predictions_for_bb): Check that profile is nonzero.
2719 (propagate_freq): Do not set frequency.
2720 (drop_profile): Simplify.
2721 (counts_to_freqs): Simplify.
2722 (expensive_function_p): Use to_frequency.
2723 (propagate_unlikely_bbs_forward): Simplify.
2724 (determine_unlikely_bbs): Simplify.
2725 (estimate_bb_frequencies): Add hack to silence graphite issues.
2726 (compute_function_frequency): Use ipa counts.
2727 (pass_profile::execute): Update.
2728 (rebuild_frequencies): Use counts only.
2729 (force_edge_cold): Use counts only.
2730 * profile-count.c (profile_count::dump): Dump new count types.
2731 (profile_count::differs_from_p): Check compatiblity.
2732 (profile_count::to_frequency): New function.
2733 (profile_count::to_cgraph_frequency): New function.
2734 * profile-count.h (struct function): Declare.
2735 (enum profile_quality): Add profile_guessed_local and
2736 profile_guessed_global0.
2737 (class profile_proability): Decrease number of bits to 29;
2738 update from_reg_br_prob_note and to_reg_br_prob_note.
2739 (class profile_count: Update comment; decrease number of bits
2740 to 61. Check compatibility.
2741 (profile_count::compatible_p): New private member function.
2742 (profile_count::ipa_p): New member function.
2743 (profile_count::operator<): Handle global zero correctly.
2744 (profile_count::operator>): Handle global zero correctly.
2745 (profile_count::operator<=): Handle global zero correctly.
2746 (profile_count::operator>=): Handle global zero correctly.
2747 (profile_count::nonzero_p): New member function.
2748 (profile_count::force_nonzero): New member function.
2749 (profile_count::max): New member function.
2750 (profile_count::apply_scale): Handle IPA scalling.
2751 (profile_count::guessed_local): New member function.
2752 (profile_count::global0): New member function.
2753 (profile_count::ipa): New member function.
2754 (profile_count::to_frequency): Declare.
2755 (profile_count::to_cgraph_frequency): Declare.
2756 * profile.c (OVERLAP_BASE): Delete.
2757 (compute_frequency_overlap): Delete.
2758 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2759 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2760 * sched-ebb.c (rank): Use counts only.
2761 * shrink-wrap.c (handle_simple_exit): Use counts only.
2762 (try_shrink_wrapping): Use counts only.
2763 (place_prologue_for_one_component): Use counts only.
2764 * tracer.c (find_best_predecessor): Use to_frequency.
2765 (find_trace): Use to_frequency.
2766 (tail_duplicate): Use to_frequency.
2767 * trans-mem.c (expand_transaction): Do not update frequency.
2768 * tree-call-cdce.c: Do not update frequency.
2769 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2770 (gimple_merge_blocks): Likewise.
2771 (gimple_split_edge): Likewise.
2772 (gimple_duplicate_sese_region): Likewise.
2773 (gimple_duplicate_sese_tail): Likewise.
2774 (move_sese_region_to_fn): Likewise.
2775 (gimple_account_profile_record): Likewise.
2776 (insert_cond_bb): Likewise.
2777 * tree-complex.c (expand_complex_div_wide): Likewise.
2778 * tree-eh.c (lower_resx): Update profile.
2779 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2781 (initialize_cfun): Do not initialize frequencies
2782 (freqs_to_counts): Delete.
2783 (copy_cfg_body): Ignore count parameter.
2784 (copy_body): Update.
2785 (expand_call_inline): Update count_max.
2786 (optimize_inline_calls): Update count_max.
2787 (tree_function_versioning): Update count_max.
2788 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2789 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2791 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2792 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2793 (try_peel_loop): Likewise.
2794 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2796 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2797 (tree_transform_and_unroll_loop): Do not use frequencies
2798 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2799 Use reliable prediction only.
2800 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2801 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2802 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2803 probability scaling.
2804 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2805 not update frequency
2806 (any_remaining_duplicated_blocks): Likewise.
2807 (update_profile): Likewise.
2808 (estimated_freqs_path): Delete.
2809 (freqs_to_counts_path): Delete.
2810 (clear_counts_path): Delete.
2811 (ssa_fix_duplicate_block_edges): Likewise.
2812 (duplicate_thread_path): Likewise.
2813 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2814 * tree-tailcall.c (decrease_profile): Do not update frequency.
2815 (eliminate_tail_call): Likewise.
2816 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2817 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2818 (optimize_mask_stores): Likewise.
2819 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2820 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2821 (ubsan_expand_ptr_ifn): Update profile.
2822 * value-prof.c (gimple_ic): Simplify.
2823 * value-prof.h (gimple_ic): Update prototype.
2824 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2825 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2827 (want_inline_self_recursive_call_p): Likewise.
2828 (resolve_noninline_speculation): Only cummulate defined counts.
2829 (inline_small_functions): Use nonzero_p.
2830 (ipa_inline): Do not access freed node.
2832 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2834 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2835 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2837 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2839 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2840 non-legitimate address.
2842 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2844 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2845 (*lt0_<mode>di, *lt0_<mode>si): New.
2847 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2849 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2850 TARGET_PAIRED_FLOAT.
2852 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2853 Jim Wilson <jim.wilson@linaro.org>
2855 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2856 * config/aarch64/aarch64-tune.md: Regenerated.
2857 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2858 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2860 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2862 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2863 (arc_expand_prologue): Restore blink for millicode.
2864 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2866 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2869 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2870 gen_vec_duplicate after forcing the scalar into a register.
2872 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2874 * combine (try_combine): Print the insns input to try_combine to the
2877 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2880 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2881 Remove second argument from aarch64_process_target_attr call.
2882 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2884 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2886 (aarch64_handle_attr_arch): Remove second argument.
2887 (aarch64_handle_attr_cpu): Ditto.
2888 (aarch64_handle_attr_tune): Ditto.
2889 (aarch64_handle_attr_isa_flags): Ditto.
2890 (aarch64_process_one_target_attr): Ditto.
2891 (aarch64_process_target_attr): Ditto.
2892 (aarch64_option_valid_attribute_p): Remove second argument.
2893 on aarch64_process_target_attr call.
2895 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2897 * diagnostic.c: Include "selftest-diagnostic.h".
2898 (selftest::assert_location_text): New function.
2899 (selftest::test_diagnostic_get_location_text): New function.
2900 (selftest::diagnostic_c_tests): Call it.
2902 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2904 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2905 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2906 (class selftest::test_diagnostic_context): Move to...
2907 * selftest-diagnostic.c: New file.
2908 * selftest-diagnostic.h: New file.
2910 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2912 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2913 offset range for FT32B.
2914 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2915 * config/ft32/ft32.md: Add TARGET_NOPM.
2916 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2917 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2919 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2921 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2923 2017-11-02 Jeff Law <law@redhat.com>
2925 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2926 virtual keyword on FINAL OVERRIDE members.
2928 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2929 virtuals together. Add virtual destructor.
2930 (substitute_and_fold_engine): Similarly.
2932 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2934 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2936 2017-11-02 Richard Biener <rguenther@suse.de>
2938 PR tree-optimization/82795
2939 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2941 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2943 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2945 (gcc_GAS_CHECK_FEATURE): Remove.
2946 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2947 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2948 * configure: Regenerate.
2950 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2952 * config/arc/arc.c (hwloop_optimize): Account for empty
2955 2017-11-02 Richard Biener <rguenther@suse.de>
2958 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2959 Truncate ARRAY_REF index and element size.
2961 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2963 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2965 2017-11-01 Jeff Law <law@redhat.com>
2967 * tree-ssa-ccp.c (ccp_folder): New class derived from
2968 substitute_and_fold_engine.
2969 (ccp_folder::get_value): New member function.
2970 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2971 (ccp_fold_stmt): Remove prototype.
2972 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2973 * tree-ssa-copy.c (copy_folder): New class derived from
2974 substitute_and_fold_engine.
2975 (copy_folder::get_value): Renamed from get_value.
2976 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2977 * tree-vrp.c (vrp_folder): New class derived from
2978 substitute_and_fold_engine.
2979 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2980 (vrp_folder::get_value): New member function.
2981 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2982 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2983 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2984 provide a class interface to folder/substitute routines.
2985 (ssa_prop_fold_stmt_fn): Remove typedef.
2986 (ssa_prop_get_value_fn): Likewise.
2987 (subsitute_and_fold): Remove prototype.
2988 (replace_uses_in): Likewise.
2989 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2990 Renamed from replace_uses_in. Call the virtual member function
2991 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2992 (substitute_and_fold_dom_walker): Remove initialization of
2993 data member entries for calbacks. Add substitute_and_fold_engine
2994 member and initialize it.
2995 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2996 member functions for get_value, replace_phi_args_in c
2997 replace_uses_in, and fold_stmt calls.
2998 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2999 substitute_and_fold. Remove assert. Update ctor call.
3001 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3002 (ssa_prop_visit_phi_fn): Likewise.
3003 (class ssa_propagation_engine): New class to provide an interface
3005 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3007 (ssa_prop_visit_phi): Likewise.
3008 (ssa_propagation_engine::simulate_stmt): Moved into class.
3009 Call visit_phi/visit_stmt from the class rather than via
3010 file scoped static variables.
3011 (ssa_propagation_engine::simulate_block): Moved into class.
3012 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3013 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3014 set file scoped statics for the visit_stmt/visit_phi callbacks.
3015 * tree-complex.c (complex_propagate): New class derived from
3016 ssa_propagation_engine.
3017 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3018 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3019 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3021 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3022 ssa_propagation_engine.
3023 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3024 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3025 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3026 * tree-ssa-copy.c (copy_prop): New class derived from
3027 ssa_propagation_engine.
3028 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3029 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3030 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3031 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3032 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3033 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3034 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3036 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3038 PR rtl-optimization/82778
3039 PR rtl-optimization/82597
3040 * compare-elim.c (struct comparison): Add in_a_setter field.
3041 (find_comparison_dom_walker::before_dom_children): Remove killed
3042 bitmap and df_simulate_find_defs call, instead walk the defs.
3043 Compute last_setter and initialize in_a_setter. Merge definitions
3044 with first initialization for a few variables.
3045 (try_validate_parallel): Use insn_invalid_p instead of
3046 recog_memoized. Return insn rather than just the pattern.
3047 (try_merge_compare): Fix up comment. Don't uselessly test if
3048 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3050 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3054 Alan Hayward <alan.hayward@arm.com>
3055 David Sherwood <david.sherwood@arm.com>
3057 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3058 aarch64_hard_regno_nregs to get the number of registers
3061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3062 Alan Hayward <alan.hayward@arm.com>
3063 David Sherwood <david.sherwood@arm.com>
3065 * config/aarch64/constraints.md (Upl): Rename to...
3067 * config/aarch64/aarch64.md
3068 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3071 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3072 Alan Hayward <alan.hayward@arm.com>
3073 David Sherwood <david.sherwood@arm.com>
3075 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3076 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3079 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3080 Alan Hayward <alan.hayward@arm.com>
3081 David Sherwood <david.sherwood@arm.com>
3083 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3084 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3085 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3087 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3088 of a single element.
3089 * config/aarch64/iterators.md: Add a comment above the permute
3090 unspecs to say that they are generated directly by
3091 aarch64_expand_vec_perm_const.
3092 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3094 2017-11-01 Nathan Sidwell <nathan@acm.org>
3096 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3098 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3100 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3101 and medany code models, and describe what they do.
3103 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3105 Revert accidental duplicate:
3107 * combine.c (can_change_dest_mode): Reject changes in
3108 REGMODE_NATURAL_SIZE.
3110 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3112 PR rtl-optimization/64682
3113 PR rtl-optimization/69567
3114 PR rtl-optimization/69737
3115 PR rtl-optimization/82683
3116 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3117 register mentioned in the note, drop the note, unless it came from I3,
3118 in which case it should go to I3 again.
3120 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3122 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3123 and return false if not.
3124 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3126 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3128 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3129 range is known to be empty.
3131 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3132 Alan Hayward <alan.hayward@arm.com>
3133 David Sherwood <david.sherwood@arm.com>
3135 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3136 and CONST_VECTOR_NUNITS instead of computing the number of units from
3137 the byte sizes of the vector and element.
3138 (simplify_binary_operation_1): Likewise.
3139 (simplify_const_binary_operation): Likewise.
3140 (simplify_ternary_operation): Likewise.
3142 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3143 Alan Hayward <alan.hayward@arm.com>
3144 David Sherwood <david.sherwood@arm.com>
3146 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3147 (int_mem_offset): ...this new function.
3148 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3149 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3150 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3153 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3154 Alan Hayward <alan.hayward@arm.com>
3155 David Sherwood <david.sherwood@arm.com>
3157 * lower-subreg.c (interesting_mode_p): New function.
3158 (compute_costs, find_decomposable_subregs, decompose_register)
3159 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3160 (resolve_clobber, dump_choices): Use it.
3162 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3163 Alan Hayward <alan.hayward@arm.com>
3164 David Sherwood <david.sherwood@arm.com>
3166 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3168 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3169 Alan Hayward <alan.hayward@arm.com>
3170 David Sherwood <david.sherwood@arm.com>
3172 * alias.c (find_base_value, find_base_term): Only process integer
3173 truncations. Check the precision rather than the size.
3175 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3176 Alan Hayward <alan.hayward@arm.com>
3177 David Sherwood <david.sherwood@arm.com>
3179 * machmode.h (is_narrower_int_mode): New function
3180 * optabs.c (expand_float, expand_fix): Use it.
3181 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3183 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3184 Alan Hayward <alan.hayward@arm.com>
3185 David Sherwood <david.sherwood@arm.com>
3187 * rtl.h (narrower_subreg_mode): New function.
3188 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
3195 (convert_optab_handler): ...here.
3196 (widening_optab_handler): Delete.
3197 (find_widening_optab_handler): Remove permit_non_widening parameter.
3198 (find_widening_optab_handler_and_mode): Likewise. Provide an
3199 override that operates on mode class wrappers.
3200 * optabs-query.c (widening_optab_handler): Delete.
3201 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3202 parameter. Assert that the two modes are the same class and that
3203 the "from" mode is narrower than the "to" mode. Use
3204 convert_optab_handler instead of widening_optab_handler.
3205 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3206 instead of widening_optab_handler.
3207 * expr.c (expand_expr_real_2): Update calls to
3208 find_widening_optab_handler.
3209 * optabs.c (expand_widen_pattern_expr): Likewise.
3210 (expand_binop_directly): Take the insn_code as a parameter.
3211 (expand_binop): Only call find_widening_optab_handler for
3212 conversion optabs; use optab_handler otherwise. Update calls
3213 to find_widening_optab_handler and expand_binop_directly.
3214 Use convert_optab_handler instead of widening_optab_handler.
3215 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3216 find_widening_optab_handler and use scalar_mode rather than
3218 (convert_plusminus_to_widen): Likewise.
3220 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3221 Alan Hayward <alan.hayward@arm.com>
3222 David Sherwood <david.sherwood@arm.com>
3224 * machmode.h (fixed_size_mode): New class.
3225 * rtl.h (get_pool_mode): Return fixed_size_mode.
3226 * gengtype.c (main): Add fixed_size_mode.
3227 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3228 (get_raw_arg_mode): Likewise.
3229 * doc/tm.texi: Regenerate.
3230 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3231 * targhooks.c (default_get_reg_raw_mode): Likewise.
3232 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3233 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3234 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3235 (msp430_get_raw_result_mode): Likewise.
3236 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3237 * dbxout.c (dbxout_parms): Require fixed-size modes.
3238 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3239 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3240 * omp-low.c (lower_oacc_reductions): Likewise.
3241 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3242 (simplify_subreg): Update accordingly.
3243 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3244 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3245 that aren't fixed-size.
3246 (get_pool_mode): Return a fixed_size_mode.
3247 (output_constant_pool_2): Take a fixed_size_mode.
3249 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3250 Alan Hayward <alan.hayward@arm.com>
3251 David Sherwood <david.sherwood@arm.com>
3253 * doc/rtl.texi (vec_series): Document.
3254 (const): Say that the operand can be a vec_series.
3255 * rtl.def (VEC_SERIES): New rtx code.
3256 * rtl.h (const_vec_series_p_1): Declare.
3257 (const_vec_series_p): New function.
3258 * emit-rtl.h (gen_const_vec_series): Declare.
3259 (gen_vec_series): Likewise.
3260 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3261 (gen_vec_series): Likewise.
3262 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3263 * simplify-rtx.c (simplify_unary_operation): Handle negations
3265 (simplify_binary_operation_series): New function.
3266 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3267 (test_vector_ops_series): New function.
3268 (test_vector_ops): Call it.
3269 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3270 gen_const_vec_series.
3271 (altivec_lvsr): Likewise.
3272 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3274 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3275 Alan Hayward <alan.hayward@arm.com>
3276 David Sherwood <david.sherwood@arm.com>
3278 * doc/rtl.texi (const): Update description of address constants.
3279 Say that vector constants are allowed too.
3280 * common.md (E, F): Use CONSTANT_P instead of checking for
3282 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3283 checking for CONST_VECTOR.
3284 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3286 * expr.c (expand_expr_real_2): Check for vector modes instead
3287 of checking for CONST_VECTOR.
3288 * rtl.h (const_vec_p): New function.
3289 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3290 (unwrap_const_vec_duplicate): Handle them here too.
3292 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3293 David Malcolm <dmalcolm@redhat.com>
3294 Alan Hayward <alan.hayward@arm.com>
3295 David Sherwood <david.sherwood@arm.com>
3297 * rtl.h (vec_duplicate_p): New function.
3298 * selftest-rtl.c (assert_rtx_eq_at): New function.
3299 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3300 (assert_rtx_eq_at): Declare.
3301 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3302 * selftest-run-tests.c (selftest::run_tests): Call it.
3303 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3304 (simplify_unary_operation_1): Recursively handle vector duplicates.
3305 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3307 (simplify_subreg): Handle subregs of vector duplicates.
3308 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3309 (selftest::simplify_rtx_c_tests): New functions.
3311 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3312 Alan Hayward <alan.hayward@arm.com>
3313 David Sherwood <david.sherwood@arm.com>
3315 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3316 (gen_vec_duplicate): Likewise.
3317 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3319 (gen_const_vector): ...here.
3320 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3321 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3322 whose elements are all equal.
3323 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3324 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3325 (simplify_relational_operation): Likewise.
3326 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3328 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3329 (aarch64_expand_vector_init): Likewise.
3330 * config/arm/arm.c (neon_vdup_constant): Likewise.
3331 (neon_expand_vector_init): Likewise.
3332 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3333 (arm_block_set_unaligned_vect): Likewise.
3334 (arm_block_set_aligned_vect): Likewise.
3335 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3336 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3337 (expand_vec_perm_even_odd_pack): Likewise.
3338 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3339 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3340 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3341 gen_const_vec_duplicate.
3342 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3343 * config/mips/mips.c (mips_gen_const_int_vector): Use
3344 gen_const_vec_duplicate.
3345 (mips_expand_vector_init): Use CONST0_RTX.
3346 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3347 (define_split): Use gen_const_vec_duplicate.
3348 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3349 (define_split): Use gen_const_vec_duplicate.
3350 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3351 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3352 * config/spu/spu.c (spu_const): Likewise.
3354 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3355 Alan Hayward <alan.hayward@arm.com>
3356 David Sherwood <david.sherwood@arm.com>
3358 * combine.c (can_change_dest_mode): Reject changes in
3359 REGMODE_NATURAL_SIZE.
3361 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3363 * configure.ac (--enable-libssp): New.
3364 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3365 trying to determine target-specific default. Adjust indentation.
3366 * configure: Regenerated.
3367 * doc/install.texi (Configuration): Expand --disable-libssp
3370 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3372 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3375 2017-10-31 Martin Jambor <mjambor@suse.cz>
3378 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3380 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3382 * auto-profile.c (autofdo_source_profile::read): Use
3383 UNKNOWN_LOCATION rather than 0.
3384 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3385 (warning_at): ...this overload.
3386 (warning_at_rich_loc_n): Rename to...
3387 (warning_n): ...this overload.
3388 (error_at_rich_loc): Rename to...
3389 (error_at): ...this overload.
3390 (pedwarn_at_rich_loc): Rename to...
3391 (pedwarn): ...this overload.
3392 (permerror_at_rich_loc): Rename to...
3393 (permerror): ...this overload.
3394 (inform_at_rich_loc): Rename to...
3395 (inform): ...this overload.
3396 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3397 (diagnostic_n_impl_richloc): Rename to...
3398 (diagnostic_n_impl): ...this rich_location *-based decl.
3399 (inform_at_rich_loc): Rename to...
3400 (inform): ...this, and add an assertion.
3401 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3402 (warning_at_rich_loc): Rename to...
3403 (warning_at): ...this, and add an assertion.
3404 (warning_at_rich_loc_n): Rename to...
3405 (warning_n): ...this, and add an assertion.
3406 (warning_n): Update location_t-based implementation for removal of
3407 location_t-based diagnostic_n_impl.
3408 (pedwarn_at_rich_loc): Rename to...
3409 (pedwarn): ...this, and add an assertion.
3410 (permerror_at_rich_loc): Rename to...
3411 (permerror): ...this, and add an assertion.
3412 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3413 (error_at_rich_loc): Rename to...
3414 (error_at): ...this, and add an assertion.
3415 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3416 (driver::do_spec_on_infiles): Likewise.
3417 * substring-locations.c (format_warning_va): Update for renaming
3418 of inform_at_rich_loc.
3420 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3422 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3423 _Float<N>X built-in functions so that the variant without the
3424 "__builtin_" prefix is only enabled for the GNU C and Objective C
3425 languages when they are in non-strict ANSI/ISO mode.
3426 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3427 * target.def (floatn_builtin_p): Add a target hook to control
3428 whether _Float<N> and _Float<N>X built-in functions without the
3429 "__builtin_" prefix are enabled, and return true for C and
3430 Objective C in the default hook. Include langhooks.h in
3432 * targhooks.h (default_floatn_builtin_p): Likewise.
3433 * targhooks.c (default_floatn_builtin_p): Likewise.
3434 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3435 floatn_builtin_p target hook.
3436 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3438 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3439 Eric Botcazou <ebotcazou@adacore.com>
3441 PR rtl-optimization/81803
3442 * lra-constraints.c (curr_insn_transform): Also reload the whole
3443 register for a strict subreg no wider than a word if this is for
3444 a WORD_REGISTER_OPERATIONS target.
3446 2017-10-31 Jason Merrill <jason@redhat.com>
3448 * gdbinit.in: Skip over inlines from timevar.h.
3450 2017-10-31 Martin Liska <mliska@suse.cz>
3452 * doc/gcov.texi: Document new option.
3453 * gcov.c (print_usage): Likewise print it.
3454 (process_args): Support the argument.
3455 (format_count): New function.
3456 (format_gcov): Use the function.
3458 2017-10-31 Martin Liska <mliska@suse.cz>
3460 * gcov.c (struct name_map): do not use typedef.
3461 Define operator== and operator<.
3462 (name_search): Remove.
3463 (name_sort): Remove.
3464 (main): Do not allocate names.
3465 (process_file): Add vertical space.
3466 (generate_results): Use std::find.
3467 (release_structures): Do not release memory.
3468 (find_source): Use std::find.
3470 2017-10-31 Martin Liska <mliska@suse.cz>
3472 * gcov.c (struct line_info): Remove it's typedef.
3473 (line_info::line_info): Add proper ctor.
3474 (line_info::has_block): Do not use a typedef.
3475 (struct source_info): Do not use typedef.
3476 (circuit): Likewise.
3477 (get_cycles_count): Likewise.
3478 (output_intermediate_file): Iterate via vector iterator.
3479 (add_line_counts): Use std::vector methods.
3480 (accumulate_line_counts): Likewise.
3481 (output_lines): Likewise.
3483 2017-10-31 Martin Liska <mliska@suse.cz>
3485 * gcov.c (struct source_info): Remove typedef.
3486 (source_info::source_info): Add proper ctor.
3487 (accumulate_line_counts): Use struct, not it's typedef.
3488 (output_gcov_file): Likewise.
3489 (output_lines): Likewise.
3490 (main): Do not allocate an array.
3491 (output_intermediate_file): Use size of vector container.
3492 (process_file): Resize the vector.
3493 (generate_results): Do not preallocate, use newly added vector
3495 (release_structures): Do not release sources.
3496 (find_source): Use vector methods.
3497 (add_line_counts): Do not use typedef.
3499 2017-10-31 Martin Liska <mliska@suse.cz>
3501 * doc/gcov.texi: Document that.
3502 * gcov.c (add_line_counts): Mark lines with a non-executed
3504 (output_line_beginning): Handle such lines.
3505 (output_lines): Pass new argument.
3506 (output_intermediate_file): Print it in intermediate format.
3508 2017-10-31 Martin Liska <mliska@suse.cz>
3510 * color-macros.h: New file.
3511 * diagnostic-color.c: Factor out color related to macros to
3513 * doc/gcov.texi: Document -k option.
3514 * gcov.c (INCLUDE_STRING): Include string.h.
3515 (print_usage): Add -k option.
3516 (process_args): Parse it.
3517 (pad_count_string): New function.
3518 (output_line_beginning): Likewise.
3519 (DEFAULT_LINE_START): New macro.
3520 (output_lines): Support color output.
3522 2017-10-31 Martin Liska <mliska@suse.cz>
3524 PR gcov-profile/82633
3525 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3526 their interaction with GCOV infrastructure.
3527 * configure.ac: Add -fkeep-{inline,static}-functions to
3529 * configure: Regenerate.
3531 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3534 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3536 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3539 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3540 into a register if it does not fit into an immediate offset field.
3542 2017-10-31 Olivier Hainque <hainque@adacore.com>
3544 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3546 2017-10-31 Julia Koval <julia.koval@intel.com>
3548 * config.gcc: Add gfniintrin.h.
3549 * config/i386/gfniintrin.h: New.
3550 * config/i386/i386-builtin-types.def
3551 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3552 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3553 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3554 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3555 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3556 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3557 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3558 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3559 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3560 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3561 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3562 * config/i386/immintrin.h: Include gfniintrin.h.
3563 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3565 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3567 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3569 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3571 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3572 (arm_ashldi3_1bit): Remove pattern.
3573 (ashrdi3): Remove shift by 1 expansion.
3574 (arm_ashrdi3_1bit): Remove pattern.
3575 (lshrdi3): Remove shift by 1 expansion.
3576 (arm_lshrdi3_1bit): Remove pattern.
3577 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3578 cost of ashldi3 by 1.
3579 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3580 (<shift>di3_neon): Likewise.
3582 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3584 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3585 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3586 and (*aarch64_simd_mov<VQ:mode>).
3587 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3588 pattern alternative.
3589 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3590 attributes to match pattern alternative.
3592 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3594 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3595 * config/rs6000/emmintrin.h: New file.
3596 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3598 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3600 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3601 * (movdi_vfp_cortexa8): Remove pattern.
3603 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3605 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3607 (Specific, alpha*-dec-osf5.1): Remove.
3608 (Specific, mips-sgi-irix5): Remove.
3609 (Specific, mips-sgi-irix6): Remove.
3611 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3614 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3615 arguments to clear_bit_region_be.
3617 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3619 * gimplify.c: Include memmodel.h.
3621 2017-10-30 Martin Jambor <mjambor@suse.cz>
3623 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3624 condition whether loop should be executed at all.
3626 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3628 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3629 gimple folding of vec_madd() intrinsics.
3630 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3631 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3632 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3634 2017-10-30 Richard Biener <rguenther@suse.de>
3636 PR tree-optimization/82762
3638 2017-10-23 Richard Biener <rguenther@suse.de>
3640 PR tree-optimization/82129
3642 2017-08-01 Richard Biener <rguenther@suse.de>
3644 PR tree-optimization/81181
3645 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3646 (compute_antic): ... end of iteration here.
3648 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3650 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3651 -std=iso9899:2017 and -std=gnu17.
3652 * doc/standards.texi (C Language): Document C17 support.
3653 * doc/cpp.texi (Overview): Mention -std=c17.
3654 (Standard Predefined Macros): Document C11 and C17 values of
3655 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3656 * doc/extend.texi (Inline): Do not list individual options for
3657 standards newer than C99.
3658 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3660 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3663 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3665 * asan.c (asan_finish_file): Align asan globals array by shadow
3668 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3671 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3672 (struct store_immediate_info): Add bitregion_start and bitregion_end
3674 (store_immediate_info::store_immediate_info): Add brs and bre
3675 arguments and initialize bitregion_{start,end} from those.
3676 (struct merged_store_group): Add bitregion_start, bitregion_end,
3677 align_base and mask fields. Drop unnecessary struct keyword from
3678 struct store_immediate_info. Add do_merge method.
3679 (clear_bit_region_be): Use memset instead of loop storing zeros.
3680 (merged_store_group::do_merge): New method.
3681 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3682 stores as long as the surrounding bitregions have no gaps.
3683 (merged_store_group::merge_overlapping): Use do_merge.
3684 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3685 is byte aligned, rather than requiring that start and width are
3686 byte aligned. Drop unnecessary struct keyword from
3687 struct store_immediate_info. Allocate and populate also mask array.
3688 Make start of the arrays relative to bitregion_start rather than
3689 start and size them according to bitregion_{end,start} difference.
3690 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3691 struct store_immediate_info.
3692 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3693 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3694 struct keyword from struct store_immediate_info.
3695 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3696 between stores as long as the surrounding bitregions have no gaps.
3698 (struct split_store): Add orig non-static data member.
3699 (split_store::split_store): Initialize orig to false.
3700 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3701 if there is exactly a single original stmt. Change stmts argument
3702 to pointer from reference, if NULL, don't push anything to it. Add
3703 first argument, use it to optimize skipping over orig stmts that
3704 are known to be before bitpos already. Simplify.
3705 (split_group): Return unsigned int count how many stores are or
3706 would be needed rather than a bool. Add allow_unaligned argument.
3707 Change split_stores argument from reference to pointer, if NULL,
3708 only do a dry run computing how many stores would be produced.
3709 Rewritten algorithm to use both alignment and misalign if
3710 !allow_unaligned and handle bitfield stores with gaps.
3711 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3712 from bitregion_start instead of start. Compute allow_unaligned
3713 here, if true, do 2 split_group dry runs to compute which one
3714 produces fewer stores and prefer aligned if equal. Punt if
3715 new count is bigger or equal than original before emitting any
3716 statements, rather than during that. Remove no longer needed
3717 new_ssa_names tracking. Replace num_stmts with
3718 split_stores.length (). Use 32-bit stack allocated entries
3719 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3720 if possible. Handle bitfields with gaps.
3721 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3722 Compute bitregion_{start,end} for the stores and construct
3723 store_immediate_info with that. Formatting fixes.
3725 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3728 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3729 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3731 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3733 * gimplify.c: Include tm_p.h.
3735 * common.opt (gcoff): Re-add as ignored option.
3736 (gcoff1, gcoff2, gcoff3): Likewise.
3738 * Makefile.in (OBJS): Delete sdbout.o.
3739 (GTFILES): Delete $(srcdir)/sdbout.c.
3740 * debug.h: Delete sdb_debug_hooks.
3741 * final.c: Delete sdbout.h include.
3742 (final_scan_insn): Delete SDB_DEBUG check.
3743 (rest_of_clean_state): Likewise.
3744 * output.h: Delete sdb_begin_function_line.
3747 * toplev.c: Delete sdbout.h include.
3748 (process_options): Delete SDB_DEBUG check.
3749 * tree-core.h (tree_type_common): Delete pointer field of
3751 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3752 TYPE_SYMTAB_POINTER.
3753 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3754 (TYPE_SYMTAB_IS_POINTER): Delete.
3755 (TYPE_SYMTAB_IS_DIE): Renumber.
3756 * xcoffout.c: Refer to former sdbout.c file.
3757 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3759 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3760 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3761 (Debugging Options): Delete -gcoff.
3762 (-gstabs): Delete SDB reference.
3764 (-gcoff@var{level}): Delete.
3765 * doc/passes.texi (Debugging information output): Delete SDB and
3766 sdbout.c references.
3767 * doc/tm.texi: Regenerate.
3768 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3769 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3771 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3772 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3773 (SDB_DEBUGGING_INFO): Delete.
3774 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3775 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3776 * target.def (output_source_filename): Delete COFF reference.
3778 * common.opt (gcoff): Delete.
3779 (gxcoff+): Update Negative chain.
3780 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3782 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3783 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3784 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3786 (expand_function_start): Change sdb reference to past tense.
3787 (expand_function_end): Change sdb reference to past tense.
3788 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3789 * opts.c (debug_type_names): Delete coff entry.
3790 (common_handle_option): Delete OPT_gcoff case.
3791 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3793 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3794 * config/cris/cris.h: Delete SDB reference in comment.
3795 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3796 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3797 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3798 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3800 (ix86_expand_prologue): Likewise.
3801 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3802 * config/ia64/ia64.h: Likewise.
3803 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3804 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3806 * config/mmix/mmix.h: Likewise.
3807 * config/nds32/nds32.c: Likewise.
3808 * config/stormy/storym16.h: Likewise.
3809 * config/visium/visium.h: Likewise.
3810 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3812 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3814 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3815 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3816 FRAME_POINTER_REGNUM point at high end of local var area.
3818 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3820 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3821 Move comment around. Do not reset best_edge for a copiable
3822 destination if the copy would cause a partition change.
3823 (better_edge_p): Remove redundant check.
3825 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3827 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3829 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3831 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3832 for math functions that have _Float<N> and _Float<N>X variants.
3833 (mathfn_built_in_2): Add support for math functions that have
3834 _Float<N> and _Float<N>X variants.
3835 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3836 (expand_builtin_mathfn_ternary): Add support for fma with
3837 _Float<N> and _Float<N>X variants.
3838 (expand_builtin): Likewise.
3839 (fold_builtin_3): Likewise.
3840 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3841 create math function _Float<N> and _Float<N>X variants as external
3843 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3844 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3845 the __builtin_ prefix and if not strict ansi, without the prefix.
3846 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3847 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3848 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3849 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3850 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3851 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3852 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3853 function signatures for fma _Float<N> and _Float<N>X variants.
3854 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3855 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3856 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3857 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3858 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3859 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3860 * gencfn-macros.c (print_case_cfn): Add support for math functions
3861 that have _Float<N> and _Float<N>X variants.
3862 (print_define_operator_list): Likewise.
3863 (fltfn_suffixes): Likewise.
3865 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3866 for math functions that have _Float<N> and _Float<N>X variants.
3867 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3868 and _Float<N>X variants.
3869 (COPYSIGN): Likewise.
3872 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3873 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3875 (integer_valued_read_call_p): Likewise.
3876 * fold-const-call.c (fold_const_call_ss): Likewise.
3877 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3878 _Float<N> and _Float<N>X variants.
3879 (fold_const_call_ssss): Add support for fma _Float<N> and
3880 _Float<N>X variants.
3881 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3882 support for copysign and fma _Float<N> and _Float<N>X variants.
3883 (backprop::process_builtin_call_use): Likewise.
3884 * tree-call-cdce.c (can_test_argument_range); Add support for
3885 sqrt _Float<N> and _Float<N>X variants.
3886 (edom_only_function): Likewise.
3887 (get_no_error_domain): Likewise.
3888 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3889 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3890 copysign _Float<N> and _Float<N>X variants.
3891 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3892 handled by machine independent code.
3893 (FMAF128): Likewise.
3894 * doc/cpp.texi (Common Predefined Macros): Document defining
3895 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3896 fma _Float<N> and _Float<N>X variants.
3898 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3901 * config/i386/i386-modes.def (CCFPU): Remove definition.
3902 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3903 (ix86_cc_modes_compatible): Ditto.
3904 (ix86_expand_carry_flag_compare): Ditto.
3905 (ix86_expand_int_movcc): Ditto.
3906 (ix86_expand_int_addcc): Ditto.
3907 (ix86_reverse_condition): Ditto.
3908 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3909 Return true/false for unordered/ordered fp comparisons.
3910 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3911 (ix86_prepare_fp_compare_args): Update for rename.
3912 (ix86_expand_fp_compare): Update for rename. Generate unordered
3913 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3914 (ix86_expand_sse_compare_and_jump): Ditto.
3915 * config/i386/predicates.md (fcmov_comparison_operator):
3916 Remove CCFPU mode handling.
3917 (ix86_comparison_operator): Ditto.
3918 (ix86_carry_flag_operator): Ditto.
3919 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3920 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3921 (*cmpu<mode>_cc_i387): Ditto.
3922 (FPCMP): Remove mode iterator.
3923 (unord): Remove mode attribute.
3924 (unord_subst): New define_subst transformation
3925 (unord): New define_subst attribute.
3927 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3928 (*cmpi<unord>xf_i387): Ditto.
3929 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3930 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3931 using unord_subst transformation.
3932 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3933 (round_saeonly): Also handle CCFP mode.
3934 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3935 Remove UNSPEC_SAHF unspec handling.
3937 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3939 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3941 2017-10-27 Jeff Law <law@redhat.com>
3943 * gimple-ssa-sprintf.c: Include domwalk.h.
3944 (class sprintf_dom_walker): New class, derived from dom_walker.
3945 (sprintf_dom_walker::before_dom_children): New function.
3946 (struct call_info): Moved into sprintf_dom_walker class
3947 (compute_formath_length, handle_gimple_call): Likewise.
3948 (sprintf_length::execute): Call the dominator walker rather
3949 than walking the statements.
3951 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3952 gimple statement locations.
3953 (check_array_bounds): Corresponding changes. Get the statement's
3954 location directly from wi->stmt.
3956 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3959 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3961 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3963 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3964 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3966 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3969 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3970 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3971 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3972 maybe_get_pool_constant.
3973 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3976 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3978 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3979 2.26 caveat. Update gas and gld versions.
3980 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3983 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3985 * cgraph.h (set_malloc_flag): Declare.
3986 * cgraph.c (set_malloc_flag_1): New function.
3987 (set_malloc_flag): Likewise.
3988 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3989 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3991 (read_ipa_call_summary): Add support for reading is_return_callee.
3992 (write_ipa_call_summary): Stream is_return_callee.
3993 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3994 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3995 ipa-prop.h, ipa-fnsummary.h.
3996 (pure_const_names): Change to static.
3997 (malloc_state_e): Define.
3998 (malloc_state_names): Define.
3999 (funct_state_d): Add field malloc_state.
4000 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4001 (check_retval_uses): New function.
4002 (malloc_candidate_p): Likewise.
4003 (analyze_function): Add support for malloc attribute.
4004 (pure_const_write_summary): Stream malloc_state.
4005 (pure_const_read_summary): Add support for reading malloc_state.
4006 (dump_malloc_lattice): New function.
4007 (propagate_malloc): New function.
4008 (warn_function_malloc): New function.
4009 (ipa_pure_const::execute): Call propagate_malloc and
4010 ipa_free_fn_summary.
4011 (pass_local_pure_const::execute): Add support for malloc attribute.
4012 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4013 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4015 2017-10-27 Martin Liska <mliska@suse.cz>
4017 PR gcov-profile/82457
4018 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4019 for fork-like functions to be properly instrumented.
4021 2017-10-27 Richard Biener <rguenther@suse.de>
4024 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4025 info when we redirected EH.
4027 2017-10-26 Michael Collison <michael.collison@arm.com>
4029 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4031 (<optab>_trunchf<GPI:mode>2: New pattern.
4032 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4033 * config/aarch64/iterators.md (wv): New mode attribute.
4034 (vf, VF): New mode attributes.
4035 (vgp, VGP): New mode attributes.
4036 (s): Update attribute with SImode and DImode prefixes.
4038 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4040 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4041 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4042 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4043 (nios2_option_overide): Initialize it. Don't allow R0-relative
4044 addressing with PIC.
4045 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4046 (nios2_symbolic_constant_p): Likewise.
4047 (nios2_legitimate_address_p): Likewise.
4048 (nios2_r0rel_section_name_p): New.
4049 (nios2_symbol_ref_in_r0rel_data_p): New.
4050 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4051 (r0rel_constant_p): New.
4052 (nios2_print_operand_address): Handle r0rel_constant_p.
4053 (nios2_cdx_narrow_form_p): Likewise.
4054 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4055 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4056 (Nios II Options): Document -mr0rel-sec.
4058 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4060 * config/nios2/nios2.c: Include xregex.h.
4061 (nios2_gprel_sec_regex): New.
4062 (nios2_option_overide): Initialize it. Don't allow GP-relative
4063 addressing with PIC.
4064 (nios2_small_section_name_p): Check for regex match.
4065 * config/nios2/nios2.opt (mgprel-sec=): New option.
4066 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4067 (Nios II Options): Document -mgprel-sec.
4069 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4071 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4073 2017-10-26 Tom de Vries <tom@codesourcery.com>
4075 PR tree-optimization/82707
4076 * gimple.c (gimple_copy): Fix unsharing of
4077 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4079 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4081 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4082 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4083 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4084 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4085 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4086 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4087 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4088 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4089 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4090 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4091 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4092 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4093 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4094 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4095 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4096 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4097 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4099 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4101 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4103 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4104 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4105 warning to rs6000.c. Remove the Undocumented flag, since it has
4107 (-mabi=ibmlongdouble): Likewise.
4108 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4109 already set, set the default format for long double.
4110 (rs6000_debug_reg_global): Print whether long double is IBM or
4112 (rs6000_option_override_internal): Rework setting long double
4113 format. Only warn if the user is changing the long double default
4114 and they did not use -Wno-psabi.
4115 * doc/invoke.texi (PowerPC options): Update the documentation for
4116 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4118 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4119 Alan Hayward <alan.hayward@arm.com>
4120 David Sherwood <david.sherwood@arm.com>
4122 * rtl.h (wider_subreg_mode): New function.
4123 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4124 rather than an unsigned int *.
4125 * ira-color.c (regno_max_ref_width): Replace with...
4126 (regno_max_ref_mode): ...this new variable.
4127 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4128 Use wider_subreg_mode.
4129 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4130 rather than an unsigned int *.
4131 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4132 (process_alt_operands): Likewise.
4133 (invariant_p): Likewise.
4134 * lra-spills.c (assign_mem_slot): Likewise.
4135 (add_pseudo_to_slot): Likewise.
4136 * lra.c (collect_non_operand_hard_regs): Likewise.
4137 (add_regs_to_insn_regno_info): Likewise.
4138 * reload1.c (regno_max_ref_width): Replace with...
4139 (regno_max_ref_mode): ...this new variable.
4140 (reload): Update accordingly. Update call to
4141 ira_sort_regnos_for_alter_reg.
4142 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4143 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4144 (scan_paradoxical_subregs): Likewise.
4146 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4148 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4149 (aarch64_frame): Add emit_frame_chain boolean.
4150 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4151 Move eh_return case to aarch64_layout_frame.
4152 (aarch64_layout_frame): Initialize emit_frame_chain.
4153 (aarch64_expand_prologue): Use emit_frame_chain.
4155 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4157 * config/aarch64/aarch64.c (aarch64_layout_frame):
4158 Ensure LR is always stored at the bottom of the callee-saves.
4159 Remove rarely used frame layout which saves callee-saves at top of
4160 frame, so the store of LR can be used as a valid probe in all cases.
4162 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4164 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4165 Improve unaligned TImode/TFmode base/offset split.
4167 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4168 Alan Hayward <alan.hayward@arm.com>
4169 David Sherwood <david.sherwood@arm.com>
4171 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4172 * combine.c (find_single_use_1): Likewise.
4173 (expand_field_assignment): Likewise.
4174 (move_deaths): Likewise.
4175 * lra-constraints.c (simplify_operand_subreg): Likewise.
4176 (curr_insn_transform): Likewise.
4177 * lra.c (collect_non_operand_hard_regs): Likewise.
4178 (add_regs_to_insn_regno_info): Likewise.
4179 * rtlanal.c (reg_referenced_p): Likewise.
4180 (covers_regno_no_parallel_p): Likewise.
4182 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4184 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4185 Don't print any bits outside the precision of the value.
4186 * wide-int.cc (test_printing): Add some new tests.
4188 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4190 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4191 supports -xbrace_comment option.
4192 * configure: Regenerate.
4193 * config.in: Regenerate.
4194 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4195 (ASM_CPU_SPEC): Use it.
4197 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4199 * target.def (static_rtx_alignment): New hook.
4200 * targhooks.h (default_static_rtx_alignment): Declare.
4201 * targhooks.c (default_static_rtx_alignment): New function.
4202 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4203 * doc/tm.texi: Regenerate.
4204 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4205 instead of targetm.constant_alignment. Remove call to
4207 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4208 (cris_preferred_mininum_alignment): New function, split out from...
4209 (cris_constant_alignment): ...here.
4210 (cris_static_rtx_alignment): New function.
4211 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4213 (ix86_constant_alignment): ...here.
4214 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4215 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4216 (mmix_static_rtx_alignment): New function.
4217 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4218 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4220 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4223 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4224 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4226 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4228 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4229 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4230 mark as a sign-extending load.
4231 (local_pic_load_u): Define.
4233 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4236 * fold-const.c (operand_equal_for_comparison_p): Also return true
4237 if ARG0 is a simple variant of ARG1 with narrower precision.
4238 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4240 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4242 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4244 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4245 scatter_static, scatter_per_elt.
4246 * x86-tune-costs.h: Add new cost entries.
4248 2017-10-25 Richard Biener <rguenther@suse.de>
4250 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4251 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4252 class pass_fre): Move to ...
4253 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4254 class pass_fre): ... here and adjust for statistics.
4256 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4259 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4260 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4261 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4262 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4265 2017-10-25 Richard Biener <rguenther@suse.de>
4267 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4268 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4269 eliminate_push_avail, eliminate_insert): Move inside...
4270 (class eliminate_dom_walker): ... this class in preparation
4272 (fini_eliminate): Remove by merging with ...
4273 (eliminate): ... this function. Adjust for class changes.
4274 (pass_pre::execute): Remove fini_eliminate call.
4275 (pass_fre::execute): Likewise.
4277 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4280 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4281 (VPERMI2, VPERMI2I): New mode iterators.
4282 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4283 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4285 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4286 mode iterator. Remove 3 old define_insn patterns.
4287 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4288 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4289 VPERMI2 mode iterator, remove the other two expanders.
4290 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4291 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4292 instructions, remove the other two patterns.
4293 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4294 mode iterator, remove the other two patterns.
4295 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4296 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4297 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4298 and adjust argument order accordingly.
4299 (ix86_expand_vec_perm): Adjust caller.
4300 (expand_vec_perm_1): Likewise.
4301 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4302 (expand_vec_perm_vpermt2_vpshub2): ... this.
4303 (ix86_expand_vec_perm_const_1): Adjust caller.
4304 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4307 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4308 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4309 (vec_shl_<mode>): Remove unused expander.
4310 (avx512bw_<shift_insn><mode>3): New define_insn.
4311 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4312 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4314 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4317 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4320 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4322 PR rtl-optimization/82396
4323 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4324 (autopref_multipass_init): Simplify initialization.
4325 (autopref_rank_data): Simplify sort order.
4326 * gcc/sched-int.h (autopref_multipass_data_): Remove
4327 multi_mem_insn_p, min_offset and max_offset.
4329 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4332 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4333 Check special value of flag_omit_frame_pointer.
4334 (aarch64_can_eliminate): Likewise.
4335 (aarch64_override_options_after_change_1): Simplify handling of
4336 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4338 2017-10-24 Richard Biener <rguenther@suse.de>
4340 PR tree-optimization/82697
4341 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4342 zero for conditional load and unconditional store.
4344 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4346 * doc/install.texi: Document bootstrap-cet.
4348 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4351 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4352 ENDBR instruction at function entrance if function is only
4355 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4358 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4359 patterns to better describe from which operation the CF is computed.
4360 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4361 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4362 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4363 is 0, use _0 suffixed expanders instead of emitting a comparison
4366 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4368 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4369 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4371 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4374 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4375 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4376 * loop-iv.c (iv_get_reaching_def): Likewise.
4377 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4378 variable is promoted and the partition contains undefined values.
4380 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4382 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4384 (nios2_address_cost): Define.
4385 (nios2_legitimize_address): Recognize (exp + constant) directly.
4386 (TARGET_ADDRESS_COST): Define.
4388 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4390 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4391 (nios2_symbolic_memory_operand_p): Declare.
4392 (nios2_split_large_constant): Declare.
4393 (nios2_split_symbolic_memory_operand): Declare.
4394 * config/nios2/nios2.c: Adjust includes.
4395 (nios2_symbolic_constant_allowed): New.
4396 (nios2_symbolic_constant_p): New.
4397 (nios2_plus_symbolic_constant_p): New.
4398 (nios2_valid_addr_expr_p): Recognize addresses involving
4400 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4401 (nios2_symbolic_memory_operand_p): New.
4402 (nios2_large_constant_p): New.
4403 (nios2_split_large_constant): New.
4404 (nios2_split_plus_large_constant): New.
4405 (nios2_split_symbolic_memory_operand): New.
4406 (nios2_legitimize_address): Code refactoring. Handle addresses
4407 involving symbolic constants.
4408 (nios2_emit_move_sequence): Likewise.
4409 (nios2_print_operand): Improve error output.
4410 (nios2_print_operand_address): Handle LO_SUM.
4411 (nios2_cdx_narrow_form_p): Likewise.
4412 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4413 operands involving symbolic constants.
4414 (movhi_internal, movsi_internal): Likewise.
4415 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4416 (extendhisi2, extendqi<mode>2): Likewise.
4418 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4420 * tree-pass.h (PROP_rtl_split_insns): Define.
4421 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4423 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4425 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4427 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4430 * target.def (const_not_ok_for_debug_p): Default to
4431 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4432 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4433 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4434 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4435 which targetm.const_not_ok_for_debug_p returned true.
4436 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4438 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4440 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4441 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4442 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4444 (ix86_const_not_ok_for_debug_p): New function.
4445 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4446 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4448 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4451 * system.h: Conditionally include "unique-ptr.h" if
4452 INCLUDE_UNIQUE_PTR is defined.
4453 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4454 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4456 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4458 * config/rl78/rl78.md: New define_expand "subdi3".
4460 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4463 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4464 DF_REF_INSN if DF_REF_INSN_INFO is false.
4466 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4468 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4469 xmm_move instead of sse_move.
4470 (sse_store_index): New function.
4471 (ix86_register_move_cost): Be more sensible about mismatch stall;
4472 model AVX moves correctly; make difference between sse->integer and
4474 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4475 moves; make difference between SSE and AVX.
4476 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4477 and zmm_move. Increase size of sse load and store tables;
4478 add unaligned load and store tables; add ssemmx_to_integer.
4479 * x86-tune-costs.h: Update all entries according to real
4480 move latencies from Agner Fog's manual and chip documentation.
4482 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4485 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4486 * config/i386/constraints.md (Wf): New constraint.
4487 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4488 (cmp<dwi>_doubleword): Removed.
4489 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4490 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4491 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4492 expand with cmp<dwi>_doubleword. For LTU and GEU use
4493 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4495 * common.opt (gcolumn-info): Enable by default.
4496 * doc/invoke.texi (gcolumn-info): Document new default.
4498 2017-10-23 Richard Biener <rguenther@suse.de>
4500 PR tree-optimization/82672
4501 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4502 Fold the stmt if we propagated into it.
4504 2017-10-23 Richard Biener <rguenther@suse.de>
4506 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4507 (bitmap_remove_expr_from_set): ... this. All callers call this
4508 for non-constant values.
4509 (bitmap_set_subtract): Rename to...
4510 (bitmap_set_subtract_expressions): ... this. Adjust and
4512 (bitmap_set_contains_value): Remove superfluous check.
4513 (bitmap_set_replace_value): Inline into single caller ...
4514 (bitmap_value_replace_in_set): ... here and simplify.
4515 (dependent_clean): Merge into ...
4516 (clean): ... this using an overload. Adjust.
4517 (prune_clobbered_mems): Adjust.
4518 (compute_antic_aux): Likewise.
4519 (compute_partial_antic_aux): Likewise.
4521 2017-10-23 Richard Biener <rguenther@suse.de>
4523 PR tree-optimization/82129
4525 2017-08-01 Richard Biener <rguenther@suse.de>
4527 PR tree-optimization/81181
4528 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4529 (compute_antic): ... end of iteration here.
4531 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4533 * target.def (starting_frame_offset): New hook.
4534 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4535 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4536 * doc/tm.texi.in: Regenerate.
4537 * hooks.h (hook_hwi_void_0): Declare.
4538 * hooks.c (hook_hwi_void_0): New function.
4539 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4540 STARTING_FRAME_OFFSET.
4541 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4542 * reload1.c (reload): Likewise.
4543 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4544 instead of STARTING_FRAME_OFFSET.
4545 * function.c (try_fit_stack_local): Likewise.
4546 (assign_stack_local_1): Likewise
4547 (instantiate_virtual_regs): Likewise.
4548 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4549 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4550 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4551 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4552 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4553 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4554 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4555 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4556 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4557 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4558 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4559 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4560 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4561 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4562 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4563 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4564 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4565 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4566 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4567 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4568 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4569 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4570 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4571 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4572 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4573 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4574 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4575 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4576 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4577 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4578 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4579 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4580 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4581 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4582 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4583 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4584 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4585 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4586 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4587 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4588 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4589 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4590 return a HOST_WIDE_INT.
4591 (avr_builtin_setjmp_frame_value): Use it instead of
4592 STARTING_FRAME_OFFSET.
4593 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4594 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4595 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4597 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4598 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4599 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4600 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4601 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4602 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4603 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4604 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4605 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4606 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4607 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4608 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4610 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4611 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4612 * config/mips/mips.c (mips_compute_frame_info): Refer to
4613 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4614 (mips_starting_frame_offset): New function.
4615 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4616 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4617 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4618 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4619 and return a HOST_WIDE_INT.
4620 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4621 (mmix_initial_elimination_offset): Refer to
4622 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4623 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4624 * config/pa/pa.c (pa_starting_frame_offset): New function.
4625 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4626 (pa_expand_prologue): Likewise.
4627 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4628 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4629 !FRAME_GROWS_DOWNWARD handling to...
4630 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4631 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4632 !FRAME_GROWS_DOWNWARD handling to...
4633 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4634 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4635 !FRAME_GROWS_DOWNWARD handling to...
4636 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4637 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4639 (rs6000_starting_frame_offset): New function.
4640 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4641 !FRAME_GROWS_DOWNWARD handling to...
4642 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4643 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4644 !FRAME_GROWS_DOWNWARD handling to...
4645 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4646 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4647 !FRAME_GROWS_DOWNWARD handling to...
4648 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4649 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4650 (rs6000_starting_frame_offset): New function.
4651 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4652 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4653 * config/vax/vax.c (vax_starting_frame_offset): New function.
4654 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4655 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4656 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4657 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4658 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4659 * system.h (STARTING_FRAME_OFFSET): Poison.
4661 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4663 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4664 SCALAR_TYPE_MODE instead of TYPE_MODE.
4666 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4667 Alan Hayward <alan.hayward@arm.com>
4668 David Sherwood <david.sherwood@arm.com>
4670 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4672 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4673 Alan Hayward <alan.hayward@arm.com>
4674 David Sherwood <david.sherwood@arm.com>
4676 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4678 2017-10-23 Richard Biener <rguenther@suse.de>
4680 PR tree-optimization/82129
4681 * tree-ssa-pre.c (bitmap_set_and): Remove.
4682 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4683 canonicalizing expressions in the set to those with lowest
4684 ID rather than taking that from the first edge.
4686 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4688 * combine.c (rtx_equal_for_field_assignment_p): Use
4689 byte_lowpart_offset.
4691 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4692 Alan Hayward <alan.hayward@arm.com>
4693 David Sherwood <david.sherwood@arm.com>
4695 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4696 to a SUBREG_PROMOTED_VAR.
4698 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4699 Alan Hayward <alan.hayward@arm.com>
4700 David Sherwood <david.sherwood@arm.com>
4702 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4703 (expand_debug_source_expr): Likewise.
4704 * combine.c (combine_simplify_rtx): Likewise.
4705 * cse.c (fold_rtx): Likewise.
4706 * optabs.c (expand_float): Likewise.
4707 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4708 (simplify_binary_operation_1): Likewise.
4710 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4711 Alan Hayward <alan.hayward@arm.com>
4712 David Sherwood <david.sherwood@arm.com>
4714 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4715 (record_promoted_value): Likewise.
4716 * expr.c (expand_expr_real_2): Likewise.
4717 * ree.c (update_reg_equal_equiv_notes): Likewise.
4718 (combine_set_extension): Likewise.
4719 * rtlanal.c (low_bitmask_len): Likewise.
4720 * simplify-rtx.c (neg_const_int): Likewise.
4721 (simplify_binary_operation_1): Likewise.
4723 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4724 Alan Hayward <alan.hayward@arm.com>
4725 David Sherwood <david.sherwood@arm.com>
4727 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4728 * regcprop.c (maybe_mode_change): Likewise.
4729 * reload1.c (alter_reg): Likewise.
4731 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4733 * inchash.h (inchash::hash::add_wide_int): New function.
4734 * lto-streamer-out.c (hash_tree): Use it.
4736 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4738 * inchash.h (inchash::hash::add_wide_int): Rename to...
4739 (inchash::hash::add_hwi): ...this.
4740 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4741 (polymorphic_call_target_hasher::hash): Likewise.
4742 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4743 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4744 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4745 * lto-streamer-out.c (hash_tree): Likewise.
4746 * optc-save-gen.awk: Likewise.
4747 * tree.c (add_expr): Likewise.
4749 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4752 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4753 for ordered inequality comparisons even with TARGET_IEEE_FP.
4755 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4758 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4759 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4760 Expand with cmp<dwi>_doubleword.
4762 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4764 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4765 List CET intrinsics.
4766 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4767 specific to -fcf-protection option.
4769 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4771 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4772 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4773 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4774 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4775 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4776 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4777 (extra_objs): Add cet.o for Linux/x86 targets.
4778 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4779 * config/i386/cet.c: New file.
4780 * config/i386/cetintrin.h: Likewise.
4781 * config/i386/t-cet: Likewise.
4782 * config/i386/cpuid.h (bit_SHSTK): New.
4783 (bit_IBT): Likewise.
4784 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4785 pass IBT and SHSTK bits.
4786 * config/i386/i386-builtin-types.def
4787 (VOID_FTYPE_UNSIGNED_PVOID): New.
4788 (VOID_FTYPE_UINT64_PVOID): Likewise.
4789 * config/i386/i386-builtin.def: Add CET intrinsics.
4790 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4791 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4792 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4793 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4795 * config/i386/i386.c (rest_of_insert_endbranch): New.
4796 (pass_data_insert_endbranch): Likewise.
4797 (pass_insert_endbranch): Likewise.
4798 (make_pass_insert_endbranch): Likewise.
4799 (ix86_notrack_prefixed_insn_p): Likewise.
4800 (ix86_target_string): Add -mibt, -mshstk flags.
4801 (ix86_option_override_internal): Add flag_cf_protection
4803 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4804 (ix86_print_operand): Add 'notrack' prefix output.
4805 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4806 (ix86_expand_builtin): Expand CET intrinsics.
4807 (x86_output_mi_thunk): Add 'endbranch' instruction.
4808 * config/i386/i386.h (TARGET_IBT): New.
4809 (TARGET_IBT_P): Likewise.
4810 (TARGET_SHSTK): Likewise.
4811 (TARGET_SHSTK_P): Likewise.
4812 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4813 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4814 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4815 (builtin_setjmp_setup): New pattern.
4816 (builtin_longjmp): Likewise.
4817 (rdssp<mode>): Likewise.
4818 (incssp<mode>): Likewise.
4819 (saveprevssp): Likewise.
4820 (rstorssp): Likewise.
4821 (wrss<mode>): Likewise.
4822 (wruss<mode>): Likewise.
4823 (setssbsy): Likewise.
4824 (clrssbsy): Likewise.
4825 (nop_endbr): Likewise.
4826 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4828 * config/i386/immintrin.h: Include <cetintrin.h>.
4829 * config/i386/linux-common.h
4830 (file_end_indicate_exec_stack_and_cet): New prototype.
4831 (TARGET_ASM_FILE_END): New.
4833 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4835 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4836 latencies instead of having separate table; make difference between
4837 integer and float costs.
4838 * i386.h (processor_costs): Remove scalar_stmt_cost,
4839 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4840 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4842 * x86-tune-costs.h: Remove entries which has been removed in
4843 procesor_costs from all tables; make cond_taken_branch_cost
4844 and cond_not_taken_branch_cost COST_N_INSNS based.
4846 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4848 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4850 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4852 * config/i386/i386.md (isa): Remove fma_avx512f.
4853 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4854 <avx512>_fmadd_<mode>_mask3<round_name>,
4855 <avx512>_fmsub_<mode>_mask<round_name>,
4856 <avx512>_fmsub_<mode>_mask3<round_name>,
4857 <avx512>_fnmadd_<mode>_mask<round_name>,
4858 <avx512>_fnmadd_<mode>_mask3<round_name>,
4859 <avx512>_fnmsub_<mode>_mask<round_name>,
4860 <avx512>_fnmsub_<mode>_mask3<round_name>,
4861 <avx512>_fmaddsub_<mode>_mask<round_name>,
4862 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4863 <avx512>_fmsubadd_<mode>_mask<round_name>,
4864 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4865 (*vec_widen_umult_even_v16si<mask_name>,
4866 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4867 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4869 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4871 * extend.texi: Add 'nocf_check' documentation.
4872 * gimple.texi: Add second parameter to
4873 gimple_build_call_from_tree.
4874 * invoke.texi: Add -fcf-protection documentation.
4875 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4877 2017-10-20 Richard Biener <rguenther@suse.de>
4879 PR tree-optimization/82473
4880 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4881 the largest input type.
4883 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4885 * c-attribs.c (handle_nocf_check_attribute): New function.
4886 (c_common_attribute_table): Add 'nocf_check' handling.
4887 * gimple-parser.c: Add second argument NULL to
4888 gimple_build_call_from_tree.
4889 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4890 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4892 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4893 * common.opt: Add fcf-protection flag.
4894 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4895 * flag-types.h: Add enum cf_protection_level.
4896 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4897 Add 'nocf_check' attribute propagation to gimple call.
4898 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4899 (gimple_build_call_from_tree): Update prototype.
4900 (gimple_call_nocf_check_p): New function.
4901 (gimple_call_set_nocf_check): Likewise.
4902 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4903 * ipa-icf.c: Add nocf_check attribute in statement hash.
4904 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4905 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4906 * toplev.c (process_options): Add flag_cf_protection handling.
4908 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4910 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4912 2017-10-20 Richard Biener <rguenther@suse.de>
4914 PR tree-optimization/82603
4915 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4916 remove false predicated stores.
4918 2017-10-20 Richard Biener <rguenther@suse.de>
4920 * graphite-isl-ast-to-gimple.c
4921 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4922 Remove return value and simplify, dump copied stmt after lhs
4924 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4925 Reduce dump verbosity.
4926 (gsi_insert_earliest): Likewise.
4927 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4928 * graphite.c (print_global_statistics): Adjust dumping.
4929 (print_graphite_scop_statistics): Likewise.
4930 (print_graphite_statistics): Do not dump loops here.
4931 (graphite_transform_loops): But here.
4933 2017-10-20 Nicolas Roche <roche@adacore.com>
4935 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4936 * configure: Regenerate.
4938 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4941 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4942 functions when optimizing replace GIMPLE_RETURN stmts with
4943 calls to __builtin_unreachable ().
4946 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4947 for -fsanitize=thread link of executables.
4948 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4949 link of executables.
4952 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4954 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4955 define_insns for logical vector shifts to use VI248_AVX512BW
4956 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4957 condition, useless isa and prefix attributes. Change the first
4958 2 of these define_insns to ...
4959 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4960 define_insn for avx512vl.
4961 (<shift_insn><mode>3): ... and this, new define_insn without
4962 masking for non-avx512vl.
4965 * config/i386/sse.md (*andnot<mode>3,
4966 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4967 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4968 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4969 not applied use empty suffix even for TARGET_AVX512VL.
4970 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4971 is applied, supply evex,evex or evex,evex,evex instead of just
4974 2017-10-20 Julia Koval <julia.koval@intel.com>
4976 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4977 (OPTION_MASK_ISA_GFNI_UNSET): New.
4978 (ix86_handle_option): Handle OPT_mgfni.
4979 * config/i386/cpuid.h (bit_GFNI): New.
4980 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4981 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4982 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4983 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4984 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4985 * config/i386/i386.opt: Add mgfni.
4987 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4989 * config/msp430/msp430.c (msp430_option_override): Disable
4990 -fdelete-null-pointer-checks.
4991 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4993 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4995 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4996 of x87 and SSE instructions.
4998 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5000 * asan.c (create_cond_insert_point): Do not update edge count.
5001 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5002 (afdo_propagate_circuit): Likewise.
5003 (afdo_calculate_branch_prob): Likewise.
5004 (afdo_annotate_cfg): Likewise.
5005 * basic-block.h (struct edge_def): Remove count.
5006 (edge_def::count): New accessor.
5007 * bb-reorder.c (rotate_loop): Update.
5008 (find_traces_1_round): Update.
5009 (connect_traces): Update.
5010 (sanitize_hot_paths): Update.
5011 * cfg.c (unchecked_make_edge): Update.
5012 (make_single_succ_edge): Update.
5013 (check_bb_profile): Update.
5014 (dump_edge_info): Update.
5015 (update_bb_profile_for_threading): Update.
5016 (scale_bbs_frequencies_int): Update.
5017 (scale_bbs_frequencies_gcov_type): Update.
5018 (scale_bbs_frequencies_profile_count): Update.
5019 (scale_bbs_frequencies): Update.
5020 * cfganal.c (connect_infinite_loops_to_exit): Update.
5021 * cfgbuild.c (compute_outgoing_frequencies): Update.
5022 (find_many_sub_basic_blocks): Update.
5023 * cfgcleanup.c (try_forward_edges): Update.
5024 (try_crossjump_to_edge): Update
5025 * cfgexpand.c (expand_gimple_cond): Update
5026 (expand_gimple_tailcall): Update
5027 (construct_exit_block): Update
5028 * cfghooks.c (verify_flow_info): Update
5029 (redirect_edge_succ_nodup): Update
5030 (split_edge): Update
5031 (make_forwarder_block): Update
5032 (duplicate_block): Update
5033 (account_profile_record): Update
5034 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5035 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5036 * cfgloopmanip.c (scale_loop_profile): Update.
5038 (lv_adjust_loop_entry_edge): Update.
5039 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5040 (force_nonfallthru_and_redirect): Update.
5041 (purge_dead_edges): Update.
5042 (rtl_flow_call_edges_add): Update.
5043 * cgraphunit.c (init_lowered_empty_function): Update.
5044 (cgraph_node::expand_thunk): Update.
5045 * gimple-pretty-print.c (dump_probability): Update.
5046 (dump_edge_probability): Update.
5047 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5048 * haifa-sched.c (sched_create_recovery_edges): Update.
5049 * hsa-gen.c (convert_switch_statements): Update.
5050 * ifcvt.c (dead_or_predicable): Update.
5051 * ipa-inline-transform.c (inline_transform): Update.
5052 * ipa-split.c (split_function): Update.
5053 * ipa-utils.c (ipa_merge_profiles): Update.
5054 * loop-doloop.c (add_test): Update.
5055 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5056 * lto-streamer-in.c (input_cfg): Update.
5057 (input_function): Update.
5058 * lto-streamer-out.c (output_cfg): Update.
5059 * modulo-sched.c (sms_schedule): Update.
5060 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5061 * predict.c (maybe_hot_edge_p): Update.
5062 (unlikely_executed_edge_p): Update.
5063 (probably_never_executed_edge_p): Update.
5064 (dump_prediction): Update.
5065 (drop_profile): Update.
5066 (propagate_unlikely_bbs_forward): Update.
5067 (determine_unlikely_bbs): Update.
5068 (force_edge_cold): Update.
5069 * profile.c (compute_branch_probabilities): Update.
5070 * reg-stack.c (better_edge): Update.
5071 * shrink-wrap.c (handle_simple_exit): Update.
5072 * tracer.c (better_p): Update.
5073 * trans-mem.c (expand_transaction): Update.
5074 (split_bb_make_tm_edge): Update.
5075 * tree-call-cdce.c: Update.
5076 * tree-cfg.c (gimple_find_sub_bbs): Update.
5077 (gimple_split_edge): Update.
5078 (gimple_duplicate_sese_region): Update.
5079 (gimple_duplicate_sese_tail): Update.
5080 (gimple_flow_call_edges_add): Update.
5081 (insert_cond_bb): Update.
5082 (execute_fixup_cfg): Update.
5083 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5084 * tree-complex.c (expand_complex_div_wide): Update.
5085 * tree-eh.c (lower_resx): Update.
5086 (unsplit_eh): Update.
5087 (cleanup_empty_eh_move_lp): Update.
5088 * tree-inline.c (copy_edges_for_bb): Update.
5089 (freqs_to_counts): Update.
5090 (copy_cfg_body): Update.
5091 * tree-ssa-dce.c (remove_dead_stmt): Update.
5092 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5093 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5094 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5095 (unloop_loops): Update.
5096 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5097 * tree-ssa-loop-split.c (connect_loops): Update.
5098 (split_loop): Update.
5099 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5100 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5101 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5102 * tree-ssa-reassoc.c (branch_fixup): Update.
5103 * tree-ssa-tail-merge.c (replace_block_by): Update.
5104 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5105 (compute_path_counts): Update.
5106 (update_profile): Update.
5107 (recompute_probabilities): Update.
5108 (update_joiner_offpath_counts): Update.
5109 (estimated_freqs_path): Update.
5110 (freqs_to_counts_path): Update.
5111 (clear_counts_path): Update.
5112 (ssa_fix_duplicate_block_edges): Update.
5113 (duplicate_thread_path): Update.
5114 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5115 (case_bit_test_cmp): Update.
5116 (collect_switch_conv_info): Update.
5117 (gen_inbound_check): Update.
5118 (do_jump_if_equal): Update.
5119 (emit_cmp_and_jump_insns): Update.
5120 * tree-tailcall.c (decrease_profile): Update.
5121 (eliminate_tail_call): Update.
5122 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5123 (vect_do_peeling): Update.
5124 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5125 * ubsan.c (ubsan_expand_null_ifn): Update.
5126 (ubsan_expand_ptr_ifn): Update.
5127 * value-prof.c (gimple_divmod_fixed_value): Update.
5128 (gimple_mod_pow2): Update.
5129 (gimple_mod_subtract): Update.
5130 (gimple_ic): Update.
5131 (gimple_stringop_fixed_value): Update.
5133 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5136 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5138 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5140 PR rtl-optimization/82395
5141 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5142 based on non_spilled_static_chain_regno_p.
5144 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5146 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5147 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5148 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5150 2017-10-19 Martin Sebor <msebor@redhat.com>
5152 PR tree-optimization/82596
5153 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5155 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5157 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5158 (get_mem_refs_of_builtin_call): Likewise.
5159 * builtins.c (expand_builtin_apply): Adjust call to
5160 allocate_dynamic_stack_space.
5161 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5162 the third argument to allocate_dynamic_stack_space, otherwise -1.
5163 (expand_builtin): Deal with all alloca variants.
5164 (is_inexpensive_builtin): Likewise.
5165 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5166 * calls.c (special_function_p): Deal with all alloca variants.
5167 (initialize_argument_information): Adjust call to
5168 allocate_dynamic_stack_space.
5169 (expand_call): Likewise.
5170 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5171 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5172 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5173 use it for the stack usage computation.
5174 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5175 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5176 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5177 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5178 (in_loop_p): Remove first argument and useless check.
5179 (pass_walloca::execute): Remove useless test and adjust call to above.
5180 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5181 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5182 (gimplify_call_expr): Deal with all alloca variants.
5183 * hsa-gen.c (gen_hsa_alloca): Likewise.
5184 (gen_hsa_insns_for_call): Likewise.
5185 * ipa-pure-const.c (special_builtin_state): Likewise.
5186 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5187 * tree-object-size.c (alloc_object_size): Likewise.
5188 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5189 (call_may_clobber_ref_p_1): Likewise.
5190 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5191 (ccp_fold_stmt): Likewise.
5192 (optimize_stack_restore): Likewise.
5193 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5194 (mark_all_reaching_defs_necessary_1): Likewise.
5195 (propagate_necessity): Likewise.
5196 (eliminate_unnecessary_stmts): Likewise.
5197 * tree.c (build_common_builtin_nodes): Build
5198 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5199 (build_alloca_call_expr): New function.
5200 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5201 (CASE_BUILT_IN_ALLOCA): Likewise.
5202 (build_alloca_call_expr): Declare.
5203 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5205 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5208 * dwarf2out.c (new_die_raw): New static inline function.
5209 (new_die): Use it to create the DIE.
5210 (add_AT_external_die_ref): Likewise.
5211 (clone_die): Likewise.
5212 (clone_as_declaration): Likewise.
5213 (dwarf2out_vms_debug_main_pointer): Likewise.
5214 (base_type_die): Likewise. Remove early return for corner cases.
5215 Do not call add_pubtype on the DIE here.
5216 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5217 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5218 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5219 native order exists for base types, attach the DIE manually and call
5220 add_pubtype on it. Do not equate a reverse order DIE to the type.
5222 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5224 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5225 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5227 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5229 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5231 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5232 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5233 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5234 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5235 * opts.c (sanitizer_opts): Add builtin.
5236 * ubsan.c (instrument_builtin): New function.
5237 (pass_ubsan::execute): Call it.
5238 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5239 * doc/invoke.texi: Document -fsanitize=builtin.
5241 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5242 builtins, store max (log2 (align), 0) into uchar field instead of
5243 align into uptr field.
5244 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5245 store uchar 0 field instead of uptr 0 field.
5246 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5247 instead of passing one address of struct with 2 locations pass
5248 two addresses of structs with 1 location each.
5249 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5250 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5251 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5252 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5253 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5254 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5255 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5256 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5258 2017-10-19 Martin Liska <mliska@suse.cz>
5261 * file-find.c (remove_prefix): Remove.
5262 * file-find.h (remove_prefix): Likewise.
5263 * gcc-ar.c: Remove smartness of lookup.
5265 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5267 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5268 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5269 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5271 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5274 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5275 (setcc + and to xor + setcc): New peephole2.
5277 2017-10-19 Tom de Vries <tom@codesourcery.com>
5279 * doc/sourcebuild.texi (Test Directives, Variants of
5280 dg-require-support): Add dg-require-stack-size.
5282 2017-10-19 Martin Liska <mliska@suse.cz>
5285 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5286 that have a large alignment.
5287 (gimplify_target_expr): Likewise.
5289 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5291 PR rtl-optimization/82602
5292 * ira.c (rtx_moveable_p): Return false for volatile asm.
5294 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5297 * config/i386/i386-modes.def (CCGZ): New CC mode.
5298 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5299 * config/i386/predicates.md (ix86_comparison_operator):
5301 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5302 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5303 with double-word subtraction.
5304 (put_condition_code): Handle CCGZmode.
5306 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5308 * wide-int.cc (debug (const wide_int &)): New.
5309 (debug (const wide_int *)): New.
5310 (debug (const widest_int &)): New.
5311 (debug (const widest_int *)): New.
5313 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5316 * lra-constraints.c (curr_insn_transform): Use non-input operand
5317 instead of output one for matched reload.
5319 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5321 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5322 (tree-ssa-loop-ivopts.h): New header file.
5323 (struct builtin_info): New fields.
5324 (classify_builtin_1): Compute and record base and offset parts for
5325 memset builtin partition by calling strip_offset.
5326 (offset_cmp, fuse_memset_builtins): New functions.
5327 (finalize_partitions): Fuse adjacent memset partitions by calling
5329 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5330 Expose the interface.
5331 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5333 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5335 PR tree-optimization/82574
5336 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5337 that data reference must be executed exactly once per iteration
5338 against the outermost loop in nest.
5339 (classify_partition): Update call to above function.
5341 2017-10-18 Richard Biener <rguenther@suse.de>
5343 PR tree-optimization/82591
5344 * graphite.c (graphite_transform_loops): Move code gen message
5346 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5347 Here. Handle scop_to_isl_ast failing.
5348 (scop_to_isl_ast): Limit the number of ISL operations.
5350 2017-10-18 Richard Biener <rguenther@suse.de>
5352 * graphite-isl-ast-to-gimple.c
5353 (translate_isl_ast_to_gimple::set_rename): Simplify.
5354 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5355 (graphite_copy_stmts_from_block): ... here.
5356 (copy_bb_and_scalar_dependences): Simplify.
5357 (add_parameters_to_ivs_params): Canonicalize.
5358 (generate_entry_out_of_ssa_copies): Simplify.
5359 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5360 by passing in ISL dimension.
5361 (parameter_index_in_region_1): Rename to ...
5362 (parameter_index_in_region): ... this.
5363 (extract_affine): Adjust assert, pass down parameter index.
5364 (add_param_constraints): Use range-info when available.
5365 (build_scop_context): Adjust.
5366 * sese.c (new_sese_info): Adjust.
5367 (free_sese_info): Likewise.
5368 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5369 Remove unused typedefs.
5370 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5372 2017-10-18 Martin Liska <mliska@suse.cz>
5374 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5376 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5378 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5379 (s390_sched_init): Do not reset s390_sched_state if we entered the
5380 current basic block via a fallthru edge and all others are unlikely.
5382 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5384 * config/s390/s390.c (NUM_SIDES): New variable.
5385 (LONGRUNNING_THRESHOLD): New variable.
5386 (LATENCY_FACTOR): New variable.
5387 (s390_sched_score): Decrease score for long-running instructions on
5389 (s390_sched_variable_issue): Perform bookkeeping for long-running
5392 2017-10-18 Richard Biener <rguenther@suse.de>
5394 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5395 Simplify with removal of the parameter rename map.
5396 (set_rename): Likewise.
5397 (should_copy_to_new_region): Likewise.
5398 (graphite_copy_stmts_from_block): Likewise.
5399 (copy_bb_and_scalar_dependences): Remove initialization of
5400 unused copied_bb_map.
5402 (copy_internal_parameters): Likewise.
5403 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5404 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5405 Use INTEGRAL_TYPE_P.
5406 (parameter_index_in_region_1): Rename to ...
5407 (assign_parameter_index_in_region): ... this. Assert we have
5408 a parameter we handle.
5409 (scan_tree_for_params): Adjust.
5410 * sese.h (parameter_rename_map_t): Remove.
5411 (struct sese_info_t): Remove unused parameter_rename_map and
5412 copied_bb_map members.
5413 * sese.c (new_sese_info): Adjust.
5414 (free_sese_info): Likewise.
5416 2017-10-18 Martin Liska <mliska@suse.cz>
5419 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5420 on an abnormal edge.
5422 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5424 * doc/invoke.texi (ffunction-sections and fdata-sections):
5427 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5429 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5430 the use statement can throw internally.
5432 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5434 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5435 any RTX present on the RHS of a SET.
5436 * compare-elim.c (try_eliminate_compare): Restore comment.
5438 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5440 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5441 may be also called on tcc_type nodes.
5442 * langhooks.c (lhd_tree_size): Likewise.
5444 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5446 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5447 format_warning_at_substring.
5448 (maybe_warn): Convert source_range * param to a location_t. Pass
5449 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5450 (format_directive): Remove code to extract source_ranges and
5451 source_range * in favor of just a location_t.
5452 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5454 * substring-locations.c (format_warning_va): Convert
5455 source_range * param to a location_t.
5456 (format_warning_at_substring): Likewise.
5457 * substring-locations.h (format_warning_va): Likewise.
5458 (format_warning_at_substring): Likewise.
5460 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5462 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5464 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5465 and scatter/gather ops.
5467 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5468 vec_gather_load and vec_scatter_store.
5469 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5470 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5471 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5472 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5473 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5474 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5476 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5478 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5479 Detect FTST instruction and handle its register pops. Only pop
5480 second operand if can_pop_second_op is true.
5481 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5482 set can_pop_second_op to false in the compare_for_stack_reg call.
5484 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5485 output_fp_compare for stack register operands.
5486 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5487 instructions here. Do not emit stack register pops here. Assert
5488 that FCOMPP pops next to top stack register. Rewrite function.
5490 2017-10-17 Nathan Sidwell <nathan@acm.org>
5493 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5494 use DECL_ASSEMBLER_NAME_RAW.
5497 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5500 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5501 Wilco Dijkstra <wilco.dijkstra@arm.com>
5503 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5504 converstion to Pmode from the buf_addr.
5506 2017-10-17 Richard Biener <rguenther@suse.de>
5508 * graphite-dependences.c (scop_get_reads_and_writes): Change
5509 output parameters to references.
5511 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5513 PR 71026/tree-optimization
5514 * fold-const.c (distribute_real_division): Removed.
5515 (fold_binary_loc): Remove calls to distribute_real_divison.
5517 2017-10-17 Richard Biener <rguenther@suse.de>
5519 * graphite-scop-detection.c
5520 (scop_detection::stmt_has_simple_data_refs_p): Always use
5521 the full nest as region.
5522 (try_generate_gimple_bb): Likewise.
5523 * sese.c (scalar_evolution_in_region): Simplify now that
5524 SCEV can handle instantiation in regions.
5525 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5526 in the non-loop part of a function if requested.
5528 2017-10-17 Richard Biener <rguenther@suse.de>
5530 PR tree-optimization/82563
5531 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5533 (graphite_regenerate_ast_isl): Call it.
5534 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5536 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5538 PR tree-optimization/82549
5539 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5540 Formatting fixes. Instead of calling make_bit_field_ref with negative
5543 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5545 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5546 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5547 _mm_maskz_reduce_ss): New.
5548 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5549 __builtin_ia32_reducess_mask): Ditto..
5550 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5551 * config/i386/sse.md (reduces<mode>): Renamed to ...
5552 (reduces<mode><mask_scalar_name>): ... this.
5553 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5555 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5556 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5558 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5560 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5561 * selftest-run-tests.c (selftest::run_tests): Call
5562 selftest::unique_ptr_tests_cc_tests.
5563 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5564 * unique-ptr-tests.cc: New file.
5566 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5569 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5571 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5572 (make_hard_regno_born, make_hard_regno_dead): Update
5573 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5575 2017-10-16 Jeff Law <law@redhat.com>
5577 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5579 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5581 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5583 2017-10-16 Olivier Hainque <hainque@adacore.com>
5585 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5586 with_cpu if we were configured for an e500v2 target cpu name.
5588 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5590 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5591 * doc/invoke.texi: Document +nodsp as a valid extension for
5594 2017-10-16 Martin Liska <mliska@suse.cz>
5596 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5597 (test_set_range): Likewise.
5598 (test_range_functions): Rename to ...
5599 (test_bit_in_range): ... this.
5600 (sbitmap_c_tests): Add new test.
5602 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5604 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5606 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5607 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5609 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5611 * config/aarch64/aarch64-builtins.c
5612 (aarch64_types_quadopu_lane_qualifiers): New.
5613 (TYPES_QUADOPU_LANE): New.
5614 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5615 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5616 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5617 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5618 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5619 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5620 (Vdottype, DOTPROD): New.
5621 (sur): Add SDOT and UDOT.
5623 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5625 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5626 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5627 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5629 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5630 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5631 Enable TARGET_DOTPROD.
5632 (cortex-a75.cortex-a55): Likewise.
5633 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5635 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5637 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5638 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5640 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5642 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5643 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5644 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5645 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5646 * config/arm/types.md (neon_dot, neon_dot_q): New.
5647 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5649 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5651 * config/arm/arm.h (TARGET_DOTPROD): New.
5652 * config/arm/arm.c (arm_arch_dotprod): New.
5653 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5654 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5655 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5656 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5657 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5658 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5659 * doc/invoke.texi (armv8.2-a): Document dotprod
5661 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5663 * i386.c (ix86_vec_cost): New function.
5664 (ix86_rtx_costs): Handle vector operations better.
5665 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5666 * x86-tune-costs.h: Add new costs to all tables.
5668 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5670 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5672 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5673 divsd, sqrtss and sqrtsd
5674 * x86-tune-costs.h: Add new entries to all costs.
5675 (znver1_cost): Fix to match real instruction latencies.
5677 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5678 Michael Collison <michael.collison@arm.com>
5680 * compare-elim.c: Include emit-rtl.h.
5681 (can_merge_compare_into_arith): New function.
5682 (try_validate_parallel): Likewise.
5683 (try_merge_compare): Likewise.
5684 (try_eliminate_compare): Call the above when no previous clobber
5686 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5689 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5693 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5694 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5698 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5699 to be any operand_equal_p operands. For & (B - 1) require
5700 B to be power of 2. Recognize
5701 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5703 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5706 * optabs.c (expand_memory_blockage): Fix call of
5707 targetm.have_memory_blockage.
5709 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5712 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5713 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5714 objects to extra_objs instead of overwriting it.
5716 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5718 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5719 Use any_fp_register_operand as operand[3] predicate. Simplify
5720 equality test for operands[2] and operands[4] memory location.
5721 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5722 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5723 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5724 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5725 any_fp_register_operand as operand[1] predicate. Simplify
5726 equality test for operands[0] and operands[3] memory location.
5727 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5728 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5729 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5731 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5733 * target-insns.def: Add memory_blockage.
5734 * optabs.c (expand_memory_blockage): New function.
5735 (expand_asm_memory_barrier): Rename ...
5736 (expand_asm_memory_blockage): ... to this.
5737 (expand_mem_thread_fence): Call expand_memory_blockage
5738 instead of expand_asm_memory_barrier.
5739 (expand_mem_singnal_fence): Ditto.
5740 (expand_atomic_load): Ditto.
5741 (expand_atomic_store): Ditto.
5742 * doc/md.texi (Standard Pattern Names For Generation):
5743 Document memory_blockage instruction pattern.
5745 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5747 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5748 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5749 * config/rl78/rl78.md: New define_expand "adddi3".
5751 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5753 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5756 2017-10-13 Jeff Law <law@redhat.com>
5758 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5760 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5763 * internal-fn.c (expand_mul_overflow): If both operands have
5764 the same highpart of -1 or 0 and the topmost bit of lowpart
5765 is different, overflow is if res <= 0 rather than res < 0.
5767 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5769 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5770 TARGET_P9_VECTOR code for unaligned_load case.
5772 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5774 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5776 2017-10-13 Nathan Sidwell <nathan@acm.org>
5778 * tree-core.h (tree_contains_struct): Make bool.
5779 * tree.c (tree_contains_struct): Likewise.
5780 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5781 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5782 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5783 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5784 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5786 2017-10-13 Richard Biener <rguenther@suse.de>
5788 * graphite-isl-ast-to-gimple.c
5789 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5790 parameters and dominance check.
5791 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5792 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5793 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5794 Do not update SSA form here or do intermediate IL verification.
5795 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5796 (graphite_initialize): Remove check on the number of loops in
5797 the function and inline into graphite_transform_loops.
5798 (graphite_finalize): Inline into graphite_transform_loops.
5799 (graphite_transform_loops): Perform SSA update and IL verification
5801 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5803 2017-10-13 Richard Biener <rguenther@suse.de>
5805 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5806 graphite_expression_type_precision): Avoid global constructor
5808 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5809 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5810 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5811 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5812 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5813 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5815 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5818 * config/i386/i386.h (ix86_red_zone_size): New.
5819 * config/i386/i386.md (push peephole2s): Replace
5820 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5822 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5823 Alan Hayward <alan.hayward@arm.com>
5824 David Sherwood <david.sherwood@arm.com>
5826 * combine.c (can_change_dest_mode): Reject changes in
5827 REGMODE_NATURAL_SIZE.
5829 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5830 Alan Hayward <alan.hayward@arm.com>
5831 David Sherwood <david.sherwood@arm.com>
5833 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5834 (expand_debug_source_expr): Likewise.
5835 * combine.c (combine_simplify_rtx): Likewise.
5836 * cse.c (fold_rtx): Likewise.
5837 * fwprop.c (canonicalize_address): Likewise.
5838 * targhooks.c (default_shift_truncation_mask): Likewise.
5840 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5841 Alan Hayward <alan.hayward@arm.com>
5842 David Sherwood <david.sherwood@arm.com>
5844 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5845 (widened_mode): Likewise.
5846 (expand_unop): Likewise.
5847 * ree.c (transform_ifelse): Likewise.
5848 (merge_def_and_ext): Likewise.
5849 (combine_reaching_defs): Likewise.
5850 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5852 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5853 Alan Hayward <alan.hayward@arm.com>
5854 David Sherwood <david.sherwood@arm.com>
5856 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5857 * combine.c (gen_lowpart_for_combine): Likewise.
5858 * dwarf2out.c (rtl_for_decl_location): Likewise.
5859 * final.c (alter_subreg): Likewise.
5860 * rtlhooks.c (gen_lowpart_general): Likewise.
5861 (gen_lowpart_if_possible): Likewise.
5863 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5864 Alan Hayward <alan.hayward@arm.com>
5865 David Sherwood <david.sherwood@arm.com>
5867 * calls.c (expand_call): Use subreg_lowpart_offset.
5868 * cse.c (cse_insn): Likewise.
5869 * regcprop.c (copy_value): Likewise.
5870 (copyprop_hardreg_forward_1): Likewise.
5872 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5875 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5876 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5877 =Q constraints to +Q and into insn condition add check
5878 that operands[0] and operands[1] are equal.
5879 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5880 =Q constraints to +Q and into insn condition add check
5881 that operands[0] is equal to either operands[1] or operands[2].
5884 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5885 instead of handling MINUS_EXPR twice (once for each argument),
5886 canonicalize operand order and handle just once, use rtype where
5887 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5890 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5891 any values of __C while still being pattern recognizable as a simple
5894 2017-10-13 Richard Biener <rguenther@suse.de>
5896 PR tree-optimization/82451
5898 2017-10-02 Richard Biener <rguenther@suse.de>
5900 PR tree-optimization/82355
5901 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5902 a mapping for the enclosing loop but avoid generating one for
5904 (copy_bb_and_scalar_dependences): Remove premature codegen
5905 error on PHIs in blocks duplicated into multiple places.
5906 * graphite-scop-detection.c
5907 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5908 in the region use it as loop and nest to analyze the DR in.
5909 (try_generate_gimple_bb): Likewise.
5910 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5911 (add_loop_constraints): For blocks in a loop not in the region
5912 create a dimension with a single iteration.
5913 * sese.h (gbb_loop_at_index): Remove assert.
5915 * cfgloop.c (loop_preheader_edge): For the loop tree root
5916 return the single successor of the entry block.
5917 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5918 Reset the SCEV hashtable and niters.
5919 * graphite-scop-detection.c
5920 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5921 assert that we only have POLYNOMIAL_CHREC that vary in loops
5922 contained in the region.
5923 (scop_detection::graphite_can_represent_expr): Adjust.
5924 (scop_detection::stmt_has_simple_data_refs_p): For loops
5925 not in the region set loop to NULL. The nest is now the
5926 entry edge to the region.
5927 (try_generate_gimple_bb): Likewise.
5928 * sese.c (scalar_evolution_in_region): Adjust for
5929 instantiate_scev change.
5930 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5931 Make nest parameter the edge into the region.
5932 (create_data_ref): Likewise.
5933 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5934 entry edge into a region and adjust instantiate_scev calls.
5935 (create_data_ref): Likewise.
5936 (graphite_find_data_references_in_stmt): Likewise.
5937 (find_data_references_in_stmt): Pass the loop preheader edge
5938 from the nest argument.
5939 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5940 parameter the edge into the region.
5941 (instantiate_parameters): Use the loop preheader edge as entry.
5942 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5944 (get_instantiated_value_entry): Make instantiate_below parameter
5945 the edge into the region.
5946 (instantiate_scev_name): Likewise. Adjust dominance checks,
5947 when we cannot use loop-based instantiation instantiate by
5948 walking use-def chains.
5949 (instantiate_scev_poly): Adjust.
5950 (instantiate_scev_binary): Likewise.
5951 (instantiate_scev_convert): Likewise.
5952 (instantiate_scev_not): Likewise.
5953 (instantiate_array_ref): Remove.
5954 (instantiate_scev_3): Likewise.
5955 (instantiate_scev_2): Likewise.
5956 (instantiate_scev_1): Likewise.
5957 (instantiate_scev_r): Do not blindly handle N-operand trees.
5958 Do not instantiate array-refs. Handle all constants and invariants.
5959 (instantiate_scev): Make instantiate_below parameter
5960 the edge into the region.
5961 (resolve_mixers): Use the loop preheader edge for the region
5962 parameter to instantiate_scev_r.
5963 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5965 2017-10-13 Richard Biener <rguenther@suse.de>
5967 PR tree-optimization/82525
5968 * graphite-isl-ast-to-gimple.c
5969 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5971 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5972 Fail code generation when we cannot represent the isl integer.
5973 (binary_op_to_tree): Elide modulo operations that are no-ops
5974 in the type we code generate. Remove now superfluous code
5977 2017-10-13 Richard Biener <rguenther@suse.de>
5979 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5980 (scop_detection::harmful_loop_in_region): Remove premature
5981 IV type restriction.
5982 (scop_detection::graphite_can_represent_scev): We can handle
5983 pointer IVs just fine.
5985 2017-10-13 Alan Modra <amodra@gmail.com>
5987 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5988 "Clobbers and Scratch Registers". Add paragraph on
5989 alternative to clobbers for scratch registers and OpenBLAS
5992 2017-10-13 Alan Modra <amodra@gmail.com>
5994 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5995 example of a memory input for a string of known length. Move
5996 commentary out of table. Add a number of new examples
5997 covering array memory inputs.
5999 2017-10-12 Martin Liska <mliska@suse.cz>
6001 PR tree-optimization/82493
6002 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6003 (test_range_functions): New function.
6004 (sbitmap_c_tests): Likewise.
6005 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6006 * selftest.h (sbitmap_c_tests): New function.
6008 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6010 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6012 * config/rs6000/amo.h: Fix spacing issue.
6014 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6017 * config/i386/i386.md (*ashl<mode>3_mask_1,
6018 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6019 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6022 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6024 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6025 (profile_probability): Set max_probability
6026 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6028 * profile-count.c (profile_probability::differs_from_p): Do not
6029 rely on max_probaiblity == 10000
6031 2017-10-12 Jeff Law <law@redhat.com>
6033 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6036 2017-10-12 Martin Sebor <msebor@redhat.com>
6040 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6041 (handle_alias_pairs): Call it.
6042 * common.opt (-Wattribute-alias): New option.
6043 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6044 * doc/invoke.texi (-Wattribute-alias): Document.
6046 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6049 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6051 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6053 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6054 (make_hard_regno_born, make_hard_regno_dead): Update
6055 bb_killed_pseudos and bb_gen_pseudos.
6057 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6059 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6061 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6063 * config/alpha/alpha.c (alpha_split_conditional_move):
6064 Use std::swap instead of manually swapping.
6065 (alpha_stdarg_optimize_hook): Ditto.
6066 (alpha_canonicalize_comparison): Ditto.
6068 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6070 * tree-loop-distribution.c (struct builtin_info): New struct.
6071 (struct partition): Refactor fields into struct builtin_info.
6072 (partition_free): Free struct builtin_info.
6073 (build_size_arg_loc, build_addr_arg_loc): Delete.
6074 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6075 information from struct builtin_info.
6076 (find_single_drs): New function refactored from classify_partition.
6077 Also moved builtin validity checks to this function.
6078 (compute_access_range, alloc_builtin): New functions.
6079 (classify_builtin_st, classify_builtin_ldst): New functions.
6080 (classify_partition): Refactor code into functions find_single_drs,
6081 classify_builtin_st and classify_builtin_ldst.
6082 (distribute_loop): Don't do runtime alias check when distributing
6084 (find_seed_stmts_for_distribution): New function.
6085 (pass_loop_distribution::execute): Refactor code finding seed
6086 stmts into above function. Support distribution for the innermost
6087 two-level loop nest. Adjust dump information.
6089 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6091 * tree-loop-distribution.c: Adjust the general comment.
6092 (NUM_PARTITION_THRESHOLD): New macro.
6093 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6094 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6095 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6096 in call to build_partition_graph.
6097 (finalize_partitions): New parameter. Make loop distribution more
6098 conservative by fusing more partitions.
6099 (distribute_loop): Don't do runtime alias check in case of loop nest
6101 (find_seed_stmts_for_distribution): New function.
6102 (prepare_perfect_loop_nest): New function.
6103 (pass_loop_distribution::execute): Refactor code finding seed stmts
6104 and loop nest into above functions. Support loop nest distribution.
6105 Adjust dump information accordingly.
6107 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6109 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6110 and set PTYPE_SEQUENTIAL for merged partition.
6112 2017-10-12 Richard Biener <rguenther@suse.de>
6114 PR tree-optimization/69728
6116 2017-09-19 Richard Biener <rguenther@suse.de>
6118 PR tree-optimization/69728
6119 * graphite-sese-to-poly.c (schedule_error): New global.
6120 (add_loop_schedule): Handle empty domain by failing the
6122 (build_original_schedule): Handle schedule_error.
6124 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6125 domain by returning an unchanged schedule.
6127 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6129 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6130 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6132 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6134 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6137 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6140 * expr.c (store_field): Don't optimize away bitsize == 0 store
6141 from CALL_EXPR with addressable return type.
6143 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6145 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6146 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6147 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6148 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6149 TARGET_ISEL instead of TARGET_ISEL<sel>.
6151 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6153 * config/rs6000/rs6000.c
6154 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6156 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6158 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6160 (reg_or_cint_operand): Fix comment.
6161 (reg_or_zero_operand): New predicate.
6162 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6163 * config/rs6000/rs6000.c (output_isel): Delete.
6164 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6165 instead of reg_or_cint_operand. Output instruction directly (not via
6167 (isel_unsigned_<mode>): Ditto.
6168 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6169 gpc_reg_operand. Add an instruction alternative for this. Output
6170 instruction directly.
6171 (*isel_reversed_unsigned_<mode>): Ditto.
6173 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6175 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6176 (TARGET_CANONICALIZE_COMPARISON): Define.
6178 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6181 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6182 Check whether the dest is REG before adding REG_EQUIV note.
6184 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6187 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6189 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6190 (make_hard_regno_born, make_hard_regno_dead): Update
6191 bb_killed_pseudos and bb_gen_pseudos.
6193 2017-10-11 Nathan Sidwell <nathan@acm.org>
6195 * incpath.h (enum incpath_kind): Name enum, prefix values.
6196 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6197 * incpath.c (heads, tails): Use INC_MAX.
6198 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6199 (merge_include_chains, split_quote_chain,
6200 register_include_chains): Update incpath_kind names.
6201 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6202 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6204 (add_framework_path, darwin_register_objc_includes): Likewise.
6205 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6207 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6209 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6210 Do not use float_operator operator predicate.
6211 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6212 * config/i386/predicates.md (float_operator): Remove predicate.
6214 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6216 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6217 (*jccxf_i387): Ditto.
6218 (*jcc<mode>_i387): Ditto.
6219 (*jccu<mode>_i387): Ditto.
6220 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6221 (*jcc_*_i387 splitters): Remove.
6222 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6223 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6224 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6227 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6229 * profile-count.h (slow_safe_scale_64bit): New function.
6230 (safe_scale_64bit): New inline.
6231 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6232 * profile-count.c: Include wide-int.h
6233 (slow_safe_scale_64bit): New.
6235 2017-10-11 Nathan Sidwell <nathan@acm.org>
6237 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6238 HAS_DECL_ASSEMBLER_NAME_P.
6239 * gimple-expr.c (gimple_decl_printable_name: Check
6240 HAS_DECL_ASSEMBLER_NAME_P too.
6241 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6242 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6243 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6244 * passes.c (rest_of_decl_compilation): Check
6245 HAS_DECL_ASSEMBLER_NAME_P too.
6246 * recog.c (verify_changes): Likewise.
6247 * tree-pretty-print.c (dump_decl_name): Likewise.
6248 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6250 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6251 (SET_DECL_ASSEMBLER_NAME): Use it.
6252 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6253 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6254 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6256 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6258 * config.gcc (i386, x86_64): Add extra objects.
6259 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6260 (ix86_min_insn_size): Declare.
6261 (ix86_issue_rate): Declare.
6262 (ix86_adjust_cost): Declare.
6263 (ia32_multipass_dfa_lookahead): Declare.
6264 (ix86_macro_fusion_p): Declare.
6265 (ix86_macro_fusion_pair_p): Declare.
6266 (ix86_bd_has_dispatch): Declare.
6267 (ix86_bd_do_dispatch): Declare.
6268 (ix86_core2i7_init_hooks): Declare.
6269 (ix86_atom_sched_reorder): Declare.
6270 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6271 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6272 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6273 (rip_relative_addr_p): Rename to ...
6274 (ix86_rip_relative_addr_p): ... this one; export.
6275 (memory_address_length): Update.
6276 (ix86_issue_rate): Move to x86-tune-sched.c.
6277 (ix86_flags_dependent): Move to x86-tune-sched.c.
6278 (ix86_agi_dependent): Move to x86-tune-sched.c.
6279 (exact_dependency_1): Move to x86-tune-sched.c.
6280 (exact_store_load_dependency): Move to x86-tune-sched.c.
6281 (ix86_adjust_cost): Move to x86-tune-sched.c.
6282 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6283 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6284 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6285 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6286 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6287 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6288 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6289 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6290 (min_insn_size): Rename to ...
6291 (ix86_min_insn_size): ... this one; export.
6292 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6293 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6294 (core2i7_first_cycle_multipass_backtrack): Move to
6295 x86-tune-sched-core.c.
6296 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6297 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6298 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6299 (ix86_avoid_jump_mispredicts): Update.
6300 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6301 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6302 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6303 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6304 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6305 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6306 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6307 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6308 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6309 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6310 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6311 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6312 (BIG): Move to ix86-tune-sched-bd.c.
6313 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6314 (enum insn_path): Move to ix86-tune-sched-bd.c.
6315 (get_mem_group): Move to ix86-tune-sched-bd.c.
6316 (is_cmp): Move to ix86-tune-sched-bd.c.
6317 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6318 (is_branch): Move to ix86-tune-sched-bd.c.
6319 (is_prefetch): Move to ix86-tune-sched-bd.c.
6320 (init_window): Move to ix86-tune-sched-bd.c.
6321 (allocate_window): Move to ix86-tune-sched-bd.c.
6322 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6323 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6324 (process_end_window): Move to ix86-tune-sched-bd.c.
6325 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6326 (find_constant): Move to ix86-tune-sched-bd.c.
6327 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6328 (has_immediate): Move to ix86-tune-sched-bd.c.
6329 (get_insn_path): Move to ix86-tune-sched-bd.c.
6330 (get_insn_group): Move to ix86-tune-sched-bd.c.
6331 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6332 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6333 (add_insn_window): Move to ix86-tune-sched-bd.c.
6334 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6335 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6336 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6337 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6338 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6339 (do_dispatch): Move to ix86-tune-sched-bd.c.
6340 (has_dispatch): Move to ix86-tune-sched-bd.c.
6341 * config/i386/t-i386: Add new object files.
6342 * config/i386/x86-tune-costs.h: New file.
6343 * config/i386/x86-tune-sched-atom.c: New file.
6344 * config/i386/x86-tune-sched-bd.c: New file.
6345 * config/i386/x86-tune-sched-core.c: New file.
6346 * config/i386/x86-tune-sched.c: New file.
6348 2017-10-11 Liu Hao <lh_mouse@126.com>
6350 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6351 the generic version below instead.
6352 (should_colorize): Recognize Windows consoles as terminals
6354 * pretty-print.c [__MINGW32__] (write_all): New function.
6355 [__MINGW32__] (find_esc_head): Likewise.
6356 [__MINGW32__] (find_esc_terminator): Likewise.
6357 [__MINGW32__] (eat_esc_sequence): Likewise.
6358 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6360 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6363 2017-10-11 Richard Biener <rguenther@suse.de>
6365 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6366 Properly call analyze_scalar_evolution with the loop of the stmt.
6368 2017-10-11 Richard Biener <rguenther@suse.de>
6370 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6371 * tree-core.h (tree_base): Add chrec_var union member.
6372 * tree.h (CHREC_VAR): Remove.
6373 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6374 * tree-chrec.h (build_polynomial_chrec): Adjust.
6375 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6376 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6378 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6380 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6381 * match.pd: ... here.
6382 ((T) X == (T) Y): Relax condition.
6384 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6386 PR tree-optimization/82472
6387 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6389 (break_alias_scc_partitions): Update postorder number.
6391 2017-10-11 Martin Liska <mliska@suse.cz>
6394 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6396 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6398 2017-10-11 Martin Liska <mliska@suse.cz>
6400 * print-rtl.c (print_insn): Move declaration of idbuf
6401 to same scope as name.
6403 2017-10-11 Martin Liska <mliska@suse.cz>
6408 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6411 2017-10-11 Martin Liska <mliska@suse.cz>
6414 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6417 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6419 * config.gcc (default_gnu_indirect_function): Default to yes for
6420 arm*-*-linux* with glibc.
6422 2017-10-11 Richard Biener <rguenther@suse.de>
6424 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6425 default-defs and types we do not want to analyze.
6426 (interpret_loop_phi): Replace unreachable code with an assert.
6427 (compute_scalar_evolution_in_loop): Remove and inline ...
6428 (analyze_scalar_evolution_1): ... here, replacing condition with
6429 what makes the intent clearer. Remove handling of cases
6430 get_scalar_evolution now handles.
6432 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6434 PR rtl-optimization/81434
6435 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6436 comment for main loop. In sched_group_found if, also add checks for
6437 pass and min_cost_group.
6439 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6441 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6442 (rs6000_insn_cost): New function.
6443 * config/rs6000/rs6000.md (cost): New attribute.
6445 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6446 H.J. Lu <hongjiu.lu@intel.com>
6450 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6451 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6452 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6453 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6454 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6455 OPTION_MASK_ISA_MMX and some other bit set require both
6456 mmx and the ISAs without the mmx bit.
6457 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6458 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6459 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6460 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6461 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6462 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6463 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6464 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6465 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6466 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6467 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6468 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6469 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6470 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6471 Add OPTION_MASK_ISA_MMX.
6473 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6475 * config.gcc (armv7*-*-freebsd*): New target.
6476 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6478 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6480 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6481 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6484 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6486 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6487 a template parameter.
6488 (WIDE_INT_REF_FOR): Update accordingly.
6489 * tree.h (wi::int_traits <const_tree>): Delete.
6490 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6491 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6492 (wi::tree_to_wide_ref): New typedef.
6493 (wi::to_wide): New function.
6494 * calls.c (get_size_range): Use wi::to_wide when operating on
6496 * cgraph.c (cgraph_node::create_thunk): Likewise.
6497 * config/i386/i386.c (ix86_data_alignment): Likewise.
6498 (ix86_local_alignment): Likewise.
6499 * dbxout.c (stabstr_O): Likewise.
6500 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6501 * expr.c (const_vector_from_tree): Likewise.
6502 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6503 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6504 (fold_negate_expr_1, int_const_binop_1, const_binop)
6505 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6506 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6507 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6508 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6509 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6510 (fold_not_const, round_up_loc): Likewise.
6511 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6512 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6513 (alloca_call_type): Likewise.
6514 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6515 * godump.c (go_output_typedef): Likewise.
6516 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6517 * internal-fn.c (get_min_precision): Likewise.
6518 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6519 * ipa-polymorphic-call.c
6520 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6521 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6522 (ipa_modify_call_arguments): Likewise.
6523 * match.pd: Likewise.
6524 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6525 * print-tree.c (print_node_brief, print_node): Likewise.
6526 * stmt.c (expand_case): Likewise.
6527 * stor-layout.c (layout_type): Likewise.
6528 * tree-affine.c (tree_to_aff_combination): Likewise.
6529 * tree-cfg.c (group_case_labels_stmt): Likewise.
6530 * tree-data-ref.c (dr_analyze_indices): Likewise.
6531 (prune_runtime_alias_test_list): Likewise.
6532 * tree-dump.c (dequeue_and_dump): Likewise.
6533 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6534 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6535 * tree-pretty-print.c (dump_generic_node): Likewise.
6536 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6537 (simple_iv_with_niters): Likewise.
6538 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6539 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6540 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6541 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6542 (refine_value_range_using_guard, number_of_iterations_ne_max)
6543 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6544 (get_cst_init_from_scev, record_nonwrapping_iv)
6545 (scev_var_range_cant_overflow): Likewise.
6546 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6547 * tree-ssa-pre.c (compute_avail): Likewise.
6548 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6549 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6550 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6551 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6552 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6553 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6554 (dump_case_nodes, try_switch_expansion): Likewise.
6555 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6556 (vect_do_peeling): Likewise.
6557 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6558 * tree-vect-stmts.c (vectorizable_load): Likewise.
6559 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6560 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6561 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6562 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6563 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6564 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6565 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6566 (evrp_dom_walker::before_dom_children): Likewise.
6567 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6568 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6569 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6570 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6571 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6572 (get_range_pos_neg): Likewise.
6573 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6574 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6575 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6576 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6577 * config/avr/avr.c (avr_fold_builtin): Likewise.
6578 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6579 * config/msp430/msp430.c (msp430_attr): Likewise.
6580 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6581 * config/powerpcspe/powerpcspe-c.c
6582 (altivec_resolve_overloaded_builtin): Likewise.
6583 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6584 (rs6000_expand_ternop_builtin): Likewise.
6585 * config/rs6000/rs6000-c.c
6586 (altivec_resolve_overloaded_builtin): Likewise.
6587 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6588 (rs6000_expand_ternop_builtin): Likewise.
6589 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6591 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6593 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6594 when copying loop nest with only one inner loop.
6596 2017-10-10 Richard Biener <rguenther@suse.de>
6598 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6599 blocks if SCEV is active.
6600 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6602 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6603 (scev_initialize): Assert we are not yet initialized.
6605 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6607 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6608 inner loop's exit stmt by making it always exit the loop, otherwise
6609 we would generate an infinite empty loop.
6611 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6613 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6614 renaming variables in new preheader if it's deleted.
6616 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6618 * tree-loop-distribution.c (struct partition): Remove unused field
6619 loops of the structure.
6620 (partition_alloc, partition_free): Ditto.
6621 (build_rdg_partition_for_vertex): Ditto.
6623 2017-10-09 Jeff Law <law@redhat.com>
6625 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6626 return type to match prototype and documentation.
6628 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6630 * config/rs6000/rs6000.c (processor_costs): Move to ...
6631 * config/rs6000/rs6000.h: ... here.
6632 (rs6000_cost): Declare.
6634 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6636 * except.c (setjmp_fn): New global variable.
6637 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6638 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6639 if DONT_USE_BUILTIN_SETJMP is defined.
6641 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6643 * target.def (insn_cost): New hook.
6644 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6645 * doc/tm.texi: Regenerate.
6646 * rtlanal.c (insn_cost): Use the new hook.
6648 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6650 * combine.c (combine_validate_cost): Compute the new insn_cost,
6651 not just pattern_cost.
6652 (try_combine): Adjust comment.
6654 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6656 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6658 * combine.c (uid_insn_cost): Adjust comment.
6659 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6661 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6662 * dse.c (find_shift_sequence): Ditto.
6663 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6664 (bb_valid_for_noce_process_p): Use pattern_cost.
6665 * rtl.h (insn_rtx_cost): Delete.
6666 (pattern_cost): New prototype.
6667 (insn_cost): New prototype.
6668 * rtlanal.c (insn_rtx_cost): Rename to...
6669 (pattern_cost): ... this.
6672 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6674 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6675 (*jcc<mode>_0_r_i387): Ditto.
6676 (*jccxf_r_i387): Ditto.
6677 (*jcc<mode>_r_i387): Ditto.
6678 (*jccu<mode>_r_i387): Ditto.
6679 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6680 (*jcc): Rename from *jcc_1.
6682 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6684 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6685 deferred rescans after the lvx/stvx recombination pre-pass.
6687 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6689 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6690 memory operation instruction support.
6691 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6692 (rs6000-ibm-aix[789]*): Likewise.
6693 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6694 Document new functions.
6696 2017-10-09 Richard Biener <rguenther@suse.de>
6698 PR tree-optimization/82397
6699 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6700 equality only for semantically equal trees.
6702 2017-10-09 Richard Biener <rguenther@suse.de>
6704 PR tree-optimization/82449
6705 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6706 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6707 allow constant addresses.
6708 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6711 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6713 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6716 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6719 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6722 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6725 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6727 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6730 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6731 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6733 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6735 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6736 (WI_BINARY_PREDICATE_RESULT): Likewise.
6737 (wi::binary_traits::operator_result): New type.
6738 (wi::binary_traits::predicate_result): Likewise.
6739 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6740 (generic_wide_int::operator==, generic_wide_int::operator!=)
6741 (generic_wide_int::operator&, generic_wide_int::and_not)
6742 (generic_wide_int::operator|, generic_wide_int::or_not)
6743 (generic_wide_int::operator^, generic_wide_int::operator+
6744 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6746 (operator~, unary operator-, operator==, operator!=, operator&)
6747 (operator|, operator^, operator+, binary operator-, operator*): New
6749 * expr.c (get_inner_reference): Use wi::bit_and_not.
6750 * fold-const.c (fold_binary_loc): Likewise.
6751 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6752 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6753 (bit_value_binop): Likewise.
6754 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6755 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6756 (extract_range_from_binary_expr_1): Likewise.
6757 (masked_increment): Likewise.
6758 (simplify_bit_ops_using_ranges): Likewise.
6760 2017-10-09 Martin Jambor <mjambor@suse.cz>
6763 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6764 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6765 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6767 (hsa_fixup_mov_insn_type): New function.
6768 (hsa_op_with_type::get_in_type): Use it.
6769 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6770 immediates in an assert.
6771 (hsa_op_with_type::extend_int_to_32bit): New method.
6772 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6773 types. Convert to dest type if necessary.
6774 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6775 (reg_for_gimple_ssa): Pass false as min32int to
6776 hsa_type_for_scalar_tree_type.
6777 (gen_hsa_addr): Fixup type when creating addresable temporary.
6778 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6779 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6780 necessary. Call hsa_fixup_mov_insn_type.
6781 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6782 extend operands and convert to dest type if necessary.
6783 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6784 to dest type if necessary.
6785 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6786 if conversion nt necessary and size matches.
6787 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6788 to dest type if necessary.
6789 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6790 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6792 (gen_hsa_clrsb): Likewise.
6793 (gen_hsa_ffs): Likewise.
6794 (gen_hsa_divmod): Extend operands and convert to dest type if
6796 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6798 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6800 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6801 Remove empty default arguments. Use a brace block as output
6803 (conditional return): Ditto.
6805 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6806 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6808 (group_ending_nop): Ditto.
6809 (doloop_end): Ditto.
6810 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6811 (splitters for those): Ditto.
6813 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6815 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6816 a conditional jump (and the compare for it) so that pc_rtx is the
6818 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6819 for the deleted and renamed ctr<mode>_internal[234] patterns.
6820 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6821 Delete second conditional return pattern.
6822 (ctr<mode>_internal2): Delete this second bdnz pattern.
6823 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6824 (ctr<mode>_internal4): Delete this second bdz pattern.
6826 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6828 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6829 (always_initialized_rtx_for_ssa_name_p): New predicate.
6830 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6831 (finish_out_of_ssa): Free new field of SA.
6832 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6833 * tree-ssa-coalesce.c: Include tree-ssa.h.
6834 (get_parm_default_def_partitions): Remove extern keyword.
6835 (get_undefined_value_partitions): New function.
6836 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6837 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6839 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6841 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6843 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6845 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6847 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6848 for targets that preffer 128bit.
6850 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6852 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6854 2017-10-08 Olivier Hainque <hainque@adacore.com>
6856 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6857 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6859 (thumb_set_return_address): Likewise.
6861 2017-10-08 Olivier Hainque <hainque@adacore.com>
6863 * common/config/arm/arm-common.c (arm_except_unwind_info):
6864 Handle DWARF2_UNWIND_INFO.
6866 2017-10-07 Michael Collison <michael.collison@arm.com>
6868 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6871 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6873 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6874 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6875 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6876 defined, force the creation of a new block for a dispatch label.
6878 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6880 * invoke.texi (Wsuggest-attribute=cold): Document.
6881 * common.opt (Wsuggest-attribute=cold): New
6882 * ipa-pure-const.c (warn_function_cold): New function.
6883 * predict.c (compute_function_frequency): Use it.
6884 * predict.h (warn_function_cold): Declare.
6886 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6888 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6891 2017-10-06 Martin Liska <mliska@suse.cz>
6893 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6894 keyword for member functions.
6895 (struct sanopt_tree_couple): New struct.
6896 (struct sanopt_tree_couple_hash): New function.
6897 (struct sanopt_ctx): Add new hash_map.
6898 (has_dominating_ubsan_ptr_check): New function.
6899 (record_ubsan_ptr_check_stmt): Likewise.
6900 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6901 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6902 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6904 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6907 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6908 aarch64_simd_valid_immediate on CONST_VECTORs.
6909 (aarch64_reg_or_bic_imm): Likewise.
6911 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6913 PR rtl-optimization/82396
6914 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6916 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6918 * graphite-dependences.c (scop_get_reads): Move code to...
6919 (scop_get_must_writes): Move code to...
6920 (scop_get_may_writes): Move code to...
6921 (scop_get_reads_and_writes): ... here.
6922 (scop_get_dependences): Call scop_get_reads_and_writes.
6924 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6926 PR tree-optimization/82434
6927 * fold-const.h (can_native_encode_type_p,
6928 can_native_encode_string_p): Remove.
6929 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6930 don't encode anything, just return what would be otherwise returned.
6931 (native_encode_fixed, native_encode_complex, native_encode_vector):
6933 (native_encode_string): Likewise. Inline by hand
6934 can_native_encode_string_p.
6935 (can_native_encode_type_p): Remove.
6936 (can_native_encode_string_p): Remove.
6937 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6938 STRING_CSTs using can_native_encode_string_p, test all
6939 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6940 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6941 argument from native_encode_expr.
6942 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6943 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6946 2017-10-06 Richard Biener <rguenther@suse.de>
6948 PR tree-optimization/82397
6949 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6950 operand_equal_p but rely on data_ref_compare_tree for detecting
6952 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6953 to match up with dr_group_sort_cmp.
6955 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6958 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6960 * config/s390/s390-builtin-types.def: Regenerate.
6962 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6965 * config/s390/s390-builtin-types.def: Regenerate.
6966 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6967 Change flag from B_VXE to B_VX.
6968 (s390_vec_min_dbl): Remove B_VXE flag.
6970 2017-10-06 Richard Biener <rguenther@suse.de>
6972 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6973 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6974 translate_isl_ast_to_gimple::is_valid_rename,
6975 translate_isl_ast_to_gimple::get_rename,
6976 translate_isl_ast_to_gimple::get_def_bb_for_const,
6977 translate_isl_ast_to_gimple::get_new_name,
6978 translate_isl_ast_to_gimple::collect_all_ssa_names,
6979 translate_isl_ast_to_gimple::copy_loop_phi_args,
6980 translate_isl_ast_to_gimple::collect_all_ssa_names,
6981 translate_isl_ast_to_gimple::copy_loop_phi_args,
6982 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6983 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6984 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6985 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6986 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6987 translate_isl_ast_to_gimple::copy_cond_phi_args,
6988 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6989 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6990 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6991 translate_isl_ast_to_gimple::rename_uses,
6992 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6993 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6994 (set_rename_for_each_def): Likewise.
6995 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6996 here. Handle rewriting SCEV analyzable uses here.
6997 (copy_bb_and_scalar_dependences): Generate code for PHI
6999 (graphite_regenerate_ast_isl): Adjust.
7000 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7001 (add_write, add_read): New functions.
7002 (build_cross_bb_scalars_def): Use it and simplify.
7003 (build_cross_bb_scalars_use): Likewise.
7004 (graphite_find_cross_bb_scalar_vars): Inline into...
7005 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7006 simulating out-of-SSA. Compute liveout and add dependencies.
7007 (build_scops): Force an empty entry block.
7008 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7010 (sese_build_liveouts): Declare.
7011 (sese_trivially_empty_bb_p): Likewise.
7012 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7013 compute liveout and debug_liveout.
7014 (sese_bad_liveouts_use): Remove.
7015 (sese_reset_debug_liveouts_bb): Likewise.
7016 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7017 (sese_build_liveouts): Build liveout and debug_liveout and store
7019 (new_sese_info): Adjust.
7020 (free_sese_info): Likewise.
7021 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7022 do not build liveout here.
7023 (move_sese_in_condition): Adjust region entry.
7024 (scev_analyzable_p): Match up with chrec_apply requirements.
7025 (sese_trivially_empty_bb_p): New.
7026 * tree-into-ssa.c (get_reaching_def): Properly support generating
7027 default-defs for incremental rewrite of anonymous names.
7029 2017-10-06 Richard Biener <rguenther@suse.de>
7031 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7032 precision do not perform modulo reduction.
7034 2017-10-06 Richard Biener <rguenther@suse.de>
7036 PR tree-optimization/82436
7037 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7038 conservatively choose the vectorization factor when checking
7039 whether we can perform the required load permutation.
7040 (vect_transform_slp_perm_load): Assert when we may not fail.
7042 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7044 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7045 message for incompatible -msdata=* and -mcall-* options.
7047 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7049 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7050 rate for post-reload scheduling.
7052 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7054 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7056 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7058 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7059 to improve monte carlo in scimark.
7061 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7063 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7064 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7065 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7066 pentium4_cost, nocona_cost): Set reassociation width to 1.
7067 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7068 width to 2 for fp operations and 1 otherwise.
7069 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7071 (atom_cost): Set reassociation width to 2.
7072 (slm_cost, generic_cost): Set fp reassociation width
7073 to 2 and 1 otherwise.
7074 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7075 (core_cost): Set fp reassociation width to 4 and vector to 2.
7076 (ix86_reassociation_width): Rewrite using cost table; special case
7077 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7078 and TARGET_AVX128_OPTIMAL.
7079 * config/i386/i386.h (processor_costs): Add
7080 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7081 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7082 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7083 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7084 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7085 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7087 2017-10-05 Nathan Sidwell <nathan@acm.org>
7089 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7091 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7093 * config/arm/arm.c (arm_test_fpu_data): New.
7094 (arm_run_selftests): Call arm_test_fpu_data.
7096 2017-10-04 Nathan Sidwell <nathan@acm.org>
7098 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7100 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7102 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7104 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7105 check type for aarch64_simd_valid_immediate.
7106 (aarch64_output_simd_mov_immediate): Update prototype.
7107 (aarch64_simd_valid_immediate): Update prototype.
7108 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7109 support for ORR-immediate.
7110 (and<mode>3): modified pattern to add support for BIC-immediate.
7111 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7112 now checks for valid immediate for BIC and ORR based on new enum
7114 (aarch64_output_simd_mov_immediate): Function now used to output
7115 BIC/ORR imm as well based on new enum argument.
7116 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7118 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7119 (aarch64_reg_or_bic_imm): Likewise.
7121 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7123 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7124 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7126 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7130 PR rtl-optimization/82396
7131 * haifa-sched.c (autopref_multipass_init): Simplify
7133 (autopref_rank_data): Simplify sort order.
7134 * sched-int.h (autopref_multipass_data_): Remove
7135 multi_mem_insn_p, min_offset and max_offset.
7137 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7139 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7141 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7143 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7146 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7148 * doc/sourcebuild.texi: Document vect_long_mult.
7150 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7152 PR tree-optimization/82413
7153 * fold-const.c (build_range_check): Use widest_int when comparing
7154 the maximum ETYPE value with HIGH.
7156 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7158 PR rtl-optimization/82396
7159 * haifa-sched.c (autopref_multipass_init): Simplify
7161 (autopref_rank_data): Simplify sort order.
7162 * sched-int.h (autopref_multipass_data_): Remove
7163 multi_mem_insn_p, min_offset and max_offset.
7165 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7167 PR tree-optimization/82381
7168 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7169 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7172 PR tree-optimization/82374
7173 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7174 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7175 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7176 current_function_decl to the new decl.
7178 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7180 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7181 helper macro for IEEE float128 hardware built-in functions.
7182 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7184 (TRUNCF128_ODD): Likewise.
7185 (ADDF128_ODD): Likewise.
7186 (SUBF128_ODD): Likewise.
7187 (MULF128_ODD): Likewise.
7188 (DIVF128_ODD): Likewise.
7189 (FMAF128_ODD): Likewise.
7190 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7191 UNSPEC_TRUNC_ROUND_TO_ODD.
7192 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7193 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7194 floating point round to odd instructions.
7195 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7196 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7197 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7198 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7199 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7200 (trunc<mode>sf2_hw): Change the truncate with round to odd
7201 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7202 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7203 to odd hardware instructions.
7204 (sub<mode>3_odd): Likewise.
7205 (mul<mode>3_odd): Likewise.
7206 (div<mode>3_odd): Likewise.
7207 (sqrt<mode>2_odd): Likewise.
7208 (fma<mode>4_odd): Likewise.
7209 (fms<mode>4_odd): Likewise.
7210 (nfma<mode>4_odd): Likewise.
7211 (nfms<mode>4_odd): Likewise.
7212 (trunc<mode>df2_odd): Change the truncate with round to odd
7213 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7215 * doc/extend.texi (PowerPC built-in functions): Update documentation
7216 for existing IEEE float128-bit built-in functions. Add built-in
7217 functions that generate the IEEE 128-bit floating point round to
7220 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7222 PR rtl-optimization/77729
7223 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7224 to (X&(C1&~C2))|C2 transformations.
7226 2017-10-03 Martin Jambor <mjambor@suse.cz>
7228 PR tree-optimization/82363
7229 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7230 mismatch, mark lacc written regardless of racc.
7232 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7234 PR tree-optimization/82381
7235 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7236 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7237 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7238 fallthrough into reassoc_stmt_dominates_stmt_p.
7241 * combine.c (combine_instructions): Don't combine in unreachable
7244 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7247 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7248 function to not use the have_cpu variable. Do not set cpu_index,
7249 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7251 (rs6000_valid_attribute_p): Remove duplicate initializations of
7252 old_optimize and func_optimize.
7253 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7254 (rs6000_activate_target_options): Make global.
7255 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7258 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7260 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7261 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7262 Set *poffset to 0 and *psize and *pmax_size to -1 if
7263 *poffset + *psize overflows in HOST_WIDE_INT.
7265 PR tree-optimization/82387
7266 PR tree-optimization/82388
7267 PR tree-optimization/82389
7268 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7269 instead of live_bytes non-NULL.
7271 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7274 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7277 2017-10-02 Richard Biener <rguenther@suse.de>
7279 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7280 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7281 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7283 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7285 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7286 requested precision matches the type's.
7287 * calls.c (alloc_max_size): Calculate the new candidate size as
7288 a widest_int and use wi::to_widest when comparing it with the
7289 current candidate size.
7290 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7291 zero rather than integer_zero_node.
7292 * match.pd: Check for a no-op conversion before using wi::add
7293 rather than after. Use tree_to_uhwi when summing small shift
7294 counts into an unsigned int.
7296 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7297 Alan Hayward <alan.hayward@arm.com>
7298 David Sherwood <david.sherwood@arm.com>
7301 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7302 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7303 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7304 POINTER_AND_FP_REGS.
7306 2017-10-02 Richard Biener <rguenther@suse.de>
7308 PR tree-optimization/82355
7309 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7310 a mapping for the enclosing loop but avoid generating one for
7312 (copy_bb_and_scalar_dependences): Remove premature codegen
7313 error on PHIs in blocks duplicated into multiple places.
7314 * graphite-scop-detection.c
7315 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7316 in the region use it as loop and nest to analyze the DR in.
7317 (try_generate_gimple_bb): Likewise.
7318 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7319 (add_loop_constraints): For blocks in a loop not in the region
7320 create a dimension with a single iteration.
7321 * sese.h (gbb_loop_at_index): Remove assert.
7323 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7325 * omp-expand.c (adjust_context_scope): New function.
7326 (expand_parallel_call): Call adjust_context_scope.
7328 2017-10-01 Jeff Law <law@redhat.com>
7330 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7331 dom_opt_dom_walker class with direct access to private members.
7332 Add comments. Call test_for_singularity.
7333 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7334 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7335 m_dummy_cond anymore.
7336 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7338 (pass_dominator:execute): Build the dummy_cond here and pass it
7339 to the dom_opt_dom_walker ctor.
7340 (test_for_singularity): New function.
7342 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7343 Maya Rashish <coypu@sdf.org>
7345 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7346 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7347 (alpha*-*-netbsd*) Use nbsd_tm_file.
7348 (arm*-*-netbsdelf*) Likewise.
7349 (i[34567]86-*-netbsdelf*) Likewise.
7350 (x86_64-*-netbsd*) Likewise.
7351 (mips*-*-netbsd*) Likewise.
7352 (powerpc-*-netbsd*) Likewise.
7353 (sh*-*-netbsd*) Likewise.
7354 (sparc-*-netbsdelf*) Likewise.
7355 (sparc64-*-netbsd*) Likewise.
7356 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7358 (vax-*-netbsdelf*) Likewise.
7359 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7360 (UINT_FAST8_TYPE) Likewise.
7361 (INT_FAST16_TYPE) Check CHAR_FAST16.
7362 (UINT_FAST16_TYPE) Likewise.
7364 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7367 * config/i386/i386.md
7368 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7369 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7370 *divmodsi4_zext_2): New define_insn_and_split.
7371 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7372 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7373 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7374 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7375 New define_insn_and_split.
7376 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7377 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7378 operands[1] having DImode when mode is SImode.
7380 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7381 always SImode for DIV and MOD in REG_EQUAL notes.
7383 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7386 * match.pd: Fix handling of NaNs in pattern.
7388 2017-09-29 Jeff Law <law@redhat.com>
7390 * sbitmap.c (bitmap_bit_in_range_p): New function.
7391 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7392 * tree-ssa-dse.c (live_bytes_read): New function.
7393 (dse_classify_store): Ignore reads of dead bytes.
7395 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7396 typos and whitespace errors.
7397 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7398 * config/s390/s390.c (s390_emit_prologue): Likewise.
7400 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7403 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7404 with a symbol for LRA.
7406 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7408 PR rtl-optimization/82338
7409 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7411 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7413 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7414 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7415 (qsort_chk): Declare.
7416 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7417 (qsort_chk): New function.
7419 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7421 PR tree-optimization/82337
7422 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7423 phi definition if the PHI result appears in an abnormal PHI.
7424 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7425 basis appears in an abnormal PHI.
7427 2017-09-29 Richard Biener <rguenther@suse.de>
7429 * graphite-isl-ast-to-gimple.c
7430 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7431 (binary_op_to_tree): Use it.
7432 (get_rename_from_scev): Likewise.
7433 (copy_loop_phi_nodes): Likewise.
7434 (copy_bb_and_scalar_dependences): Likewise.
7435 (translate_pending_phi_nodes): Likewise.
7437 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7440 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7441 for movabsq $(i32 << shift), r64.
7443 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7445 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7446 index when encoding %esp as %rsp to avoid 0x67 prefix.
7448 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7450 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7451 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7453 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7455 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7456 Extensions with more than 16 double VFP registers.
7457 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7458 to_clear_mask and all code related to it. Replace the remaining
7459 entry by a sbitmap and adapt code accordingly.
7461 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7463 * brig-builtins.def: Change pure attributes to const.
7465 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7467 * config.gcc (default_gnu_indirect_function): Default to yes for
7468 sparc*-*-linux* with glibc.
7470 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7472 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7473 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7474 when creating .init_array and .fini_array sections with priority
7477 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7480 * config/aarch64/aarch64.c
7481 (aarch64_builtin_support_vector_misalignment): Always return false
7482 when misalignment is unknown.
7484 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7486 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7487 this function to return false if the definition used by the swap
7488 instruction is artificial, or if the memory address from which the
7489 constant value is loaded is not represented by a base address held
7490 in a register or if the base address register is a frame or stack
7491 pointer. Additionally, return false if the base address of the
7492 loaded constant is a SYMBOL_REF but is not considered to be a
7494 (replace_swapped_load_constant): New function.
7495 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7496 loaded constant vector with a load of a swapped constant vector.
7498 2017-09-27 Carl Love <cel@us.ibm.com>
7500 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7501 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7502 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7505 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7507 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7508 first, always call autopref_rank_data otherwise.
7510 2017-09-27 Richard Biener <rguenther@suse.de>
7512 * graphite-scop-detection.c (find_scop_parameters): Move
7513 loop bound handling ...
7514 (gather_bbs::before_dom_children): ... here, avoiding the need
7515 to build scop_info->loop_nest.
7516 (record_loop_in_sese): Remove.
7517 * sese.h (sese_info_t::loop_nest): Remove.
7518 * sese.c (new_sese_info): Do not allocate loop_nest.
7519 (free_sese_info): Do not free loop_nest.
7521 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7524 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7525 lhs from calls if the lhs has addressable type.
7527 2017-09-27 Richard Biener <rguenther@suse.de>
7529 * graphite.h (scop::max_alias_set): New member.
7530 * graphite-scop-detection.c: Remove references to non-existing
7531 --param in comments.
7532 (build_alias_sets): Record the maximum alias set used for drs.
7533 (build_scops): Support zero as unlimited for
7534 --param graphite-max-arrays-per-scop.
7535 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7537 (build_poly_sr_1): ... here. Compute alias set based on the
7538 maximum alias set used for drs rather than
7539 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7541 2017-09-27 Richard Biener <rguenther@suse.de>
7543 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7544 --param loop-block-tile-size=0 to disable tiling.
7546 2017-09-27 Richard Biener <rguenther@suse.de>
7548 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7549 (graphite-max-nb-scop-params): Document special value zero.
7550 * domwalk.h (dom_walker::STOP): New symbolical constant.
7551 (dom_walker::dom_walker): Add optional parameter for bb to
7553 (dom_walker::~dom_walker): Declare.
7554 (dom_walker::before_dom_children): Document STOP return value.
7555 (dom_walker::m_user_bb_to_rpo): New member.
7556 (dom_walker::m_bb_to_rpo): Likewise.
7557 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7558 mapping here if not provided by the user.
7559 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7560 provided by the user.
7561 (dom_walker::STOP): Define.
7562 (dom_walker::walk): Do not compute bb to RPO mapping here.
7563 Support STOP return value from before_dom_children to stop
7565 * graphite-optimize-isl.c (optimize_isl): If the schedule
7566 is the same still generate code if -fgraphite-identity
7567 or -floop-parallelize-all are given.
7568 * graphite-scop-detection.c: Include cfganal.h.
7569 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7571 (gather_bbs::before_dom_children): Return STOP for BBs
7573 (build_scops): Compute bb to RPO mapping and pass it to
7574 the domwalk. Treat --param graphite-max-nb-scop-params=0
7575 as not limiting the number of params.
7576 * graphite.c (graphite_initialize): Remove limit on the
7577 number of basic-blocks in a function.
7578 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7579 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7580 default value of 10.
7582 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7584 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7585 Adjust code to eliminate needing to do the shift right 32-bits
7586 operation after XSCVDPSPN.
7588 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7590 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7591 ((X / Y) != 0 -> X >= Y): Likewise.
7593 2017-09-26 Carl Love <cel@us.ibm.com>
7595 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7596 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7597 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7598 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7599 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7600 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7601 definitions and overloading.
7602 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7603 statement for P9V_BUILTIN_XST_LEN_R.
7604 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7605 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7606 define_expand and define_insn for the instructions and builtins.
7607 * doc/extend.texi: Update the built-in documentation file for the new
7609 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7610 define_insn for the instructions
7612 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7615 * gcc/config/netbsd-protos.h: New file.
7616 * gcc/config/netbsd.c: New file.
7617 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7618 * gcc/config/t-netbsd: New file.
7619 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7620 (tmake_file) Add t-netbsd.
7621 (extra_objs) Add netbsd.o.
7623 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7627 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7629 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7631 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7632 sign extension from a vector register to a GPR by doing a 32-bit
7633 direct move and then an EXTSW.
7634 (extendsi<mode>2 splitter): Likewise.
7635 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7636 right or vector extract after doing XSCVDPSPN. Use
7637 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7639 (movdi_from_sf_zero_ext): Likewise.
7640 (reload_gpr_from_vsxsf): Likewise.
7641 (p8_mfvsrd_4_disf): Delete, no longer used.
7642 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7643 then needing to move the SFmode to a GPR to use the XSCVDPSP
7644 instruction instead of FRSP and XSCVDPSPN.
7645 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7646 it is adjacent to the other XSCVSPDP insns.
7647 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7648 SFmode to be in traditional Altivec registers.
7649 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7650 (vsx_xscvspdpn): Likewise.
7651 (vsx_xscvspdpn_scalar): Likewise.
7653 2017-09-26 Martin Jambor <mjambor@suse.cz>
7655 * tree-sra.c (compare_access_positions): Put integral types first,
7656 stabilize sorting of integral types, remove conditions putting
7657 non-full-precision integers last.
7658 (sort_and_splice_var_accesses): Disable scalarization if a
7659 non-integert would be represented by a non-full-precision integer.
7661 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7663 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7664 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7665 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7666 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7667 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7668 conditionals inside the function instead of around it. Call
7669 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7670 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7672 2017-09-26 Richard Biener <rguenther@suse.de>
7674 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7676 (scop_detection::build_scop_breadth): ... this. Removed.
7677 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7678 (scop_detection::harmful_stmt_in_bb): Likewise.
7679 (scop_detection::graphite_can_represent_stmt): Likewise.
7680 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7681 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7682 (scop_detection::can_represent_loop_1): ... this. Removed.
7683 (scop_detection::harmful_loop_in_region): Simplify after inlining
7684 the above and remove more quadraticness.
7685 (build_scops): Adjust.
7686 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7689 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7692 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7693 REGNO (base) == SP_REG if base is a REG.
7696 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7697 if it is different SSA_NAME.
7698 (optimize_range_tests_cmp_bitwise): New function.
7699 (optimize_range_tests): Call it.
7701 2017-09-26 Richard Biener <rguenther@suse.de>
7703 PR tree-optimization/82321
7704 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7705 for the def being inside the loop.
7707 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7709 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7711 * config/s390/s390-builtins.def: Fix constraint on op4.
7713 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7715 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7716 independent expanders.
7717 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7718 ("vec_ordered", "vec_unordered"): New expanders.
7720 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7722 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7725 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7727 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7728 vec_unpacks_lo_v16qi.
7729 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7731 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7733 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7734 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7735 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7737 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7739 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7741 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7743 ("*vec_slb<mode>"): New insn pattern.
7744 ("vec_shr_<mode>"): New expander.
7745 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7746 and force the shift count operand to V16QImode.
7747 ("vec_srb<mode>"): Set shift count mode to V16QI.
7749 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7751 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7752 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7753 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7755 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7758 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7760 2017-09-26 Richard Biener <rguenther@suse.de>
7762 PR tree-optimization/82320
7763 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7766 2017-09-25 Jeff Law <law@redhat.com>
7768 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7769 prototype for new argument.
7770 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7771 mostly extracted from rs6000_emit_allocate_stack.
7772 (rs6000_emit_probe_stack_range_stack_clash): New function.
7773 (rs6000_emit_allocate_stack): Call
7774 rs6000_emit_probe_stack_range_stack_clash as needed.
7775 (rs6000_emit_probe_stack_range): Add additional argument
7776 to call to gen_probe_stack_range{si,di}.
7777 (output_probe_stack_range): New.
7778 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7779 (output_probe_stack_range_stack_clash): New.
7780 (rs6000_emit_prologue): Emit notes into dump file as requested.
7781 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7782 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7783 Add additional operand and pass it to output_probe_stack_range.
7785 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7787 PR tree-optimization/82163
7788 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7789 (checking_verify_loop_closed_ssa): New parameter.
7790 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7791 (check_loop_closed_ssa_stmt): Delete.
7792 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7793 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7794 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7797 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7799 * brig-builtins.def: Treat HSAIL barrier builtins as
7800 setjmp/longjump style functions.
7802 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7804 * target.def (constant_alignment): New hook.
7805 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7806 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7807 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7808 * doc/tm.texi: Regenerate.
7809 * targhooks.h (default_constant_alignment): Declare.
7810 (constant_alignment_word_strings): Likewise.
7811 * targhooks.c (default_constant_alignment): New function.
7812 (constant_alignment_word_strings): Likewise.
7813 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7814 instead of CONSTANT_ALIGNMENT.
7815 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7816 (force_const_mem): Likewise.
7817 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7818 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7819 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7820 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7821 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7823 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7824 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7825 constant_alignment_word_strings.
7826 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7827 (CONSTANT_ALIGNMENT): Likewise.
7828 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7829 (arm_constant_alignment): New function.
7830 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7831 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7832 constant_alignment_word_strings.
7833 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7834 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7835 constant_alignment_word_strings.
7836 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7837 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7838 (cris_constant_alignment): New function.
7839 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7840 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7841 (epiphany_constant_alignment): New function.
7842 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7843 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7844 constant_alignment_word_strings.
7845 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7846 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7847 constant_alignment_word_strings.
7848 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7849 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7850 constant_alignment_word_strings.
7851 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7852 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7853 * config/i386/i386.c (ix86_constant_alignment): Make static.
7854 Use the same interface as the target hook.
7855 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7856 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7857 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7858 constant_alignment_word_strings.
7859 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7860 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7861 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7862 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7863 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7864 constant_alignment_word_strings.
7865 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7866 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7867 constant_alignment_word_strings.
7868 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7869 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7870 constant_alignment_word_strings.
7871 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7872 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7874 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7875 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7876 * config/mips/mips.c (mips_constant_alignment): New function.
7877 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7878 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7879 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7880 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7881 (mmix_constant_alignment): Make static. Use the same interface
7883 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7884 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7885 constant_alignment_word_strings.
7886 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7887 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7888 constant_alignment_word_strings.
7889 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7890 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7891 constant_alignment_word_strings.
7892 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7893 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7894 (rs6000_constant_alignment): New function.
7895 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7896 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7897 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7898 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7899 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7900 (rs6000_constant_alignment): New function.
7901 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7902 * config/s390/s390.c (s390_constant_alignment): New function.
7903 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7904 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7905 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7906 constant_alignment_word_strings.
7907 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7908 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7909 (sparc_constant_alignment): New function.
7910 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7911 * config/spu/spu.c (spu_constant_alignment): New function.
7912 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7913 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7914 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7915 constant_alignment_word_strings.
7916 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7917 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7918 constant_alignment_word_strings.
7919 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7920 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7921 constant_alignment_word_strings.
7922 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7923 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7924 (visium_constant_alignment): New function.
7925 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7926 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7927 (xtensa_constant_alignment): New function.
7928 * system.h (CONSTANT_ALIGNMENT): Poison.
7930 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7932 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7933 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7934 (rs6000_builtin_valid_without_lhs): New helper function.
7935 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7936 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7938 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7940 * target.h (vec_perm_indices): Use unsigned short rather than
7942 (auto_vec_perm_indices): Likewise.
7943 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7944 Use unsigned int rather than unsigned char.
7945 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7947 2017-09-25 Richard Biener <rguenther@suse.de>
7949 * cfgloop.h (sort_sibling_loops): Declare.
7950 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7951 (sort_sibling_loops): New function sorting the sibling loop list
7953 * graphite.c (graphite_transform_loops): Sort sibling loops.
7955 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7957 * target.def (vec_perm_const_ok): Change sel parameter to
7959 * optabs-query.c (can_vec_perm_p): Update accordingly.
7960 * doc/tm.texi: Regenerate.
7961 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7962 auto_vec_perm_indices and remove separate nelt field.
7963 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7964 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7965 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7966 (aarch64_expand_vec_perm_const): Update accordingly.
7967 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7968 to vec_perm_indices.
7969 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7970 auto_vec_perm_indices and remove separate nelt field.
7971 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7972 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7973 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7975 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7976 to vec_perm_indices.
7977 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7978 sel to vec_perm_indices.
7979 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7980 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7981 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7983 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7986 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7989 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7990 on the FUNCTION_DECL function context if it has a DIE that is a
7993 2017-09-25 Richard Biener <rguenther@suse.de>
7995 PR tree-optimization/82285
7996 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7999 2017-09-25 Tom de Vries <tom@codesourcery.com>
8003 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8006 2017-09-25 Richard Biener <rguenther@suse.de>
8008 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8009 ISL errors other than isl_error_quota happen. Dump if the
8010 schedule is the same.
8011 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8012 errors instead of aborting inside ISL.
8014 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8017 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8018 of libgcc_eh for m64.
8019 * config/i386/darwin64.h: Likewise.
8021 2017-09-25 Richard Biener <rguenther@suse.de>
8024 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8025 attribute for incomplete types nor twice for complete ones.
8027 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8030 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8031 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8034 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8037 * config/i386/i386.opt (mprefer-avx256): Use
8038 ix86_target_flags variable.
8039 * config/i386/i386.c (ix86_target_string): Move
8040 -mprefer-avx256 to flag2_opts.
8042 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8045 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8046 and x != -1 | y != -1 into (x & y) != -1.
8048 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8050 * config.gcc: Add new case statement to set
8051 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8052 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8053 s390x-*-linux* case statements. Added aarch64 to the list of
8054 supported architectures.
8056 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8058 PR tree-optimization/82289
8059 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8060 STMT_VINFO_RELEVANT_P.
8062 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8063 Alan Hayward <alan.hayward@arm.com>
8064 David Sherwood <david.sherwood@arm.com>
8066 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8067 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8068 (extract_range_from_binary_expr_1): Don't call
8069 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8071 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8072 Alan Hayward <alan.hayward@arm.com>
8073 David Sherwood <david.sherwood@arm.com>
8075 * target.def (preferred_vector_alignment): New hook.
8076 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8078 * doc/tm.texi: Regenerate.
8079 * targhooks.h (default_preferred_vector_alignment): Declare.
8080 * targhooks.c (default_preferred_vector_alignment): New function.
8081 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8083 (DR_TARGET_ALIGNMENT): New macro.
8084 (aligned_access_p): Update commentary.
8085 (vect_known_alignment_in_bytes): New function.
8086 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8088 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8089 Calculate the misalignment based on the target alignment rather than
8091 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8092 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8093 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8094 the target alignment, rather than masking the element misalignment
8095 with the number of elements in a vector. Also use the target
8096 alignment when calculating the maximum number of peels.
8097 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8098 instead of TYPE_ALIGN_UNIT.
8099 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8100 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8101 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8102 (vect_create_data_ref_ptr): Likewise.
8103 (vect_setup_realignment): Realign by ANDing with
8104 -DR_TARGET_MISALIGNMENT.
8105 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8106 the number of peels based on DR_TARGET_ALIGNMENT.
8107 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8108 with the guaranteed alignment boundary when deciding whether
8110 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8111 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8112 (ensure_base_align): Remove stmt_info parameter. Get the
8113 target base alignment from DR_TARGET_ALIGNMENT.
8114 (vectorizable_store): Update call accordingly. Interpret
8115 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8117 (vectorizable_load): Likewise.
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 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8124 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8125 (vect_enhance_data_refs_alignment): Likewise.
8127 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8129 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8130 error. Only quit immediately if parsing is complete.
8131 (BEGIN): Initialize fatal_err and parse_done.
8132 (begin fpu, end fpu): Check number of arguments.
8133 (begin arch, end arch): Likewise.
8134 (begin cpu, end cpu): Likewise.
8135 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8136 (optalias): Likewise.
8138 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8140 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8141 * config/arm/arm-isa.h: Delete. Move definitions to ...
8142 * arm-cpus.in: ... here. Use new feature and fgroup values.
8143 * config/arm/arm.c (arm_option_override): Use lower case for feature
8145 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8146 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8147 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8149 (print_isa_bits_for): New function.
8150 (gen_isa): New function.
8151 (gen_comm_data): Use print_isa_bits_for.
8152 (define feature): New keyword.
8153 (define fgroup): New keyword.
8154 * config/arm/t-arm (TM_H): Remove.
8155 (GTM_H): Add arm-isa.h.
8156 (arm-isa.h): Add rule to generate file.
8157 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8158 case for feature bit names.
8160 2017-09-22 Richard Biener <rguenther@suse.de>
8162 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8164 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8165 print of no dependency loops ...
8166 * graphite.c (graphite_transform_loops): ... here.
8167 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8169 (same_close_phi_node, remove_duplicate_close_phi,
8170 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8171 (canonicalize_loop_closed_ssa): ... here and simplify.
8172 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8173 (optimize_isl): Use dump_printf_loc to tell when we stopped
8174 optimizing because of an ISL timeout.
8176 2017-09-22 Richard Biener <rguenther@suse.de>
8178 PR tree-optimization/82291
8179 * tree-if-conv.c (predicate_mem_writes): Make sure to
8180 remove writes in blocks predicated with false.
8182 2017-09-22 Richard Biener <rguenther@suse.de>
8184 * sese.c: Include cfganal.h.
8185 (if_region_set_false_region): Remove.
8186 (create_if_region_on_edge): Likewise.
8187 (move_sese_in_condition): Re-implement without destroying
8190 2017-09-22 Richard Biener <rguenther@suse.de>
8192 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8193 Verify both BBs contain loop PHI nodes before dispatching to
8195 (graphite_regenerate_ast_isl): Do not recompute dominators,
8196 do not verify three times. Restructure for clarity.
8197 * graphite-scop-detection.c (same_close_phi_node,
8198 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8199 defined_in_loop_p, canonicalize_loop_closed_ssa,
8200 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8201 checking and SSA rewrite, move to ...
8202 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8203 (graphite_initialize): Do not pass in ctx, do not reset the
8204 SCEV cache, compute only dominators.
8205 (graphite_transform_loops): Allocate ISL ctx after
8206 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8207 Maintain post-dominators only around build_scops.
8208 * sese.c (if_region_set_false_region): Make static. Free
8209 and recompute dominators.
8210 (move_sese_in_condition): Assert we don't get called with
8211 post-dominators computed.
8212 * sese.h (if_region_set_false_region): Remove.
8214 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8216 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8217 mode attribute for TARGET_AVX512VL.
8219 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8221 * config/i386/i386.opt (mprefer-avx256): New option.
8222 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8224 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8225 for TARGET_PREFER_AVX256.
8226 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8228 2017-09-21 Jeff Law <law@redhat.com>
8230 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8231 Fix dump output if the only stack space is for pushed registers.
8233 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8235 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8238 2017-09-21 Martin Sebor <msebor@redhat.com>
8241 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8242 code (in C++) or code that triggers warnings.
8244 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8246 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8248 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8250 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8251 * sched-rgn.c: Ditto.
8252 * sel-sched-ir.c: Ditto.
8254 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8256 * toplev.h (set_random_seed): Adjust return type.
8257 * toplev.c (init_local_tick): Move eager initialization of random_seed
8258 to get_random_seed. Adjust comment.
8259 (init_random_seed): Inline to get_random_seed, delete.
8260 (get_random_seed): Initialize random_seed lazily.
8261 (set_random_seed): Do not return previous value.
8262 (print_switch_value): Do not call get_random_seed.
8264 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8266 * cgraph.c (delete_function_version): New, broken out from...
8267 (cgraph_node::delete_function_version): ...here. Rename to
8268 cgraph_node::delete_function_version_by_decl. Update all uses.
8269 (cgraph_node::remove): Call delete_function_version.
8271 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8274 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8275 VAR_DECLs to which addressable non-volatile parameters are mapped
8276 and for id->retvar after the return value assignment. Clear
8277 id->retval and id->retbnd after inlining.
8279 2017-09-21 Richard Biener <rguenther@suse.de>
8281 PR tree-optimization/82276
8282 PR tree-optimization/82244
8283 * tree-vrp.c (build_assert_expr_for): Set
8284 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8286 (remove_range_assertions): Revert earlier change.
8288 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8291 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8293 2017-09-21 Richard Biener <rguenther@suse.de>
8295 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8296 Restore valid IL after code generation errors.
8297 * graphite.c (graphite_transform_loops): Diagnose code
8298 generation issues as MSG_MISSED_OPTIMIZATION and continue
8299 with processing SCOPs.
8301 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8302 Alan Hayward <alan.hayward@arm.com>
8303 David Sherwood <david.sherwood@arm.com>
8305 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8306 rather than choosing between plus_constant and gen_rtx_<CODE>.
8307 * expr.c (emit_push_insn): Likewise.
8308 (expand_expr_real_2): Likewise.
8310 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8311 Alan Hayward <alan.hayward@arm.com>
8312 David Sherwood <david.sherwood@arm.com>
8314 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8316 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8317 Alan Hayward <alan.hayward@arm.com>
8318 David Sherwood <david.sherwood@arm.com>
8320 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8321 calling tree_to_uhwi.
8323 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8324 Alan Hayward <alan.hayward@arm.com>
8325 David Sherwood <david.sherwood@arm.com>
8327 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8328 INTEGER_CST rather than a negative test for ADDR_EXPR.
8330 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8335 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8337 2017-09-21 Richard Biener <rguenther@suse.de>
8339 PR tree-optimization/71351
8340 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8341 graphite_create_new_loop_guard): Remove, fold remaining parts
8343 (translate_isl_ast_node_for): ... here and simplify.
8345 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8348 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8349 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8350 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8351 alternative always use QI mode, for -Os imov (=R,R) alternative
8352 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8355 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8356 Jeff Law <law@redhat.com>
8358 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8359 (allocate_stack_space): New function, partially extracted from
8361 (s390_emit_prologue): Track offset to most recent stack probe.
8362 Code to allocate space moved into allocate_stack_space.
8363 Dump actions when no stack is allocated.
8364 (s390_prologue_plus_offset): New function.
8365 (s390_emit_stack_probe): Likewise.
8367 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8369 * common.opt (Wa, Wl, Wp, g, gz=): Add
8371 (gno-column-info): Remove.
8372 (gcolumn-info): Drop RejectNegative.
8374 (gno-record-gcc-switches): Remove.
8375 (grecord-gcc-switches): Drop RejectNegative.
8376 (gno-split-dwarf): Remove.
8377 (gsplit-dwarf): Drop RejectNegative.
8378 (gno-strict-dwarf): Remove.
8379 (gstrict-dwarf): Drop RejectNegative.
8380 * config/darwin.opt (gfull, gused): Add RejectNegative.
8381 * dwarf2out.c (gen_producer_string): Drop
8382 gno-record-gcc-switches handler.
8383 * optc-gen.awk: Add g to prefixes with negative forms.
8384 * opts-common.c (remapping_prefix_p): New.
8385 (find_opt): Check it.
8386 (generate_canonical_option): Test g prefix.
8387 (option_map): Add -gno- mapping.
8388 (add_misspelling_candidates): Check remapping_prefix_p.
8390 2017-09-20 Jeff Law <law@redhat.com>
8392 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8393 thinko in stack clash protection support.
8395 * explow.c (compute_stack_clash_protection_loop_data): Use
8396 CONST_INT_P instead of explicit test. Verify object is a
8397 CONST_INT_P before looking at INTVAL.
8398 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8399 instead of explicit test.
8401 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8404 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8405 address instead of to r1 and r11.
8407 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8409 * config.gcc: Support "knm".
8410 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8411 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8413 * config/i386/i386.c (m_KNM): Define.
8414 (processor_target_table): Add "knm".
8416 (ix86_option_override_internal): Add "knm".
8417 (ix86_issue_rate): Add PROCESSOR_KNM.
8418 (ix86_adjust_cost): Ditto.
8419 (ia32_multipass_dfa_lookahead): Ditto.
8420 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8421 (fold_builtin_cpu): Add M_INTEL_KNM.
8422 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8423 (processor_type): Add PROCESSOR_KNM.
8424 * config/i386/x86-tune.def: Add m_KNM.
8425 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8427 2017-09-20 Richard Biener <rguenther@suse.de>
8429 PR tree-optimization/80213
8430 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8431 are allowed in empty BBs as well.
8432 (canonicalize_loop_closed_ssa): Also look for other complex
8434 (scop_detection::get_sese): Include the loop-closed PHI block
8436 (scop_detection::merge_sese): Remove code adding extra blocks.
8437 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8438 (build_scops): Assert the final returned scop is invalid.
8440 2017-09-20 Richard Biener <rguenther@suse.de>
8442 PR tree-optimization/82264
8443 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8445 (vn_phi_lookup): Likewise.
8446 (vn_phi_insert): Likewise.
8448 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8450 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8451 that fits into uhwi or shwi, add DW_AT_const_value regardless
8452 of early_dwarf without going through RTL, using add_AT_unsigned
8455 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8456 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8457 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8458 (DEBUG_MACRO_SECTION): Likewise.
8459 (DEBUG_LTO_MACRO_SECTION): Likewise.
8460 (DEBUG_STR_DWO_SECTION): Likewise.
8461 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8462 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8463 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8464 (DEBUG_LTO_LINE_STR_SECTION): Define.
8465 (init_sections_and_labels): Initialize debug_line_str_section
8466 variable. Initialize debug_loc_section for -gdwarf-5 to
8467 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8469 2017-09-20 Richard Biener <rguenther@suse.de>
8471 * graphite-sese-to-poly.c (extract_affine): Properly handle
8472 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8474 2017-09-20 Richard Biener <rguenther@suse.de>
8476 PR tree-optimization/81373
8477 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8478 Force SESE live-out defs to be handled even if they are
8481 2017-09-19 Jeff Law <law@redhat.com>
8483 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8484 nothing for stack adjustments with REG_STACK_CHECK.
8485 * sched-deps.c (parse_add_or_inc): Reject insns with
8486 REG_STACK_CHECK from dependency breaking.
8487 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8488 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8489 * reg-notes.def (STACK_CHECK): New note.
8491 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8492 (ix86_expand_prologue): Dump stack clash info as needed.
8493 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8495 * function.c (dump_stack_clash_frame_info): New function.
8496 * function.h (dump_stack_clash_frame_info): Prototype.
8497 (enum stack_clash_probes): New enum.
8499 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8500 flag_stack_clash_protection.
8501 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8502 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8503 (arm_frame_pointer_required): Likewise.
8504 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8505 (ia64_expand_prologue): Likewise.
8506 * config/mips/mips.c (mips_expand_prologue): Likewise.
8507 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8508 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8509 (sparc_flat_expand_prologue): Likewise.
8510 * config/spu/spu.c (spu_expand_prologue): Likewise.
8512 * explow.c: Include "params.h".
8513 (anti_adjust_stack_and_probe_stack_clash): New function.
8514 (get_stack_check_protect): Likewise.
8515 (compute_stack_clash_protection_loop_data): Likewise.
8516 (emit_stack_clash_protection_loop_start): Likewise.
8517 (emit_stack_clash_protection_loop_end): Likewise.
8518 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8519 Use anti_adjust_stack_and_probe_stack_clash.
8520 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8521 (emit_stack_clash_protection_loop_start): Likewise.
8522 (emit_stack_clash_protection_loop_end): Likewise.
8523 * rtl.h (get_stack_check_protect): Prototype.
8524 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8525 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8526 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8528 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8530 * doc/tm.texi: Rebuilt.
8531 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8532 get_stack_check_protect.
8533 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8534 * config/arm/arm.c (arm_expand_prologue): Likewise.
8535 (arm_frame_pointer_required): Likewise.
8536 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8537 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8538 * config/mips/mips.c (mips_expand_prologue): Likewise.
8539 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8540 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8541 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8542 (sparc_flat_expand_prologue): Likewise.
8544 * common.opt (-fstack-clash-protection): New option.
8545 * flag-types.h (enum stack_check_type): Note difference between
8546 -fstack-check= and -fstack-clash-protection.
8547 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8548 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8549 * toplev.c (process_options): Issue warnings/errors for cases
8550 not handled with -fstack-clash-protection.
8551 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8552 (-fstack-check): Note additional problem with -fstack-check=generic.
8553 Note that -fstack-check is primarily for Ada and refer users
8554 to -fstack-clash-protection for stack-clash-protection.
8555 Document new params for stack clash protection.
8557 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8559 * config/i386/i386.c (ix86_split_long_move): Do not handle
8560 address used for LEA in a special way.
8562 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8564 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8566 2017-09-19 Martin Sebor <msebor@redhat.com>
8569 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8570 of incompatible types.
8572 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8574 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8575 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8576 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8577 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8579 2017-09-19 Richard Biener <rguenther@suse.de>
8581 PR tree-optimization/82244
8582 * tree-vrp.c (remove_range_assertions): Do not propagate
8583 a constant to abnormals but replace the assert with a copy.
8585 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8587 PR rtl-optimization/57878
8588 PR rtl-optimization/68988
8589 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8590 avoidance test involving non_reload_pseudos. Move frequency test
8591 below the general fragmentation avoidance test.
8593 2017-09-19 Richard Biener <rguenther@suse.de>
8595 PR tree-optimization/69728
8596 * graphite-sese-to-poly.c (schedule_error): New global.
8597 (add_loop_schedule): Handle empty domain by failing the
8599 (build_original_schedule): Handle schedule_error.
8601 2017-09-19 Richard Biener <rguenther@suse.de>
8603 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8604 Do not iterate to sibling loops but only to siblings of inner
8607 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8610 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8611 that the registers are different.
8613 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8615 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8616 to processor_model and "amdfam17h" to arch_names_table.
8617 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8619 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8622 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8624 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8625 Alan Hayward <alan.hayward@arm.com>
8626 David Sherwood <david.sherwood@arm.com>
8628 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8630 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8632 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8633 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8634 vect_schedule_slp_instance.
8635 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8636 Update call to vect_slp_analyze_node_operations. Simplify return
8638 (vect_slp_analyze_bb_1): Update call accordingly.
8639 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8640 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8641 (vect_schedule_slp): Update call accordingly.
8643 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8644 Alan Hayward <alan.hayward@arm.com>
8645 David Sherwood <david.sherwood@arm.com>
8647 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8648 with types that aren't in fact scalar.
8650 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8652 * tree-vect-slp.c (vect_record_max_nunits): New function,
8654 (vect_build_slp_tree_1): ...here.
8655 (vect_build_slp_tree_2): Call it for phis too.
8657 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8659 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8660 to vect_get_vec_def_for_operand when getting the mask operand.
8662 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8663 Alan Hayward <alan.hayward@arm.com>
8664 David Sherwood <david.sherwood@arm.com>
8666 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8669 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8670 Alan Hayward <alan.hayward@arm.com>
8671 David Sherwood <david.sherwood@arm.com>
8673 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8674 calculation for vector booleans.
8676 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8677 Alan Hayward <alan.hayward@arm.com>
8678 David Sherwood <david.sherwood@arm.com>
8680 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8682 (vect_transform_stmt): ...here.
8683 (vect_analyze_stmt): Use it instead of calling
8684 vectorizable_live_operation directly.
8686 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8688 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8689 non-SIMT targets in acc vector loops.
8691 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8693 * configure.ac: Add arc and check if assembler supports gdwarf2.
8694 * configure: Regenerate.
8696 2017-09-18 Richard Biener <rguenther@suse.de>
8698 PR tree-optimization/82220
8699 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8700 epilogue niters from the min_profitable_iters compute.
8702 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8705 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8706 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8707 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8709 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8712 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8713 switching to a new text section.
8715 2017-09-18 Richard Biener <rguenther@suse.de>
8717 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8719 (build_alias_set): Reject aliases with no access function.
8721 2017-09-18 Richard Biener <rguenther@suse.de>
8723 PR tree-optimization/79622
8724 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8726 (build_cross_bb_scalars_use): Likewise.
8728 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8730 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8732 2017-09-18 Alan Modra <amodra@gmail.com>
8735 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8736 stack_pointer_rtx for count 0. Update comments. Break up
8737 large rtl expression.
8739 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8741 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8742 Increase to 20 bytes.
8743 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8744 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8745 or avx version of the stub.
8747 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8750 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8751 compute the minimum stack alignment. Also update preferred stack
8752 boundary for leaf functions.
8754 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8756 PR tree-optimization/82228
8757 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8760 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8762 * common/config/nds32/nds32-common.c
8763 (nds32_option_optimization_table): Refine formatting.
8764 (nds32_option_optimization_table): Use -fsched-pressure and
8765 -fomit-frame-pointer for specific optimization level.
8767 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8769 * config/nds32/nds32.c: Refine formatting and comments.
8770 * config/nds32/nds32.h: Likewise.
8771 * config/nds32/nds32.md: Likewise.
8772 * config/nds32/nds32-cost.c: Likewise.
8773 * config/nds32/nds32-isr.c: Likewise.
8774 * config/nds32/nds32-md-auxiliary.c: Likewise.
8775 * config/nds32/nds32-multiple.md: Likewise.
8776 * config/nds32/nds32-predicates.c: Likewise.
8778 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8779 Jakub Jelinek <jakub@redhat.com>
8781 Add support for -std=c++2a.
8782 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8784 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8786 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8789 * doc/extend.texi (Common Function Attributes): Add
8790 references to ARM, AArch64, and S/390 specific attributes.
8791 (Function Specific Option Pragmas): Add AArch64 and S/390
8792 to list of back ends that support the target pragma.
8794 2017-09-15 Nathan Sidwell <nathan@acm.org>
8796 * doc/standards.texi: Fix C++17 description. Update URLs for
8799 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8801 * common.opt (Wcast-align=strict): New warning option.
8802 * doc/invoke.texi: Document -Wcast-align=strict.
8804 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8806 * cgraph.h (cgraph_thunk_info): Add comments.
8807 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8808 assert for VIRTUAL_* arguments stricter.
8810 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8812 PR tree-optimization/71026
8813 * match.pd: Move RDIV patterns from fold-const.c
8814 * fold-const.c (distribute_real_division): Removed.
8815 (fold_binary_loc): Remove calls to distribute_real_divison.
8817 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8819 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8820 c++1z and gnu++1z as deprecated. Change other references to
8821 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8822 Change -Wc++1z-compat to -Wc++17-compat.
8823 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8824 * dwarf2out.c (highest_c_language): Handle C++17.
8825 (gen_compile_unit_die): Likewise.
8827 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8829 PR rtl-optimization/82192
8830 * combine.c (make_extraction): Don't look through non-paradoxical
8831 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8834 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8835 Alan Hayward <alan.hayward@arm.com>
8836 David Sherwood <david.sherwood@arm.com>
8838 * target.def (function_arg_offset): New hook.
8839 * targhooks.h (default_function_arg_offset): Declare.
8840 * targhooks.c (default_function_arg_offset): New function.
8841 * function.c (locate_and_pad_parm): Use
8842 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8843 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8844 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8845 * doc/tm.texi: Regenerate.
8846 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8847 * config/spu/spu.c (spu_function_arg_offset): New function.
8848 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8849 * system.h (FUNCTION_ARG_OFFSET): Poison.
8851 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8852 Alan Hayard <alan.hayward@arm.com>
8853 David Sherwood <david.sherwood@arm.com>
8855 * target.def (truly_noop_truncation): New hook.
8856 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8857 than TRULY_NOOP_TRUNCATION.
8858 * hooks.h (hook_bool_uint_uint_true): Declare.
8859 * hooks.c (hook_bool_uint_uint_true): New function.
8860 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8861 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8862 * doc/tm.texi: Regenerate.
8863 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8864 rather than TRULY_NOOP_TRUNCATION in comments.
8865 (simplify_comparison): Likewise.
8866 (record_truncated_value): Likewise.
8867 * expmed.c (extract_bit_field_1): Likewise.
8868 (extract_split_bit_field): Likewise.
8869 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8870 instead of TRULY_NOOP_TRUNCATION.
8871 * function.c (assign_parm_setup_block): Likewise.
8872 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8873 * rtlhooks.c: Include target.h.
8874 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8875 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8876 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8877 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8878 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8879 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8880 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8881 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8882 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8883 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8884 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8885 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8886 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8887 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8888 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8889 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8890 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8891 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8892 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8893 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8894 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8895 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8896 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8897 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8898 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8899 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8900 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8901 rather than TRULY_NOOP_TRUNCATION in comments.
8902 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8903 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8904 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8905 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8906 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8907 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8908 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8909 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8910 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8911 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8912 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8913 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8914 rather than TRULY_NOOP_TRUNCATION in comments.
8915 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8916 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8917 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8918 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8919 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8920 TRULY_NOOP_TRUNCATION condition.
8921 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8922 (TRULY_NOOP_TRUNCATION): Delete.
8923 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8924 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8925 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8926 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8927 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8928 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8929 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8930 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8931 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8932 rather than TRULY_NOOP_TRUNCATION in comments.
8933 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8934 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8935 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8936 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8937 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8938 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8940 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8943 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8944 (*cmp_ior): Likewise.
8945 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8946 (*ior_scc_scc_cmp): Likewise.
8947 (*and_scc_scc): Likewise.
8948 (*and_scc_scc_cmp): Likewise.
8950 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8951 Alan Hayard <alan.hayward@arm.com>
8952 David Sherwood <david.sherwood@arm.com>
8954 * target.def (can_change_mode_class): New hook.
8955 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8956 (hard_regno_nregs): Likewise.
8957 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8958 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8959 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8960 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8961 (LOAD_EXTEND_OP): Update accordingly.
8962 * doc/tm.texi: Regenerate.
8963 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8964 CANNOT_CHANGE_MODE_CLASS.
8965 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8966 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8967 * combine.c (simplify_set): Update accordingly.
8968 * emit-rtl.c (validate_subreg): Likewise.
8969 * recog.c (general_operand): Likewise.
8970 * regcprop.c (mode_change_ok): Likewise.
8971 * reload1.c (choose_reload_regs): Likewise.
8972 (inherit_piecemeal_p): Likewise.
8973 * rtlanal.c (simplify_subreg_regno): Likewise.
8974 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8975 instead of CANNOT_CHANGE_MODE_CLASS.
8976 (reload_cse_simplify_operands): Likewise.
8977 * reload.c (push_reload): Use targetm.can_change_mode_class
8978 instead of CANNOT_CHANGE_MODE_CLASS.
8979 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8980 REG_CANNOT_CHANGE_MODE_P.
8981 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8982 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8983 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8984 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8985 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8986 (arm_can_change_mode_class): New function.
8987 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8988 than CANNOT_CHANGE_MODE_CLASS in comments.
8989 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8990 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8991 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8992 (ix86_can_change_mode_class): ...this new function, inverting the
8993 sense of the return value.
8994 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8995 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8996 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8997 (ia64_can_change_mode_class): New function.
8998 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8999 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9000 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9001 (m32c_can_change_mode_class): ...this new function, inverting the
9002 sense of the return value.
9003 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9004 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9005 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9006 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9007 (mips_can_change_mode_class): ...this new function, inverting the
9008 sense of the return value.
9009 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9010 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9011 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9012 (msp430_can_change_mode_class): New function.
9013 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9014 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9015 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9016 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9017 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9018 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9019 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9020 (pa_cannot_change_mode_class): Replace with...
9021 (pa_can_change_mode_class): ...this new function, inverting the
9022 sense of the return value.
9023 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9024 than CANNOT_CHANGE_MODE_CLASS in comments.
9025 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9026 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9027 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9028 (pdp11_cannot_change_mode_class): Replace with...
9029 (pdp11_can_change_mode_class): ...this new function, inverting the
9030 sense of the return value.
9031 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9032 * config/powerpcspe/powerpcspe-protos.h
9033 (rs6000_cannot_change_mode_class_ptr): Delete.
9034 * config/powerpcspe/powerpcspe.c
9035 (rs6000_cannot_change_mode_class_ptr): Delete.
9036 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9037 (rs6000_option_override_internal): Assign to
9038 targetm.can_change_mode_class instead of
9039 rs6000_cannot_change_mode_class_ptr.
9040 (rs6000_cannot_change_mode_class): Replace with...
9041 (rs6000_can_change_mode_class): ...this new function, inverting the
9042 sense of the return value.
9043 (rs6000_debug_cannot_change_mode_class): Replace with...
9044 (rs6000_debug_can_change_mode_class): ...this new function.
9045 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9046 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9047 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9048 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9049 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9051 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9052 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9053 (rs6000_option_override_internal): Assign to
9054 targetm.can_change_mode_class instead of
9055 rs6000_cannot_change_mode_class_ptr.
9056 (rs6000_cannot_change_mode_class): Replace with...
9057 (rs6000_can_change_mode_class): ...this new function, inverting the
9058 sense of the return value.
9059 (rs6000_debug_cannot_change_mode_class): Replace with...
9060 (rs6000_debug_can_change_mode_class): ...this new function.
9061 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9062 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9063 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9064 (s390_can_change_mode_class): ...this new function, inverting the
9065 sense of the return value.
9066 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9067 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9068 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9069 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9070 (sh_cannot_change_mode_class): Replace with...
9071 (sh_can_change_mode_class): ...this new function, inverting the
9072 sense of the return value.
9073 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9074 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9075 (sparc_can_change_mode_class): New function.
9076 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9077 * config/spu/spu.c (spu_can_change_mode_class): New function.
9078 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9079 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9080 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9081 (visium_can_change_mode_class): New function.
9082 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9084 2017-09-15 Richard Biener <rguenther@suse.de>
9086 PR tree-optimization/82217
9087 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9088 but not undefined case.
9090 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9093 * postreload.c (reload_cse_simplify_operands): Skip
9094 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9096 2017-09-15 Richard Biener <rguenther@suse.de>
9098 PR tree-optimization/68823
9099 * graphite-scop-detection.c (build_alias_set): If we have a
9100 possible dependence check whether we can handle them by just
9101 looking at the DRs DR_ACCESS_FNs.
9102 (build_scops): If build_alias_set fails, fail the SCOP.
9104 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9106 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9107 to support float128 built-in functions that require the ISA 3.0
9109 (BU_FLOAT128_3_HW): Likewise.
9110 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9112 (FMAF128): Likewise.
9114 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9115 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9116 floating point instructions.
9117 (rs6000_invalid_builtin): Likewise.
9118 (rs6000_builtin_mask_names): Likewise.
9119 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9120 (RS6000_BTM_FLOAT128_HW): Likewise.
9121 (RS6000_BTM_COMMON): Likewise.
9122 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9124 * doc/extend.texi (RS/6000 built-in functions): Document the
9125 IEEE 128-bit floating point square root and fused multiply-add
9128 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9130 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9131 reg (r2) isn't in the set of registers defined in the prologue.
9133 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9134 Alan Hayward <alan.hayward@arm.com>
9135 David Sherwood <david.sherwood@arm.com>
9137 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9138 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9139 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9140 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9141 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9143 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9144 max_vectorization_factor.
9145 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9147 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9148 Alan Hayward <alan.hayward@arm.com>
9149 David Sherwood <david.sherwood@arm.com>
9151 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9152 (vect_worthwhile_without_simd_p): Declare.
9153 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9154 (vectorizable_reduction): Use it.
9155 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9156 (vectorizable_operation): Likewise.
9158 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9159 Alan Hayward <alan.hayward@arm.com>
9160 David Sherwood <david.sherwood@arm.com>
9162 * tree-vectorizer.h (vect_get_num_copies): New function.
9163 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9164 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9165 (vectorizable_induction): Likewise.
9166 (vectorizable_live_operation): Likewise.
9167 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9168 (vectorizable_bswap): Likewise.
9169 (vectorizable_call): Likewise.
9170 (vectorizable_conversion): Likewise.
9171 (vectorizable_assignment): Likewise.
9172 (vectorizable_shift): Likewise.
9173 (vectorizable_operation): Likewise.
9174 (vectorizable_store): Likewise.
9175 (vectorizable_load): Likewise.
9176 (vectorizable_condition): Likewise.
9177 (vectorizable_comparison): Likewise.
9178 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9180 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9181 Alan Hayward <alan.hayward@arm.com>
9182 David Sherwood <david.sherwood@arm.com>
9184 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9185 of vect_init_vector.
9187 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9188 Alan Hayward <alan.hayward@arm.com>
9189 David Sherwood <david.sherwood@arm.com>
9191 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9192 an inline wrapper that provides a location.
9193 (gimple_build_vector): Likewise.
9194 * gimple-fold.c (gimple_build_vector_from_val): New function.
9195 (gimple_build_vector): Likewise.
9196 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9197 functions to build the initial value. Always return a gimple value.
9198 (get_initial_defs_for_reduction): Likewise. Only compute
9200 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9201 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9202 (vectorizable_induction): Use gimple_build_vector rather than
9205 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9206 Alan Hayward <alan.hayward@arm.com>
9207 David Sherwood <david.sherwood@arm.com>
9209 * target.h (vec_perm_indices): New typedef.
9210 (auto_vec_perm_indices): Likewise.
9211 * optabs-query.h: Include target.h
9212 (can_vec_perm_p): Take a vec_perm_indices *.
9213 * optabs-query.c (can_vec_perm_p): Likewise.
9214 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9215 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9216 * tree-vect-generic.c (lower_vec_perm): Likewise.
9217 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9218 (vect_grouped_load_supported): Likewise.
9219 (vect_shift_permute_load_chain): Likewise.
9220 (vect_permute_store_chain): Use auto_vec_perm_indices.
9221 (vect_permute_load_chain): Likewise.
9222 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9223 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9224 Update uses of can_vec_perm_p.
9225 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9226 mode with a number of elements. Take a vec_perm_indices *.
9227 (vect_create_epilog_for_reduction): Update accordingly.
9228 Use auto_vec_perm_indices.
9229 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9230 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9231 (vect_transform_slp_perm_load): Likewise.
9232 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9233 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9234 (vect_gen_perm_mask_checked): Likewise.
9235 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9236 (vect_gen_perm_mask_checked): Likewise.
9237 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9238 (vectorizable_store): Likewise.
9239 (vectorizable_load): Likewise.
9240 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9241 (vectorizable_bswap): Likewise.
9243 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9244 Alan Hayward <alan.hayward@arm.com>
9245 David Sherwood <david.sherwood@arm.com>
9247 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9248 * tree.c (build_vector): Likewise.
9249 (build_vector_from_ctor): Update accordingly.
9250 (build_vector_from_val): Likewise.
9251 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9252 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9253 * tree-vect-generic.c (add_rshift): Likewise.
9254 (expand_vector_divmod): Likewise.
9255 (optimize_vector_constructor): Likewise.
9256 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9257 (vect_transform_slp_perm_load): Likewise.
9258 (vect_schedule_slp_instance): Likewise.
9259 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9260 (vectorizable_call): Likewise.
9261 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9262 * expmed.c (make_tree): Likewise.
9263 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9264 a vector passed to build_vector.
9265 (fold_convert_const): Likewise.
9266 (exact_inverse): Likewise.
9267 (fold_ternary_loc): Likewise.
9268 (fold_relational_const): Likewise.
9269 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9270 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9271 (const_unop): Likewise. Store the reduction accumulator in a
9272 variable rather than an array.
9273 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9274 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9275 the new vector, rather than constructing it after the input arrays.
9276 (native_interpret_vector): Use auto_vec<tree> when building
9277 a vector passed to build_vector. Add elements in order.
9278 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9279 auto_vec<tree> when building a vector passed to build_vector.
9280 (vect_create_epilog_for_reduction): Likewise.
9281 (vectorizable_induction): Likewise.
9282 (get_initial_def_for_reduction): Likewise. Fix indentation of
9284 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9286 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9287 passed to build_vector.
9289 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9290 Alan Hayward <alan.hayward@arm.com>
9291 David Sherwood <david.sherwood@arm.com>
9293 * tree-core.h (tree_base::u): Add an "nelts" field.
9294 (tree_vector): Use VECTOR_CST_NELTS as the length.
9295 * tree.c (tree_size): Likewise.
9296 (make_vector): Initialize VECTOR_CST_NELTS.
9297 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9298 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9299 TYPE_VECTOR_SUBPARTS.
9300 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9301 as the number of units, setting it from VECTOR_CST_NELTS.
9302 (const_vector_from_tree): Likewise.
9303 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9304 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9305 (fold_negate_expr_1): Likewise.
9306 (fold_convert_const): Likewise.
9307 (const_binop): Likewise. Differentiate the number of output and
9309 (const_unop): Likewise.
9310 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9311 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9312 in cases that did the opposite.
9314 2017-09-14 Richard Biener <rguenther@suse.de>
9316 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9319 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9321 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9323 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9326 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9327 if and where to split a bb, except for splitting before debug insn
9328 sequences followed by non-label real insn. Delete debug insns
9329 in between basic blocks.
9331 * combine.c (make_compound_operation_int): Formatting fixes.
9333 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9334 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9335 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9336 * config/sol2.h (LINK_EH_SPEC): Likewise.
9337 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9338 * config/s390/linux.h (LINK_SPEC): Likewise.
9339 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9340 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9341 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9342 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9343 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9344 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9345 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9346 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9348 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9350 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9352 (ENDFILE_LINUX_SPEC): Likewise.
9353 (LINK_EH_SPEC): Likewise.
9354 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9355 (LINK_OS_LINUX_SPEC32): Likewise.
9356 (LINK_OS_LINUX_SPEC64): Likewise.
9357 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9358 (LINK_OS_LINUX_SPEC): Likewise.
9360 2017-09-13 Martin Liska <mliska@suse.cz>
9363 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9364 CASE_HIGH is NULL_TREE.
9366 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9367 Alan Hayward <alan.hayward@arm.com>
9368 David Sherwood <david.sherwood@arm.com>
9370 * target.def (secondary_memory_needed): New hook.
9371 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9372 instead of SECONDARY_MEMORY_NEEDED.
9373 (secondary_memory_needed_mode): Likewise.
9374 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9375 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9376 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9377 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9378 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9379 * doc/tm.texi: Regenerate.
9380 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9381 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9382 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9383 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9384 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9385 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9386 mode argument first and change the reg_class arguments to reg_class_t.
9387 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9388 Make static. Update the call to inline_secondary_memory_needed.
9389 (ix86_register_move_cost): Update the call to
9390 inline_secondary_memory_needed.
9391 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9392 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9394 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9395 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9397 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9398 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9399 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9400 and match hook interface. Add comment from mips.h.
9401 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9402 * config/mmix/mmix.md (truncdfsf2): Refer to
9403 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9405 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9406 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9407 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9408 (pa_secondary_memory_needed): New function.
9409 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9410 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9411 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9412 (pdp11_secondary_memory_needed): Make static and match hook interface.
9413 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9414 * config/powerpcspe/powerpcspe-protos.h
9415 (rs6000_secondary_memory_needed_ptr): Delete.
9416 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9418 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9419 (rs6000_option_override_internal): Assign to
9420 targetm.secondary_memory_needed rather than
9421 rs6000_secondary_memory_needed_ptr.
9422 (rs6000_secondary_memory_needed): Match hook interface.
9423 (rs6000_debug_secondary_memory_needed): Likewise.
9424 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9425 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9426 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9427 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9428 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9429 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9431 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9432 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9433 (rs6000_option_override_internal): Assign to
9434 targetm.secondary_memory_needed rather than
9435 rs6000_secondary_memory_needed_ptr.
9436 (rs6000_secondary_memory_needed): Match hook interface.
9437 (rs6000_debug_secondary_memory_needed): Likewise.
9438 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9439 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9440 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9441 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9442 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9443 (sparc_secondary_memory_needed): New function.
9444 * lra-constraints.c (check_and_process_move): Refer to
9445 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9447 (curr_insn_transform): Likewise.
9448 (process_alt_operands): Use targetm.secondary_memory_needed
9449 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9450 (check_secondary_memory_needed_p): Likewise.
9451 (choose_split_class): Likewise.
9452 * reload.c: Unconditionally include code that was previously
9453 conditional on SECONDARY_MEMORY_NEEDED.
9454 (push_secondary_reload): Use targetm.secondary_memory_needed
9455 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9456 (push_reload): Likewise.
9457 * reload1.c: Unconditionally include code that was previously
9458 conditional on SECONDARY_MEMORY_NEEDED.
9459 (choose_reload_regs): Use targetm.secondary_memory_needed
9460 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9461 (gen_reload): Likewise.
9462 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9464 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9465 Alan Hayward <alan.hayward@arm.com>
9466 David Sherwood <david.sherwood@arm.com>
9468 * target.def (secondary_memory_needed_mode): New hook:
9469 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9470 * targhooks.h (default_secondary_memory_needed_mode): New function.
9471 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9472 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9473 * doc/tm.texi: Regenerate.
9474 * lra-constraints.c (check_and_process_move): Use
9475 targetm.secondary_memory_needed_mode instead of
9476 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9477 (curr_insn_transform): Likewise.
9478 * reload.c (get_secondary_mem): Likewise.
9479 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9480 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9482 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9483 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9484 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9485 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9486 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9488 * config/powerpcspe/powerpcspe-protos.h
9489 (rs6000_secondary_memory_needed_mode): Delete.
9490 * config/powerpcspe/powerpcspe.c
9491 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9492 (rs6000_secondary_memory_needed_mode): Make static.
9493 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9494 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9496 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9498 (rs6000_secondary_memory_needed_mode): Make static.
9499 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9500 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9501 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9502 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9503 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9505 (sparc_secondary_memory_needed_mode): New function.
9506 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9508 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9510 * config/aarch64/constraints.md (Umq): New constraint.
9511 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9513 (mov<mode>): Update condition.
9515 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9517 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9518 when bitposition is the same.
9520 2017-09-13 Richard Biener <rguenther@suse.de>
9522 * dwarf2out.c (output_die_symbol): Remove.
9523 (output_die): Do not output a DIEs symbol.
9525 2017-09-13 Richard Biener <rguenther@suse.de>
9528 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9529 default-def to avoid breaking iterator update with the weird
9530 interaction with cgraph_update_edges_for_call_stmt_node.
9532 2017-09-13 Richard Biener <rguenther@suse.de>
9534 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9535 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9536 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9537 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9539 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9541 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9542 Disable pc relative literal load irrespective of
9543 TARGET_FIX_ERR_A53_84341 for default.
9545 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9547 * config/sparc/sparc.c (output_return): Output the source location of
9548 the insn in the delay slot, if any.
9549 (output_sibcall): Likewise.
9551 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9554 * common.opt (-static-pie): New alias.
9555 (shared): Negate static-pie.
9556 (-no-pie): Update help text.
9558 (static-pie): New option.
9559 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9560 -static-pie support.
9561 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9562 (LINK_EH_SPEC): Likewise.
9563 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9564 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9565 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9566 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9567 (init_gcc_specs): Likewise.
9568 (init_spec): Likewise.
9569 (display_help): Update help message for -pie.
9570 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9573 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9575 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9576 (movdi_aarch64): Likewise.
9577 (movti_aarch64): Likewise.
9579 2017-09-12 Simon Wright <simon@pushface.org>
9582 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9583 calculation of the minor version, always output as 0.
9585 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9588 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9589 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9590 on it early, rather than manual conversion late. For
9591 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9592 instead of performing manual conversion.
9594 2017-09-12 Carl Love <cel@us.ibm.com>
9596 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9597 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9599 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9600 VMULOSW): Add definitions.
9601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9602 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9603 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9604 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9605 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9607 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9609 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9611 (movti_aarch64): Likewise.
9612 (movdf_aarch64): Likewise.
9613 (movtf_aarch64): Likewise.
9614 (load_pairdi): Likewise.
9615 (store_pairdi): Likewise.
9616 (load_pairdf): Likewise.
9617 (store_pairdf): Likewise.
9618 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9619 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9620 (ldr_got_small_<mode>): Likewise.
9621 (ldr_got_small_28k_<mode>): Likewise.
9622 (ldr_got_tiny): Likewise.
9623 * config/aarch64/iterators.md (ldst_sz): New.
9624 (ldpstp_sz): Likewise.
9625 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9627 (thunderx_load): Split load_8 to load_16.
9628 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9630 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9631 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9632 (xgene1_store_pair): Split store_8 to store_16.
9633 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9634 (falkor_st_0_st_sd): Split store_8 to store_16.
9636 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9638 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9639 and store1/2/3/4 to store_4/8/12/16.
9640 * config/aarch64/aarch64.md: Update for rename.
9641 * config/arm/arm.md: Likewise.: Likewise.
9642 * config/arm/arm.c: Likewise.
9643 * config/arm/thumb1.md: Likewise.
9644 * config/arm/thumb2.md: Likewise.
9645 * config/arm/vfp.md: Likewise.
9646 * config/arm/arm-generic.md: Likewise.
9647 * config/arm/arm1020e.md: Likewise.
9648 * config/arm/arm1026ejs.md: Likewise.
9649 * config/arm/arm1136jfs.md: Likewise.
9650 * config/arm/arm926ejs.md: Likewise.
9651 * config/arm/cortex-a15.md: Likewise.
9652 * config/arm/cortex-a17.md: Likewise.
9653 * config/arm/cortex-a5.md: Likewise.
9654 * config/arm/cortex-a53.md: Likewise.
9655 * config/arm/cortex-a57.md: Likewise.
9656 * config/arm/cortex-a7.md: Likewise.
9657 * config/arm/cortex-a8.md: Likewise.
9658 * config/arm/cortex-a9.md: Likewise.
9659 * config/arm/cortex-m4.md: Likewise.
9660 * config/arm/cortex-m7.md: Likewise.
9661 * config/arm/cortex-r4.md: Likewise.
9662 * config/arm/exynos-m1.md: Likewise.
9663 * config/arm/fa526.md: Likewise.
9664 * config/arm/fa606te.md: Likewise.
9665 * config/arm/fa626te.md: Likewise.
9666 * config/arm/fa726te.md: Likewise.
9667 * config/arm/fmp626.md: Likewise.
9668 * config/arm/iwmmxt.md: Likewise.
9669 * config/arm/ldmstm.md: Likewise.
9670 * config/arm/marvell-pj4.md: Likewise.
9671 * config/arm/xgene1.md: Likewise.
9672 * config/aarch64/thunderx.md: Likewise.
9673 * config/aarch64/thunderx2t99.md: Likewise.
9674 * config/aarch64/falkor.md: Likewise.
9676 2017-09-12 Martin Liska <mliska@suse.cz>
9678 * attribs.c (private_lookup_attribute): New function.
9679 * attribs.h (private_lookup_attribute): Declared here.
9680 (lookup_attribute): Called from this place.
9682 2017-09-12 Richard Biener <rguenther@suse.de>
9684 PR tree-optimization/82157
9685 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9686 stmts with side-effects.
9688 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9689 Alan Hayward <alan.hayward@arm.com>
9690 David Sherwood <david.sherwood@arm.com>
9692 * target.def (hard_regno_nregs): New hook.
9693 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9694 * targhooks.h (default_hard_regno_nregs): Declare.
9695 * targhooks.c (default_hard_regno_nregs): New function.
9696 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9697 (TARGET_HARD_REGNO_NREGS): ...this hook.
9698 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9699 (CLASS_MAX_NREGS): Likewise.
9700 * doc/tm.texi: Regenerate.
9701 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9702 instead of HARD_REGNO_NREGS.
9703 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9704 HARD_REGNO_NREGS in the comment.
9705 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9706 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9707 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9708 Return an unsigned int.
9709 (TARGET_HARD_REGNO_NREGS): Redefine.
9710 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9711 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9712 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9713 (arc_hard_regno_nregs): New function.
9714 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9715 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9716 (arm_hard_regno_nregs): New function.
9717 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9718 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9719 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9720 (TARGET_HARD_REGNO_NREGS): Redefine.
9721 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9722 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9723 (HARD_REGNO_NREGS): Delete.
9724 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9725 (cr16_hard_regno_nregs): New function.
9726 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9727 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9728 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9729 (cris_hard_regno_nregs): New function.
9730 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9731 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9732 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9733 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9734 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9735 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9736 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9737 (frv_hard_regno_nregs): Make static. Take and return an
9739 (frv_class_max_nregs): Remove outdated copy of documentation.
9740 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9741 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9742 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9743 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9744 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9745 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9746 (TARGET_HARD_REGNO_NREGS): Redefine.
9747 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9748 (CLASS_MAX_NREGS): Update comment.
9749 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9750 (ia64_hard_regno_nregs): New function.
9751 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9752 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9753 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9754 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9755 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9757 (m32c_hard_regno_nregs): Likewise. Make static.
9758 (TARGET_HARD_REGNO_NREGS): Redefine.
9759 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9760 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9761 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9762 (m68k_hard_regno_nregs): New function.
9763 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9764 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9765 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9766 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9767 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9768 Take and return an unsigned int.
9769 (TARGET_HARD_REGNO_NREGS): Redefine.
9770 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9771 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9772 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9773 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9774 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9775 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9776 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9777 (msp430_hard_regno_nregs): Make static. Take and return an
9779 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9780 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9781 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9782 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9783 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9784 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9785 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9786 (TARGET_HARD_REGNO_NREGS): Redefine.
9787 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9788 (PA_HARD_REGNO_NREGS): ...this.
9789 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9790 (PA_HARD_REGNO_NREGS): ...this.
9791 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9792 (pa_hard_regno_nregs): New function.
9793 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9794 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9795 (pdp11_hard_regno_nregs): New function.
9796 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9797 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9798 (rs6000_hard_regno_nregs_hook): New function.
9799 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9800 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9801 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9802 Take and return an unsigned int. Move earlier in file.
9803 (TARGET_HARD_REGNO_NREGS): Redefine.
9804 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9805 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9806 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9807 (rl78_hard_regno_nregs): Make static. Take and return an
9809 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9810 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9811 (rs6000_hard_regno_nregs_hook): New function.
9812 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9813 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9814 (TARGET_HARD_REGNO_NREGS): Redefine.
9815 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9816 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9817 instead of HARD_REGNO_NREGS.
9818 (s390_hard_regno_nregs): New function.
9819 (s390_hard_regno_mode_ok): Add comment from s390.h.
9820 (TARGET_HARD_REGNO_NREGS): Redefine.
9821 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9822 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9823 (sh_hard_regno_nregs): New function.
9824 (sh_pass_in_reg_p): Use it.
9825 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9826 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9827 (sparc_hard_regno_nregs): New function.
9828 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9829 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9830 (spu_function_arg_advance): Use it, supplying a valid register number.
9831 (TARGET_HARD_REGNO_NREGS): Redefine.
9832 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9833 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9834 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9835 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9836 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9837 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9838 (CLASS_MAX_NREGS): Remove copy of old documentation.
9839 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9840 (visium_hard_regno_nregs): New function.
9841 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9842 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9843 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9844 xtensa_hard_regno_nregs): New function.
9845 * system.h (HARD_REGNO_NREGS): Poison.
9847 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9849 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9850 hard_regno_nregs instead of HARD_REGNO_NREGS.
9851 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9852 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9853 (c6x_expand_epilogue): Likewise.
9854 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9855 (frv_read_iacc_argument): Likewise.
9856 * config/sh/sh.c: Include regs.h.
9857 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9858 (regs_used): Likewise.
9859 (output_stack_adjust): Likewise.
9860 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9861 * expmed.c: Include regs.h.
9862 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9863 * ree.c: Include regs.h.
9864 (combine_reaching_defs): Use hard_regno_nregs instead of
9866 (add_removable_extension): Likewise.
9868 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9870 * regs.h (hard_regno_nregs): Turn into a function.
9871 (end_hard_regno): Update accordingly.
9872 * caller-save.c (setup_save_areas): Likewise.
9873 (save_call_clobbered_regs): Likewise.
9874 (replace_reg_with_saved_mem): Likewise.
9875 (insert_restore): Likewise.
9876 (insert_save): Likewise.
9877 * combine.c (can_change_dest_mode): Likewise.
9878 (move_deaths): Likewise.
9879 (distribute_notes): Likewise.
9880 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9881 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9882 (rs6000_split_multireg_move): Likewise.
9883 (rs6000_register_move_cost): Likewise.
9884 (rs6000_memory_move_cost): Likewise.
9885 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9886 (rs6000_split_multireg_move): Likewise.
9887 (rs6000_register_move_cost): Likewise.
9888 (rs6000_memory_move_cost): Likewise.
9889 * cselib.c (cselib_reset_table): Likewise.
9890 (cselib_lookup_1): Likewise.
9891 * emit-rtl.c (set_mode_and_regno): Likewise.
9892 * function.c (aggregate_value_p): Likewise.
9893 * ira-color.c (setup_profitable_hard_regs): Likewise.
9894 (check_hard_reg_p): Likewise.
9895 (calculate_saved_nregs): Likewise.
9896 (assign_hard_reg): Likewise.
9897 (improve_allocation): Likewise.
9898 (calculate_spill_cost): Likewise.
9899 * ira-emit.c (modify_move_list): Likewise.
9900 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9901 (ira_hard_reg_in_set_p): Likewise.
9902 * ira.c (setup_reg_mode_hard_regset): Likewise.
9903 (clarify_prohibited_class_mode_regs): Likewise.
9904 (check_allocation): Likewise.
9905 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9906 (lra_setup_reg_renumber): Likewise.
9907 (setup_try_hard_regno_pseudos): Likewise.
9908 (spill_for): Likewise.
9909 (assign_hard_regno): Likewise.
9910 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9911 * lra-constraints.c (in_class_p): Likewise.
9912 (lra_constraint_offset): Likewise.
9913 (simplify_operand_subreg): Likewise.
9914 (lra_constraints): Likewise.
9915 (split_reg): Likewise.
9916 (split_if_necessary): Likewise.
9917 (invariant_p): Likewise.
9918 (inherit_in_ebb): Likewise.
9919 * lra-lives.c (process_bb_lives): Likewise.
9920 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9921 (get_hard_regs): Likewise.
9922 (do_remat): Likewise.
9923 * lra-spills.c (assign_spill_hard_regs): Likewise.
9924 * mode-switching.c (create_pre_exit): Likewise.
9925 * postreload.c (reload_combine_recognize_pattern): Likewise.
9926 * recog.c (peep2_find_free_register): Likewise.
9927 * regcprop.c (kill_value_regno): Likewise.
9928 (set_value_regno): Likewise.
9929 (copy_value): Likewise.
9930 (maybe_mode_change): Likewise.
9931 (find_oldest_value_reg): Likewise.
9932 (copyprop_hardreg_forward_1): Likewise.
9933 * regrename.c (check_new_reg_p): Likewise.
9934 (regrename_do_replace): Likewise.
9935 * reload.c (push_reload): Likewise.
9936 (combine_reloads): Likewise.
9937 (find_dummy_reload): Likewise.
9938 (operands_match_p): Likewise.
9939 (find_reloads): Likewise.
9940 (find_equiv_reg): Likewise.
9941 (reload_adjust_reg_for_mode): Likewise.
9942 * reload1.c (count_pseudo): Likewise.
9943 (count_spilled_pseudo): Likewise.
9944 (find_reg): Likewise.
9945 (clear_reload_reg_in_use): Likewise.
9946 (free_for_value_p): Likewise.
9947 (allocate_reload_reg): Likewise.
9948 (choose_reload_regs): Likewise.
9949 (reload_adjust_reg_for_temp): Likewise.
9950 (emit_reload_insns): Likewise.
9951 (delete_output_reload): Likewise.
9952 * rtlanal.c (subreg_get_info): Likewise.
9953 * sched-deps.c (sched_analyze_reg): Likewise.
9954 * sel-sched.c (init_regs_for_mode): Likewise.
9955 (mark_unavailable_hard_regs): Likewise.
9956 (choose_best_reg_1): Likewise.
9957 (verify_target_availability): Likewise.
9958 * valtrack.c (dead_debug_insert_temp): Likewise.
9959 * var-tracking.c (track_loc_p): Likewise.
9960 (emit_note_insn_var_location): Likewise.
9961 * varasm.c (make_decl_rtl): Likewise.
9962 * reginfo.c (choose_hard_reg_mode): Likewise.
9963 (init_reg_modes_target): Refer directly to
9964 this_target_regs->x_hard_regno_nregs.
9966 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9968 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9969 instead of hard_regno_nregs.
9971 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9973 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9974 end_hard_regno instead of hard_regno_nregs.
9975 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9976 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9977 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9978 * ira-color.c (improve_allocation): Likewise.
9979 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9980 * lra-lives.c (mark_regno_live): Likewise.
9981 (mark_regno_dead): Likewise.
9982 * lra-remat.c (operand_to_remat): Likewise.
9983 * lra.c (collect_non_operand_hard_regs): Likewise.
9984 * postreload.c (reload_combine_note_store): Likewise.
9985 (move2add_valid_value_p): Likewise.
9986 * reload.c (regno_clobbered_p): Likewise.
9988 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9990 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9992 * config/v850/v850.c (v850_reorg): Likewise.
9993 * reload.c (refers_to_regno_for_reload_p): Likewise.
9994 (find_equiv_reg): Likewise.
9995 * reload1.c (reload_reg_reaches_end_p): Likewise.
9997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9999 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10001 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10002 * config/arm/arm.c (output_move_neon): Likewise.
10003 (arm_attr_length_move_neon): Likewise.
10004 (neon_split_vcombine): Likewise.
10005 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10006 (c6x_mark_reg_written): Likewise.
10007 (c6x_dwarf_register_span): Likewise.
10008 * config/i386/i386.c (ix86_save_reg): Likewise.
10009 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10010 (rws_access_reg): Likewise.
10011 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10012 * mode-switching.c (create_pre_exit): Likewise.
10013 * ree.c (combine_reaching_defs): Likewise.
10014 (add_removable_extension): Likewise.
10015 * regcprop.c (find_oldest_value_reg): Likewise.
10016 (copyprop_hardreg_forward_1): Likewise.
10017 * reload.c (reload_inner_reg_of_subreg): Likewise.
10018 (push_reload): Likewise.
10019 (combine_reloads): Likewise.
10020 (find_dummy_reload): Likewise.
10021 (reload_adjust_reg_for_mode): Likewise.
10022 * reload1.c (find_reload_regs): Likewise.
10023 (forget_old_reloads_1): Likewise.
10024 (reload_reg_free_for_value_p): Likewise.
10025 (reload_adjust_reg_for_temp): Likewise.
10026 (emit_reload_insns): Likewise.
10027 (delete_output_reload): Likewise.
10028 * sel-sched.c (choose_best_reg_1): Likewise.
10029 (choose_best_pseudo_reg): Likewise.
10031 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10032 Alan Hayward <alan.hayward@arm.com>
10033 David Sherwood <david.sherwood@arm.com>
10035 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10036 * target.def (slow_unaligned_access): New hook.
10037 * targhooks.h (default_slow_unaligned_access): Declare.
10038 * targhooks.c (default_slow_unaligned_access): New function.
10039 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10040 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10041 * doc/tm.texi: Regenerate.
10042 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10043 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10044 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10046 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10047 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10049 (rs6000_slow_unaligned_access): New function.
10050 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10051 (expand_block_compare): Likewise.
10052 (expand_strn_compare): Likewise.
10053 (rs6000_rtx_costs): Likewise.
10054 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10055 (riscv_slow_unaligned_access): Likewise.
10056 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10057 (riscv_slow_unaligned_access_p): ...this and make static.
10058 (riscv_option_override): Update accordingly.
10059 (riscv_slow_unaligned_access): New function.
10060 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10061 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10062 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10063 (rs6000_slow_unaligned_access): New function.
10064 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10065 (rs6000_rtx_costs): Likewise.
10066 * config/rs6000/rs6000-string.c (expand_block_compare)
10067 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10068 of SLOW_UNALIGNED_ACCESS.
10069 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10070 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10071 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10072 of SLOW_UNALIGNED_ACCESS.
10073 * expmed.c (simple_mem_bitfield_p): Likewise.
10074 * expr.c (alignment_for_piecewise_move): Likewise.
10075 (emit_group_load_1): Likewise.
10076 (emit_group_store): Likewise.
10077 (copy_blkmode_from_reg): Likewise.
10078 (emit_push_insn): Likewise.
10079 (expand_assignment): Likewise.
10080 (store_field): Likewise.
10081 (expand_expr_real_1): Likewise.
10082 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10083 * lra-constraints.c (simplify_operand_subreg): Likewise.
10084 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10085 * gimple-ssa-store-merging.c: Likewise in block comment at start
10087 * tree-ssa-strlen.c: Include target.h.
10088 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10089 of SLOW_UNALIGNED_ACCESS.
10090 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10092 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10094 PR rtl-optimization/82185
10095 * expmed.c (emit_store_flag_int): Only test tem if it has been
10098 2017-09-12 Richard Biener <rguenther@suse.de>
10100 PR middle-end/82149
10101 * match.pd ((FTYPE) N CMP CST): Fix typo.
10103 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10105 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10107 (mips_near_type_p): Add 'short_call' attribute as a synonym
10109 * doc/extend.texi (short_call): Document new function attribute.
10111 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10114 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10115 assertion check that in the condition.
10116 (get_atomic_generic_size): Likewise. Before testing if parameter
10117 has pointer type, if it has array type, call for C++
10118 default_conversion to perform array-to-pointer conversion.
10120 2017-09-12 Richard Biener <rguenther@suse.de>
10122 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10123 for operations we cannot scalarize.
10125 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10127 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10128 vectors heap vectors. Clean up comments.
10129 Make visited_bbs a reference.
10130 (profitable_jump_thread_path): Make GC
10131 vectors heap vectors. Clean up comments.
10133 (convert_and_register_jump_thread_path): Make GC vectors heap
10135 (check_subpath_and_update_thread_path): Same. Clean up comments.
10136 Make visited_bbs a reference.
10137 (handle_phi): Abstract common code to to
10138 register_jump_thread_path_if_profitable.
10139 Rename VAR_BB to DEF_BB.
10141 Make GC vectors heap vectors.
10142 Make visited_bbs a reference.
10143 (handle_assignment): Same.
10144 (register_jump_thread_path_if_profitable): New.
10145 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10147 Make GC vectors heap vectors. Clean up comments.
10148 Make visited_bbs a reference.
10149 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10150 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10153 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10156 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10157 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10159 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10161 Revert r251800 and r251799.
10163 2017-09-11 Martin Jambor <mjambor@suse.cz>
10166 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10167 arguments in advance.
10168 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10169 use it to find predecessor edges.
10170 (naive_outof_ssa): Collect vector of predecessors.
10172 2017-09-08 Jason Merrill <jason@redhat.com>
10174 PR c++/70029 - ICE with ref-qualifier and -flto
10175 * langhooks.h (struct lang_hooks_for_types): Add
10176 copy_lang_qualifiers.
10177 * attribs.c (build_type_attribute_qual_variant): Use it.
10178 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10180 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10181 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10183 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10186 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10187 (*mulsi3_sp64): New instruction.
10188 (mulsi3): New expander.
10190 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10192 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10194 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10196 * sancov.c: Include memmodel.h.
10198 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10201 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10204 2017-09-07 Carl Love <cel@us.ibm.com>
10206 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10207 the sldi instruction.
10209 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10211 * sancov.c: Include tm_p.h.
10213 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10216 * output.h (switch_to_other_text_partition): New declaration.
10217 * varasm.c (switch_to_other_text_partition): New function.
10218 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10219 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10220 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10221 to the other text partition before emitting LCL label and switch back
10222 after emitting the word after it.
10224 2017-09-07 Richard Biener <rguenther@suse.de>
10226 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10227 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10228 critical edges here, after loop init.
10229 (pass_data_pre): Remove PROP_no_crit_edges flags.
10230 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10231 for valueization of call args to avoid leaking VN_TOP.
10232 (visit_use): Assert we do not visit default defs.
10233 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10234 Use error_mark_node to more easily detect leaking VN_TOP.
10235 All default-defs are varying, not VN_TOP. Mark them visited.
10236 (run_scc_vn): Make code match comment.
10238 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10240 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10241 OPTION_MASK_FLOAT128_KEYWORD.
10242 (POWERPC_MASKS): Likewise.
10243 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10244 support for the -mfloat128-type option, and make -mfloat128
10245 default on PowerPC Linux systems. Define or undefine
10246 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10247 Define __float128 to be __ieee128 if IEEE 128-bit support is
10248 enabled, or undefine it.
10249 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10250 Delete defining __FLOAT128_TYPE__.
10251 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10252 -mfloat128-type option and make -mfloat128 default on PowerPC
10254 (TARGET_FLOAT128_TYPE): Likewise.
10255 (-mfloat128-type): Likewise.
10256 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10257 Delete the -mfloat128-type option and make -mfloat128 default on
10258 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10259 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10260 128-bit floating point is enabled. Change tests from using
10261 -mfloat128-type to -mfloat128.
10262 (rs6000_mangle_type): Use the correct mangling for the __float128
10263 type even if normal long double is restricted to 64-bits.
10264 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10266 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10267 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10268 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10270 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10271 documentation for -mfloat128.
10273 2017-09-06 Olivier Hainque <hainque@adacore.com>
10275 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10277 2017-09-06 Wish Wu <wishwu007@gmail.com>
10278 Jakub Jelinek <jakub@redhat.com>
10280 * asan.c (initialize_sanitizer_builtins): Add
10281 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10282 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10283 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10284 BT_FN_VOID_UINT64_PTR variables.
10285 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10286 (BT_FN_VOID_UINT16_UINT16): Likewise.
10287 (BT_FN_VOID_UINT32_UINT32): Likewise.
10288 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10289 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10290 (BT_FN_VOID_UINT64_PTR): Likewise.
10291 * common.opt (flag_sanitize_coverage): New variable.
10292 (fsanitize-coverage=trace-pc): Remove.
10293 (fsanitize-coverage=): Add.
10294 * flag-types.h (enum sanitize_coverage_code): New enum.
10295 * fold-const.c (fold_range_test): Disable non-short-circuit
10296 optimization if flag_sanitize_coverage.
10297 (fold_truth_andor): Likewise.
10298 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10299 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10300 (coverage_sanitizer_opts): New array.
10301 (get_closest_sanitizer_option): Add OPTS argument, handle also
10302 OPT_fsanitize_coverage_.
10303 (parse_sanitizer_options): Adjusted to also handle
10304 OPT_fsanitize_coverage_.
10305 (common_handle_option): Add OPT_fsanitize_coverage_.
10306 * sancov.c (instrument_comparison, instrument_switch): New function.
10307 (sancov_pass): Add trace-cmp support.
10308 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10309 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10310 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10311 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10312 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10313 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10314 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10315 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10316 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10317 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10319 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10321 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10322 error. Only quit immediately if parsing is complete.
10323 (BEGIN): Initialize fatal_err and parse_done.
10324 (begin fpu, end fpu): Check number of arguments.
10325 (begin arch, end arch): Likewise.
10326 (begin cpu, end cpu): Likewise.
10327 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10328 (optalias): Likewise.
10330 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10332 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10333 * config/arm/arm-isa.h: Delete. Move definitions to ...
10334 * arm-cpus.in: ... here. Use new feature and fgroup values.
10335 * config/arm/arm.c (arm_option_override): Use lower case for feature
10337 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10338 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10339 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10341 (print_isa_bits_for): New function.
10342 (gen_isa): New function.
10343 (gen_comm_data): Use print_isa_bits_for.
10344 (define feature): New keyword.
10345 (define fgroup): New keyword.
10346 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10347 (arm-isa.h): Add rule to generate file.
10348 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10349 case for feature bit names.
10351 2017-09-06 Richard Biener <rguenther@suse.de>
10353 * tree-ssa-pre.c (NECESSARY): Remove.
10354 (create_expression_by_pieces): Do not touch pass-local flags.
10355 (insert_into_preds_of_block): Likewise.
10356 (do_pre_regular_insertion): Likewise.
10357 (eliminate_insert): Likewise.
10358 (eliminate_dom_walker::before_dom_children): Likewise.
10359 (fini_eliminate): Do not look at inserted_exprs.
10360 (mark_operand_necessary): Remove.
10361 (remove_dead_inserted_code): Replace with simple work-list
10362 algorithm based on inserted_exprs and SSA uses.
10363 (pass_pre::execute): Re-order fini_eliminate and
10364 remove_dead_inserted_code.
10366 2017-09-06 Olivier Hainque <hainque@adacore.com>
10368 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10369 for VxWorks 7. Adjust surrounding comments.
10371 2017-09-06 Richard Biener <rguenther@suse.de>
10373 * gimple-ssa-strength-reduction.c
10374 (find_candidates_dom_walker::before_dom_children): Also allow
10377 2017-09-06 Richard Biener <rguenther@suse.de>
10379 PR tree-optimization/82108
10380 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10381 for gap in the non-permutation SLP case.
10383 2017-09-06 Martin Jambor <mjambor@suse.cz>
10385 PR tree-optimization/82078
10386 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10387 add_access_to_work_queue...
10388 (build_accesses_from_assign): ...here.
10389 (propagate_all_subaccesses): Make sure racc is the group
10390 representative, if there is one.
10392 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10394 PR middle-end/82095
10395 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10398 2017-09-06 Richard Biener <rguenther@suse.de>
10400 * gimple-ssa-strength-reduction.c
10401 (find_candidates_dom_walker::before_doom_children): Use a
10402 type and not a mode check.
10404 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10407 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10409 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10411 2017-09-05 Jeff Law <law@redhat.com>
10413 PR tree-optimization/64910
10414 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10415 cases where we have 3 or more operands.
10417 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10419 PR middle-end/81768
10420 * omp-low.c (lower_omp_for): Recompute tree invariant if
10421 gimple_omp_for_initial/final is ADDR_EXPR.
10423 PR middle-end/81768
10424 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10425 into gimple val before gimplification fo the COND_EXPR.
10427 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10429 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10430 REGION_COPY argument.
10431 (thread_through_all_blocks): Remove unused argument to
10432 duplicate_thread_path.
10434 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10435 Alan Hayward <alan.hayward@arm.com>
10436 David Sherwood <david.sherwood@arm.com>
10438 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10439 Take a scalar_mode rather than a machine_mode.
10440 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10441 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10442 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10443 (aarch64_gen_adjusted_ldpstp): Likewise.
10444 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10446 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10447 Alan Hayward <alan.hayward@arm.com>
10448 David Sherwood <david.sherwood@arm.com>
10450 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10451 Take a scalar_int_mode instead of a machine_mode.
10452 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10453 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10454 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10455 (aarch64_simd_attr_length_rglist): Delete.
10456 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10457 a scalar_int_mode instead of a machine_mode.
10458 (aarch64_add_offset): Likewise.
10459 (aarch64_internal_mov_immediate): Likewise
10460 (aarch64_add_constant_internal): Likewise.
10461 (aarch64_add_constant): Likewise.
10462 (aarch64_movw_imm): Likewise.
10463 (aarch64_rtx_arith_op_extract_p): Likewise.
10464 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10465 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10466 Remove assert that the mode isn't a vector.
10467 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10468 (aarch64_expand_mov_immediate): Update calls after above changes.
10469 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10470 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10471 (aarch64_move_imm): Likewise.
10472 (aarch64_can_const_movi_rtx_p): Likewise.
10473 (aarch64_strip_extend): Likewise.
10474 (aarch64_extr_rtx_p): Likewise.
10475 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10476 a CONST_INT when the mode parameter is VOIDmode.
10477 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10479 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10481 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10482 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10483 (bitwise_type_for_mode): Update accordingly.
10485 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10487 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10488 * stor-layout.c (mode_for_size_tree): Likewise.
10489 (mode_for_array): Update accordingly.
10490 (layout_decl): Likewise.
10491 (compute_record_mode): Likewise. Only set the mode once.
10493 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10495 * target.def (get_mask_mode): Change return type to opt_mode.
10497 * doc/tm.texi: Regenerate.
10498 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10499 * targhooks.c (default_get_mask_mode): Likewise.
10500 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10501 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10502 targetm.get_mask_mode.
10503 * tree.c (build_truth_vector_type): Likewise.
10505 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10507 * machmode.h (mode_for_vector): Return an opt_mode.
10508 * stor-layout.c (mode_for_vector): Likewise.
10509 (mode_for_int_vector): Update accordingly.
10510 (layout_type): Likewise.
10511 * config/i386/i386.c (emit_memmov): Likewise.
10512 (ix86_expand_set_or_movmem): Likewise.
10513 (ix86_expand_vector_init): Likewise.
10514 (ix86_get_mask_mode): Likewise.
10515 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10517 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10518 * expmed.c (extract_bit_field_1): Likewise.
10519 * expr.c (expand_expr_real_2): Likewise.
10520 * optabs-query.c (can_vec_perm_p): Likewise.
10521 (can_vec_mask_load_store_p): Likewise.
10522 * optabs.c (expand_vec_perm): Likewise.
10523 * targhooks.c (default_get_mask_mode): Likewise.
10524 * tree-vect-stmts.c (vectorizable_store): Likewise.
10525 (vectorizable_load): Likewise.
10526 (get_vectype_for_scalar_type_and_size): Likewise.
10528 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10530 * machmode.h (mode_for_int_vector): New function.
10531 * stor-layout.c (mode_for_int_vector): Likewise.
10532 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10533 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10534 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10535 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10536 (s390_expand_vcond): Likewise.
10538 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10540 * machmode.h (opt_machine_mode): New type.
10541 (opt_mode<T>): Allow construction from anything that can be
10543 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10544 (mode_for_size): Return an opt_machine_mode.
10545 * stor-layout.c (mode_for_size): Likewise.
10546 (mode_for_size_tree): Update call accordingly.
10547 (bitwise_mode_for_mode): Likewise.
10548 (make_fract_type): Likewise.
10549 (make_accum_type): Likewise.
10550 * caller-save.c (replace_reg_with_saved_mem): Update call
10552 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10553 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10554 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10555 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10556 * expmed.c (extract_bit_field_1): Likewise.
10557 * reload.c (get_secondary_mem): Likewise.
10558 * varasm.c (assemble_integer): Likewise.
10559 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10562 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10564 * machmode.h (decimal_float_mode_for_size): New function.
10565 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10566 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10567 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10568 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10569 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10570 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10572 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10574 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10575 (get_builtin_sync_mode): Likewise.
10576 (expand_ifn_atomic_compare_exchange): Likewise.
10577 (expand_builtin_atomic_clear): Likewise.
10578 (expand_builtin_atomic_test_and_set): Likewise.
10579 (fold_builtin_atomic_always_lock_free): Likewise.
10580 * calls.c (compute_argument_addresses): Likewise.
10581 (emit_library_call_value_1): Likewise.
10582 (store_one_arg): Likewise.
10583 * combine.c (combine_instructions): Likewise.
10584 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10585 * config/arm/arm.c (arm_function_value): Likewise.
10586 (aapcs_allocate_return_reg): Likewise.
10587 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10588 * config/i386/i386.c (construct_container): Likewise.
10589 (ix86_gimplify_va_arg): Likewise.
10590 (ix86_expand_sse_cmp): Likewise.
10591 (emit_memmov): Likewise.
10592 (emit_memset): Likewise.
10593 (expand_small_movmem_or_setmem): Likewise.
10594 (ix86_expand_pextr): Likewise.
10595 (ix86_expand_pinsr): Likewise.
10596 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10597 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10599 * config/mips/mips.c (mips_function_value_1) Likewise.
10600 (mips_block_move_straight): Likewise.
10601 (mips_expand_ins_as_unaligned_store): Likewise.
10602 * config/powerpcspe/powerpcspe.c
10603 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10604 (rs6000_darwin64_record_arg_flush): Likewise.
10605 * config/rs6000/rs6000.c
10606 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10607 (rs6000_darwin64_record_arg_flush): Likewise.
10608 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10609 (sparc_function_value_1): Likewise.
10610 * config/spu/spu.c (adjust_operand): Likewise.
10611 (spu_emit_branch_or_set): Likewise.
10612 (arith_immediate_p): Likewise.
10613 * emit-rtl.c (gen_lowpart_common): Likewise.
10614 * expr.c (expand_expr_real_1): Likewise.
10615 * function.c (assign_parm_setup_block): Likewise.
10616 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10617 * reload1.c (alter_reg): Likewise.
10618 * stor-layout.c (mode_for_vector): Likewise.
10619 (layout_type): Likewise.
10621 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10623 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10624 (spu_convert_move): Likewise.
10625 * lower-subreg.c (resolve_simple_move): Likewise.
10627 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10630 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10631 define_insn to a define_expand.
10632 (altivec_vsum2sws_direct): New define_insn.
10633 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10635 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10637 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10640 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10645 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10646 realign stack if stack alignment needed is less than incoming
10649 2017-09-05 Marek Polacek <polacek@redhat.com>
10652 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10655 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10657 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10659 2017-09-05 Richard Biener <rguenther@suse.de>
10661 PR tree-optimization/82084
10662 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10664 2017-09-05 Richard Biener <rguenther@suse.de>
10666 PR tree-optimization/82102
10667 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10669 2017-09-05 Martin Liska <mliska@suse.cz>
10671 PR tree-optimization/82032
10672 * tree-cfg.c (generate_range_test): New function.
10673 * tree-cfg.h (generate_range_test): Declared here.
10674 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10675 (cleanup_control_expr_graph): Use it.
10676 * tree-switch-conversion.c (try_switch_expansion): Remove
10678 (emit_case_nodes): Use generate_range_test.
10680 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10683 * config/i386/i386.md (*<btsc><mode>_mask): Add
10684 TARGET_USE_BT to insn constraint.
10685 (*btr<mode>_mask): Ditto.
10687 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10689 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10690 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10692 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10695 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10696 TARGET_NEON and TARGET_IWMMXT.
10697 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10698 TARGET_NEON and TARGET_IWMMXT.
10699 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10701 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10703 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10704 (ix86_rewrite_tls_address): Ditto.
10705 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10706 (ix86_rewrite_tls_address_1): Ditto.
10707 (ix86_rewrite_tls_address): Ditto.
10708 * config/i386/predicates.md (tls_address_pattern): New predicate.
10709 * config/i386/i386.md (TLS address splitter): New splitter.
10711 2017-09-04 Richard Biener <rguenther@suse.de>
10713 PR tree-optimization/82084
10714 * fold-const.h (can_native_encode_string_p): Declare.
10715 * fold-const.c (can_native_encode_string_p): Factor out from ...
10716 (native_encode_string): ... here.
10717 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10718 vectorizing stores from constants we later cannot handle.
10720 2017-09-04 Marek Polacek <polacek@redhat.com>
10723 * doc/invoke.texi: Update -Wtautological-compare documentation.
10725 2017-09-04 Jeff Law <law@redhat.com>
10727 PR tree-optimization/64910
10728 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10729 swap the first and last operand if the last is a constant.
10731 2017-09-04 Marek Polacek <polacek@redhat.com>
10734 * convert.c (do_narrow): When sanitizing signed integer overflows,
10735 bail out for signed types.
10736 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10738 2017-09-04 Richard Biener <rguenther@suse.de>
10740 PR tree-optimization/82060
10741 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10742 Move devirtualization after stmt folding and before EH/AB/noreturn
10743 cleanup to get the stmt refs canonicalized. Use a bool instead
10744 of gimple_modified_p since that doesn't work for NOPs. Schedule
10745 NOPs generated by folding for removal.
10747 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10748 Alan Hayward <alan.hayward@arm.com>
10749 David Sherwood <david.sherwood@arm.com>
10751 * coretypes.h (pad_direction): New enum.
10752 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10753 (FUNCTION_ARG_PADDING): Likewise.
10754 * target.def (function_arg_padding): New hook.
10755 * targhooks.h (default_function_arg_padding): Declare.
10756 * targhooks.c (default_function_arg_padding): New function.
10757 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10758 (TARGET_FUNCTION_ARG_PADDING): ...this.
10759 * doc/tm.texi: Regenerate.
10760 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10761 instead of direction.
10762 (compute_argument_addresses): Likewise.
10763 (load_register_parameters): Likewise.
10764 (emit_library_call_value_1): Likewise.
10765 (store_one_arg): Use targetm.calls.function_arg_padding instead
10766 of FUNCTION_ARG_PADDING.
10767 (must_pass_in_stack_var_size_or_pad): Likewise.
10768 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10769 (emit_group_store): Likewise.
10770 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10771 instead of FUNCTION_ARG_PADDING.
10772 (emit_push_insn): Likewise, and propagate enum change throughout
10774 * function.h (direction): Delete.
10775 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10777 * function.c (assign_parm_find_stack_rtl): Likewise.
10778 (assign_parm_setup_block_p): Likewise.
10779 (assign_parm_setup_block): Likewise.
10780 (gimplify_parameters): Likewise.
10781 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10782 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10784 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10785 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10786 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10787 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10788 (aarch64_function_arg_padding): ...this new function.
10789 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10790 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10791 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10792 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10793 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10794 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10795 (arm_pad_arg_upward): Replace with...
10796 (arm_function_arg_padding): ...this new function.
10797 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10799 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10800 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10801 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10802 (ia64_hpux_function_arg_padding): Replace with...
10803 (ia64_function_arg_padding): ...this new function. Use pad_direction
10804 instead of direction. Check for TARGET_HPUX.
10805 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10806 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10807 (iq2000_function_arg_padding): New function.
10808 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10809 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10810 (mips_function_arg_padding): ...this new function.
10811 (mips_pad_reg_upward): Update accordingly.
10812 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10813 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10814 targetm.calls.function_arg_padding.
10815 (FUNCTION_ARG_PADDING): Delete.
10816 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10817 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10818 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10819 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10820 (nios2_block_reg_padding): Return pad_direction instead of direction.
10821 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10822 instead of direction.
10823 (nios2_function_arg_padding): Likewise. Make static.
10824 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10825 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10826 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10827 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10828 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10829 (pa_function_arg_padding): Make static. Return pad_direction instead
10831 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10832 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10833 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10834 instead of direction. Use targetm.calls.function_arg_padding.
10835 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10836 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10837 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10838 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10839 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10841 (function_arg_padding): Rename to...
10842 (rs6000_function_arg_padding): ...this. Make static. Return
10843 pad_direction instead of direction.
10844 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10845 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10846 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10847 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10848 instead of direction. Use targetm.calls.function_arg_padding.
10849 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10850 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10851 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10852 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10853 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10854 (function_arg_padding): Rename to...
10855 (rs6000_function_arg_padding): ...this. Make static. Return
10856 pad_direction instead of direction.
10857 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10858 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10859 * config/s390/s390.c (s390_function_arg_padding): New function.
10860 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10861 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10862 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10863 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10864 (function_arg_padding): Rename to...
10865 (sparc_function_arg_padding): ...this. Make static. Return
10866 pad_direction instead of direction.
10867 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10868 * config/spu/spu.c (spu_function_arg_padding): New function.
10869 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10870 * system.h (FUNCTION_ARG_PADDING): Poison.
10872 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10873 Alan Hayward <alan.hayward@arm.com>
10874 David Sherwood <david.sherwood@arm.com>
10876 * target.def (modes_tieable_p): New hook.
10877 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10878 (TARGET_MODES_TIEABLE_P): ...this.
10879 * doc/tm.texi.in: Regenerate.
10880 * hooks.h (hook_bool_mode_mode_true): Declare.
10881 * hooks.c (hook_bool_mode_mode_true): New function.
10882 * combine.c (subst): Use targetm.modes_tieable_p instead of
10884 * dse.c (find_shift_sequence): Likewise.
10885 * expmed.c (extract_low_bits): Likewise.
10886 * lower-subreg.c: Include target.h.
10887 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10889 * rtlanal.c (rtx_cost): Likewise.
10890 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10891 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10892 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10893 (TARGET_MODES_TIEABLE_P): Redefine.
10894 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10895 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10896 (TARGET_MODES_TIEABLE_P): Redefine.
10897 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10898 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10899 (arc_modes_tieable_p): New function.
10900 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10901 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10902 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10903 (arm_modes_tieable_p): Make static.
10904 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10905 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10906 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10907 (TARGET_MODES_TIEABLE_P): Redefine.
10908 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10909 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10910 (TARGET_MODES_TIEABLE_P): Redefine.
10911 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10912 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10913 (cr16_modes_tieable_p): New function.
10914 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10915 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10916 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10917 (TRULY_NOOP_TRUNCATION): Update comment.
10918 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10919 (TRULY_NOOP_TRUNCATION): Update comment.
10920 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10921 (frv_modes_tieable_p): New function.
10922 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10923 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10924 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10925 (TARGET_MODES_TIEABLE_P): Redefine.
10926 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10927 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10928 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10929 (TARGET_MODES_TIEABLE_P): Redefine.
10930 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10931 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10932 (ia64_modes_tieable_p): New function.
10933 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10934 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10935 (iq2000_modes_tieable_p): New function.
10936 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10937 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10938 (lm32_modes_tieable_p): New function.
10939 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10940 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10941 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10942 (TARGET_MODES_TIEABLE_P): Redefine.
10943 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10944 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10945 (m32r_modes_tieable_p): New function.
10946 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10947 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10948 (m68k_modes_tieable_p): New function.
10949 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10950 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10951 (mcore_modes_tieable_p): New function.
10952 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10953 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10955 (TARGET_MODES_TIEABLE_P): Redefine.
10956 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10957 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10958 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10959 (TARGET_MODES_TIEABLE_P): Redefine.
10960 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10961 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10962 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10963 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10964 (mn10300_modes_tieable_p): ...this and make static.
10965 (TARGET_MODES_TIEABLE_P): Redefine.
10966 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10967 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10968 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10969 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10970 (msp430_modes_tieable_p): Make static.
10971 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10972 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10973 (TARGET_MODES_TIEABLE_P): Redefine.
10974 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10975 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10976 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10977 (TARGET_MODES_TIEABLE_P): Redefine.
10978 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10979 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10980 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10981 (TARGET_MODES_TIEABLE_P): Redefine.
10982 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10983 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10984 (pdp11_modes_tieable_p): New function.
10985 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10986 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10987 (rs6000_modes_tieable_p): New function.
10988 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10989 * config/powerpcspe/powerpcspe.md: Update comment.
10990 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10991 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10992 (TARGET_MODES_TIEABLE_P): Redefine.
10993 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10994 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10995 (rl78_modes_tieable_p): New function.
10996 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10997 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10998 (rs6000_modes_tieable_p): New function.
10999 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11000 * config/rs6000/rs6000.md: Update comment.
11001 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11002 * config/rx/rx.c (rx_modes_tieable_p): New function.
11003 (TARGET_MODES_TIEABLE_P): Redefine.
11004 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11005 * config/s390/s390.c (s390_modes_tieable_p): New function.
11006 (TARGET_MODES_TIEABLE_P): Redefine.
11007 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11008 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11009 (sh_modes_tieable_p): New function.
11010 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11011 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11012 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11013 (sparc_modes_tieable_p): Make static.
11014 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11015 * config/spu/spu.c (spu_modes_tieable_p): New function.
11016 (TARGET_MODES_TIEABLE_P): Redefine.
11017 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11018 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11019 (TARGET_MODES_TIEABLE_P): Redefine.
11020 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11021 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11022 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11023 * config/v850/v850.c (v850_modes_tieable_p): New function.
11024 (TARGET_MODES_TIEABLE_P): Redefine.
11025 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11026 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11027 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11028 (visium_modes_tieable_p): New function.
11029 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11030 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11031 (xtensa_modes_tieable_p): New function.
11032 * system.h (MODES_TIEABLE_P): Poison.
11034 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11035 Alan Hayward <alan.hayward@arm.com>
11036 David Sherwood <david.sherwood@arm.com>
11038 * target.def (hard_regno_mode_ok): New hook.
11039 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11040 (TARGET_HARD_REGNO_MODE_OK): ...this.
11041 * doc/tm.texi.in: Regenerate.
11042 * hooks.h (hook_bool_uint_mode_true): Declare.
11043 * hooks.c (hook_bool_uint_mode_true): New function.
11044 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11045 HARD_REGNO_MODE_OK.
11046 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11047 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11048 instead of HARD_REGNO_MODE_OK.
11049 * caller-save.c: Include target.h.
11050 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11051 HARD_REGNO_MODE_OK.
11052 * combine.c (can_combine_p): Likewise.
11053 (combinable_i3pat): Likewise.
11054 (can_change_dest_mode): Likewise.
11055 * expr.c (init_expr_target): Likewise.
11056 (convert_move): Likewise.
11057 (convert_modes): Likewise.
11058 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11059 (setup_prohibited_mode_move_regs): Likewise.
11060 * ira.h (target_ira): Likewise.
11061 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11062 * lra-constraints.c (process_alt_operands): Likewise.
11063 (split_reg): Likewise.
11064 * recog.c (peep2_find_free_register): Likewise.
11065 * ree.c (combine_reaching_defs): Likewise.
11066 * regcprop.c (maybe_mode_change): Likewise.
11067 * reginfo.c (init_reg_sets_1): Likewise.
11068 (choose_hard_reg_mode): Likewise.
11069 (simplifiable_subregs): Likewise.
11070 * regrename.c (check_new_reg_p): Likewise.
11071 * reload.c (find_valid_class): Likewise.
11072 (find_valid_class_1): Likewise.
11073 (reload_inner_reg_of_subreg): Likewise.
11074 (push_reload): Likewise.
11075 (combine_reloads): Likewise.
11076 (find_dummy_reload): Likewise.
11077 (find_reloads): Likewise.
11078 * reload1.c (find_reg): Likewise.
11079 (set_reload_reg): Likewise.
11080 (allocate_reload_reg): Likewise.
11081 (choose_reload_regs): Likewise.
11082 (reload_adjust_reg_for_temp): Likewise.
11083 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11084 (simplify_subreg_regno): Likewise.
11085 * sel-sched.c (init_regs_for_mode): Likewise.
11086 * varasm.c (make_decl_rtl): Likewise.
11087 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11088 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11089 HARD_REGNO_MODE_OK.
11090 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11091 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11092 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11093 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11094 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11096 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11097 (arc_mode_class): Delete.
11098 (HARD_REGNO_MODE_OK): Delete.
11099 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11100 (arc_hard_regno_mode_ok): Rename old array to...
11101 (arc_hard_regno_mode_ok_modes): ...this.
11102 (arc_conditional_register_usage): Update accordingly.
11103 (arc_mode_class): Make static.
11104 (arc_hard_regno_mode_ok): New function.
11105 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11106 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11107 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11108 (arm_hard_regno_mode_ok): Make static.
11109 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11110 HARD_REGNO_MODE_OK.
11111 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11112 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11113 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11115 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11116 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11117 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11118 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11119 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11120 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11121 * config/bfin/predicates.md (valid_reg_operand): Use
11122 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11123 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11124 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11125 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11126 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11127 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11128 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11129 (cr16_hard_regno_mode_ok): Make static and return a bool.
11130 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11131 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11132 (cris_hard_regno_mode_ok): New function.
11133 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11134 (epiphany_mode_class): Delete.
11135 (HARD_REGNO_MODE_OK): Delete.
11136 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11137 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11138 (hard_regno_mode_ok): Rename to...
11139 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11140 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11141 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11142 HARD_REGNO_MODE_OK.
11143 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11144 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11145 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11146 (frv_hard_regno_mode_ok): Make static and return a bool.
11147 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11148 HARD_REGNO_MODE_OK.
11149 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11150 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11151 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11152 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11154 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11155 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11156 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11157 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11159 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11160 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11161 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11162 (ia64_hard_regno_mode_ok): New function.
11163 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11164 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11165 (iq2000_hard_regno_mode_ok): New function.
11166 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11167 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11168 (lm32_hard_regno_mode_ok): New function.
11169 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11170 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11171 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11172 instead of HARD_REGNO_MODE_OK.
11173 (m32c_hard_regno_ok): Rename to...
11174 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11175 (m32c_cannot_change_mode_class): Update accordingly.
11176 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11177 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11178 (m32r_mode_class): Delete.
11179 (HARD_REGNO_MODE_OK): Delete.
11180 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11181 (m32r_hard_regno_mode_ok): Rename to...
11182 (m32r_hard_regno_modes): ...this.
11183 (m32r_mode_class): Make static.
11184 (m32r_hard_regno_mode_ok): New function.
11185 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11186 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11187 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11188 (m68k_hard_regno_mode_ok): Make static.
11189 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11190 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11191 (mcore_hard_regno_mode_ok): New function.
11192 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11193 (HARD_REGNO_MODE_OK): Delete.
11194 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11196 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11197 (microblaze_hard_regno_mode_ok): New function.
11198 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11199 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11200 (mips_hard_regno_mode_ok): Delete.
11201 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11202 (mips_hard_regno_mode_ok_p): ...this and make static.
11203 (mips_hard_regno_mode_ok_p): Rename to...
11204 (mips_hard_regno_mode_ok_uncached): ...this.
11205 (mips_hard_regno_mode_ok): New function.
11206 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11207 of HARD_REGNO_MODE_OK.
11208 (mips_option_override): Update after above name changes.
11209 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11210 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11211 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11212 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11213 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11214 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11215 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11216 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11217 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11218 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11219 (msp430_hard_regno_mode_ok): Make static and return a bool.
11220 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11221 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11222 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11224 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11225 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11226 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11227 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11228 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11229 (PA_HARD_REGNO_MODE_OK): ...this
11230 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11231 (PA_HARD_REGNO_MODE_OK): ...this.
11232 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11233 (pa_hard_regno_mode_ok): New function.
11234 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11235 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11236 (pdp11_hard_regno_mode_ok): New function.
11237 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11238 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11240 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11243 (rs6000_hard_regno_mode_ok): Rename to...
11244 (rs6000_hard_regno_mode_ok_uncached): ...this.
11245 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11246 (rs6000_hard_regno_mode_ok): New function.
11247 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11248 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11249 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11250 (riscv_hard_regno_mode_ok): ...this and make static.
11251 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11252 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11253 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11254 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11255 (rl78_hard_regno_mode_ok): Make static and return bool.
11256 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11257 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11259 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11260 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11261 (rs6000_hard_regno_mode_ok): Rename to...
11262 (rs6000_hard_regno_mode_ok_uncached): ...this.
11263 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11264 (rs6000_hard_regno_mode_ok): New function.
11265 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11266 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11268 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11269 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11270 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11271 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11272 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11273 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11274 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11275 (sh_hard_regno_mode_ok): Make static.
11276 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11277 instead of HARD_REGNO_MODE_OK.
11278 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11279 (sparc_mode_class): Delete.
11280 (HARD_REGNO_MODE_OK): Delete.
11281 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11282 (hard_regno_mode_classes): Make static.
11283 (sparc_mode_class): Likewise.
11284 (sparc_hard_regno_mode_ok): New function.
11285 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11286 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11287 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11289 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11290 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11291 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11292 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11293 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11294 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11295 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11296 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11297 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11298 (visium_hard_regno_mode_ok): New function.
11299 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11300 instead of HARD_REGNO_MODE_OK.
11301 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11302 (HARD_REGNO_MODE_OK): Delete.
11303 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11304 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11305 (xtensa_option_override): Update accordingly.
11306 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11307 (xtensa_hard_regno_mode_ok): New function.
11308 * system.h (HARD_REGNO_MODE_OK): Poison.
11310 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11311 Alan Hayward <alan.hayward@arm.com>
11312 David Sherwood <david.sherwood@arm.com>
11314 * target.def (hard_regno_call_part_clobbered): New hook.
11315 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11316 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11317 * doc/tm.texi: Regenerate.
11318 * hooks.h (hook_bool_uint_mode_false): Declare.
11319 * hooks.c (hook_bool_uint_mode_false): New function.
11320 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11321 * cselib.c (cselib_process_insn): Use
11322 targetm.hard_regno_call_part_clobbered instead of
11323 HARD_REGNO_CALL_PART_CLOBBERED.
11324 * ira-conflicts.c (ira_build_conflicts): Likewise.
11325 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11326 * lra-constraints.c (need_for_call_save_p): Likewise.
11327 * lra-lives.c: Include target.h.
11328 (check_pseudos_live_through_calls): Use
11329 targetm.hard_regno_call_part_clobbered instead of
11330 HARD_REGNO_CALL_PART_CLOBBERED.
11331 * regcprop.c: Include target.h.
11332 (copyprop_hardreg_forward_1): Use
11333 targetm.hard_regno_call_part_clobbered instead of
11334 HARD_REGNO_CALL_PART_CLOBBERED.
11335 * reginfo.c (choose_hard_reg_mode): Likewise.
11336 * regrename.c (check_new_reg_p): Likewise.
11337 * reload.c (find_equiv_reg): Likewise.
11338 * reload1.c (emit_reload_insns): Likewise.
11339 * sched-deps.c (deps_analyze_insn): Likewise.
11340 * sel-sched.c (init_regs_for_mode): Likewise.
11341 (mark_unavailable_hard_regs): Likewise.
11342 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11343 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11344 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11346 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11347 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11348 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11350 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11352 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11353 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11354 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11356 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11357 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11358 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11360 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11361 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11363 * config/powerpcspe/powerpcspe.c
11364 (rs6000_hard_regno_call_part_clobbered): New function.
11365 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11366 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11367 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11369 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11370 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11371 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11373 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11374 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11375 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11377 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11378 Alan Hayward <alan.hayward@arm.com>
11379 David Sherwood <david.sherwood@arm.com>
11381 * rtl.h (subreg_memory_offset): Declare.
11382 * emit-rtl.c (subreg_memory_offset): New function.
11383 * expmed.c (store_bit_field_1): Use it.
11384 * expr.c (undefined_operand_subword_p): Likewise.
11385 * simplify-rtx.c (simplify_subreg): Likewise.
11387 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11389 PR rtl-optimization/57448
11392 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11393 using atomic_load pattern.
11394 (expand_atomic_store): Likewise.
11396 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11399 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11400 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11401 handling. Use replace_call_with_value with NULL instead of
11402 gsi_replace, unlink_stmt_vdef and release_defs.
11404 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11407 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11409 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11412 * rtl.h (emit_library_call_value_1): Declare.
11413 (emit_library_call): Replace declaration with a series of overloads.
11414 Remove the parameter count argument.
11415 (emit_library_call_value): Likewise.
11416 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11417 with an "rtx_mode_t *".
11418 (emit_library_call_value): Delete.
11419 (emit_library_call): Likewise.
11420 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11421 (asan_emit_allocas_unpoison): Likewise.
11422 * builtins.c (expand_builtin_powi): Likewise.
11423 (expand_asan_emit_allocas_unpoison): Likewise.
11424 * cfgexpand.c (expand_main_function): Likewise.
11425 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11426 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11427 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11428 * config/arm/arm.c (arm_trampoline_init): Likewise.
11429 (arm_call_tls_get_addr): Likewise.
11430 (arm_expand_divmod_libfunc): Likewise.
11431 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11432 (smulsi3_highpart): Likewise.
11433 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11434 (c6x_expand_compare): Likewise.
11435 (c6x_expand_movmem): Likewise.
11436 * config/frv/frv.c (frv_trampoline_init): Likewise.
11437 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11438 (ix86_expand_divmod_libfunc): Likewise.
11439 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11440 (ia64_expand_compare): Likewise.
11441 (ia64_profile_hook): Likewise.
11442 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11443 (nonlocal_goto): Likewise.
11444 (restore_stack_nonlocal): Likewise.
11445 * config/m32r/m32r.c (block_move_call): Likewise.
11446 (m32r_trampoline_init): Likewise.
11447 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11448 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11449 (m68k_call_m68k_read_tp): Likewise.
11450 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11451 (microblaze_expand_divide): Likewise.
11452 * config/mips/mips.h (mips_args): Likewise.
11453 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11454 (MIPS_ICACHE_SYNC): Likewise.
11455 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11456 (nios2_trampoline_init): Likewise.
11457 * config/pa/pa.c (hppa_tls_call): Likewise.
11458 (pa_trampoline_init): Likewise.
11459 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11460 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11461 (expand_strn_compare): Likewise.
11462 (rs6000_generate_compare): Likewise.
11463 (rs6000_expand_float128_convert): Likewise.
11464 (output_profile_hook): Likewise.
11465 (rs6000_trampoline_init): Likewise.
11466 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11467 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11468 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11469 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11470 (rs6000_generate_compare): Likewise.
11471 (rs6000_expand_float128_convert): Likewise.
11472 (output_profile_hook): Likewise.
11473 (rs6000_trampoline_init): Likewise.
11474 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11475 * config/sh/sh.c (sh_trampoline_init): Likewise.
11476 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11477 (sparc_emit_float_lib_cmp): Likewise.
11478 (sparc32_initialize_trampoline): Likewise.
11479 (sparc64_initialize_trampoline): Likewise.
11480 (sparc_profile_hook): Likewise.
11481 * config/spu/spu.c (ea_load_store): Likewise.
11482 * config/spu/spu.md (floatunssidf2): Likewise.
11483 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11484 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11485 * config/visium/visium.c (expand_block_move_4): Likewise.
11486 (expand_block_move_2): Likewise.
11487 (expand_block_move_1): Likewise.
11488 (expand_block_set_4): Likewise.
11489 (expand_block_set_2): Likewise.
11490 (expand_block_set_1): Likewise.
11491 (visium_trampoline_init): Likewise.
11492 (visium_profile_hook): Likewise.
11493 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11494 (xtensa_setup_frame_addresses): Likewise.
11495 (xtensa_trampoline_init): Likewise.
11496 * except.c (sjlj_emit_function_enter): Likewise.
11497 (sjlj_emit_function_exit): Likewise.
11498 * explow.c (allocate_dynamic_stack_space): Likewise.
11499 (probe_stack_range): Likewise.
11500 * expr.c (convert_mode_scalar): Likewise.
11501 * optabs.c (expand_binop): Likewise.
11502 (expand_twoval_binop_libfunc): Likewise.
11503 (expand_unop): Likewise.
11504 (prepare_cmp_insn): Likewise.
11505 (prepare_float_lib_cmp): Likewise.
11506 (expand_float): Likewise.
11507 (expand_fix): Likewise.
11508 (expand_fixed_convert): Likewise.
11509 (maybe_emit_sync_lock_test_and_set): Likewise.
11510 (expand_atomic_compare_and_swap): Likewise.
11511 (expand_mem_thread_fence): Likewise.
11512 (expand_atomic_fetch_op): Likewise.
11514 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11516 * doc/generic.texi (OpenACC): Adjust URL.
11517 * doc/invoke.texi (C Dialect Options): Ditto.
11519 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11521 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11522 predicate for operand 1. Add (m,<S>) constraint.
11523 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11524 Prevent memory operand 1 with register operand 2.
11526 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11528 PR rtl-optimization/82024
11529 * combine.c (try_combine): If the combination result is a PARALLEL,
11530 and we only need to retain the SET in there that would be placed
11531 at I2, check that we can place that at I3 instead, before doing so.
11533 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11536 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11538 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11539 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11542 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11543 Jeff Law <law@redhat.com>
11545 * varasm.c (bss_initializer_p): Do not put constants into .bss
11546 (categorize_decl_for_section): Handle bss_initializer_p returning
11547 false when DECL_INITIAL is NULL.
11549 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11552 * config/s390/s390.c (s390_can_inline_p): New function.
11554 2017-09-01 Jeff Law <law@redhat.com>
11556 PR tree-optimization/82052
11557 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11558 Always initialize the returned slot after a hash table miss
11559 when INSERT is true.
11561 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11563 * config/s390/s390.md (mem_signal_fence): Remove.
11564 * doc/md.texi (mem_signal_fence): Remove.
11565 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11567 * target-insns.def (mem_signal_fence): Remove.
11569 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11572 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11575 * asan.c (create_odr_indicator): Strip name encoding from assembler
11576 name before appending it after __odr_asan_.
11578 2017-09-01 Martin Liska <mliska@suse.cz>
11580 PR tree-optimization/82059
11581 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11582 frequency only when an edge is redirected.
11584 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11586 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11587 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11588 (arc_conditional_register_usage): Remove ARC600 lp_count
11590 (arc_file_start): Emit Tag_ARC_CPU_variation.
11591 (arc_can_use_doloop_p): New conditions to use ZOLs.
11592 (hwloop_fail): New function.
11593 (hwloop_optimize): Likewise.
11594 (hwloop_pattern_reg): Likewise.
11595 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11596 (arc_reorg_loops): New function, calls reorg_loops.
11597 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11598 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11600 (arc_loop_hazard): Remove function, functionality moved into
11602 (arc_hazard): Remove arc_loop_hazard call.
11603 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11604 into hwloop_optimize.
11605 (arc_label_align): Remove ZOL handling.
11606 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11607 * config/arc/arc.md (doloop_begin): Remove pattern.
11608 (doloop_begin_i): Likewise.
11609 (doloop_end_i): Likewise.
11610 (doloop_fallback): Likewise.
11611 (doloop_fallback_m): Likewise.
11612 (doloop_end): Reimplement expand.
11613 (arc_lp): New pattern for LP instruction.
11614 (loop_end): New pattern.
11615 (loop_fail): Likewise.
11616 (decrement_and_branch_until_zero): Likewise.
11617 * config/arc/arc.opt (mlpc-width): New option.
11618 * doc/invoke.texi (mlpc-width): Document option.
11620 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11622 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11623 (arc_ccfsm_advance): Fix checking for delay slots.
11624 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11626 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11628 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11630 (movhi_insn): Update store instruction constraint which are saving
11631 6-bit short immediates.
11632 (movsi_insn): Consider also short scaled load operations.
11633 (zero_extendhisi2_i): Use Usd constraint instead of T.
11634 (extendhisi2_i): Add q constraint.
11635 (arc_clzsi2): Add type and length attributes.
11636 (arc_ctzsi2): Likewise.
11637 * config/arc/constraints.md (Usc): Update constraint, the
11638 assembler can parse two relocations for a single instruction.
11640 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11642 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11643 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11645 2017-08-31 Olivier Hainque <hainque@adacore.com>
11647 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11648 match as powerpc-wrs-vxworks*.
11650 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11652 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11653 register constraint for by-element operand.
11654 (aarch64_mls_elt_merge<mode>): Likewise.
11656 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11658 * config/arc/arc.c (arc_can_follow_jump): Check for short
11661 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11663 * config.gcc: Use g.opt for arc.
11664 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11665 functionality moved to ...
11666 (legitimate_scaled_address_p): New function, ...here.
11667 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11668 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11669 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11671 (arc_override_options): Handle G option.
11672 (arc_output_pic_addr_const): Correct function definition.
11673 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11674 (arc_decl_anon_ns_mem_p): Delete.
11675 (arc_in_small_data_p): Overhaul this function to take into
11676 consideration the value given via G option.
11677 (arc_rewrite_small_data_1): Renamed and corrected old
11678 arc_rewrite_small_data function.
11679 (arc_rewrite_small_data): New function.
11680 (small_data_pattern): Don't use pic_offset_table_rtx.
11681 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11682 * config/arc/simdext.md (movmisalignv2hi): Use
11683 prepare_move_operands function.
11685 (movmisalign*): Likewise.
11686 * doc/invoke.texi (ARC options): Document -G option.
11688 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11690 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11692 * config/arc/arc.c (arc_print_operand): Output scalled address for
11693 sdata whenever is possible.
11694 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11695 load/stores are available.
11696 (compact_sda_memory_operand): Check for the alignment required by
11697 code density instructions.
11698 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11700 * config/arc/constraints.md (Usd): Update constraint.
11701 (Us0): New constraint.
11702 (Usc): Update constraint.
11704 2017-08-31 Richard Biener <rguenther@suse.de>
11706 PR middle-end/82054
11707 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11708 function only once.
11710 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11712 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11713 Resize type_signature.
11715 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11716 Alan Hayward <alan.hayward@arm.com>
11717 David Sherwood <david.sherwood@arm.com>
11719 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11720 subregs whose inner modes can be stored in GPRs.
11721 (aarch64_classify_index): Likewise.
11723 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11724 Alan Hayward <alan.hayward@arm.com>
11725 David Sherwood <david.sherwood@arm.com>
11727 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11728 (V_INT_EQUIV): ...this.
11729 (v_cmp_result): Rename to...
11730 (v_int_equiv): ...this.
11731 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11732 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11733 (copysign<mode>3): Likewise.
11734 (aarch64_simd_bsl<mode>_internal): Likewise.
11735 (aarch64_simd_bsl<mode>): Likewise.
11736 (vec_cmp<mode><mode>): Likewise.
11737 (vcond<mode><mode>): Likewise.
11738 (vcond<v_cmp_mixed><mode>): Likewise.
11739 (vcondu<mode><v_cmp_mixed>): Likewise.
11740 (aarch64_cm<optab><mode>): Likewise.
11741 (aarch64_cmtst<mode>): Likewise.
11742 (aarch64_fac<optab><mode>): Likewise.
11743 (vec_perm_const<mode>): Likewise.
11744 (vcond_mask_<mode><v_cmp_result>): Rename to...
11745 (vcond_mask_<mode><v_int_equiv>): ...this.
11746 (vec_cmp<mode><v_cmp_result>): Rename to...
11747 (vec_cmp<mode><v_int_equiv>): ...this.
11749 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11750 Alan Hayward <alan.hayward@arm.com>
11751 David Sherwood <david.sherwood@arm.com>
11753 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11755 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11756 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11757 (UNSPEC_LD4_DREG): New unspecs.
11758 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11759 (aarch64_ld2<mode>_dreg_be): Replace with...
11760 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11762 (aarch64_ld3<mode>_dreg_le)
11763 (aarch64_ld3<mode>_dreg_be): Replace with...
11764 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11766 (aarch64_ld4<mode>_dreg_le)
11767 (aarch64_ld4<mode>_dreg_be): Replace with...
11768 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11771 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11773 PR tree-optimization/81987
11774 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11775 insert an initializer in a location not dominated by the stride
11778 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11780 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11781 on the entire header of the finally block in the fallthru case.
11783 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11785 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11787 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11789 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11790 rs6000_emit_move_from_cr and call renamed function.
11791 (rs6000_emit_prologue): Call renamed functions.
11792 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11793 movesi_from_cr, remove volatile CRs.
11795 2017-08-30 Jon Beniston <jon@beniston.com>
11796 Richard Biener <rguenther@suse.de>
11798 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11799 of VECTOR_MODE_P check.
11800 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11801 element vector types.
11803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11805 * df.h (df_read_modify_subreg_p): Remove in favor of...
11806 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11807 const_rtx instead of an rtx.
11808 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11809 * df-problems.c (df_word_lr_mark_ref): Likewise.
11810 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11811 (mark_pseudo_reg_dead): Likewise.
11812 (mark_ref_dead): Likewise.
11813 * reginfo.c (init_subregs_of_mode): Likewise.
11814 * sched-deps.c (sched_analyze_1): Likewise.
11815 * df-scan.c (df_def_record_1): Likewise.
11816 (df_uses_record): Likewise.
11817 (df_read_modify_subreg_p): Remove in favor of...
11818 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11819 const_rtx instead of an rtx.
11821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11822 Alan Hayward <alan.hayward@arm.com>
11823 David Sherwood <david.sherwood@arm.com>
11825 * rtl.h (partial_subreg_p): New function.
11826 * caller-save.c (save_call_clobbered_regs): Use it.
11827 * calls.c (expand_call): Likewise.
11828 * combine.c (combinable_i3pat): Likewise.
11829 (simplify_set): Likewise.
11830 (make_extraction): Likewise.
11831 (make_compound_operation_int): Likewise.
11832 (gen_lowpart_or_truncate): Likewise.
11833 (force_to_mode): Likewise.
11834 (make_field_assignment): Likewise.
11835 (reg_truncated_to_mode): Likewise.
11836 (record_truncated_value): Likewise.
11837 (move_deaths): Likewise.
11838 * cse.c (record_jump_cond): Likewise.
11839 (cse_insn): Likewise.
11840 * cselib.c (cselib_lookup_1): Likewise.
11841 * expmed.c (extract_bit_field_using_extv): Likewise.
11842 * function.c (assign_parm_setup_reg): Likewise.
11843 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11844 * ira-build.c (create_insn_allocnos): Likewise.
11845 * lra-coalesce.c (merge_pseudos): Likewise.
11846 * lra-constraints.c (match_reload): Likewise.
11847 (simplify_operand_subreg): Likewise.
11848 (curr_insn_transform): Likewise.
11849 * lra-lives.c (process_bb_lives): Likewise.
11850 * lra.c (new_insn_reg): Likewise.
11851 (lra_substitute_pseudo): Likewise.
11852 * regcprop.c (mode_change_ok): Likewise.
11853 (maybe_mode_change): Likewise.
11854 (copyprop_hardreg_forward_1): Likewise.
11855 * reload.c (push_reload): Likewise.
11856 (find_reloads): Likewise.
11857 (find_reloads_subreg_address): Likewise.
11858 * reload1.c (alter_reg): Likewise.
11859 (eliminate_regs_1): Likewise.
11860 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11862 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11864 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11865 back to if statements, including unpack.
11867 2017-08-30 Martin Liska <mliska@suse.cz>
11869 PR inline-asm/82001
11870 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11872 (func_checker::compare_asm_inputs_outputs): ... this function.
11873 (func_checker::compare_gimple_asm): Use the function to compare
11874 also ASM constrains.
11875 * ipa-icf-gimple.h: Rename the function.
11877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11878 Alan Hayward <alan.hayward@arm.com>
11879 David Sherwood <david.sherwood@arm.com>
11881 * coretypes.h (complex_mode): New type.
11882 * gdbhooks.py (build_pretty_printer): Handle it.
11883 * machmode.h (complex_mode): New class.
11884 (complex_mode::includes_p): New function.
11885 (is_complex_int_mode): Likewise.
11886 (is_complex_float_mode): Likewise.
11887 * genmodes.c (get_mode_class): Handle complex mode classes.
11888 * function.c (expand_function_end): Use is_complex_int_mode.
11890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11891 Alan Hayward <alan.hayward@arm.com>
11892 David Sherwood <david.sherwood@arm.com>
11894 * coretypes.h (scalar_mode_pod): New typedef.
11895 * gdbhooks.py (build_pretty_printer): Handle it.
11896 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11897 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11898 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11899 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11900 as_a <scalar_mode>.
11902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11903 Alan Hayward <alan.hayward@arm.com>
11904 David Sherwood <david.sherwood@arm.com>
11906 * machmode.h (mode_for_vector): Take a scalar_mode instead
11908 * stor-layout.c (mode_for_vector): Likewise.
11909 * explow.c (promote_mode): Use as_a <scalar_mode>.
11910 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11913 Alan Hayward <alan.hayward@arm.com>
11914 David Sherwood <david.sherwood@arm.com>
11916 * target.def (preferred_simd_mode): Take a scalar_mode
11917 instead of a machine_mode.
11918 * targhooks.h (default_preferred_simd_mode): Likewise.
11919 * targhooks.c (default_preferred_simd_mode): Likewise.
11920 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11921 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11922 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11923 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11924 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11925 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11926 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11927 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11929 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11930 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11931 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11932 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11933 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11934 * doc/tm.texi: Regenerate.
11935 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 * target.def (scalar_mode_supported_p): Take a scalar_mode
11943 instead of a machine_mode.
11944 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11945 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11946 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11947 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11948 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11949 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11950 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11951 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11952 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11953 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11954 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11955 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11956 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11957 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11959 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11960 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11961 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11962 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11963 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11965 * doc/tm.texi: Regenerate.
11967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11968 Alan Hayward <alan.hayward@arm.com>
11969 David Sherwood <david.sherwood@arm.com>
11971 * coretypes.h (opt_scalar_mode): New typedef.
11972 * gdbhooks.py (build_pretty_printers): Handle it.
11973 * machmode.h (mode_iterator::get_2xwider): Add overload for
11975 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11977 * expr.c (convert_mode_scalar): Likewise.
11978 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11979 * optabs.c (expand_float): Likewise.
11980 (expand_fix): Likewise.
11981 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11984 Alan Hayward <alan.hayward@arm.com>
11985 David Sherwood <david.sherwood@arm.com>
11987 * optabs.c (expand_float): Explicitly check for scalars before
11988 using a branching expansion.
11989 (expand_fix): Likewise.
11991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11992 Alan Hayward <alan.hayward@arm.com>
11993 David Sherwood <david.sherwood@arm.com>
11995 * expr.c (convert_mode): Split scalar handling out into...
11996 (convert_mode_scalar): ...this new function. Treat the modes
11999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12000 Alan Hayward <alan.hayward@arm.com>
12001 David Sherwood <david.sherwood@arm.com>
12003 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12005 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12008 Alan Hayward <alan.hayward@arm.com>
12009 David Sherwood <david.sherwood@arm.com>
12011 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12012 rather than a machine_mode.
12013 (fixed_from_string): Likewise.
12014 (fixed_convert): Likewise.
12015 (fixed_convert_from_int): Likewise.
12016 (fixed_convert_from_real): Likewise.
12017 (real_convert_from_fixed): Likewise.
12018 * fixed-value.c (fixed_from_double_int): Likewise.
12019 (fixed_from_string): Likewise.
12020 (fixed_convert): Likewise.
12021 (fixed_convert_from_int): Likewise.
12022 (fixed_convert_from_real): Likewise.
12023 (real_convert_from_fixed): Likewise.
12024 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12027 Alan Hayward <alan.hayward@arm.com>
12028 David Sherwood <david.sherwood@arm.com>
12030 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12031 of separate mode class checks. Do not allow vector modes here.
12032 (immed_wide_int_const): Use as_a <scalar_mode>.
12033 * explow.c (trunc_int_for_mode): Likewise.
12034 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12035 (wi::shwi): Likewise.
12036 (wi::min_value): Likewise.
12037 (wi::max_value): Likewise.
12038 * dwarf2out.c (loc_descriptor): Likewise.
12039 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12040 for CONST_WIDE_INT.
12042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12043 Alan Hayward <alan.hayward@arm.com>
12044 David Sherwood <david.sherwood@arm.com>
12046 * tree.h (SCALAR_TYPE_MODE): New macro.
12047 * expr.c (expand_expr_addr_expr_1): Use it.
12048 (expand_expr_real_2): Likewise.
12049 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12050 (fold_convert_const_fixed_from_int): Likewise.
12051 (fold_convert_const_fixed_from_real): Likewise.
12052 (native_encode_fixed): Likewise
12053 (native_encode_complex): Likewise
12054 (native_encode_vector): Likewise.
12055 (native_interpret_fixed): Likewise.
12056 (native_interpret_real): Likewise.
12057 (native_interpret_complex): Likewise.
12058 (native_interpret_vector): Likewise.
12059 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12060 (simd_clone_adjust_argument_types): Likewise.
12061 (simd_clone_init_simd_arrays): Likewise.
12062 (simd_clone_adjust): Likewise.
12063 * stor-layout.c (layout_type): Likewise.
12064 * tree.c (build_minus_one_cst): Likewise.
12065 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12066 * tree-inline.c (estimate_move_cost): Likewise.
12067 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12068 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12069 (vectorizable_reduction): Likewise.
12070 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12071 (vect_recog_mixed_size_cond_pattern): Likewise.
12072 (check_bool_pattern): Likewise.
12073 (adjust_bool_pattern): Likewise.
12074 (search_type_for_mask_1): Likewise.
12075 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12076 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12077 (vectorizable_load): Likewise.
12078 (vectorizable_store): Likewise.
12079 * ubsan.c (ubsan_encode_value): Likewise.
12080 * varasm.c (output_constant): Likewise.
12082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12083 Alan Hayward <alan.hayward@arm.com>
12084 David Sherwood <david.sherwood@arm.com>
12086 * coretypes.h (scalar_mode): New class.
12087 * machmode.h (scalar_mode): Likewise.
12088 (scalar_mode::includes_p): New function.
12089 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12090 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12091 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12092 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12093 * expmed.c (store_bit_field_1): Likewise.
12094 (extract_bit_field_1): Likewise.
12095 * expr.c (write_complex_part): Likewise.
12096 (read_complex_part): Likewise.
12097 (emit_move_complex_push): Likewise.
12098 (expand_expr_real_2): Likewise.
12099 * function.c (assign_parm_setup_reg): Likewise.
12100 (assign_parms_unsplit_complex): Likewise.
12101 * optabs.c (expand_binop): Likewise.
12102 * rtlanal.c (subreg_get_info): Likewise.
12103 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12104 * varasm.c (output_constant_pool_2): Likewise.
12106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12107 Alan Hayward <alan.hayward@arm.com>
12108 David Sherwood <david.sherwood@arm.com>
12110 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12112 (expmed_mult_highpart_optab): Likewise.
12113 (expmed_mult_highpart): Likewise.
12115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12116 Alan Hayward <alan.hayward@arm.com>
12117 David Sherwood <david.sherwood@arm.com>
12119 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12120 instead of a machine_mode.
12121 (builtin_memset_read_str): Likewise.
12122 * builtins.c (c_readstr): Likewise.
12123 (builtin_memcpy_read_str): Likewise.
12124 (builtin_strncpy_read_str): Likewise.
12125 (builtin_memset_read_str): Likewise.
12126 (builtin_memset_gen_str): Likewise.
12127 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12128 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12129 instead of a machine_mode.
12130 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12132 (make_extraction): Likewise.
12133 (try_widen_shift_mode): Take and return scalar_int_modes instead
12135 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12136 a scalar_int_mode instead of a machine_mode.
12137 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12138 (avr_addr_space_pointer_mode): Likewise.
12139 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12140 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12141 (msp430_unwind_word_mode): Likewise.
12142 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12143 (spu_addr_space_pointer_mode): Likewise.
12144 (spu_addr_space_address_mode): Likewise.
12145 (spu_libgcc_cmp_return_mode): Likewise.
12146 (spu_libgcc_shift_count_mode): Likewise.
12147 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12148 (rl78_addr_space_pointer_mode): Likewise.
12149 (fl78_unwind_word_mode): Likewise.
12150 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12152 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12153 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12154 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12155 (mips_valid_pointer_mode): Likewise.
12156 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12157 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12158 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12160 (ft32_addr_space_address_mode): Likewise.
12161 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12162 scalar_int_mode instead of a machine_mode.
12163 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12165 (m32c_addr_space_address_mode): Likewise.
12166 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12167 (rs6000_eh_return_filter_mode): Likewise.
12168 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12169 (rs6000_eh_return_filter_mode): Likewise.
12170 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12171 (s390_libgcc_shift_count_mode): Likewise.
12172 (s390_unwind_word_mode): Likewise.
12173 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12175 * target.def (mode_rep_extended): Likewise.
12176 (valid_pointer_mode): Likewise.
12177 (addr_space.valid_pointer_mode): Likewise.
12178 (eh_return_filter_mode): Return a scalar_int_mode rather than
12180 (libgcc_cmp_return_mode): Likewise.
12181 (libgcc_shift_count_mode): Likewise.
12182 (unwind_word_mode): Likewise.
12183 (addr_space.pointer_mode): Likewise.
12184 (addr_space.address_mode): Likewise.
12185 * doc/tm.texi: Regenerate.
12186 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12188 (do_jump): Use scalar_int_mode for local variables.
12189 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12190 rather than a machine_mode.
12191 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12192 (scompare_loc_descriptor_wide): Likewise.
12193 (scompare_loc_descriptor_narrow): Likewise.
12194 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12196 * except.c (sjlj_emit_dispatch_table): Likewise.
12197 (expand_builtin_eh_copy_values): Likewise.
12198 * explow.c (convert_memory_address_addr_space_1): Likewise.
12199 Take a scalar_int_mode rather than a machine_mode.
12200 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12201 than a machine_mode.
12202 (memory_address_addr_space): Use scalar_int_mode for local variables.
12203 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12204 rather than a machine_mode.
12205 * expmed.c (mask_rtx): Likewise.
12206 (init_expmed_one_conv): Likewise.
12207 (expand_mult_highpart_adjust): Likewise.
12208 (extract_high_half): Likewise.
12209 (expmed_mult_highpart_optab): Likewise.
12210 (expmed_mult_highpart): Likewise.
12211 (expand_smod_pow2): Likewise.
12212 (expand_sdiv_pow2): Likewise.
12213 (emit_store_flag_int): Likewise.
12214 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12216 (extract_low_bits): Likewise.
12217 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12219 * expr.c (pieces_addr::adjust): Likewise.
12220 (can_store_by_pieces): Likewise.
12221 (store_by_pieces): Likewise.
12222 (clear_by_pieces_1): Likewise.
12223 (expand_expr_addr_expr_1): Likewise.
12224 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12225 (expand_expr_real_1): Likewise.
12226 (try_casesi): Likewise.
12227 * final.c (shorten_branches): Likewise.
12228 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12229 type of "mode" to machine_mode.
12230 * internal-fn.c (expand_arith_overflow_result_store): Take a
12231 scalar_int_mode rather than a machine_mode.
12232 (expand_mul_overflow): Use scalar_int_mode for local variables.
12233 * loop-doloop.c (doloop_modify): Likewise.
12234 (doloop_optimize): Likewise.
12235 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12236 than a machine_mode.
12237 (expand_doubleword_shift_condmove): Likewise.
12238 (expand_doubleword_shift): Likewise.
12239 (expand_doubleword_clz): Likewise.
12240 (expand_doubleword_popcount): Likewise.
12241 (expand_doubleword_parity): Likewise.
12242 (expand_absneg_bit): Use scalar_int_mode for local variables.
12243 (prepare_float_lib_cmp): Likewise.
12244 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12245 rather than a machine_mode.
12246 (convert_memory_address_addr_space): Likewise.
12247 (get_mode_bounds): Likewise.
12248 (get_address_mode): Return a scalar_int_mode rather than a
12250 * rtlanal.c (get_address_mode): Likewise.
12251 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12252 than a machine_mode.
12253 * targhooks.c (default_mode_rep_extended): Likewise.
12254 (default_valid_pointer_mode): Likewise.
12255 (default_addr_space_valid_pointer_mode): Likewise.
12256 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12257 than a machine_mode.
12258 (default_libgcc_cmp_return_mode): Likewise.
12259 (default_libgcc_shift_count_mode): Likewise.
12260 (default_unwind_word_mode): Likewise.
12261 (default_addr_space_pointer_mode): Likewise.
12262 (default_addr_space_address_mode): Likewise.
12263 * targhooks.h (default_eh_return_filter_mode): Likewise.
12264 (default_libgcc_cmp_return_mode): Likewise.
12265 (default_libgcc_shift_count_mode): Likewise.
12266 (default_unwind_word_mode): Likewise.
12267 (default_addr_space_pointer_mode): Likewise.
12268 (default_addr_space_address_mode): Likewise.
12269 (default_mode_rep_extended): Take a scalar_int_mode rather than
12271 (default_valid_pointer_mode): Likewise.
12272 (default_addr_space_valid_pointer_mode): Likewise.
12273 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12275 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12276 rather than a machine_mode.
12277 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12278 for local variables.
12279 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12280 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12281 than a machine_mode.
12283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12284 Alan Hayward <alan.hayward@arm.com>
12285 David Sherwood <david.sherwood@arm.com>
12287 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12288 the mode argument to scalar_int_mode.
12289 (do_jump_by_parts_zero_rtx): Likewise.
12290 (do_jump_by_parts_equality_rtx): Likewise.
12291 (do_jump_by_parts_greater): Take a mode argument.
12292 (do_jump_by_parts_equality): Likewise.
12293 (do_jump_1): Update calls accordingly.
12295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12296 Alan Hayward <alan.hayward@arm.com>
12297 David Sherwood <david.sherwood@arm.com>
12299 * is-a.h (safe_dyn_cast): New function.
12300 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12301 (jump_table_for_label): Likewise.
12302 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12303 instead of an rtx_insn *.
12304 (shorten_branches): Use dyn_cast instead of LABEL_P and
12305 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12306 rtx_jump_table_data::get_data_mode.
12307 (final_scan_insn): Likewise.
12309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12310 Alan Hayward <alan.hayward@arm.com>
12311 David Sherwood <david.sherwood@arm.com>
12313 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12314 trying to combine a full-register integer set with a subreg
12317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12318 Alan Hayward <alan.hayward@arm.com>
12319 David Sherwood <david.sherwood@arm.com>
12321 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12322 that is always either address_mode or pointer_mode.
12324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12325 Alan Hayward <alan.hayward@arm.com>
12326 David Sherwood <david.sherwood@arm.com>
12328 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12329 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
12336 is_a <scalar_int_mode> instead of checking for a nonzero
12337 precision. Forcibly convert op_mode to a scalar_int_mode
12338 in that case. More clearly differentiate the operand and
12339 result modes and use the former when deciding what the value
12340 of a count-bits operation should be. Use is_int_mode instead
12341 of checking for a MODE_INT. Remove redundant check for whether
12342 this mode has a zero precision.
12344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12345 Alan Hayward <alan.hayward@arm.com>
12346 David Sherwood <david.sherwood@arm.com>
12348 * optabs.c (widen_leading): Change the type of the mode argument
12349 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12350 (widen_bswap): Likewise.
12351 (expand_parity): Likewise.
12352 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12353 (expand_ffs): Likewise.
12354 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12363 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12370 value_mode arguments. Use scalar_int_mode internally.
12371 (store_bit_field_1): Rename the new integer mode from imode
12372 to op0_mode and use it instead of GET_MODE (op0). Update calls
12373 to store_split_bit_field, store_bit_field_using_insv and
12374 store_fixed_bit_field.
12375 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12376 Use scalar_int_mode internally. Use a bit count rather than a mode
12377 when calculating the largest bit size for get_best_mode.
12378 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12379 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12380 Remove assertion that OP0 has a scalar integer mode.
12381 (store_split_bit_field): Add op0_mode and value_mode arguments.
12382 Update calls to extract_fixed_bit_field.
12383 (extract_bit_field_using_extv): Add an op0_mode argument.
12384 Use scalar_int_mode internally.
12385 (extract_bit_field_1): Rename the new integer mode from imode to
12386 op0_mode and use it instead of GET_MODE (op0). Update calls to
12387 extract_split_bit_field, extract_bit_field_using_extv and
12388 extract_fixed_bit_field.
12389 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12390 to extract_split_bit_field and extract_fixed_bit_field_1.
12391 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12392 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12393 on the target mode.
12394 (extract_split_bit_field): Add an op0_mode argument. Update call
12395 to extract_fixed_bit_field.
12397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12398 Alan Hayward <alan.hayward@arm.com>
12399 David Sherwood <david.sherwood@arm.com>
12401 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12402 * explow.c (hard_function_value): Likewise.
12403 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12404 convert_to_mode call outside the loop.
12405 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12406 for the mode iterator. Require the mode specified by max_pieces
12408 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12410 (copy_blkmode_to_reg): Likewise.
12411 (set_storage_via_setmem): Likewise.
12412 * optabs.c (prepare_cmp_insn): Likewise.
12413 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12414 * stor-layout.c (finish_bitfield_representative): Likewise.
12416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12417 Alan Hayward <alan.hayward@arm.com>
12418 David Sherwood <david.sherwood@arm.com>
12420 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12421 * expr.c (convert_move): Use them.
12422 (convert_modes): Likewise.
12423 (store_expr_with_bounds): Likewise.
12425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12426 Alan Hayward <alan.hayward@arm.com>
12427 David Sherwood <david.sherwood@arm.com>
12429 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12430 parameter for the mode of "x". Remove the "known_x", "known_mode"
12431 and "known_ret" arguments. Change the type of the mode argument
12432 to scalar_int_mode.
12433 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12434 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12435 (reg_num_sign_bit_copies_for_combine): Likewise.
12436 * rtlanal.c (nonzero_bits1): Likewise.
12437 (num_sign_bit_copies1): Likewise.
12438 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12439 (reg_num_sign_bit_copies_general): Likewise.
12440 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12441 (reg_nonzero_bits_general): Likewise.
12443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12444 Alan Hayward <alan.hayward@arm.com>
12445 David Sherwood <david.sherwood@arm.com>
12447 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12448 than in subroutines. Return 1 for non-integer modes.
12449 (cached_num_sign_bit_copies): Change the type of the mode parameter
12450 to scalar_int_mode.
12451 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12452 classes. Handle CONST_INT_P first and then check whether X also
12453 has a scalar integer mode. Check the same thing for inner registers
12454 of a SUBREG and for values that are being extended or truncated.
12456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12457 Alan Hayward <alan.hayward@arm.com>
12458 David Sherwood <david.sherwood@arm.com>
12460 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12461 in subroutines. Return the mode mask for non-integer modes.
12462 (cached_nonzero_bits): Change the type of the mode parameter
12463 to scalar_int_mode.
12464 (nonzero_bits1): Likewise. Remove early exit for other mode
12465 classes. Handle CONST_INT_P first and then check whether X
12466 also has a scalar integer mode.
12468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12469 Alan Hayward <alan.hayward@arm.com>
12470 David Sherwood <david.sherwood@arm.com>
12472 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12473 Return a scalar_int_mode and assert that the size is greater than
12475 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12476 (op_by_pieces_d::op_by_pieces_d): Likewise.
12477 (op_by_pieces_d::run): Likewise.
12478 (can_store_by_pieces): Likewise.
12480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12481 Alan Hayward <alan.hayward@arm.com>
12482 David Sherwood <david.sherwood@arm.com>
12484 * combine.c (extract_left_shift): Add a mode argument and update
12486 (make_compound_operation_int): Change the type of the mode parameter
12487 to scalar_int_mode and update the call to extract_left_shift.
12489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12490 Alan Hayward <alan.hayward@arm.com>
12491 David Sherwood <david.sherwood@arm.com>
12493 * combine.c (simplify_and_const_int): Change the type of the mode
12494 parameter to scalar_int_mode.
12495 (simplify_and_const_int_1): Likewise. Update recursive call.
12497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12498 Alan Hayward <alan.hayward@arm.com>
12499 David Sherwood <david.sherwood@arm.com>
12501 * combine.c (simplify_compare_const): Check that the mode is a
12502 scalar_int_mode (rather than VOIDmode) before testing its
12504 (simplify_comparison): Move COMPARISON_P handling out of the
12505 loop and restrict the latter part of the loop to scalar_int_modes.
12506 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12507 and when considering SUBREG_REGs. Use is_int_mode instead of
12508 checking GET_MODE_CLASS against MODE_INT.
12510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12511 Alan Hayward <alan.hayward@arm.com>
12512 David Sherwood <david.sherwood@arm.com>
12514 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12515 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12516 shift_unit_mode and for modes involved in scalar shifts.
12518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12519 Alan Hayward <alan.hayward@arm.com>
12520 David Sherwood <david.sherwood@arm.com>
12522 * combine.c (force_int_to_mode): New function, split out from...
12523 (force_to_mode): ...here. Keep xmode up-to-date and use it
12524 instead of GET_MODE (x).
12526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12527 Alan Hayward <alan.hayward@arm.com>
12528 David Sherwood <david.sherwood@arm.com>
12530 * optabs-query.h (extraction_insn::struct_mode): Change type to
12531 opt_scalar_int_mode and update comment.
12532 (extraction_insn::field_mode): Change type to scalar_int_mode.
12533 (extraction_insn::pos_mode): Likewise.
12534 * combine.c (make_extraction): Update accordingly.
12535 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12536 (get_optab_extraction_insn): Likewise.
12537 * recog.c (simplify_while_replacing): Likewise.
12538 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12539 parameter to opt_scalar_int_mode.
12541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12542 Alan Hayward <alan.hayward@arm.com>
12543 David Sherwood <david.sherwood@arm.com>
12545 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12546 to a scalar_int_mode instead of a machine_mode.
12547 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12548 (get_best_mode): Return a boolean and use a pointer argument to store
12549 the selected mode. Replace the limit mode parameter with a bit limit.
12550 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12551 for the values returned by bit_field_mode_iterator::next_mode.
12552 (store_bit_field): Update call to get_best_mode.
12553 (store_fixed_bit_field): Likewise.
12554 (extract_fixed_bit_field): Likewise.
12555 * expr.c (optimize_bitfield_assignment_op): Likewise.
12556 * fold-const.c (optimize_bit_field_compare): Likewise.
12557 (fold_truth_andor_1): Likewise.
12558 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12559 Update for new type of m_mode.
12560 (get_best_mode): As above.
12562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12563 Alan Hayward <alan.hayward@arm.com>
12564 David Sherwood <david.sherwood@arm.com>
12566 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12567 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12568 (store_bit_field): Check is_a <scalar_int_mode> before calling
12569 strict_volatile_bitfield_p.
12570 (extract_bit_field): Likewise.
12572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12573 Alan Hayward <alan.hayward@arm.com>
12574 David Sherwood <david.sherwood@arm.com>
12576 * target.def (cstore_mode): Return a scalar_int_mode.
12577 * doc/tm.texi: Regenerate.
12578 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12579 * targhooks.h (default_cstore_mode): Likewise.
12580 * targhooks.c (default_cstore_mode): Likewise, using a forced
12582 * expmed.c (emit_cstore): Expect the target of the cstore to be
12585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12586 Alan Hayward <alan.hayward@arm.com>
12587 David Sherwood <david.sherwood@arm.com>
12589 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12591 (niter_desc): Likewise mode.
12592 (iv_analyze): Add a mode parameter.
12594 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12595 and change its type to scalar_int_mode.
12596 * loop-iv.c: Update commentary at head of file.
12597 (iv_constant): Pass the mode paraeter before the rtx it describes
12598 and change its type to scalar_int_mode. Remove VOIDmode handling.
12599 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12600 (iv_extend): Likewise.
12601 (shorten_into_mode): Likewise.
12602 (iv_add): Use scalar_int_mode.
12603 (iv_mult): Likewise.
12604 (iv_shift): Likewise.
12605 (canonicalize_iv_subregs): Likewise.
12606 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12607 it describes and change its mode to scalar_int_mode. Also change
12608 the type of the returned inner_mode to scalar_int_mode.
12609 (get_biv_step): Likewise, turning outer_mode from a pointer
12610 into a direct parameter. Update call to get_biv_step_1.
12611 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12612 iv_constant and get_biv_step.
12613 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12614 and change its type to scalar_int_mode. Don't initialise iv->mode
12615 to VOIDmode and remove later checks for its still being VOIDmode.
12616 Update calls to iv_analyze_op and iv_analyze_expr. Check
12617 is_a <scalar_int_mode> when changing the mode under consideration.
12618 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12619 Update call to iv_analyze_expr.
12620 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12621 inner register is not also a scalar_int_mode. Update call to
12623 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12624 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12625 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12626 separate mode class checks. Update calls to iv_analyze. Remove
12627 fix-up of VOIDmodes after iv_analyze_biv.
12628 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12629 don't have a scalar_int_mode. Update call to biv_p.
12631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12632 Alan Hayward <alan.hayward@arm.com>
12633 David Sherwood <david.sherwood@arm.com>
12635 * cfgexpand.c (convert_debug_memory_address): Use
12636 as_a <scalar_int_mode>.
12637 * combine.c (expand_compound_operation): Likewise.
12638 (make_extraction): Likewise.
12639 (change_zero_ext): Likewise.
12640 (simplify_comparison): Likewise.
12641 * cse.c (cse_insn): Likewise.
12642 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12643 (mem_loc_descriptor): Likewise.
12644 (loc_descriptor): Likewise.
12645 * expmed.c (init_expmed_one_mode): Likewise.
12646 (synth_mult): Likewise.
12647 (emit_store_flag_1): Likewise.
12648 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12649 of a comparison with size.
12650 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12651 (reduce_to_bit_field_precision): Likewise.
12652 * function.c (expand_function_end): Likewise.
12653 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12654 * loop-doloop.c (doloop_modify): Likewise.
12655 * optabs.c (expand_binop): Likewise.
12656 (expand_unop): Likewise.
12657 (expand_copysign_absneg): Likewise.
12658 (prepare_cmp_insn): Likewise.
12659 (maybe_legitimize_operand): Likewise.
12660 * recog.c (const_scalar_int_operand): Likewise.
12661 * rtlanal.c (get_address_mode): Likewise.
12662 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12663 (simplify_cond_clz_ctz): Likewise.
12664 * tree-nested.c (get_nl_goto_field): Likewise.
12665 * tree.c (build_vector_type_for_mode): Likewise.
12666 * var-tracking.c (use_narrower_mode): Likewise.
12668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12669 Alan Hayward <alan.hayward@arm.com>
12670 David Sherwood <david.sherwood@arm.com>
12672 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12673 * builtins.c (expand_builtin_signbit): Use it.
12674 * cfgexpand.c (expand_debug_expr): Likewise.
12675 * dojump.c (do_jump): Likewise.
12676 (do_compare_and_jump): Likewise.
12677 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12678 * expmed.c (make_tree): Likewise.
12679 * expr.c (expand_expr_real_2): Likewise.
12680 (expand_expr_real_1): Likewise.
12681 (try_casesi): Likewise.
12682 * fold-const-call.c (fold_const_call_ss): Likewise.
12683 * fold-const.c (unextend): Likewise.
12684 (extract_muldiv_1): Likewise.
12685 (fold_single_bit_test): Likewise.
12686 (native_encode_int): Likewise.
12687 (native_encode_string): Likewise.
12688 (native_interpret_int): Likewise.
12689 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12690 * internal-fn.c (expand_addsub_overflow): Likewise.
12691 (expand_neg_overflow): Likewise.
12692 (expand_mul_overflow): Likewise.
12693 (expand_arith_overflow): Likewise.
12694 * match.pd: Likewise.
12695 * stor-layout.c (layout_type): Likewise.
12696 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12697 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12698 * tree-ssanames.c (get_range_info): Likewise.
12699 * tree-switch-conversion.c (array_value_type) Likewise.
12700 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12701 (vect_recog_divmod_pattern): Likewise.
12702 (vect_recog_mixed_size_cond_pattern): Likewise.
12703 * tree-vrp.c (extract_range_basic): Likewise.
12704 (simplify_float_conversion_using_ranges): Likewise.
12705 * tree.c (int_fits_type_p): Likewise.
12706 * ubsan.c (instrument_bool_enum_load): Likewise.
12707 * varasm.c (mergeable_string_section): Likewise.
12708 (narrowing_initializer_constant_valid_p): Likewise.
12709 (output_constant): Likewise.
12711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12712 Alan Hayward <alan.hayward@arm.com>
12713 David Sherwood <david.sherwood@arm.com>
12715 * machmode.h (NARROWEST_INT_MODE): New macro.
12716 * expr.c (alignment_for_piecewise_move): Use it instead of
12717 GET_CLASS_NARROWEST_MODE (MODE_INT).
12718 (push_block): Likewise.
12719 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12721 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12724 Alan Hayward <alan.hayward@arm.com>
12725 David Sherwood <david.sherwood@arm.com>
12727 * postreload.c (move2add_valid_value_p): Change the type of the
12728 mode parameter to scalar_int_mode.
12729 (move2add_use_add2_insn): Add a mode parameter and use it instead
12731 (move2add_use_add3_insn): Likewise.
12732 (reload_cse_move2add): Update accordingly.
12734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12735 Alan Hayward <alan.hayward@arm.com>
12736 David Sherwood <david.sherwood@arm.com>
12738 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12740 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12741 * optabs.c (expand_unop): Likewise.
12743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12744 Alan Hayward <alan.hayward@arm.com>
12745 David Sherwood <david.sherwood@arm.com>
12747 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12748 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12749 (popcount_loc_descriptor): Likewise.
12750 (bswap_loc_descriptor): Likewise.
12751 (rotate_loc_descriptor): Likewise.
12752 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12753 calling the functions above.
12755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12756 Alan Hayward <alan.hayward@arm.com>
12757 David Sherwood <david.sherwood@arm.com>
12759 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12761 (try_combine): Likewise.
12762 (simplify_if_then_else): Likewise.
12763 * cse.c (cse_insn): Likewise.
12764 * dwarf2out.c (mem_loc_descriptor): Likewise.
12765 * emit-rtl.c (gen_lowpart_common): Likewise.
12766 * simplify-rtx.c (simplify_truncation): Likewise.
12767 (simplify_binary_operation_1): Likewise.
12768 (simplify_const_relational_operation): Likewise.
12769 (simplify_ternary_operation): Likewise.
12770 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12773 Alan Hayward <alan.hayward@arm.com>
12774 David Sherwood <david.sherwood@arm.com>
12776 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12777 * reload.c (push_reload): Likewise.
12778 (find_reloads): Likewise.
12780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12781 Alan Hayward <alan.hayward@arm.com>
12782 David Sherwood <david.sherwood@arm.com>
12784 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12785 (make_compound_operation_int): Likewise.
12786 (change_zero_ext): Likewise.
12787 * expr.c (convert_move): Likewise.
12788 (convert_modes): Likewise.
12789 * fwprop.c (forward_propagate_subreg): Likewise.
12790 * loop-iv.c (get_biv_step_1): Likewise.
12791 * optabs.c (widen_operand): Likewise.
12792 * postreload.c (move2add_valid_value_p): Likewise.
12793 * recog.c (simplify_while_replacing): Likewise.
12794 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12795 (simplify_binary_operation_1): Likewise. Remove redundant
12796 mode equality check.
12798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12799 Alan Hayward <alan.hayward@arm.com>
12800 David Sherwood <david.sherwood@arm.com>
12802 * combine.c (combine_simplify_rtx): Add checks for
12803 is_a <scalar_int_mode>.
12804 (simplify_if_then_else): Likewise.
12805 (make_field_assignment): Likewise.
12806 (simplify_comparison): Likewise.
12807 * ifcvt.c (noce_try_bitop): Likewise.
12808 * loop-invariant.c (canonicalize_address_mult): Likewise.
12809 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12812 Alan Hayward <alan.hayward@arm.com>
12813 David Sherwood <david.sherwood@arm.com>
12815 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12816 is_a <scalar_int_mode> instead of != BLKmode.
12818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12819 Alan Hayward <alan.hayward@arm.com>
12820 David Sherwood <david.sherwood@arm.com>
12822 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12823 instead of != VOIDmode.
12824 * combine.c (if_then_else_cond): Likewise.
12825 (change_zero_ext): Likewise.
12826 * dwarf2out.c (mem_loc_descriptor): Likewise.
12827 (loc_descriptor): Likewise.
12828 * rtlanal.c (canonicalize_condition): Likewise.
12829 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12832 Alan Hayward <alan.hayward@arm.com>
12833 David Sherwood <david.sherwood@arm.com>
12835 * simplify-rtx.c (simplify_binary_operation_1): Use
12836 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12839 Alan Hayward <alan.hayward@arm.com>
12840 David Sherwood <david.sherwood@arm.com>
12842 * wide-int.h (int_traits<unsigned char>) New class.
12843 (int_traits<unsigned short>) Likewise.
12844 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12845 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12847 * combine.c (set_nonzero_bits_and_sign_copies): Use
12848 is_a <scalar_int_mode>.
12849 (find_split_point): Likewise.
12850 (combine_simplify_rtx): Likewise.
12851 (simplify_logical): Likewise.
12852 (expand_compound_operation): Likewise.
12853 (expand_field_assignment): Likewise.
12854 (make_compound_operation): Likewise.
12855 (extended_count): Likewise.
12856 (change_zero_ext): Likewise.
12857 (simplify_comparison): Likewise.
12858 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12859 (ucompare_loc_descriptor): Likewise.
12860 (minmax_loc_descriptor): Likewise.
12861 (mem_loc_descriptor): Likewise.
12862 (loc_descriptor): Likewise.
12863 * expmed.c (init_expmed_one_mode): Likewise.
12864 * lra-constraints.c (lra_constraint_offset): Likewise.
12865 * optabs.c (prepare_libcall_arg): Likewise.
12866 * postreload.c (move2add_note_store): Likewise.
12867 * reload.c (operands_match_p): Likewise.
12868 * rtl.h (load_extend_op): Likewise.
12869 * rtlhooks.c (gen_lowpart_general): Likewise.
12870 * simplify-rtx.c (simplify_truncation): Likewise.
12871 (simplify_unary_operation_1): Likewise.
12872 (simplify_binary_operation_1): Likewise.
12873 (simplify_const_binary_operation): Likewise.
12874 (simplify_const_relational_operation): Likewise.
12875 (simplify_subreg): Likewise.
12876 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12877 * var-tracking.c (adjust_mems): Likewise.
12878 (prepare_call_arguments): Likewise.
12880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12881 Alan Hayward <alan.hayward@arm.com>
12882 David Sherwood <david.sherwood@arm.com>
12884 * machmode.h (is_int_mode): New fuction.
12885 * combine.c (find_split_point): Use it.
12886 (combine_simplify_rtx): Likewise.
12887 (simplify_if_then_else): Likewise.
12888 (simplify_set): Likewise.
12889 (simplify_shift_const_1): Likewise.
12890 (simplify_comparison): Likewise.
12891 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12892 * cse.c (notreg_cost): Likewise.
12893 (cse_insn): Likewise.
12894 * cselib.c (cselib_lookup_1): Likewise.
12895 * dojump.c (do_jump_1): Likewise.
12896 (do_compare_rtx_and_jump): Likewise.
12897 * dse.c (get_call_args): Likewise.
12898 * dwarf2out.c (rtl_for_decl_init): Likewise.
12899 (native_encode_initializer): Likewise.
12900 * expmed.c (emit_store_flag_1): Likewise.
12901 (emit_store_flag): Likewise.
12902 * expr.c (convert_modes): Likewise.
12903 (store_field): Likewise.
12904 (expand_expr_real_1): Likewise.
12905 * fold-const.c (fold_read_from_constant_string): Likewise.
12906 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12907 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12908 * optabs.c (expand_binop): Likewise.
12909 (expand_unop): Likewise.
12910 (expand_abs_nojump): Likewise.
12911 (expand_one_cmpl_abs_nojump): Likewise.
12912 * simplify-rtx.c (mode_signbit_p): Likewise.
12913 (val_signbit_p): Likewise.
12914 (val_signbit_known_set_p): Likewise.
12915 (val_signbit_known_clear_p): Likewise.
12916 (simplify_relational_operation_1): Likewise.
12917 * tree.c (vector_type_mode): Likewise.
12919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12920 Alan Hayward <alan.hayward@arm.com>
12921 David Sherwood <david.sherwood@arm.com>
12923 * machmode.h (smallest_mode_for_size): Fix formatting.
12924 (smallest_int_mode_for_size): New function.
12925 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12926 instead of smallest_mode_for_size.
12927 * combine.c (make_extraction): Likewise.
12928 * config/arc/arc.c (arc_expand_movmem): Likewise.
12929 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12930 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12931 * config/s390/s390.c (s390_expand_insv): Likewise.
12932 * config/sparc/sparc.c (assign_int_registers): Likewise.
12933 * config/spu/spu.c (spu_function_value): Likewise.
12934 (spu_function_arg): Likewise.
12935 * coverage.c (get_gcov_type): Likewise.
12936 (get_gcov_unsigned_t): Likewise.
12937 * dse.c (find_shift_sequence): Likewise.
12938 * expmed.c (store_bit_field_1): Likewise.
12939 * expr.c (convert_move): Likewise.
12940 (store_field): Likewise.
12941 * internal-fn.c (expand_arith_overflow): Likewise.
12942 * optabs-query.c (get_best_extraction_insn): Likewise.
12943 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12944 * stor-layout.c (layout_type): Likewise.
12945 (initialize_sizetypes): Likewise.
12946 * targhooks.c (default_get_mask_mode): Likewise.
12947 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12950 Alan Hayward <alan.hayward@arm.com>
12951 David Sherwood <david.sherwood@arm.com>
12953 * machmode.h (opt_mode::else_blk): New function.
12954 (int_mode_for_mode): Declare.
12955 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12956 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12958 * cfgexpand.c (expand_debug_expr): Likewise.
12959 * combine.c (gen_lowpart_or_truncate): Likewise.
12960 (gen_lowpart_for_combine): Likewise.
12961 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12962 * config/avr/avr.c (avr_to_int_mode): Likewise.
12963 (avr_out_plus_1): Likewise.
12964 (avr_out_plus): Likewise.
12965 (avr_out_round): Likewise.
12966 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12967 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12968 (s390_expand_vcond): Likewise.
12969 * config/spu/spu.c (spu_split_immediate): Likewise.
12970 (spu_expand_mov): Likewise.
12971 * dse.c (get_stored_val): Likewise.
12972 * expmed.c (store_bit_field_1): Likewise.
12973 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12975 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12976 (extract_low_bits): Likewise.
12977 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12978 handling rather than repeating the check.
12979 (emit_group_store): Likewise.
12980 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12981 * optabs.c (expand_absneg_bit): Likewise.
12982 (expand_copysign_absneg): Likewise.
12983 (expand_copysign_bit): Likewise.
12984 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12985 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12986 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12987 * var-tracking.c (prepare_call_arguments): Likewise.
12988 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12989 int_mode_for_mode instead of mode_for_size.
12990 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12993 Alan Hayward <alan.hayward@arm.com>
12994 David Sherwood <david.sherwood@arm.com>
12996 * machmode.h (int_mode_for_size): New function.
12997 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12998 instead of mode_for_size.
12999 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13001 * combine.c (expand_field_assignment): Use int_mode_for_size
13002 instead of mode_for_size.
13003 (make_extraction): Likewise.
13004 (simplify_shift_const_1): Likewise.
13005 (simplify_comparison): Likewise.
13006 * dojump.c (do_jump): Likewise.
13007 * dwarf2out.c (mem_loc_descriptor): Likewise.
13008 * emit-rtl.c (init_derived_machine_modes): Likewise.
13009 * expmed.c (flip_storage_order): Likewise.
13010 (convert_extracted_bit_field): Likewise.
13011 * expr.c (copy_blkmode_from_reg): Likewise.
13012 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13013 * internal-fn.c (expand_mul_overflow): Likewise.
13014 * lower-subreg.c (simple_move): Likewise.
13015 * optabs-libfuncs.c (init_optabs): Likewise.
13016 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13017 * tree.c (vector_type_mode): Likewise.
13018 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13019 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13020 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13021 * tree-vect-stmts.c (vectorizable_load): Likewise.
13022 (vectorizable_store): Likewise.
13024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13025 Alan Hayward <alan.hayward@arm.com>
13026 David Sherwood <david.sherwood@arm.com>
13028 * coretypes.h (pod_mode): New type.
13029 (scalar_int_mode_pod): New typedef.
13030 * machmode.h (pod_mode): New class.
13031 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13032 * genmodes.c (emit_mode_int_n): Update accordingly.
13033 * lower-subreg.h (target_lower_subreg): Change type to
13034 scalar_int_mode_pod.
13035 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13036 scalar_int_mode_pod.
13038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13039 Alan Hayward <alan.hayward@arm.com>
13040 David Sherwood <david.sherwood@arm.com>
13042 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13043 machine_mode to scalar_int_mode.
13044 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13045 (rs6000_option_override_internal): Remove cast to int.
13046 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13047 machine_mode to scalar_int_mode.
13048 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13049 (rs6000_option_override_internal): Remove cast to int.
13050 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13051 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13053 * config/s390/s390.c (s390_expand_builtin): Likewise.
13054 * coretypes.h (scalar_int_mode): New type.
13055 (opt_scalar_int_mode): New typedef.
13056 * machmode.h (scalar_int_mode): New class.
13057 (scalar_int_mode::includes_p): New function.
13058 (byte_mode): Change type to scalar_int_mode.
13059 (word_mode): Likewise.
13060 (ptr_mode): Likewise.
13061 * emit-rtl.c (byte_mode): Likewise.
13062 (word_mode): Likewise.
13063 (ptr_mode): Likewise.
13064 (init_derived_machine_modes): Update accordingly.
13065 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13066 and MODE_PARTIAL_INT.
13067 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13068 opt_scalar_int_mode.
13070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13071 Alan Hayward <alan.hayward@arm.com>
13072 David Sherwood <david.sherwood@arm.com>
13074 * target.def (libgcc_floating_mode_supported_p): Take a
13076 * doc/tm.texi: Regenerate.
13077 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13079 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13080 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13084 Alan Hayward <alan.hayward@arm.com>
13085 David Sherwood <david.sherwood@arm.com>
13087 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13088 * doc/tm.texi: Regenerate.
13089 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13090 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13091 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13092 * targhooks.h (default_floatn_mode): Likewise.
13093 * targhooks.c (default_floatn_mode): Likewise.
13094 * tree.c (build_common_tree_nodes): Update accordingly.
13096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13097 Alan Hayward <alan.hayward@arm.com>
13098 David Sherwood <david.sherwood@arm.com>
13100 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13101 (mode_iterator::iterate_p): Likewise.
13102 (mode_iterator::get_wider): Likewise.
13103 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13106 Alan Hayward <alan.hayward@arm.com>
13107 David Sherwood <david.sherwood@arm.com>
13109 * coretypes.h (opt_scalar_float_mode): New typedef.
13110 * machmode.h (float_mode_for_size): New function.
13111 * emit-rtl.c (double_mode): Delete.
13112 (init_emit_once): Use float_mode_for_size.
13113 * stor-layout.c (layout_type): Likewise.
13114 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13117 Alan Hayward <alan.hayward@arm.com>
13118 David Sherwood <david.sherwood@arm.com>
13120 * output.h (assemble_real): Take a scalar_float_mode.
13121 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13122 * config/arm/arm.md (consttable_4): Likewise.
13123 (consttable_8): Likewise.
13124 (consttable_16): Likewise.
13125 * config/mips/mips.md (consttable_float): Likewise.
13126 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13127 * varasm.c (assemble_real): Take a scalar_float_mode.
13128 (output_constant_pool_2): Update accordingly.
13129 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13136 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13137 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13138 (native_encode_real): Likewise.
13139 (native_interpret_real): Likewise.
13140 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13141 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13144 Alan Hayward <alan.hayward@arm.com>
13145 David Sherwood <david.sherwood@arm.com>
13147 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13148 <scalar_float_mode>. Simplify.
13149 (gen_extend_conv_libfunc): Likewise.
13151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13152 Alan Hayward <alan.hayward@arm.com>
13153 David Sherwood <david.sherwood@arm.com>
13155 * coretypes.h (scalar_float_mode): New type.
13156 * machmode.h (mode_traits::from_int): Use machine_mode if
13157 USE_ENUM_MODES is defined.
13158 (is_a): New function.
13160 (dyn_cast): Likewise.
13161 (scalar_float_mode): New class.
13162 (scalar_float_mode::includes_p): New function.
13163 (is_float_mode): Likewise.
13164 * gdbhooks.py (MachineModePrinter): New class.
13165 (build_pretty_printer): Use it for scalar_float_mode.
13166 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13167 (format_helper::format_helper): Turn into a template.
13168 * genmodes.c (get_mode_class): New function.
13169 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13170 or machine_mode if none.
13171 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13172 as_a <scalar_float_mode>.
13173 * dwarf2out.c (mem_loc_descriptor): Likewise.
13174 (insert_float): Likewise.
13175 (add_const_value_attribute): Likewise.
13176 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13177 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13178 (expand_unop): Update accordingly.
13179 (expand_abs_nojump): Likewise.
13180 (expand_copysign_absneg): Take a scalar_float_mode.
13181 (expand_copysign_bit): Likewise.
13182 (expand_copysign): Update accordingly.
13184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13185 Alan Hayward <alan.hayward@arm.com>
13186 David Sherwood <david.sherwood@arm.com>
13188 * coretypes.h (opt_mode): New class.
13189 * machmode.h (opt_mode): Likewise.
13190 (opt_mode::else_void): New function.
13191 (opt_mode::require): Likewise.
13192 (opt_mode::exists): Likewise.
13193 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13194 (GET_MODE_2XWIDER_MODE): Likewise.
13195 (mode_iterator::get_wider): Update accordingly.
13196 (mode_iterator::get_2xwider): Likewise.
13197 (mode_iterator::get_known_wider): Likewise, turning into a template.
13198 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13199 forcing a wider mode to exist.
13200 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13201 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13202 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13203 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13204 * lower-subreg.c (init_lower_subreg): Likewise.
13205 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13206 on the final iteration.
13207 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13208 a wider mode exists before asking for a move pattern.
13209 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13210 forcing a wider mode to exist.
13211 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13212 returning false if no such mode exists.
13213 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13214 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13215 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13216 Avoid checking for a MODE_INT if we already know the mode is not a
13218 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13219 forcing a wider mode to exist.
13220 (expmed_mult_highpart_optab): Likewise.
13221 (expmed_mult_highpart): Likewise.
13222 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13224 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13225 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13226 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13227 * internal-fn.c (expand_mul_overflow): Update use of
13228 GET_MODE_2XWIDER_MODE.
13229 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13230 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13231 GET_MODE_WIDER_MODE.
13232 (convert_plusminus_to_widen): Likewise.
13233 * tree-switch-conversion.c (array_value_type): Likewise.
13234 * var-tracking.c (emit_note_insn_var_location): Likewise.
13235 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13236 Return false inside rather than outside the loop if no wider mode
13238 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13239 and GET_MODE_2XWIDER_MODE
13240 (can_compare_p): Use else_void.
13241 * gdbhooks.py (OptMachineModePrinter): New class.
13242 (build_pretty_printer): Use it for opt_mode.
13244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13245 Alan Hayward <alan.hayward@arm.com>
13246 David Sherwood <david.sherwood@arm.com>
13248 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13249 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13252 Alan Hayward <alan.hayward@arm.com>
13253 David Sherwood <david.sherwood@arm.com>
13255 * machmode.h (mode_traits): New structure.
13256 (get_narrowest_mode): New function.
13257 (mode_iterator::start): Likewise.
13258 (mode_iterator::iterate_p): Likewise.
13259 (mode_iterator::get_wider): Likewise.
13260 (mode_iterator::get_known_wider): Likewise.
13261 (mode_iterator::get_2xwider): Likewise.
13262 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13263 (FOR_EACH_MODE): Likewise.
13264 (FOR_EACH_MODE_FROM): Likewise.
13265 (FOR_EACH_MODE_UNTIL): Likewise.
13266 (FOR_EACH_WIDER_MODE): Likewise.
13267 (FOR_EACH_2XWIDER_MODE): Likewise.
13268 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13269 * combine.c (simplify_comparison): Likewise
13270 * config/i386/i386.c (type_natural_mode): Likewise.
13271 * cse.c (cse_insn): Likewise.
13272 * dse.c (find_shift_sequence): Likewise.
13273 * emit-rtl.c (init_derived_machine_modes): Likewise.
13274 (init_emit_once): Likewise.
13275 * explow.c (hard_function_value): Likewise.
13276 * expmed.c (extract_fixed_bit_field_1): Likewise.
13277 (extract_bit_field_1): Likewise.
13278 (expand_divmod): Likewise.
13279 (emit_store_flag_1): Likewise.
13280 * expr.c (init_expr_target): Likewise.
13281 (convert_move): Likewise.
13282 (alignment_for_piecewise_move): Likewise.
13283 (widest_int_mode_for_size): Likewise.
13284 (emit_block_move_via_movmem): Likewise.
13285 (copy_blkmode_to_reg): Likewise.
13286 (set_storage_via_setmem): Likewise.
13287 (compress_float_constant): Likewise.
13288 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13289 * optabs-query.c (get_best_extraction_insn): Likewise.
13290 * optabs.c (expand_binop): Likewise.
13291 (expand_twoval_unop): Likewise.
13292 (expand_twoval_binop): Likewise.
13293 (widen_leading): Likewise.
13294 (widen_bswap): Likewise.
13295 (expand_parity): Likewise.
13296 (expand_unop): Likewise.
13297 (prepare_cmp_insn): Likewise.
13298 (prepare_float_lib_cmp): Likewise.
13299 (expand_float): Likewise.
13300 (expand_fix): Likewise.
13301 (expand_sfix_optab): Likewise.
13302 * postreload.c (move2add_use_add2_insn): Likewise.
13303 * reg-stack.c (reg_to_stack): Likewise.
13304 * reginfo.c (choose_hard_reg_mode): Likewise.
13305 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13306 * stor-layout.c (mode_for_size): Likewise.
13307 (smallest_mode_for_size): Likewise.
13308 (mode_for_vector): Likewise.
13309 (finish_bitfield_representative): Likewise.
13310 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13311 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13312 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13313 * var-tracking.c (prepare_call_arguments): Likewise.
13315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13316 Alan Hayward <alan.hayward@arm.com>
13317 David Sherwood <david.sherwood@arm.com>
13319 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13320 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13321 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13322 * machmode.h (mode_size): Move earlier in file.
13323 (mode_precision): Likewise.
13324 (mode_inner): Likewise.
13325 (mode_nunits): Likewise.
13326 (mode_unit_size): Likewise.
13327 (unit_unit_precision): Likewise.
13328 (mode_wider): Likewise.
13329 (mode_2xwider): Likewise.
13330 (machine_mode): New class.
13331 (mode_to_bytes): New function.
13332 (mode_to_bits): Likewise.
13333 (mode_to_precision): Likewise.
13334 (mode_to_inner): Likewise.
13335 (mode_to_unit_size): Likewise.
13336 (mode_to_unit_precision): Likewise.
13337 (mode_to_nunits): Likewise.
13338 (GET_MODE_SIZE): Use mode_to_bytes.
13339 (GET_MODE_BITSIZE): Use mode_to_bits.
13340 (GET_MODE_PRECISION): Use mode_to_precision.
13341 (GET_MODE_INNER): Use mode_to_inner.
13342 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13343 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13344 (GET_MODE_NUNITS): Use mode_to_nunits.
13345 * system.h (ALWAYS_INLINE): New macro.
13346 * config/powerpcspe/powerpcspe-c.c
13347 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13348 int for arg1_mode and arg2_mode.
13350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13351 Alan Hayward <alan.hayward@arm.com>
13352 David Sherwood <david.sherwood@arm.com>
13354 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13355 Prefix mode names with E_ in case statements.
13356 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13357 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13358 (aarch64_split_simd_move): Likewise.
13359 (aarch64_gen_storewb_pair): Likewise.
13360 (aarch64_gen_loadwb_pair): Likewise.
13361 (aarch64_gen_store_pair): Likewise.
13362 (aarch64_gen_load_pair): Likewise.
13363 (aarch64_get_condition_code_1): Likewise.
13364 (aarch64_constant_pool_reload_icode): Likewise.
13365 (get_rsqrte_type): Likewise.
13366 (get_rsqrts_type): Likewise.
13367 (get_recpe_type): Likewise.
13368 (get_recps_type): Likewise.
13369 (aarch64_gimplify_va_arg_expr): Likewise.
13370 (aarch64_simd_container_mode): Likewise.
13371 (aarch64_emit_load_exclusive): Likewise.
13372 (aarch64_emit_store_exclusive): Likewise.
13373 (aarch64_expand_compare_and_swap): Likewise.
13374 (aarch64_gen_atomic_cas): Likewise.
13375 (aarch64_emit_bic): Likewise.
13376 (aarch64_emit_atomic_swap): Likewise.
13377 (aarch64_emit_atomic_load_op): Likewise.
13378 (aarch64_evpc_trn): Likewise.
13379 (aarch64_evpc_uzp): Likewise.
13380 (aarch64_evpc_zip): Likewise.
13381 (aarch64_evpc_ext): Likewise.
13382 (aarch64_evpc_rev): Likewise.
13383 (aarch64_evpc_dup): Likewise.
13384 (aarch64_gen_ccmp_first): Likewise.
13385 (aarch64_gen_ccmp_next): Likewise.
13386 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13387 (alpha_emit_xfloating_libcall): Likewise.
13388 (emit_insxl): Likewise.
13389 (alpha_arg_type): Likewise.
13390 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13391 (arc_preferred_simd_mode): Likewise.
13392 (arc_secondary_reload): Likewise.
13393 (get_arc_condition_code): Likewise.
13394 (arc_print_operand): Likewise.
13395 (arc_legitimate_constant_p): Likewise.
13396 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13397 * config/arc/arc.md (casesi_load): Likewise.
13398 (casesi_compact_jump): Likewise.
13399 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13400 (cc_use_register): Likewise.
13401 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13402 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13403 (arm_init_iwmmxt_builtins): Likewise.
13404 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13405 (neon_expand_vector_init): Likewise.
13406 (arm_attr_length_move_neon): Likewise.
13407 (maybe_get_arm_condition_code): Likewise.
13408 (arm_emit_vector_const): Likewise.
13409 (arm_preferred_simd_mode): Likewise.
13410 (arm_output_iwmmxt_tinsr): Likewise.
13411 (thumb1_output_casesi): Likewise.
13412 (thumb2_output_casesi): Likewise.
13413 (arm_emit_load_exclusive): Likewise.
13414 (arm_emit_store_exclusive): Likewise.
13415 (arm_expand_compare_and_swap): Likewise.
13416 (arm_evpc_neon_vuzp): Likewise.
13417 (arm_evpc_neon_vzip): Likewise.
13418 (arm_evpc_neon_vrev): Likewise.
13419 (arm_evpc_neon_vtrn): Likewise.
13420 (arm_evpc_neon_vext): Likewise.
13421 (arm_validize_comparison): Likewise.
13422 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13423 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13424 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13425 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13426 (c6x_preferred_simd_mode): Likewise.
13427 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13428 (epiphany_rtx_costs): Likewise.
13429 * config/epiphany/predicates.md (proper_comparison_operator):
13431 * config/frv/frv.c (condexec_memory_operand): Likewise.
13432 (frv_emit_move): Likewise.
13433 (output_move_single): Likewise.
13434 (output_condmove_single): Likewise.
13435 (frv_hard_regno_mode_ok): Likewise.
13436 (frv_matching_accg_mode): Likewise.
13437 * config/h8300/h8300.c (split_adds_subs): Likewise.
13438 (h8300_rtx_costs): Likewise.
13439 (h8300_print_operand): Likewise.
13440 (compute_mov_length): Likewise.
13441 (output_logical_op): Likewise.
13442 (compute_logical_op_length): Likewise.
13443 (compute_logical_op_cc): Likewise.
13444 (h8300_shift_needs_scratch_p): Likewise.
13445 (output_a_shift): Likewise.
13446 (compute_a_shift_length): Likewise.
13447 (compute_a_shift_cc): Likewise.
13448 (expand_a_rotate): Likewise.
13449 (output_a_rotate): Likewise.
13450 * config/i386/i386.c (classify_argument): Likewise.
13451 (function_arg_advance_32): Likewise.
13452 (function_arg_32): Likewise.
13453 (function_arg_64): Likewise.
13454 (function_value_64): Likewise.
13455 (ix86_gimplify_va_arg): Likewise.
13456 (ix86_legitimate_constant_p): Likewise.
13457 (put_condition_code): Likewise.
13458 (split_double_mode): Likewise.
13459 (ix86_avx256_split_vector_move_misalign): Likewise.
13460 (ix86_expand_vector_logical_operator): Likewise.
13461 (ix86_split_idivmod): Likewise.
13462 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13463 (ix86_build_const_vector): Likewise.
13464 (ix86_build_signbit_mask): Likewise.
13465 (ix86_match_ccmode): Likewise.
13466 (ix86_cc_modes_compatible): Likewise.
13467 (ix86_expand_branch): Likewise.
13468 (ix86_expand_sse_cmp): Likewise.
13469 (ix86_expand_sse_movcc): Likewise.
13470 (ix86_expand_int_sse_cmp): Likewise.
13471 (ix86_expand_vec_perm_vpermi2): Likewise.
13472 (ix86_expand_vec_perm): Likewise.
13473 (ix86_expand_sse_unpack): Likewise.
13474 (ix86_expand_int_addcc): Likewise.
13475 (ix86_split_to_parts): Likewise.
13476 (ix86_vectorize_builtin_gather): Likewise.
13477 (ix86_vectorize_builtin_scatter): Likewise.
13478 (avx_vpermilp_parallel): Likewise.
13479 (inline_memory_move_cost): Likewise.
13480 (ix86_tieable_integer_mode_p): Likewise.
13481 (x86_maybe_negate_const_int): Likewise.
13482 (ix86_expand_vector_init_duplicate): Likewise.
13483 (ix86_expand_vector_init_one_nonzero): Likewise.
13484 (ix86_expand_vector_init_one_var): Likewise.
13485 (ix86_expand_vector_init_concat): Likewise.
13486 (ix86_expand_vector_init_interleave): Likewise.
13487 (ix86_expand_vector_init_general): Likewise.
13488 (ix86_expand_vector_set): Likewise.
13489 (ix86_expand_vector_extract): Likewise.
13490 (emit_reduc_half): Likewise.
13491 (ix86_emit_i387_round): Likewise.
13492 (ix86_mangle_type): Likewise.
13493 (ix86_expand_round_sse4): Likewise.
13494 (expand_vec_perm_blend): Likewise.
13495 (canonicalize_vector_int_perm): Likewise.
13496 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13497 (expand_vec_perm_1): Likewise.
13498 (expand_vec_perm_interleave3): Likewise.
13499 (expand_vec_perm_even_odd_pack): Likewise.
13500 (expand_vec_perm_even_odd_1): Likewise.
13501 (expand_vec_perm_broadcast_1): Likewise.
13502 (ix86_vectorize_vec_perm_const_ok): Likewise.
13503 (ix86_expand_vecop_qihi): Likewise.
13504 (ix86_expand_mul_widen_hilo): Likewise.
13505 (ix86_expand_sse2_abs): Likewise.
13506 (ix86_expand_pextr): Likewise.
13507 (ix86_expand_pinsr): Likewise.
13508 (ix86_preferred_simd_mode): Likewise.
13509 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13510 * config/i386/sse.md (*andnot<mode>3): Likewise.
13511 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13512 (*<code><mode>3): Likewise.
13513 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13514 (ia64_expand_atomic_op): Likewise.
13515 (ia64_arg_type): Likewise.
13516 (ia64_mode_to_int): Likewise.
13517 (ia64_scalar_mode_supported_p): Likewise.
13518 (ia64_vector_mode_supported_p): Likewise.
13519 (expand_vec_perm_broadcast): Likewise.
13520 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13521 (iq2000_function_arg_advance): Likewise.
13522 (iq2000_function_arg): Likewise.
13523 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13524 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13525 (m68k_libcall_value): Likewise.
13526 (m68k_function_value): Likewise.
13527 (sched_attr_op_type): Likewise.
13528 * config/mcore/mcore.c (mcore_output_move): Likewise.
13529 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13531 (microblaze_function_arg): Likewise.
13532 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13533 (mips_print_operand): Likewise.
13534 (mips_mode_ok_for_mov_fmt_p): Likewise.
13535 (mips_vector_mode_supported_p): Likewise.
13536 (mips_preferred_simd_mode): Likewise.
13537 (mips_expand_vpc_loongson_even_odd): Likewise.
13538 (mips_expand_vec_unpack): Likewise.
13539 (mips_expand_vi_broadcast): Likewise.
13540 (mips_expand_vector_init): Likewise.
13541 (mips_expand_vec_reduc): Likewise.
13542 (mips_expand_msa_cmp): Likewise.
13543 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13544 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13545 (cc_flags_for_mode): Likewise.
13546 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13547 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13548 (nds32_output_casesi_pc_relative): Likewise.
13549 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13550 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13551 (nvptx_gen_unpack): Likewise.
13552 (nvptx_gen_pack): Likewise.
13553 (nvptx_gen_shuffle): Likewise.
13554 (nvptx_gen_wcast): Likewise.
13555 (nvptx_preferred_simd_mode): Likewise.
13556 * config/pa/pa.c (pa_secondary_reload): Likewise.
13557 * config/pa/predicates.md (base14_operand): Likewise.
13558 * config/powerpcspe/powerpcspe-c.c
13559 (altivec_resolve_overloaded_builtin): Likewise.
13560 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13562 (rs6000_preferred_simd_mode): Likewise.
13563 (output_vec_const_move): Likewise.
13564 (rs6000_expand_vector_extract): Likewise.
13565 (rs6000_split_vec_extract_var): Likewise.
13566 (reg_offset_addressing_ok_p): Likewise.
13567 (rs6000_legitimate_offset_address_p): Likewise.
13568 (rs6000_legitimize_address): Likewise.
13569 (rs6000_emit_set_const): Likewise.
13570 (rs6000_const_vec): Likewise.
13571 (rs6000_emit_move): Likewise.
13572 (spe_build_register_parallel): Likewise.
13573 (rs6000_darwin64_record_arg_recurse): Likewise.
13574 (swap_selector_for_mode): Likewise.
13575 (spe_init_builtins): Likewise.
13576 (paired_init_builtins): Likewise.
13577 (altivec_init_builtins): Likewise.
13578 (do_load_for_compare): Likewise.
13579 (rs6000_generate_compare): Likewise.
13580 (rs6000_expand_float128_convert): Likewise.
13581 (emit_load_locked): Likewise.
13582 (emit_store_conditional): Likewise.
13583 (rs6000_output_function_epilogue): Likewise.
13584 (rs6000_handle_altivec_attribute): Likewise.
13585 (rs6000_function_value): Likewise.
13586 (emit_fusion_gpr_load): Likewise.
13587 (emit_fusion_p9_load): Likewise.
13588 (emit_fusion_p9_store): Likewise.
13589 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13590 (fusion_gpr_mem_load): Likewise.
13591 (fusion_addis_mem_combo_load): Likewise.
13592 (fusion_addis_mem_combo_store): Likewise.
13593 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13594 (fusion_gpr_mem_load): Likewise.
13595 (fusion_addis_mem_combo_load): Likewise.
13596 (fusion_addis_mem_combo_store): Likewise.
13597 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13599 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13600 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13601 (rs6000_preferred_simd_mode): Likewise.
13602 (output_vec_const_move): Likewise.
13603 (rs6000_expand_vector_extract): Likewise.
13604 (rs6000_split_vec_extract_var): Likewise.
13605 (reg_offset_addressing_ok_p): Likewise.
13606 (rs6000_legitimate_offset_address_p): Likewise.
13607 (rs6000_legitimize_address): Likewise.
13608 (rs6000_emit_set_const): Likewise.
13609 (rs6000_const_vec): Likewise.
13610 (rs6000_emit_move): Likewise.
13611 (rs6000_darwin64_record_arg_recurse): Likewise.
13612 (swap_selector_for_mode): Likewise.
13613 (paired_init_builtins): Likewise.
13614 (altivec_init_builtins): Likewise.
13615 (rs6000_expand_float128_convert): Likewise.
13616 (emit_load_locked): Likewise.
13617 (emit_store_conditional): Likewise.
13618 (rs6000_output_function_epilogue): Likewise.
13619 (rs6000_handle_altivec_attribute): Likewise.
13620 (rs6000_function_value): Likewise.
13621 (emit_fusion_gpr_load): Likewise.
13622 (emit_fusion_p9_load): Likewise.
13623 (emit_fusion_p9_store): Likewise.
13624 * config/rx/rx.c (rx_gen_move_template): Likewise.
13625 (flags_from_mode): Likewise.
13626 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13627 (s390_slb_comparison): Likewise.
13628 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13629 (s390_vector_mode_supported_p): Likewise.
13630 (s390_cc_modes_compatible): Likewise.
13631 (s390_match_ccmode_set): Likewise.
13632 (s390_canonicalize_comparison): Likewise.
13633 (s390_emit_compare_and_swap): Likewise.
13634 (s390_branch_condition_mask): Likewise.
13635 (s390_rtx_costs): Likewise.
13636 (s390_secondary_reload): Likewise.
13637 (__SECONDARY_RELOAD_CASE): Likewise.
13638 (s390_expand_cs): Likewise.
13639 (s390_preferred_simd_mode): Likewise.
13640 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13641 * config/sh/sh.c (sh_print_operand): Likewise.
13642 (dump_table): Likewise.
13643 (sh_secondary_reload): Likewise.
13644 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13645 * config/sh/sh.md (casesi_worker_1): Likewise.
13646 (casesi_worker_2): Likewise.
13647 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13648 (fcc_comparison_operator): Likewise.
13649 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13650 (emit_soft_tfmode_cvt): Likewise.
13651 (sparc_preferred_simd_mode): Likewise.
13652 (output_cbranch): Likewise.
13653 (sparc_print_operand): Likewise.
13654 (sparc_expand_vec_perm_bmask): Likewise.
13655 (vector_init_bshuffle): Likewise.
13656 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13657 (spu_vector_mode_supported_p): Likewise.
13658 (spu_expand_insv): Likewise.
13659 (spu_emit_branch_or_set): Likewise.
13660 (spu_handle_vector_attribute): Likewise.
13661 (spu_builtin_splats): Likewise.
13662 (spu_builtin_extract): Likewise.
13663 (spu_builtin_promote): Likewise.
13664 (spu_expand_sign_extend): Likewise.
13665 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13666 (tilegx_simd_int): Likewise.
13667 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13668 (tilepro_simd_int): Likewise.
13669 * config/v850/v850.c (const_double_split): Likewise.
13670 (v850_print_operand): Likewise.
13671 (ep_memory_offset): Likewise.
13672 * config/vax/vax.c (vax_rtx_costs): Likewise.
13673 (vax_output_int_move): Likewise.
13674 (vax_output_int_add): Likewise.
13675 (vax_output_int_subtract): Likewise.
13676 * config/visium/predicates.md (visium_branch_operator): Likewise.
13677 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13678 (visium_print_operand_address): Likewise.
13679 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13680 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13681 (xtensa_expand_conditional_branch): Likewise.
13682 (xtensa_copy_incoming_a7): Likewise.
13683 (xtensa_output_literal): Likewise.
13684 * dfp.c (decimal_real_maxval): Likewise.
13685 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13688 Alan Hayward <alan.hayward@arm.com>
13689 David Sherwood <david.sherwood@arm.com>
13691 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13692 (mode_nunits_inline): Likewise.
13693 (mode_inner_inline): Likewise.
13694 (mode_unit_size_inline): Likewise.
13695 (mode_unit_precision_inline): Likewise.
13696 (emit_insn_modes_h): Likewise. Also emit a #define of the
13698 (emit_mode_wider): Add an E_ prefix to mode names.
13699 (emit_mode_complex): Likewise.
13700 (emit_mode_inner): Likewise.
13701 (emit_mode_adjustments): Likewise.
13702 (emit_mode_int_n): Likewise.
13703 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13704 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13705 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13706 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13707 (CRC32_BUILTIN, ENTRY): Likewise.
13708 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13709 (aarch64_pop_regs): Likewise.
13710 (aarch64_process_components): Likewise.
13711 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13712 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13713 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13714 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13715 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13716 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13718 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13719 (pcmpistr): Likewise.
13720 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13721 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13722 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13724 * config/rl78/rl78.c (mduc_regs): Likewise.
13725 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13726 (htm_expand_builtin): Likewise.
13727 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13728 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13729 * config/xtensa/xtensa.c (print_operand): Likewise.
13730 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13731 (NUM_MODE_VECTOR_INT): Likewise.
13732 * genoutput.c (null_operand): Likewise.
13733 (output_operand_data): Likewise.
13734 * genrecog.c (print_parameter_value): Likewise.
13735 * lra.c (debug_operand_data): Likewise.
13737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13739 * dwarf2out.c (scompare_loc_descriptor_wide)
13740 (scompare_loc_descriptor_narrow): New functions, split out from...
13741 (scompare_loc_descriptor): ...here.
13742 * expmed.c (emit_store_flag_int): New function, split out from...
13743 (emit_store_flag): ...here.
13745 2017-08-30 Richard Biener <rguenther@suse.de>
13747 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13748 (dwarf2out_early_finish): Move setting of AT_pubnames from
13749 early debug output to early finish.
13751 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13753 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13754 and -mdata-region to the assembler.
13756 2017-08-30 Richard Biener <rguenther@suse.de>
13758 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13760 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13761 (dwarf2out_early_global_decl): Only generate a DIE for the
13762 abstract origin if it doesn't already exist or is a declaration DIE.
13763 (resolve_addr): Do not add the linkage name twice when
13764 generating a stub DIE for the DW_TAG_GNU_call_site target.
13766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13769 Use machine_mode rather than int for arg1_mode.
13771 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13774 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13775 that the second argument of the built-in functions to unpack
13776 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13777 switch statement instead a lot of if statements.
13778 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13779 Allow 64-bit values to be in Altivec registers as well as
13780 traditional floating point registers.
13781 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13783 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13785 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13786 MAX_REGS_PER_ADDRESS == 1.
13788 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13790 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13791 * config/i386/i386.c (ix86_option_override_internal): Simplify
13792 setting of opts->x_flag_entry.
13794 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13795 Jakub Jelinek <jakub@redhat.com>
13796 Richard Biener <rguenther@suse.de>
13798 PR tree-optimization/81503
13799 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13800 folded constant fits in the target type; reorder tests for clarity.
13802 2017-08-29 Martin Liska <mliska@suse.cz>
13804 * passes.def: Include pass_lower_switch.
13805 * stmt.c (dump_case_nodes): Remove and move to
13806 tree-switch-conversion.
13807 (case_values_threshold): Likewise.
13808 (expand_switch_as_decision_tree_p): Likewise.
13809 (emit_case_decision_tree): Likewise.
13810 (expand_case): Likewise.
13811 (balance_case_nodes): Likewise.
13812 (node_has_low_bound): Likewise.
13813 (node_has_high_bound): Likewise.
13814 (node_is_bounded): Likewise.
13815 (emit_case_nodes): Likewise.
13816 (struct simple_case_node): New struct.
13817 (add_case_node): Remove.
13818 (emit_case_dispatch_table): Use vector instead of case_list.
13819 (reset_out_edges_aux): Remove.
13820 (compute_cases_per_edge): Likewise.
13821 (expand_case): Build list of simple_case_node.
13822 (expand_sjlj_dispatch_table): Use it.
13823 * tree-switch-conversion.c (struct case_node): Moved from
13824 stmt.c and adjusted.
13825 (emit_case_nodes): Likewise.
13826 (node_has_low_bound): Likewise.
13827 (node_has_high_bound): Likewise.
13828 (node_is_bounded): Likewise.
13829 (case_values_threshold): Likewise.
13830 (reset_out_edges_aux): Likewise.
13831 (compute_cases_per_edge): Likewise.
13832 (add_case_node): Likewise.
13833 (dump_case_nodes): Likewise.
13834 (balance_case_nodes): Likewise.
13835 (expand_switch_as_decision_tree_p): Likewise.
13836 (emit_jump): Likewise.
13837 (emit_case_decision_tree): Likewise.
13838 (try_switch_expansion): Likewise.
13839 (do_jump_if_equal): Likewise.
13840 (emit_cmp_and_jump_insns): Likewise.
13841 (fix_phi_operands_for_edge): New function.
13842 (record_phi_operand_mapping): Likewise.
13843 (class pass_lower_switch): New pass.
13844 (pass_lower_switch::execute): New function.
13845 (make_pass_lower_switch): Likewise.
13846 (conditional_probability):
13847 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13848 * tree-pass.h: Add make_pass_lower_switch.
13850 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13853 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13856 2017-08-29 Richard Biener <rguenther@suse.de>
13858 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13859 we do not add a DW_AT_inline attribute twice.
13860 (gen_subprogram_die): Remove code setting DW_AT_inline on
13861 DECL_ABSTRACT_P nodes.
13863 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13865 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13866 calls to internal functions.
13867 (gimplify_modify_expr): Likewise.
13868 * tree-call-cdce.c (use_internal_fn): Likewise.
13869 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13870 (convert_to_divmod): Set the nothrow flag.
13871 * tree-if-conv.c (predicate_mem_writes): Likewise.
13872 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13873 (vectorizable_call): Likewise.
13874 (vectorizable_store): Likewise.
13875 (vectorizable_load): Likewise.
13876 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13877 (vect_recog_mask_conversion_pattern): Likewise.
13879 2017-08-29 Martin Liska <mliska@suse.cz>
13882 * gcc.c (driver_handle_option): Add new argument.
13883 * opts-common.c (handle_option): Pass
13884 target_option_override_hook.
13885 * opts-global.c (lang_handle_option): Add new option.
13886 (set_default_handlers): Add new argument.
13887 (decode_options): Likewise.
13888 * opts.c (target_handle_option): Likewise.
13889 (common_handle_option): Call target_option_override_hook.
13890 * opts.h (struct cl_option_handler_func): Add hook for
13891 target option override.
13892 (struct cl_option_handlers): Likewise.
13893 (set_default_handlers): Add new argument.
13894 (decode_options): Likewise.
13895 (common_handle_option): Likewise.
13896 (target_handle_option): Likewise.
13897 * toplev.c (toplev::main): Pass targetm.target_option.override
13900 2017-08-29 Richard Biener <rguenther@suse.de>
13901 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13903 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13904 life to the active subtree.
13906 2017-08-28 Jeff Law <law@redhat.com>
13908 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13909 derive_equivalences.
13910 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13911 Code moved into....
13912 (edge_info::derive_equivalences): New private member function
13914 * tree-ssa-dom.c (class edge_info): Changed from a struct
13915 to a class. Add ctor/dtor, methods and data members.
13916 (edge_info::edge_info): Renamed from allocate_edge_info.
13917 Initialize additional members.
13918 (edge_info::~edge_info): New.
13919 (free_dom_edge_info): Delete the edge info.
13920 (record_edge_info): Use new class & associated member functions.
13921 Tighten forms for testing for edge equivalences.
13922 (record_temporary_equivalences): Iterate over the simple
13923 equivalences rather than assuming there's only one per edge.
13924 (cprop_into_successor_phis): Iterate over the simple
13925 equivalences rather than assuming there's only one per edge.
13926 (optimize_stmt): Use operand_equal_p rather than pointer
13927 equality for mini-DSE code.
13929 2017-08-28 Nathan Sidwell <nathan@acm.org>
13931 * gcc.c (execute): Fold SIGPIPE handling into switch
13932 statement. Adjust internal error message.
13934 2017-08-28 Richard Biener <rguenther@suse.de>
13937 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13938 Do nothing for removed DIEs.
13940 2017-08-28 Richard Biener <rguenther@suse.de>
13942 PR tree-optimization/81977
13943 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13946 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13949 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13950 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13951 using targetm.gen_mem_thread_fence.
13953 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13956 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13957 predicate to register_operand. Reorder operands.
13958 (*btr<mode>): Ditto.
13959 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13960 (*btr<mode>_mask): Ditto.
13962 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13964 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13965 * config/rs6000/xmmintrin.h: New file.
13966 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13968 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13971 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13972 parameter and_insn and return it.
13973 (recombine_lvx_pattern): Insert a copy to ensure availability of
13974 the base register of the copied masking operation at the point of
13975 the instruction replacement.
13976 (recombine_stvx_pattern): Likewise.
13978 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13980 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13981 undocumented switches.
13982 (-mpower9-dform-vector): Likewise.
13983 (-mpower9-dform): Likewise.
13984 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13985 comments to delete references to -mpower9-dform* switches.
13986 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13987 Delete reference to -mpower9-dform* switches, test for
13988 -mpower9-vector instead.
13989 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13990 (OTHER_P9_VECTOR_MASKS): Likewise.
13991 (POWERPC_MASKS): Likewise.
13992 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13993 tests against -mpower9-dform* to -mpower9-vector. Delete code
13994 that checked for -mpower9-dform* consistancy with other options.
13995 Add test for -mpower9-misc to enable other power9 switches.
13996 (rs6000_init_hard_regno_mode_ok): Likewise.
13997 (rs6000_option_override_internal): Likewise.
13998 (rs6000_emit_prologue): Likewise.
13999 (rs6000_emit_epilogue): Likewise.
14000 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14001 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14002 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14004 (emit_fusion_p9_store): Likewise.
14005 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14006 resetting these macros if the assembler does not support ISA 3.0
14008 (TARGET_P9_DFORM_VECTOR): Likewise.
14009 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14010 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14012 2017-08-25 Alan Modra <amodra@gmail.com>
14014 PR rtl-optimization/81747
14015 * cse.c (cse_extended_basic_block): Don't attempt to record
14016 equivalences for degenerate conditional branches that branch
14017 to their fall-through.
14019 2017-08-24 Martin Sebor <msebor@redhat.com>
14021 PR middle-end/81908
14022 * gimple-fold.c (size_must_be_zero_p): New function.
14023 (gimple_fold_builtin_memory_op): Call it.
14025 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14027 * config/rs6000/mm_malloc.h: New file.
14029 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14031 PR tree-optimization/81913
14032 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14033 analysis when either IVs in condition can wrap.
14035 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14037 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14038 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14040 2017-08-24 Richard Biener <rguenther@suse.de>
14043 * targhooks.c (default_target_can_inline_p): Properly
14044 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14045 is present and always compare.
14046 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14047 infer -mfpmath=sse from TARGET_SSE_P.
14048 (ix86_can_inline_p): Properly use target_option_default_node when
14049 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14051 2017-08-24 Richard Biener <rguenther@suse.de>
14054 * dwarf2out.c (output_die): Handle flag_generate_offload like
14056 (output_comp_unit): Likewise.
14057 (gen_array_type_die): Likewise.
14058 (dwarf2out_early_finish): Likewise.
14059 (note_variable_value_in_expr): Likewise.
14060 (dwarf2out_finish): Likewise. Adjust assert.
14061 * cgraphunit.c (symbol_table::compile): Move setting of
14062 flag_generate_offload earlier ...
14063 (symbol_table::finalize_compilation_unit): ... here, before
14064 early debug finalization.
14066 2017-08-24 Richard Biener <rguenther@suse.de>
14068 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14069 and ipa-fnsummary.h.
14070 (ix86_can_inline_p): When ix86_fpmath flags do not match
14071 check whether the callee uses FP math at all.
14073 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14075 PR middle-end/81931
14076 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14077 instead of TYPE_PRECISION.
14079 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14080 Alan Hayward <alan.hayward@arm.com>
14081 David Sherwood <david.sherwood@arm.com>
14083 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14085 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14087 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14089 (ix86_frame::stack_realign_allocate): New field.
14090 (struct machine_frame_state): Modify comments.
14091 (machine_frame_state::sp_realigned_fp_end): New field.
14092 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14093 layout calculation.
14094 (sp_valid_at): Add assertion to assure no attempt to access invalid
14095 offset of a realigned stack.
14096 (fp_valid_at): Likewise.
14097 (choose_baseaddr): Modify comments.
14098 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14099 ix86_expand_prologue.
14100 (ix86_expand_prologue): Modify stack realignment and allocation.
14101 (ix86_expand_epilogue): Modify comments.
14102 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14103 avx2_runtime, avx512f, and avx512f_runtime.
14105 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14107 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14108 (mstackrealign): Do not init to -1.
14109 * config/i386/i386.c (ix86_option_override_internal):
14110 Check opts_set, not opts when setting default value of
14111 opts->x_ix86_force_align_arg_pointer.
14113 2017-08-23 Richard Biener <rguenther@suse.de>
14115 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14116 lang_hooks.decl_printable_name.
14117 * print-rtl-function.c (print_rtx_function): Likewise.
14118 * tree-pretty-print.c (dump_function_header): Likewise.
14120 2017-08-23 Richard Biener <rguenther@suse.de>
14123 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14124 -g0 at compile-time.
14126 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14128 PR middle-end/19706
14129 * doc/sourcebuild.texi (Other hardware attributes):
14132 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14134 PR middle-end/19706
14135 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14136 Fix single-use check.
14138 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14140 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14142 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14144 * doc/install.texi: Modify to add more details on running selected
14147 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14149 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14150 is combined with -mabi=ms.
14151 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14154 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14156 PR tree-optimization/81488
14157 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14158 and cached_basis fields.
14159 (MAX_SPREAD): New constant.
14160 (alloc_cand_and_find_basis): Initialize new fields.
14161 (clear_visited): New function.
14162 (create_phi_basis_1): Rename from create_phi_basis, set visited
14163 and cached_basis fields.
14164 (create_phi_basis): New wrapper function.
14165 (phi_add_costs_1): Rename from phi_add_costs, add spread
14166 parameter, set visited field, short-circuit when limits reached.
14167 (phi_add_costs): New wrapper function.
14168 (record_phi_increments_1): Rename from record_phi_increments, set
14170 (record_phi_increments): New wrapper function.
14171 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14172 (phi_incr_cost): New wrapper function.
14173 (all_phi_incrs_profitable_1): Rename from
14174 all_phi_incrs_profitable, set visited field.
14175 (all_phi_incrs_profitable): New wrapper function.
14177 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14178 Alan Hayward <alan.hayward@arm.com>
14179 David Sherwood <david.sherwood@arm.com>
14181 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14182 that takes the outer and inner modes.
14183 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14184 comparison as the canonical test for a paradoxical subreg.
14185 * combine.c (simplify_set): Use paradoxical_subreg_p.
14186 (make_extraction): Likewise.
14187 (force_to_mode): Likewise.
14188 (rtx_equal_for_field_assignment_p): Likewise.
14189 (gen_lowpart_for_combine): Likewise.
14190 (simplify_comparison): Likewise.
14191 * cse.c (equiv_constant): Likewise.
14192 * expmed.c (store_bit_field_1): Likewise.
14193 * final.c (alter_subreg): Likewise.
14194 * fwprop.c (propagate_rtx): Likewise.
14195 (forward_propagate_subreg): Likewise.
14196 * ira-conflicts.c (ira_build_conflicts): Likewise.
14197 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14198 * lra-constraints.c (curr_insn_transform): Likewise.
14199 (split_reg): Likewise.
14200 * lra-eliminations.c (move_plus_up): Likewise.
14201 (lra_eliminate_regs_1): Likewise.
14202 * recog.c (general_operand): Likewise.
14203 * ree.c (combine_reaching_defs): Likewise.
14204 * reload.c (push_reload): Likewise.
14205 (find_reloads): Likewise.
14206 * reload1.c (elimination_effects): Likewise.
14207 (compute_reload_subreg_offset): Likewise.
14208 (choose_reload_regs): Likewise.
14209 * rtlanal.c (subreg_lsb_1): Likewise.
14210 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14211 (simplify_subreg): Likewise.
14212 * var-tracking.c (track_loc_p): Likewise.
14213 * emit-rtl.c (byte_lowpart_offset): Likewise.
14214 (paradoxical_subreg_p): Delete out-of-line definition.
14216 2017-08-22 Jeff Law <law@redhat.com>
14218 PR tree-optimization/81741
14219 PR tree-optimization/71947
14220 * tree-ssa-dom.c: Include tree-inline.h.
14221 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14222 equivalences if one is more expensive to compute than the other.
14223 * tree-ssa-scopedtables.h (class const_or_copies): Make
14224 record_const_or_copy_raw method private.
14225 (class avail_exprs_stack): New method simplify_binary_operation.
14226 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14227 avail_exprs_stack::simplify_binary_operation as needed.
14228 (avail_exprs_stack::simplify_binary_operation): New function.
14230 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14232 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14233 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14234 (DOT_SYMBOLS): Likewise.
14235 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14236 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14237 (RS6000_ABI_NAME): Likewise.
14238 (TARGET_CMODEL): Likewise.
14239 (TOC_SECTION_ASM_OP): Likewise.
14240 (SET_CMODEL): New macro.
14241 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14243 2017-08-22 Richard Biener <rguenther@suse.de>
14245 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14246 to allow for free-lang-data replacements similar to verify_type_variant.
14248 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14249 Alan Hayward <alan.hayward@arm.com>
14250 David Sherwood <david.sherwood@arm.com>
14252 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14253 VOIDmode for the LABEL_REF.
14255 2017-08-22 Richard Biener <rguenther@suse.de>
14257 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14259 2017-08-22 Richard Biener <rguenther@suse.de>
14261 * common.opt (feliminate-dwarf2-dups): Ignore.
14262 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14263 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14264 same_die_p_wrap, compute_section_prefix,
14265 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14266 (comdat_symbol_id, comdat_symbol_number): Likewise.
14267 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14269 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14270 (output_die): Mark unreachable path unreachable.
14271 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14272 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14273 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14274 (dwarf2out_early_finish): Likewise.
14276 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14278 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14280 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14283 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14284 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14285 (avr_attribute_table) <io, io_low, address>: Initialize
14286 .decl_required with true.
14288 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14290 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14291 undocumented debugging options.
14292 (-mvsx-scalar-double): Likewise.
14293 (-mallow-df-permute): Likewise.
14294 (-mvectorize-builtins): Likewise.
14295 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14296 (rs6000_builtin_vectorized_function): Likewise.
14297 (rs6000_builtin_md_vectorized_function): Likewise.
14298 (rs6000_opt_vars): Likewise.
14300 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14303 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14304 (*btrq_imm): Rename from *btrq.
14305 (*btcq_imm): Rename from *btcq.
14306 (btsc): New code attribute.
14307 (*<btsc><mode>): New insn pattern.
14308 (*btr<mode>): Ditto.
14309 (*<btsc><mode>_mask): New insn_and_split pattern.
14310 (*btr<mode>_mask): Ditto.
14312 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14313 Alan Hayward <alan.hayward@arm.com>
14314 David Sherwood <david.sherwood@arm.com>
14316 * function.c (pad_below): Simplify padding calculation.
14318 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14319 Alan Hayward <alan.hayward@arm.com>
14320 David Sherwood <david.sherwood@arm.com>
14322 * target.def (function_prologue): Remove frame size argument.
14323 (function_epilogue): Likewise.
14324 * doc/tm.texi: Regenerate.
14325 * final.c (final_start_function): Update call to function_prologue.
14326 (final_end_function): Update call to function_epilogue.
14327 (default_function_pro_epilogue): Remove frame size argument.
14328 * output.h (default_function_pro_epilogue): Likewise.
14329 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14330 (arm_output_function_prologue): Likewise.
14331 * config/frv/frv.c (frv_function_prologue): Likewise.
14332 (frv_function_epilogue): Likewise.
14333 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14334 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14335 (ia64_output_function_epilogue): Likewise.
14336 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14337 (m32r_output_function_epilogue): Likewise.
14338 * config/microblaze/microblaze.c (microblaze_function_prologue)
14339 (microblaze_function_epilogue): Likewise.
14340 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14341 (mips_output_function_epilogue): Likewise.
14342 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14343 (mmix_target_asm_function_epilogue): Likewise.
14344 * config/msp430/msp430.c (msp430_start_function): Likewise.
14345 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14346 (nds32_asm_function_epilogue): Likewise.
14347 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14348 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14349 (pa_output_function_epilogue): Likewise.
14350 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14351 (rs6000_output_function_epilogue): Likewise.
14352 * config/rl78/rl78.c (rl78_start_function): Likewise.
14353 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14354 (rs6000_output_function_epilogue): Likewise.
14355 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14356 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14357 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14358 (sparc_asm_function_epilogue): Likewise.
14360 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14362 * tree.h (type_has_mode_precision_p): New function.
14363 * convert.c (convert_to_integer_1): Use it.
14364 * expr.c (expand_expr_real_2): Likewise.
14365 (expand_expr_real_1): Likewise.
14366 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14367 * match.pd: Likewise.
14368 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14369 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14370 * tree-tailcall.c (process_assignment): Likewise.
14371 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14372 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14373 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14374 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14375 (vectorizable_assignment): Likewise.
14376 (vectorizable_shift): Likewise.
14377 (vectorizable_operation): Likewise.
14378 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14380 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14382 * match.pd: Add pow (C, x) simplification.
14384 2017-08-21 Richard Biener <rguenther@suse.de>
14386 PR tree-optimization/81900
14387 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14388 for blocks with abnormal predecessors.
14389 (compute_antic): Do not set visited flag prematurely.
14391 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14394 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14396 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14398 * stor-layout.h (vector_type_mode): Move to...
14399 * tree.h (vector_type_mode): ...here.
14400 * stor-layout.c (vector_type_mode): Move to...
14401 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14403 2017-08-21 Richard Biener <rguenther@suse.de>
14405 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14406 register_external_die hooks.
14407 (debug_false_tree_charstarstar_uhwistar): Declare.
14408 (debug_nothing_tree_charstar_uhwi): Likewise.
14409 * debug.c (do_nothing_debug_hooks): Adjust.
14410 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14411 (debug_nothing_tree_charstar_uhwi): Likewise.
14412 * dbxout.c (dbx_debug_hooks): Adjust.
14413 (xcoff_debug_hooks): Likewise.
14414 * sdbout.c (sdb_debug_hooks): Likewise.
14415 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14416 * dwarf2out.c (macinfo_label_base): New global.
14417 (dwarf2out_register_external_die): New function for the
14418 register_external_die hook.
14419 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14420 (dwarf2_debug_hooks): Use them.
14421 (dwarf2_lineno_debug_hooks): Adjust.
14422 (struct die_struct): Add with_offset flag.
14423 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14424 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14425 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14426 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14427 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14428 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14429 defining section names for the early LTO debug variants.
14430 (reset_indirect_string): New helper.
14431 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14432 (print_dw_val): Add support for offsetted symbol references.
14433 (get_ultimate_context): Split out from is_cxx.
14434 (is_cxx): Use get_ultimate_context.
14435 (is_fortran): Add decl overload.
14436 (compute_comp_unit_symbol): Split out worker from
14437 compute_section_prefix.
14438 (compute_section_prefix): Call compute_comp_unit_symbol and
14439 set comdat_type_p here.
14440 (output_die): Skip DIE symbol output for the LTO added one.
14441 Handle DIE symbol references with offset.
14442 (output_comp_unit): Guard section name mangling properly.
14443 For LTO debug sections emit a symbol at the section beginning
14444 which we use to refer to its DIEs.
14445 (add_abstract_origin_attribute): For DIEs registered via
14446 dwarf2out_register_external_die directly refer to the early
14447 DIE rather than indirectly through the shadow one we created.
14448 Remove obsolete call to dwarf2out_abstract_function for
14449 non-function/block origins.
14450 (gen_array_type_die): When generating early LTO debug do
14451 not emit DW_AT_string_length.
14452 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14453 late when in LTO. As suggested place a gcc_unreachable for
14454 the DECL_ABSTRACT_P case.
14455 (gen_subprogram_die): Avoid another specification DIE
14456 for early built declarations/definitions for the late LTO case.
14457 (gen_variable_die): Add type references for late duplicated VLA dies
14459 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14460 we have the abstract instance already.
14461 (process_scope_var): Adjust decl DIE contexts in LTO which
14462 first puts them in limbo.
14463 (gen_decl_die): Do not generate type DIEs late apart from
14464 types for VLAs or for decls we do not yet have a DIE. Do not
14465 call dwarf2out_abstract_function late.
14466 (dwarf2out_early_global_decl): Make sure to create DIEs
14467 for abstract instances of a decl first.
14468 (dwarf2out_late_global_decl): Adjust comment.
14469 (output_macinfo_op): With multiple macro sections use
14470 macinfo_label_base to distinguish labels.
14471 (output_macinfo): Likewise. Update macinfo_label_base.
14472 Pass in the line info label.
14473 (note_variable_value_in_expr): When generating LTO resolve
14474 all variable values here by generating DIEs as needed.
14475 (init_sections_and_labels): Add early LTO debug flag parameter
14476 and generate different sections and names if set. Add generation
14477 counter for the labels so we can have multiple of them.
14478 (reset_dies): Helper to allow DIEs to be output multiple times.
14479 (dwarf2out_finish): When outputting DIEs to the fat part of an
14480 LTO object first reset DIEs.
14481 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14482 (modified_type_die): Check for decl_ultimate_origin being self
14484 (gen_type_die_with_usage): Likewise.
14485 (gen_typedef_die): Allow decl_ultimate_origin being self.
14486 (set_decl_abstract_flags): Remove.
14487 (set_block_abstract_flags): Likewise.
14488 (dwarf2out_abstract_function): Treat the early generated DIEs
14489 as the abstract copy and only add DW_AT_inline and
14490 DW_AT_artificial here and call set_decl_origin_self.
14491 If the DIE has an abstract origin don't do anything.
14492 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14493 if we have none yet (Go fails to build one, PR78628).
14494 (variably_modified_type_p): Prevent endless recursion for Ada
14495 cyclic pointer types.
14496 * lto-streamer-in.c: Include debug.h.
14497 (dref_queue): New global.
14498 (lto_read_tree_1): Stream in DIE references.
14499 (lto_input_tree): Register DIE references.
14500 (input_function): Stream DECL_DEBUG_ARGS.
14501 * lto-streamer-out.c: Include debug.h.
14502 (lto_write_tree_1): Output DIE references.
14503 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14504 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14505 (output_function): Stream DECL_DEBUG_ARGS.
14506 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14507 Stream DECL_ABSTRACT_ORIGIN.
14508 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14509 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14510 DECL_CONTEXT for file-scope decls.
14511 * lto-streamer.h (struct dref_entry): Declare.
14512 (dref_queue): Likewise.
14513 * cfgexpand.c (pass_expand::execute): Do not call the
14514 outlining_inline_function hook here.
14515 * lto-wrapper.c (debug_obj): New global.
14516 (tool_cleanup): Unlink it if required.
14517 (debug_objcopy): New function.
14518 (run_gcc): Handle early debug sections in the IL files by
14519 extracting them to separate files, partially linkin them and
14520 feeding the result back as result to the linker.
14521 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14522 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14523 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14524 sections into a separate segment.
14525 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14527 (darwin_asm_dwarf_section): Likewise.
14528 (darwin_asm_output_dwarf_offset): Likewise.
14529 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14531 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14532 Alan Hayward <alan.hayward@arm.com>
14533 David Sherwood <david.sherwood@arm.com>
14535 * read-md.h (md_reader::record_potential_iterator_use): Replace
14536 pointer argument with an rtx and an index.
14537 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14538 (apply_mode_iterator): Likewise.
14539 (apply_code_iterator): Likewise.
14540 (apply_int_iterator): Likewise.
14541 (apply_subst_iterator): Likewise.
14542 (record_iterator_use): Likewise.
14543 (record_attribute_use): Likewise.
14544 (md_reader::record_potential_iterator_use): Likewise. Update calls
14545 to record_iterator_use and apply_iterator.
14546 (iterator_use): Replace ptr with x and index.
14547 (attribute_use): Likewise.
14548 (apply_attribute_uses): Update calls to apply_iterator.
14549 (apply_iterators): Likewise. Update initialization of iterator_use.
14550 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14551 and record_potential_iterator_use.
14552 (rtx_reader::read_rtx_operand): Likewise.
14554 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14555 Alan Hayward <alan.hayward@arm.com>
14556 David Sherwood <david.sherwood@arm.com>
14558 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14561 2017-08-21 Richard Biener <rguenther@suse.de>
14563 PR middle-end/81884
14564 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14565 at struct end conservatively when comparing common bases.
14567 2017-08-21 Richard Biener <rguenther@suse.de>
14569 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14570 (mem_ref_in_stmt): Remove.
14571 (determine_max_movement): Use ref index to get at the reference.
14572 (invariantness_dom_walker::before_dom_children): Deal with
14573 lim data already initialized.
14574 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14576 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14578 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14579 (TARGET_ISA_ROUND): Ditto.
14580 (TARGET_ROUND): Ditto.
14581 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14582 * config/i386/i386.md: Ditto.
14583 * config/i386/sse.md: Ditto.
14584 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14585 with OPTION_MASK_ISA_SSE4_1.
14587 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14590 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14591 __builtin_ia32_lzcnt_u16.
14593 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14596 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14597 (rs6000_set_current_function): Rewrite function to use it.
14599 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14602 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14603 and TYPE_WARN_IF_NOT_ALIGN.
14604 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14605 (handle_warn_if_not_align): New.
14606 (place_union_field): Call handle_warn_if_not_align.
14607 (place_field): Call handle_warn_if_not_align.
14608 Copy TYPE_WARN_IF_NOT_ALIGN.
14609 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14610 (layout_type): Likewise.
14611 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14613 (tree_decl_common): Add warn_if_not_align.
14614 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14615 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14616 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14617 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14618 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14619 * doc/extend.texi: Document warn_if_not_aligned attribute.
14620 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14622 2017-08-17 Martin Liska <mliska@suse.cz>
14625 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14626 (get_data_dependence): Use it as pointer type.
14627 (distribute_loop): Likewise.
14629 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14631 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14632 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14633 (p8_vmrgow_<mode>_direct): New define_insn.
14634 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14635 handle endianness for vmrgew and vmrgow permute patterns.
14637 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14639 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14640 * config/rs6000/rs6000-cpus.def: Remove comment.
14641 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14642 (POWERPC_MASKS): Likewise.
14643 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14644 use of TARGET_VSX_TIMODE.
14645 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14646 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14647 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14648 (rs6000_option_override_internal): Remove dead code.
14649 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14650 (rs6000_legitimize_reload_address): Likewise.
14651 (rs6000_legitimate_address_p): Likewise.
14652 (rs6000_opt_masks): Delete "vsx-timode".
14653 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14654 from function comment.
14655 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14656 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14657 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14659 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14660 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14661 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14662 (VSX_TI): Likewise.
14664 (define_peephole2): Likewise.
14666 2017-08-17 Martin Sebor <msebor@redhat.com>
14669 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14670 past the end of an array.
14671 (test_pp_format): Add test cases.
14673 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14675 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14676 missing ECF_NOTHROW flags.
14678 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14681 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14682 operands residing in integer registers.
14683 (*vsx_le_perm_load_<mode>): Likewise.
14684 (*vsx_le_perm_store_<mode>): Likewise.
14685 (define_peephole2): Add peepholes to optimize the above.
14687 2017-08-17 Marek Polacek <polacek@redhat.com>
14689 PR middle-end/81814
14690 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14691 to mimic what shorten_compare did. Change the return type to bool.
14692 (fold_cond_expr_with_comparison): Update call to
14693 operand_equal_for_comparison_p.
14694 (fold_ternary_loc): Likewise.
14696 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14698 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14700 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14702 2017-08-17 Richard Biener <rguenther@suse.de>
14704 * tree-ssa-structalias.c (solve_graph): When propagating
14705 to successors update the graphs succ edges and avoid duplicate work.
14707 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14710 * config/i386/i386.c (ix86_option_override_internal): Save target
14711 specific options after ix86_stack_protector_guard_reg was changed.
14713 2017-08-17 Richard Biener <rguenther@suse.de>
14715 PR tree-optimization/81827
14716 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14717 (new_var_info): Initialize it conservatively.
14718 (get_call_vi): Mark register vars.
14719 (new_scalar_tmp_constraint_exp): Likewise.
14720 (handle_rhs_call): Likewise.
14721 (handle_const_call): Likewise.
14722 (create_function_info_for): Likewise.
14723 (solve_constraints): Sort varinfos to separate register from
14724 non-register vars to pack points-to solution bitmaps during
14727 2017-08-17 Marek Polacek <polacek@redhat.com>
14729 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14731 2017-08-17 Richard Biener <rguenther@suse.de>
14733 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14734 to true when overflow is undefined and we saturated the result.
14736 2017-08-17 Alan Modra <amodra@gmail.com>
14739 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14740 Don't use store multiple if only one reg needs saving.
14741 (interesting_frame_related_regno): New function.
14742 (rs6000_frame_related): Don't emit frame info for regs that
14744 (rs6000_emit_epilogue): Likewise.
14746 2017-08-16 Nathan Sidwell <nathan@acm.org>
14748 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14749 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14750 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14751 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14752 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14753 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14754 (verify_type): Adjust for TYPE_BINFO move.
14755 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14756 process TYPE_BINFO directly.
14757 (hash_tree): Likewise.
14758 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14760 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14763 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14765 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14767 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14770 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14771 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14772 Add ix86_binary_operator_ok to insn constraint.
14773 (*iordi_1_bts): Ditto.
14774 (*xordi_1_btc): Ditto.
14775 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14776 Update corresponding peephole2 pattern.
14780 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14782 PR tree-optimization/81832
14783 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14784 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14786 2017-08-16 Marek Polacek <polacek@redhat.com>
14789 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14790 regarding size_zero_node.
14792 2017-08-16 Martin Liska <mliska@suse.cz>
14795 * config.gcc: Respect previously set extra_objs in case
14798 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14800 PR tree-optimization/81835
14801 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14802 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14803 not depend on the phi.
14805 2017-08-16 Alan Modra <amodra@gmail.com>
14807 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14810 2017-08-16 Alan Modra <amodra@gmail.com>
14812 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14813 (save_reg_p): ..into this. Update all callers.
14814 (first_reg_to_save): Simplify.
14816 2017-08-16 Alan Modra <amodra@gmail.com>
14818 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14821 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14825 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14826 constant count if that count is less than 32.
14828 2017-08-15 Nathan Sidwell <nathan@acm.org>
14830 * gcc.c (execute): Emit friendlier message if inferior is killed
14831 by an external cause.
14833 2017-08-15 Richard Biener <rguenther@suse.de>
14835 PR tree-optimization/81790
14836 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14837 CONSTRUCTORs from simplifying and VN.
14839 2017-08-14 Martin Sebor <msebor@redhat.com>
14841 * builtin-attrs.def: Add comments.
14843 2017-08-14 Martin Sebor <msebor@redhat.com>
14846 * doc/extend.texi (attribute nonstring): Document new attribute.
14848 2017-08-14 Martin Sebor <msebor@redhat.com>
14851 * tree-diagnostic.c (default_tree_printer): Handle %G.
14852 * gimple-pretty-print.h (percent_G_format): Declare new function.
14853 * gimple-pretty-print.c (percent_G_format): Define.
14854 * tree-pretty-print.c (percent_K_format): Add argument.
14856 2017-08-14 Martin Sebor <msebor@redhat.com>
14858 PR translation/79998
14859 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14860 Remove a stray space.
14862 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14865 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14866 (*iordi_1_bts): Ditto.
14867 (*xordi_1_btc): Ditto.
14869 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14872 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14873 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14875 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14876 quoted strings, and make more translator-friendly.
14877 (darwin_rs6000_override_options): Likewise.
14878 (rs6000_option_override_internal): Likewise.
14879 (rs6000_return_in_memory): Fix overlong line.
14880 (init_cmulative_args): Use quoted strings, and make more
14881 translator-friendly.
14882 (rs6000_pass_by_reference): Fix overlong line.
14883 (def_builtin): Use quoted strings.
14884 (altivec_expand_predicate_builtin): Use quoted strings, and make
14885 more translator-friendly.
14886 (htm_expand_builtin): Use quoted strings.
14887 (cpu_expand_builtin): Use quoted strings, and make more
14888 translator-friendly.
14889 (altivec_expand_builtin): Likewise.
14890 (paired_expand_predicate_builtin): Likewise.
14891 (rs6000_invalid_builtin): Likewise.
14892 (builtin_function_type): Use quoted strings.
14893 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14894 more translator-friendly.
14895 (rs6000_trampoline_init): Likewise.
14896 (rs6000_handle_altivec_attribute): Likewise.
14897 (rs6000_inner_target_options): Use quoted strings.
14898 (rs6000_disable_incompatible_switches): Likewise.
14899 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14900 strings, and make more translator-friendly.
14901 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14903 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14905 PR tree-optimization/81799
14906 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14907 cond_expr to simple gimple operand.
14909 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14911 PR middle-end/46932
14912 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14914 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14918 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14919 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14920 TARGET_GASISR_PROLOGUES.
14921 * config/avr/avr.c (avr_option_override): Same.
14922 (avr_pass_pre_proep::execute): Same.
14924 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14927 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14928 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14930 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14932 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14933 define_insn_and_split. Split to a memory load from 0 in
14934 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14935 using PTR mode iterator.
14936 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14937 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14938 (*add_tp_<mode>): Redefine as define_insn_and_split.
14939 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14940 address space. Merge with *add_tp_x32 using PTR mode iterator.
14941 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14942 Split to an add with a memory load from 0 in
14943 DEFAULT_TLS_SEG_REG address space.
14945 2017-08-12 Andrew Pinski <apinski@cavium.com>
14947 * config/aarch64/aarch64-option-extensions.def (rdma):
14948 Fix feature string to what Linux prints out in /proc/cpuinfo.
14950 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14953 * dwarf2out.c (modified_type_die): For C typedef types that have
14954 an ultimate origin, process the ultimate origin instead of the
14956 (gen_typedef_die): Assert that input DECLs have no ultimate
14958 (gen_type_die_with_usage): For typedef variants that have an
14959 ultimate origin, just call gen_decl_die on the original DECL.
14960 (process_scope_var): Avoid creating DIEs for local typedefs and
14961 concrete static variables.
14963 2017-08-12 Alan Modra <amodra@gmail.com>
14967 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14968 match gnu-user.h startfile.
14969 (ENDFILE_LINUX_SPEC): Similarly.
14971 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14974 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14976 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14978 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14979 * config/aarch64/aarch64.md (mov<mode>): Change.
14980 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14981 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14982 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14984 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14986 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14987 for storage order barriers.
14989 2017-08-11 Martin Liska <mliska@suse.cz>
14992 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14993 variables of void type.
14995 2017-08-11 Martin Liska <mliska@suse.cz>
14997 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14998 TARGET_SUPPORTS_ALIASES.
14999 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15000 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15001 (optimize_weakref): Likewise.
15002 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15003 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15004 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15006 2017-08-11 Martin Liska <mliska@suse.cz>
15009 * config/i386/i386.c (make_resolver_func): Do complete
15010 refactoring of the function.
15012 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15015 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15016 * config/i386/i386.c (ix86_stack_protect_guard): Use
15017 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15018 * doc/invoke.texi (x86 Options): Document
15019 -mstack-protector-guard-symbol= option.
15021 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15023 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15024 * config/i386/i386.c (ix86_split_stack_guard): New function.
15025 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15026 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15027 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15028 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15029 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15030 (split_stack_space_check): Call ix86_split_stack_guard.
15032 2017-08-10 Martin Sebor <msebor@redhat.com>
15034 * print-tree.c (print_node): Print location using the established
15036 Replace spaces with colons.
15037 (debug_raw, debug): Ditto.
15039 2017-08-10 Martin Sebor <msebor@redhat.com>
15042 * pretty-print.c (pp_format): Correct the handling of %s precision.
15044 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15047 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15049 (ix86_finalize_stack_frame_flags): This. Also clear
15050 frame_pointer_needed if -fno-omit-frame-pointer is used without
15052 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15053 with ix86_finalize_stack_frame_flags.
15054 (ix86_expand_epilogue): Likewise.
15055 (ix86_expand_split_stack_prologue): Likewise.
15056 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15058 2017-08-10 Martin Liska <mliska@suse.cz>
15061 * c-attribs.c (handle_target_attribute):
15062 Report warning for an empty string argument of target attribute.
15064 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15067 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15069 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15070 or DECL_NONLOCAL labels.
15071 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15072 or DECL_NONLOCAL labels here.
15074 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15076 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15077 to indicate when early gimple folding has been disabled.
15078 (rs6000_gimple_fold_builtin): Add debug content.
15079 (rs6000_invalid_builtin): Fix whitespace.
15080 (rs6000_expand_builtin): Fix whitespace.
15081 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15083 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15086 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15087 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15089 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15091 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15092 (qdf24xx): Likewise.
15093 * config/aarch64/aarch64.md: Include falkor.md.
15094 * config/aarch64/falkor.md: New.
15096 2017-08-09 Marek Polacek <polacek@redhat.com>
15099 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15100 * diagnostic.c (emit_diagnostic): Add a comment.
15101 (emit_diagnostic_valist): New function.
15103 2017-08-09 Marek Polacek <polacek@redhat.com>
15106 * input.c (make_location): New overload.
15107 * input.h (make_location): Declare.
15109 2017-08-08 Alan Modra <amodra@gmail.com>
15110 H.J. Lu <hongjiu.lu@intel.com>
15113 * gcc.c (NO_PIE_SPEC): Delete.
15114 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15116 (LINK_PIE_SPEC): ..to here.
15117 (LINK_COMMAND_SPEC): Support -no-pie.
15118 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15119 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15121 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15122 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15123 (ENDFILE_CRTEND_SPEC): Similarly.
15125 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15128 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15129 (mstack-protector-guard-offset=): Ditto.
15130 * config/i386/i386.c (ix86_option_override): Handle
15131 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15133 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15134 ix86_stack_protect_guard_offset variables.
15135 (TARGET_STACK_PROTECT_GUARD): Always define.
15136 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15137 and -mstack-protector-guard-offset= options.
15139 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15141 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15142 boundary case for the last candidate.
15144 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15146 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15147 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15149 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15151 PR middle-end/19706
15152 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15153 * config/aarch64/aarch64-builtins.c
15154 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15155 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15156 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15158 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15159 Andrew Pinski <pinskia@gmail.com>
15161 PR middle-end/19706
15162 * internal-fn.def (XORSIGN): New.
15163 * optabs.def (xorsign_optab): New.
15164 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15165 (convert_expand_mult_copysign): New.
15166 (pass_optimize_widening_mul::execute): Call
15167 convert_expand_mult_copysign.
15169 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15171 PR tree-optimization/81354
15172 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15173 Insert on edges rather than explicitly creating landing pads.
15174 (analyze_candidates_and_replace): Commit edge inserts.
15176 2017-08-08 Richard Biener <rguenther@suse.de>
15178 PR middle-end/81719
15179 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15180 (expand_simple_operations): Also look through ADDR_EXPRs with
15181 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15183 2017-08-08 Richard Biener <rguenther@suse.de>
15185 PR tree-optimization/81723
15186 * tree-vect-slp.c (struct bst_traits): New hash traits.
15187 (bst_fail): New global.
15188 (vect_build_slp_tree_2): New worker, split out from ...
15189 (vect_build_slp_tree): ... this now wrapping it with using
15190 bst_fail set to cache SLP tree build fails. Properly handle
15192 (vect_analyze_slp_instance): Allocate and free bst_fail.
15194 2017-08-08 Martin Liska <mliska@suse.cz>
15197 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15198 LABEL_DECLs that can be from a different function.
15200 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15202 PR tree-optimization/81744
15203 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15204 loop's number of iterations.
15206 2017-08-08 Martin Liska <mliska@suse.cz>
15208 * asan.c: Include header files.
15209 * attribs.c (build_decl_attribute_variant): New function moved
15211 (build_type_attribute_qual_variant): Likewise.
15212 (cmp_attrib_identifiers): Likewise.
15213 (simple_cst_list_equal): Likewise.
15214 (omp_declare_simd_clauses_equal): Likewise.
15215 (attribute_value_equal): Likewise.
15216 (comp_type_attributes): Likewise.
15217 (build_type_attribute_variant): Likewise.
15218 (lookup_ident_attribute): Likewise.
15219 (remove_attribute): Likewise.
15220 (merge_attributes): Likewise.
15221 (merge_type_attributes): Likewise.
15222 (merge_decl_attributes): Likewise.
15223 (merge_dllimport_decl_attributes): Likewise.
15224 (handle_dll_attribute): Likewise.
15225 (attribute_list_equal): Likewise.
15226 (attribute_list_contained): Likewise.
15227 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15228 (lookup_attribute_by_prefix): Likewise.
15229 * bb-reorder.c: Include header files.
15230 * builtins.c: Likewise.
15231 * calls.c: Likewise.
15232 * cfgexpand.c: Likewise.
15233 * cgraph.c: Likewise.
15234 * cgraphunit.c: Likewise.
15235 * convert.c: Likewise.
15236 * dwarf2out.c: Likewise.
15237 * final.c: Likewise.
15238 * fold-const.c: Likewise.
15239 * function.c: Likewise.
15240 * gimple-expr.c: Likewise.
15241 * gimple-fold.c: Likewise.
15242 * gimple-pretty-print.c: Likewise.
15243 * gimple.c: Likewise.
15244 * gimplify.c: Likewise.
15245 * hsa-common.c: Likewise.
15246 * hsa-gen.c: Likewise.
15247 * internal-fn.c: Likewise.
15248 * ipa-chkp.c: Likewise.
15249 * ipa-cp.c: Likewise.
15250 * ipa-devirt.c: Likewise.
15251 * ipa-fnsummary.c: Likewise.
15252 * ipa-inline.c: Likewise.
15253 * ipa-visibility.c: Likewise.
15255 * lto-cgraph.c: Likewise.
15256 * omp-expand.c: Likewise.
15257 * omp-general.c: Likewise.
15258 * omp-low.c: Likewise.
15259 * omp-offload.c: Likewise.
15260 * omp-simd-clone.c: Likewise.
15261 * opts-global.c: Likewise.
15262 * passes.c: Likewise.
15263 * predict.c: Likewise.
15264 * sancov.c: Likewise.
15265 * sanopt.c: Likewise.
15266 * symtab.c: Likewise.
15267 * toplev.c: Likewise.
15268 * trans-mem.c: Likewise.
15269 * tree-chkp.c: Likewise.
15270 * tree-eh.c: Likewise.
15271 * tree-into-ssa.c: Likewise.
15272 * tree-object-size.c: Likewise.
15273 * tree-parloops.c: Likewise.
15274 * tree-profile.c: Likewise.
15275 * tree-ssa-ccp.c: Likewise.
15276 * tree-ssa-live.c: Likewise.
15277 * tree-ssa-loop.c: Likewise.
15278 * tree-ssa-sccvn.c: Likewise.
15279 * tree-ssa-structalias.c: Likewise.
15280 * tree-ssa.c: Likewise.
15281 * tree-streamer-in.c: Likewise.
15282 * tree-vectorizer.c: Likewise.
15283 * tree-vrp.c: Likewise.
15284 * tsan.c: Likewise.
15285 * ubsan.c: Likewise.
15286 * varasm.c: Likewise.
15287 * varpool.c: Likewise.
15288 * tree.c: Remove functions moved to attribs.[ch].
15289 * tree.h: Likewise.
15290 * config/aarch64/aarch64.c: Add attrs.h header file.
15291 * config/alpha/alpha.c: Likewise.
15292 * config/arc/arc.c: Likewise.
15293 * config/arm/arm.c: Likewise.
15294 * config/avr/avr.c: Likewise.
15295 * config/bfin/bfin.c: Likewise.
15296 * config/c6x/c6x.c: Likewise.
15297 * config/cr16/cr16.c: Likewise.
15298 * config/cris/cris.c: Likewise.
15299 * config/darwin.c: Likewise.
15300 * config/epiphany/epiphany.c: Likewise.
15301 * config/fr30/fr30.c: Likewise.
15302 * config/frv/frv.c: Likewise.
15303 * config/ft32/ft32.c: Likewise.
15304 * config/h8300/h8300.c: Likewise.
15305 * config/i386/winnt.c: Likewise.
15306 * config/ia64/ia64.c: Likewise.
15307 * config/iq2000/iq2000.c: Likewise.
15308 * config/lm32/lm32.c: Likewise.
15309 * config/m32c/m32c.c: Likewise.
15310 * config/m32r/m32r.c: Likewise.
15311 * config/m68k/m68k.c: Likewise.
15312 * config/mcore/mcore.c: Likewise.
15313 * config/microblaze/microblaze.c: Likewise.
15314 * config/mips/mips.c: Likewise.
15315 * config/mmix/mmix.c: Likewise.
15316 * config/mn10300/mn10300.c: Likewise.
15317 * config/moxie/moxie.c: Likewise.
15318 * config/msp430/msp430.c: Likewise.
15319 * config/nds32/nds32-isr.c: Likewise.
15320 * config/nds32/nds32.c: Likewise.
15321 * config/nios2/nios2.c: Likewise.
15322 * config/nvptx/nvptx.c: Likewise.
15323 * config/pa/pa.c: Likewise.
15324 * config/pdp11/pdp11.c: Likewise.
15325 * config/powerpcspe/powerpcspe.c: Likewise.
15326 * config/riscv/riscv.c: Likewise.
15327 * config/rl78/rl78.c: Likewise.
15328 * config/rx/rx.c: Likewise.
15329 * config/s390/s390.c: Likewise.
15330 * config/sh/sh.c: Likewise.
15331 * config/sol2.c: Likewise.
15332 * config/sparc/sparc.c: Likewise.
15333 * config/spu/spu.c: Likewise.
15334 * config/stormy16/stormy16.c: Likewise.
15335 * config/tilegx/tilegx.c: Likewise.
15336 * config/tilepro/tilepro.c: Likewise.
15337 * config/v850/v850.c: Likewise.
15338 * config/vax/vax.c: Likewise.
15339 * config/visium/visium.c: Likewise.
15340 * config/xtensa/xtensa.c: Likewise.
15342 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15345 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15346 constraints since the -mupper-regs-* switches have been
15348 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15349 into a vector from a double word element that was extracted from
15350 another vector, and eliminate extra XXPERMDI instructions.
15351 (vsx_concat_<mode>_2): Likewise.
15352 (vsx_concat_<mode>_3): Likewise.
15353 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15354 concat to allow optimizing inserts from previous extracts.
15356 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15358 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15359 memory reference to a SSP offset in TLS address space.
15360 (ix86_print_operand) <case '@'>: Remove.
15361 (ix86_print_operand_punct_valid_p): Remove '@' code.
15362 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15363 UNSPEC_SP_TLS_TEST.
15364 (stack_tls_protect_set_<mode>): Remove.
15365 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15366 (stack_tls_protect_test_<mode>): Remove.
15367 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15369 2017-08-07 Olivier Hainque <hainque@adacore.com>
15372 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15374 2017-08-07 Douglas Rupp <rupp@adacore.com>
15376 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15377 variable was referenced as multidir in command.
15379 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15382 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15385 2017-08-07 Martin Liska <mliska@suse.cz>
15387 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15388 * config/rl78/rl78.c: Add include of attribs.h.
15389 * config/sh/sh.c: Likewise.
15390 * config/v850/v850.c: Likewise.
15392 2017-08-07 Tom de Vries <tom@codesourcery.com>
15394 PR middle-end/78266
15395 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15397 2017-08-07 Martin Liska <mliska@suse.cz>
15399 * config/mips/mips.c: Include attribs.h.
15401 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15404 * doc/invoke.texi: Document change in behvaior for -Ofast for
15407 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15409 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15411 (aarch64_pop_regs): Likewise.
15412 (aarch64_gen_load_pair): Likewise.
15413 (aarch64_save_callee_saves): Likewise.
15414 (aarch64_restore_callee_saves): Likewise.
15416 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15418 * config/i386/i386.c: Revert the last change.
15420 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15423 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15425 (ix86_finalize_stack_frame_flags): This. Also clear
15426 frame_pointer_needed if -fno-omit-frame-pointer is used without
15428 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15429 with ix86_finalize_stack_frame_flags.
15430 (ix86_expand_epilogue): Likewise.
15431 (ix86_expand_split_stack_prologue): Likewise.
15433 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15436 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15437 to P_AES for Westmere.
15439 2017-08-07 Jonathan Yong <10walls@gmail.com>
15441 * config/i386/mingw.opt (fset-stack-executable): Removed.
15442 * config/i386/cygming.opt (fset-stack-executable): Moved
15444 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15446 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15448 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15450 2017-08-07 Marek Polacek <polacek@redhat.com>
15452 PR middle-end/81737
15453 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15455 2017-08-07 Martin Liska <mliska@suse.cz>
15457 * attribs.h (canonicalize_attr_name): New function.
15458 (cmp_attribs): Move from c-format.c and adjusted.
15459 (is_attribute_p): Moved from tree.h.
15460 * tree-inline.c: Add new includes.
15461 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15462 (private_is_attribute_p): Remove.
15463 (private_lookup_attribute): Likewise.
15464 (private_lookup_attribute_by_prefix): Simplify.
15465 (remove_attribute): Use is_attribute_p.
15466 * tree.h: Remove removed declarations.
15468 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15470 PR middle-end/81698
15471 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15472 instead of computing it in the function. Formatting fix.
15473 (expand_case): Don't rely on default_edge being the first edge,
15474 clear it if removing it, pass default_edge to
15475 emit_case_dispatch_table.
15476 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15479 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15481 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15482 insn in the function, emit NOP after the insn.
15484 2017-08-06 Tom de Vries <tom@codesourcery.com>
15486 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15489 2017-08-06 Tom de Vries <tom@codesourcery.com>
15491 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15494 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15496 PR tree-optimization/57371
15497 * match.pd: New pattern.
15499 2017-08-04 Marek Polacek <polacek@redhat.com>
15501 PR middle-end/81695
15502 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15503 perform the computation in offset_int.
15505 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15507 PR tree-optimization/81136
15508 * tree-vectorizer.h: Include tree-hash-traits.h.
15509 (vec_base_alignments): New typedef.
15510 (vec_info): Add a base_alignments field.
15511 (vect_record_base_alignments): Declare.
15512 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15514 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15515 (create_data_ref): Add an is_conditional_in_stmt argument.
15516 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15517 the is_conditional_in_stmt field.
15518 (data_ref_loc): Add an is_conditional_in_stmt field.
15519 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15520 (find_data_references_in_stmt): Update call to create_data_ref.
15521 (graphite_find_data_references_in_stmt): Likewise.
15522 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15523 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15524 (vect_record_base_alignment): New function.
15525 (vect_record_base_alignments): Likewise.
15526 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15527 for nested statements even if we fail to compute a misalignment.
15528 Use pooled base alignments for unconditional references.
15529 (vect_find_same_alignment_drs): Compare base addresses instead
15531 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15532 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15534 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15536 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15537 Add an explicit name for the enum. Use auto_vec for slp_instances
15538 and grouped_stores.
15539 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15541 (_bb_vec_info): Add a constructor and destructor.
15542 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15543 (destroy_loop_vec_info): Delete.
15544 (vect_destroy_datarefs): Likewise.
15545 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15546 (vec_info::vec_info): New function.
15547 (vec_info::~vec_info): Likewise.
15548 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15549 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15550 destroy_loop_vec_info.
15551 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15552 (_loop_vec_info::_loop_vec_info): ...this.
15553 (destroy_loop_vec_info): Replace with...
15554 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15555 the stmt_vec_infos. Leave handling of vec_info information to its
15556 destructor. Remove explicit vector releases.
15557 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15558 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15559 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15560 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15561 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15562 (destroy_bb_vec_info): Replace with...
15563 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15564 information to its destructor.
15565 (vect_slp_analyze_bb_1): Use new and delete instead of
15566 new_bb_vec_info and destroy_bb_vec_info.
15567 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15570 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15572 * tree-data-ref.h (subscript): Add access_fn field.
15573 (data_dependence_relation): Add could_be_independent_p.
15574 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15575 (same_access_functions): Move to tree-data-ref.c.
15576 * tree-data-ref.c (ref_contains_union_access_p): New function.
15577 (access_fn_component_p): Likewise.
15578 (access_fn_components_comparable_p): Likewise.
15579 (dr_analyze_indices): Add a reference to access_fn_component_p.
15580 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15582 (constant_access_functions): Likewise.
15583 (add_other_self_distances): Likewise.
15584 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15585 (initialize_data_dependence_relation): Use XCNEW and remove
15586 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15587 of access functions that have the same type. Allow the
15588 subsequence to end with different bases in some circumstances.
15589 Record the chosen access functions in SUB_ACCESS_FN.
15590 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15591 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15592 (subscript_dependence_tester_1): Likewise dra and drb.
15593 (build_classic_dist_vector): Update calls accordingly.
15594 (subscript_dependence_tester): Likewise.
15595 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15596 DDR_COULD_BE_INDEPENDENT_P.
15597 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15598 comp_alias_ddrs instead of may_alias_ddrs.
15599 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15601 (vect_analyze_data_ref_dependence): Use it if
15602 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15603 distance vectors if that fails.
15604 (dependence_distance_ge_vf): New function.
15605 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15606 LOOP_VINFO_MAY_ALIAS_DDRS.
15608 2017-08-04 Richard Biener <rguenther@suse.de>
15610 PR middle-end/81705
15611 * fold-const.c (fold_binary_loc): Properly restrict
15612 minus_var0 && minus_var1 case when associating undefined overflow
15615 2017-08-04 Tom de Vries <tom@codesourcery.com>
15617 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15619 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15621 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15622 Don't start diagnostic messages with a capital letter.
15623 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15625 (rs6000_invalid_builtin): Likewise.
15626 (rs6000_trampoline_init): Likewise.
15628 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15631 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15632 after setting changeable df flags.
15634 2017-08-03 Richard Biener <rguenther@suse.de>
15636 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15637 up if the use is in USE - X.
15639 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15641 * toplev.c (dumpfile.h): New include.
15642 (internal_error_reentered): New static function. Use it...
15643 (internal_error_function): ...here to handle reentered internal_error.
15645 2017-08-03 Richard Biener <rguenther@suse.de>
15647 PR middle-end/81148
15648 * fold-const.c (split_tree): Add minus_var and minus_con
15649 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15650 here but always use minus_*.
15651 (associate_trees): Assert we never associate with MINUS_EXPR
15652 and NULL first operand. Do not recurse for PLUS_EXPR operands
15653 when associating as MINUS_EXPR either.
15654 (fold_binary_loc): Track minus_var and minus_con.
15656 2017-08-03 Tom de Vries <tom@codesourcery.com>
15659 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15660 ACCEL_COMPILER, apply finish_options on
15661 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15663 2017-08-03 Tom de Vries <tom@codesourcery.com>
15666 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15667 function_entry_patch_area_size > 0.
15669 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15672 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15673 instead of 10??LU, perform unit multiplication in wide_int,
15674 don't change alloc_object_size_limit if the limit is larger
15677 PR tree-optimization/81655
15678 PR tree-optimization/81588
15679 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15680 the case when ranges[i].low and high are 1 for unsigned type with
15683 PR middle-end/81052
15684 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15685 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15687 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15689 * tree-vrp.h: Add include guard.
15691 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15694 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15695 (ud2): New insn pattern.
15696 * config/i386/i386.c (ix86_expand_epilogue):
15697 For naked functions, generate ud2 instead of trap insn.
15699 2017-08-02 Marek Polacek <polacek@redhat.com>
15702 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15704 2017-08-02 Tom de Vries <tom@codesourcery.com>
15705 Cesar Philippidis <cesar@codesourcery.com>
15707 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15708 Add missing edge probabilities.
15710 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15712 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15713 Correct endianness.
15715 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15717 PR middle-end/79499
15718 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15719 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15720 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15721 prologue_seq sequences - if any.
15723 2017-08-02 Richard Biener <rguenther@suse.de>
15725 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15726 via vectors if supported, integer extracts via punning if supported
15727 or otherwise vector extracts.
15729 2017-08-02 Richard Biener <rguenther@suse.de>
15731 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15733 (bitmap_insert_into_set): ... this.
15735 2017-08-02 Richard Biener <rguenther@suse.de>
15737 PR tree-optimization/81633
15739 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15741 PR tree-optimization/71752
15742 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15744 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15746 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15747 (machine_function::call_ms2sysv_pad_out): Remove field.
15748 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15749 (ix86_compute_frame_layout): Likewise.
15751 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15754 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15755 attribute with interrupt attribute.
15757 2017-08-01 Andrew Pinski <apinski@cavium.com>
15759 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15760 BIT_INSERT_EXPR's operand 1
15761 to see if the types precision matches.
15763 2017-08-01 Martin Liska <mliska@suse.cz>
15765 PR middle-end/70140
15766 * builtins.c (expand_builtin_memcpy_args): Remove.
15767 (expand_builtin_memcpy): Call newly added function
15768 expand_builtin_memory_copy_args.
15769 (expand_builtin_memcpy_with_bounds): Likewise.
15770 (expand_builtin_mempcpy): Remove last argument.
15771 (expand_builtin_mempcpy_with_bounds): Likewise.
15772 (expand_builtin_memory_copy_args): New function created from
15773 expand_builtin_mempcpy_args with small modifications.
15774 (expand_builtin_mempcpy_args): Remove.
15775 (expand_builtin_stpcpy): Remove unused argument.
15776 (expand_builtin): Likewise.
15777 (expand_builtin_with_bounds): Likewise.
15779 2017-08-01 Martin Liska <mliska@suse.cz>
15782 Make mempcpy more optimal (PR middle-end/70140).
15784 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15787 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15788 __builtin_vec_cmpne verify both arguments are compatible vectors
15789 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15790 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15791 move computation of aligned to after checking the argument types.
15795 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15796 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15799 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15800 Doug Rupp <rupp@adacore.com>
15801 Olivier Hainque <hainque@adacore.com>
15803 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15804 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15806 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15807 for TARGET_OS_CPP_BUILTIN.
15808 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15809 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15811 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15812 passing required abi options to the assembler for EABI configurations.
15813 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15814 of .text.hot and .text.unlikely sections for kernel modules when
15815 using ARM style exceptions.
15816 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15817 options. Add EXTRA_CC1_SPEC.
15818 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15820 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15822 (ARM_TARGET2_DWARF_FORMAT): Define.
15823 * config/arm/t-vxworks: Adjust multilib control to removal of the
15824 Diab command line options.
15826 2017-08-01 Martin Liska <mliska@suse.cz>
15828 PR gcov-profile/81561
15829 * gcov.c (unblock): Make unblocking safe as we need to preserve
15830 index correspondence of blocks and block_lists.
15832 2017-08-01 Richard Biener <rguenther@suse.de>
15834 PR tree-optimization/81181
15835 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15836 (compute_antic): ... end of iteration here.
15838 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15840 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15841 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15842 (ftree-slp-vectorize): Likewise.
15843 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15844 can no longer be set independent of flag_tree_loop_vectorize.
15845 * omp-general.c (emp_max_vf): Likewise.
15846 * opts.c (enable_fdo_optimizations): Remove references to
15847 flag_tree_vectorize, these are now implicit.
15848 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15849 and leave it for the options machinery.
15851 2017-08-01 Martin Liska <mliska@suse.cz>
15853 PR middle-end/70140
15854 * builtins.c (expand_builtin_memcpy_args): Remove.
15855 (expand_builtin_memcpy): Call newly added function
15856 expand_builtin_memory_copy_args.
15857 (expand_builtin_memcpy_with_bounds): Likewise.
15858 (expand_builtin_mempcpy): Remove last argument.
15859 (expand_builtin_mempcpy_with_bounds): Likewise.
15860 (expand_builtin_memory_copy_args): New function created from
15861 expand_builtin_mempcpy_args with small modifications.
15862 (expand_builtin_mempcpy_args): Remove.
15863 (expand_builtin_stpcpy): Remove unused argument.
15864 (expand_builtin): Likewise.
15865 (expand_builtin_with_bounds): Likewise.
15867 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15870 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15871 print "ds:" only for immediates in generic address space.
15873 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15876 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15877 (ix86_function_ok_for_sibcall): Return false for naked functions.
15879 2017-08-01 Richard Biener <rguenther@suse.de>
15881 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15882 (compute_antic): Seed worklist with exit block predecessors.
15883 * cfganal.c (dfs_find_deadend): For a cycle return the source
15884 of the edge closing it.
15886 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15888 * config/aarch64/aarch64.c
15889 (aarch64_can_const_movi_rtx_p): Move 0 check.
15891 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15893 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15894 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15896 * match.pd: Ditto in address comparison pattern.
15898 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15900 PR tree-optimization/81627
15901 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15902 closed ssa form for store-store chain.
15904 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15906 PR tree-optimization/81620
15907 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15908 for store-store chain.
15910 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15912 PR tree-optimization/81588
15913 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15914 ranges[i].in_p, invert comparison code ccode. For >/>=,
15915 swap rhs1 and rhs2 and comparison code unconditionally,
15916 for </<= don't do that. Don't swap rhs1/rhs2 again if
15917 ranges[i].in_p, instead invert comparison code ccode if
15918 opcode or oe->rank is BIT_IOR_EXPR.
15921 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15922 a direct optab to conversion optab.
15923 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15924 with GET_MODE_INNER as last argument instead of optab_handler.
15925 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15926 vector extraction if possible and optab is available.
15927 * expr.c (store_constructor): Use convert_optab_handler instead
15928 of optab_handler. Use vector initialization from smaller
15929 vectors if possible and optab is available.
15930 * tree-vect-stmts.c (vectorizable_load): Likewise.
15931 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15932 now have two modes.
15933 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15934 of vec_init from half-sized vectors with the same element mode.
15935 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15936 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15937 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15938 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15939 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15940 after mode in gen_vec_extract* calls.
15941 (vec_extract<mode>): Renamed to ...
15942 (vec_extract<mode><ssescalarmodelower>): ... this.
15943 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15944 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15945 element mode after mode in gen_vec_init* calls.
15946 (VEC_INIT_HALF_MODE): New mode iterator.
15947 (vec_init<mode>): Renamed to ...
15948 (vec_init<mode><ssescalarmodelower>): ... this.
15949 (vec_init<mode><ssehalfvecmodelower>): New expander.
15950 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15951 (vec_extractv2sfsf): ... this.
15952 (vec_initv2sf): Renamed to ...
15953 (vec_initv2sfsf): ... this.
15954 (vec_extractv2si): Renamed to ...
15955 (vec_extractv2sisi): ... this.
15956 (vec_initv2si): Renamed to ...
15957 (vec_initv2sisi): ... this.
15958 (vec_extractv4hi): Renamed to ...
15959 (vec_extractv4hihi): ... this.
15960 (vec_initv4hi): Renamed to ...
15961 (vec_initv4hihi): ... this.
15962 (vec_extractv8qi): Renamed to ...
15963 (vec_extractv8qiqi): ... this.
15964 (vec_initv8qi): Renamed to ...
15965 (vec_initv8qiqi): ... this.
15966 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15967 (vec_init<mode>): Renamed to ...
15968 (vec_init<mode><VEC_base_l>): ... this.
15969 (vec_extract<mode>): Renamed to ...
15970 (vec_extract<mode><VEC_base_l>): ... this.
15971 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15972 (vec_initv2sfsf): ... this.
15973 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15974 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15975 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15976 element mode after mode in gen_vec_init* calls.
15977 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15978 (vec_init<mode><Vel>): ... this.
15979 (vec_extract<mode>): Renamed to ...
15980 (vec_extract<mode><Vel>): ... this.
15981 * config/aarch64/iterators.md (Vel): New mode attribute.
15982 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15983 Add element mode after mode in gen_vec_extract* calls.
15984 * config/s390/vector.md (non_vec_l): New mode attribute.
15985 (vec_extract<mode>): Renamed to ...
15986 (vec_extract<mode><non_vec_l>): ... this.
15987 (vec_init<mode>): Renamed to ...
15988 (vec_init<mode><non_vec_l>): ... this.
15989 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15990 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15992 * config/arm/iterators.md (V_elem_l): New mode attribute.
15993 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15994 (vec_extract<mode><V_elem_l>): ... this.
15995 (vec_extractv2di): Renamed to ...
15996 (vec_extractv2didi): ... this.
15997 (vec_init<mode>): Renamed to ...
15998 (vec_init<mode><V_elem_l>): ... this.
15999 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16000 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16001 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16002 Add element mode after gen_vec_extract* calls.
16003 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16004 (vec_init<mode><unitmode>): ... this.
16005 (vec_extract<mode>): Renamed to ...
16006 (vec_extract<mode><unitmode>): ... this.
16007 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16008 (vec_init<mode><unitmode>): ... this.
16009 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16010 (vec_initv2sfsf): ... this.
16011 (vec_extractv2sf): Renamed to ...
16012 (vec_extractv2sfsf): ... this.
16013 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16014 Add element mode after gen_vec_extract* calls.
16015 * config/mips/mips.md (unitmode): New mode iterator.
16016 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16017 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16018 * config/spu/spu.md (inner_l): New mode attribute.
16019 (vec_init<mode>): Renamed to ...
16020 (vec_init<mode><inner_l>): ... this.
16021 (vec_extract<mode>): Renamed to ...
16022 (vec_extract<mode><inner_l>): ... this.
16023 * config/sparc/sparc.md (veltmode): New mode iterator.
16024 (vec_init<VMALL:mode>): Renamed to ...
16025 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16026 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16027 (vec_initv2sisi): ... this.
16028 (vec_initv2sf): Renamed to ...
16029 (vec_initv2sfsf): ... this.
16030 (vec_extractv2sf): Renamed to ...
16031 (vec_extractv2sfsf): ... this.
16032 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16033 (vec_init<mode>): Renamed to ...
16034 (vec_init<mode><VEC_base_l>): ... this.
16035 (vec_extract<mode>): Renamed to ...
16036 (vec_extract<mode><VEC_base_l>): ... this.
16037 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16038 (vec_initv2sfsf): ... this.
16039 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16040 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16041 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16042 gen_vec_init* calls.
16044 2017-08-01 Richard Biener <rguenther@suse.de>
16046 PR tree-optimization/81297
16047 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16048 TREE_OVERFLOW from INTEGER_CSTs.
16050 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16052 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16054 2017-07-31 Carl Love <cel@us.ibm.com>
16056 * config/rs6000/rs6000-c: Add support for built-in functions
16057 vector signed char vec_xl_be (signed long long, signed char *);
16058 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16059 vector signed int vec_xl_be (signed long long, signed int *);
16060 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16061 vector signed long long vec_xl_be (signed long long, signed long long *);
16062 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16063 vector signed short vec_xl_be (signed long long, signed short *);
16064 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16065 vector double vec_xl_be (signed long long, double *);
16066 vector float vec_xl_be (signed long long, float *);
16067 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16068 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16069 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16071 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16072 (altivec_expand_builtin): Add switch statement to call
16073 altivec_expand_xl_be for each builtin.
16074 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16075 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16076 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16077 __builtin_vsx_le_be_v16qi.
16078 * doc/extend.texi: Update the built-in documentation file for the
16079 new built-in functions.
16081 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16084 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16086 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16088 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16090 * config.gcc: Add z14.
16091 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16092 CPU model numbers for z13s and z14.
16093 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16095 * config/s390/s390-opts.h (enum processor_type): Rename
16096 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16097 * config/s390/s390.c (processor_table): Add field for CPU name to
16098 be passed to Binutils.
16099 (s390_asm_output_machine_for_arch): Use the new field in
16100 processor_table for Binutils.
16101 (s390_expand_builtin): Replace arch12 with z14.
16102 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16103 (s390_get_sched_attrmask): Likewise.
16104 (s390_get_unit_mask): Likewise.
16105 * config/s390/s390.opt: Add z14 to processor_type enum.
16107 2017-07-31 Martin Jambor <mjambor@suse.cz>
16110 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16111 regardless of optimization level.
16113 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16114 Martin Liska <mliska@suse.cz>
16116 * predict.def: Remove old comment and adjust probability.
16117 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16118 PREDICT statements.
16120 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16123 * config/i386/i386.c (ix86_function_naked): New function.
16124 (ix86_can_use_return_insn_p): Return false for naked functions.
16125 (ix86_expand_prologue): Skip prologue for naked functions.
16126 (ix86_expand_epilogue): Skip epilogue for naked functions
16127 and emit trap instruction.
16128 (ix86_warn_func_return): New function.
16129 (ix86_attribute_table): Add "naked" attribute specification.
16130 (TARGET_WARN_FUNC_RETURN): Define.
16131 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16133 2017-07-31 Martin Liska <mliska@suse.cz>
16135 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16136 (dump_gimple_bb_header): Always dump BB info.
16137 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16139 2017-07-31 Martin Liska <mliska@suse.cz>
16142 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16143 also with current_function_decl non-null equality.
16145 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16148 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16149 change type to the element type, instead add eltype variable and
16150 use it where we are interested in the element type.
16152 PR tree-optimization/81603
16153 * ipa-polymorphic-call.c
16154 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16155 offset arithmetic in offset_int, bail out if the resulting bit offset
16156 doesn't fit into shwi.
16158 2017-07-31 Martin Liska <mliska@suse.cz>
16160 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16161 (gimplify_save_expr): Fix comment.
16163 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16166 * config/i386/i386.c (ix86_function_arg): Update arguments for
16168 (ix86_compute_frame_layout): Set the initial stack offset to
16169 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16170 INCOMING_FRAME_SP_OFFSET.
16171 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16172 stack before exception handler returns.
16173 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16174 the 'ERROR_CODE' for exception handler.
16176 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16178 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16179 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16180 (ASM_OUTPUT_REG_POP): Ditto.
16181 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16182 instead of asm_fprintf to output pure string.
16184 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16186 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16187 to imported_module_or_decl hook.
16188 (debug_nothing_tree_tree_tree_bool): Remove.
16189 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16190 * debug.c (do_nothing_debug_hooks): Use
16191 debug_nothing_tree_tree_tree_bool_bool instead of
16192 debug_nothing_tree_tree_tree_bool.
16193 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16194 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16195 * sdbout.c (sdb_debug_hooks): Likewise.
16196 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16197 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16199 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16200 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16201 attribute, don't add anything.
16203 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16205 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16206 (fold_build2_stat_loc): Likewise.
16207 (fold_build3_stat_loc): Likewise.
16208 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16209 (fold_build1_loc): Remove macro.
16210 (fold_build2_loc): Likewise.
16211 (fold_build3_loc): Likewise.
16213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16215 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16216 (gimple_build_debug_bind_source_stat): Likewise.
16217 * gimple.h (gimple_build_debug_bind): Remove macro.
16218 (gimple_build_debug_bind_source): Likewise.
16220 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16222 * bitmap.c (bitmap_alloc): Adjust.
16223 (bitmap_gc_alloc): Likewise.
16224 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16226 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16228 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16229 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16230 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16231 (bitmap_gc_alloc_stat): Likewise.
16232 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16234 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16236 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16237 * rtl.h (shallow_copy_rtx): Remove macro.
16239 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16241 * emit-rtl.c (gen_raw_REG): Adjust.
16242 * gengenrtl.c (gendef): Likewise.
16243 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16244 * rtl.h (rtx_alloc): Remove macro.
16246 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16248 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16249 (build_tree_list_stat): Likewise.
16250 * tree.h (build_tree_list): Remove macro.
16251 (build_tree_list_vec): Likewise.
16253 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16255 * tree.c (make_vector_stat): Remove _stat from name.
16256 (build_vector_stat): Likewise.
16257 * tree.h (make_vector_stat): Remove macro.
16258 (build_vector_stat): Likewise.
16260 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16262 * tree.h (build_var_debug_value): Remove prototype.
16264 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16266 * tree.c (tree_cons_stat): Remove _stat from name.
16267 * tree.h (tree_cons): Remove macro.
16269 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16271 * tree.c (build_vl_exp_stat): Remove _stat from name.
16272 * tree.h (build_vl_exp): Remove macro.
16274 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16276 * tree.c (build_decl_stat): Remove _stat from name.
16277 * tree.h (build_decl): Remove macro.
16279 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16281 * gimple.c (gimple_build_with_ops_stat): Adjust.
16282 (gimple_alloc_stat): Remove _stat from name.
16283 * gimple.h (gimple_alloc): Remove macro.
16285 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16287 * tree.c (make_tree_vec_stat): Remove _stat from name.
16288 (grow_tree_vec_stat): Likewise.
16289 * tree.h (make_tree_vec_stat): Adjust prototype.
16290 (grow_tree_vec_stat): Likewise.
16291 (make_tree_vec): Remove macro.
16292 (grow_tree_vec): Likewise.
16294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16296 * fold-const.c (fold_build1_stat_loc): Adjust.
16297 (fold_build2_stat_loc): Likewise.
16298 (fold_build3_stat_loc): Likewise.
16299 * tree.c (build0_stat): Remove _stat from name.
16300 (build1_stat): Likewise.
16301 (build2_stat): Likewise.
16302 (build3_stat): Likewise.
16303 (build4_stat): Likewise.
16304 (build5_stat): Likewise.
16305 * tree.h (build1_loc): Remove macro, and rename _stat function
16307 (build2_loc): Likewise.
16308 (build3_loc): Likewise.
16309 (build4_loc): Likewise.
16310 (build5_loc): Likewise.
16312 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16314 * tree.c (make_int_cst_stat): Remove _stat from name.
16315 * tree.h (make_int_cst_stat): Adjust prototype.
16316 (make_int_cst): Remove macro.
16318 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16320 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16321 * tree.h (make_tree_binfo_stat): Adjust prototype.
16322 (make_tree_binfo): Remove.
16324 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16326 * tree.c (copy_node_stat): Rename to copy_node.
16327 (build_distinct_type_copy): Adjust.
16328 * tree.h (copy_node_stat): Adjust prototype.
16329 (copy_node): Remove macro.
16331 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16333 * tree.c (make_node_stat): rename to make_node.
16334 (build_tree_list_stat): Adjust.
16335 (build0_stat): Likewise.
16336 (build2_stat): Likewise.
16337 (build3_stat): Likewise.
16338 (build4_stat): Likewise.
16339 (build5_stat): Likewise.
16340 (build_decl_stat): Likewise.
16341 * tree.h (make_node_stat): Adjust prototype.
16342 (make_node): remove macro.
16344 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16346 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16347 (PPC_FEATURE2_SCV): Likewise.
16348 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16350 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16352 * config/aarch64/aarch64.c
16353 (aarch64_internal_mov_immediate): Add new special pattern.
16354 * config/aarch64/aarch64.md (*movdi_aarch64):
16355 Add reg/32bit const mov case.
16357 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16358 Richard Sandiford <richard.sandiford@linaro.org>
16360 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16361 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16362 Add integer and movi cases.
16363 (movi-split-hf-df-sf split, fp16): New.
16364 (enabled): Added TARGET_FP_F16INST.
16365 * config/aarch64/iterators.md (GPF_HF): New.
16366 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16368 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16370 * config/aarch64/aarch64.c
16371 (aarch64_simd_container_mode): Add prototype.
16372 (aarch64_expand_mov_immediate): Add HI support.
16373 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16374 (aarch64_can_const_movi_rtx_p): New.
16375 (aarch64_preferred_reload_class):
16376 Remove restrictions of using FP registers for certain SIMD operations.
16377 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16378 (aarch64_valid_floating_const): Add integer move validation.
16379 (aarch64_simd_imm_scalar_p): Remove.
16380 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16381 (aarch64_legitimate_constant_p): Expand list of supported cases.
16382 * config/aarch64/aarch64-protos.h
16383 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16384 (aarch64_reinterpret_float_as_int): New.
16385 (aarch64_simd_imm_scalar_p): Remove.
16386 * config/aarch64/constraints.md (Uvi): New.
16387 (Dd): Split into Ds and new Dd.
16388 * config/aarch64/aarch64.md (*movsi_aarch64):
16390 (*movdi_aarch64): Add SIMD mov case.
16392 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16394 * tree-predcom.c: (struct chain): Handle store-store chain in which
16395 stores for elimination only store loop invariant values.
16396 (execute_pred_commoning_chain): Ditto.
16397 (prepare_initializers_chain_store_elim): Ditto.
16398 (prepare_finalizers): Ditto.
16399 (is_inv_store_elimination_chain): New function.
16400 (initialize_root_vars_store_elim_1): New function.
16402 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16404 * tree-predcom.c: Revise general description of the pass.
16405 (enum chain_type): New enum type for store elimination.
16406 (struct chain): New field supporting store elimination.
16407 (struct component): Ditto.
16408 (dump_chain): Dump store-stores chain.
16409 (release_chain): Release resources.
16410 (split_data_refs_to_components): Compute and create component
16411 contains only stores for elimination.
16412 (get_chain_last_ref_at): New function.
16413 (make_invariant_chain): Initialization.
16414 (make_rooted_chain): Specify chain type in parameter and record it.
16415 (add_looparound_copies): Skip for store-stores chain.
16416 (determine_roots_comp): Compute type of chain and pass it to
16418 (initialize_root_vars_store_elim_2): New function.
16419 (finalize_eliminated_stores): New function.
16420 (remove_stmt): Handle store for elimination.
16421 (execute_pred_commoning_chain): Execute predictive commoning on
16422 store-store chains.
16423 (determine_unroll_factor): Skip unroll for store-stores chain.
16424 (prepare_initializers_chain_store_elim): New function.
16425 (prepare_initializers_chain): Hanlde store-store chain.
16426 (prepare_finalizers_chain, prepare_finalizers): New function.
16427 (tree_predictive_commoning_loop): Return integer value indicating
16428 if loop is unrolled or lcssa form is corrupted.
16429 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16431 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16433 * tree-predcom.c (initialize_root): Delete.
16434 (execute_pred_commoning_chain): Initialize root vars and replace
16435 reference of non-combined chain directly, rather than call above
16438 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16440 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16441 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16443 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16445 * tree-predcom.c (struct chain): New field init_seq.
16446 (release_chain): Release init_seq.
16447 (prepare_initializers_chain): Record intialization stmts in above
16449 (insert_init_seqs): New function.
16450 (tree_predictive_commoning_loop): Call insert_init_seqs.
16452 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16454 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16456 2017-07-28 Richard Biener <rguenther@suse.de>
16458 * match.pd: Remove superfluous :c.
16459 * genmatch.c (simplify::id): Add member.
16460 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16462 (current_id): New global.
16463 (dt_node::parent): Move from ...
16464 (dt_operand::parent): ... here. Add for_id member.
16465 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16466 (decision_tree::find_node): Relax order requirement when
16467 merging DT_TRUE nodes to ones inbetween the current simplify
16468 and the one we try to merge with. Add diagnostic whenever
16469 we need to enforce pattern order by not merging.
16470 (decision_tree::insert): Set current_id.
16471 (decision_tree::print_node): Dump parent node and for_id.
16472 (parser::last_id): Add member.
16473 (parser::push_simplify): Assign unique id.
16474 (parser::parser): Initialize last_id.
16476 2017-07-28 Martin Liska <mliska@suse.cz>
16479 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16480 gimple_build_debug_bind.
16482 2017-07-28 Richard Biener <rguenther@suse.de>
16484 PR tree-optimization/81502
16485 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16487 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16489 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16490 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16491 for BIT_FIELD_REF args.
16492 * fold-const.c (make_bit_field_ref): Likewise.
16493 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16495 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16498 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16499 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16500 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16501 Or it into SANITIZER_UNDEFINED.
16502 * ubsan.c: Include gimple-fold.h and varasm.h.
16503 (ubsan_expand_ptr_ifn): New function.
16504 (instrument_pointer_overflow): New function.
16505 (maybe_instrument_pointer_overflow): New function.
16506 (instrument_object_size): Formatting fix.
16507 (pass_ubsan::execute): Call instrument_pointer_overflow
16508 and maybe_instrument_pointer_overflow.
16509 * internal-fn.c (expand_UBSAN_PTR): New function.
16510 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16511 * sanitizer.def (__ubsan_handle_pointer_overflow,
16512 __ubsan_handle_pointer_overflow_abort): New builtins.
16513 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16514 * internal-fn.def (UBSAN_PTR): New internal function.
16515 * opts.c (sanitizer_opts): Add pointer-overflow.
16516 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16517 * fold-const.c (build_range_check): Compute pointer range check in
16518 integral type if pointer arithmetics would be needed. Formatting
16521 2017-07-28 Martin Liska <mliska@suse.cz>
16524 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16525 parameters that are of a variable-length.
16527 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16529 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16531 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16532 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16533 (CRT_CALL_STATIC_FUNCTION): Likewise.
16534 (ASM_DEFAULT_SPEC): New define.
16535 (ASM_SPEC32): Likewise.
16536 (ASM_SPEC64): Likewise.
16537 (ASM_SPEC_COMMON): Likewise.
16538 (ASM_SPEC): Likewise.
16539 (INVALID_64BIT): Likewise.
16540 (LINK_OS_DEFAULT_SPEC): Likewise.
16541 (LINK_OS_SPEC32): Likewise.
16542 (LINK_OS_SPEC64): Likewise.
16543 (POWERPC_LINUX): Likewise.
16544 (PTRDIFF_TYPE): Likewise.
16545 (RESTORE_FP_PREFIX): Likewise.
16546 (RESTORE_FP_SUFFIX): Likewise.
16547 (SAVE_FP_PREFIX): Likewise.
16548 (SAVE_FP_SUFFIX): Likewise.
16549 (SIZE_TYPE): Likewise.
16550 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16551 (TARGET_64BIT): Likewise.
16552 (TARGET_64BIT): Likewise.
16553 (TARGET_AIX): Likewise.
16554 (WCHAR_TYPE_SIZE): Likewise.
16555 (WCHAR_TYPE): Undefine.
16556 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16557 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16558 (CPP_OS_RTEMS_SPEC): Delete.
16559 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16560 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16562 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16564 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16566 PR tree-optimization/81578
16567 * tree-parloops.c (build_new_reduction): Bail out if
16568 reduction_code isn't one of the standard OpenMP reductions.
16569 Move the details printing after that decision.
16571 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16573 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16574 related to reload_in_progress.
16575 (splat_input_operand): Likewise.
16576 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16578 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16580 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16581 (TARGET_INSTANTIATE_DECLS): Likewise.
16582 (legitimate_indexed_address_p): Delete reload_in_progress code.
16583 (rs6000_debug_legitimate_address_p): Likewise.
16584 (rs6000_eliminate_indexed_memrefs): Likewise.
16585 (rs6000_emit_le_vsx_store): Likewise.
16586 (rs6000_emit_move_si_sf_subreg): Likewise.
16587 (rs6000_emit_move): Likewise.
16588 (register_to_reg_type): Likewise.
16589 (rs6000_pre_atomic_barrier): Likewise.
16590 (rs6000_machopic_legitimize_pic_address): Likewise.
16591 (rs6000_allocate_stack_temp): Likewise.
16592 (rs6000_address_for_fpconvert): Likewise.
16593 (rs6000_address_for_altivec): Likewise.
16594 (rs6000_secondary_memory_needed_rtx): Delete function.
16595 (rs6000_check_sdmode): Likewise.
16596 (rs6000_alloc_sdmode_stack_slot): Likewise.
16597 (rs6000_instantiate_decls): Likewise.
16598 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16599 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16600 Delete reload_in_progress.
16601 (*vec_reload_and_plus_<mptrsize>): Likewise.
16602 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16603 (vsx_div_v2di): Likewise.
16604 (vsx_udiv_v2di): Likewise.
16606 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16608 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16609 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16610 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16611 (rs6000_debug_reg_global): Delete print of LRA status.
16612 (rs6000_option_override_internal): Delete dead LRA related code.
16613 (rs6000_lra_p): Delete function.
16614 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16616 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16618 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16619 * config/riscv/rtems.h: New file.
16621 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16622 Sudakshina Das <sudi.das@arm.com>
16624 * config/aarch64/aarch64.md
16625 (define_split for and<mode>3nr_compare): Move
16626 non aarch64_logical_operand to a register.
16627 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16628 register immediate operand to a register.
16629 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16631 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16633 PR middle-end/81564
16634 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16636 2017-07-27 Richard Biener <rguenther@suse.de>
16638 PR tree-optimization/81573
16639 PR tree-optimization/81494
16640 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16641 multi defuse cycle case.
16643 2017-07-27 Richard Biener <rguenther@suse.de>
16645 PR tree-optimization/81571
16646 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16649 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16651 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16652 earlier and only if MASK_FPU is set. Adjust formatting.
16654 2017-07-27 Martin Liska <mliska@suse.cz>
16656 * opt-functions.awk: Add validation of value of Init.
16657 * optc-gen.awk: Pass new argument.
16659 2017-07-27 Martin Liska <mliska@suse.cz>
16661 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16662 Fix wrong condition.
16664 2017-07-27 Martin Liska <mliska@suse.cz>
16666 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16667 BBs and edges seen by autoFDO.
16669 2017-07-27 Richard Biener <rguenther@suse.de>
16671 PR tree-optimization/81502
16672 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16673 with incompatible but same sized type.
16674 (execute_update_addresses_taken): Likewise.
16676 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16678 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16679 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16681 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16684 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16685 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16686 Change s_operand to memory_operand.
16688 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16690 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16691 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16692 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16693 Emit instructions rather than returning an expression. Handle TFmode
16694 and KFmode by casting to TImode.
16695 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16696 (rs6000_emit_le_vsx_store): Likewise.
16697 * config/rs6000/vsx.md (VSX_TI): New iterator.
16698 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16699 (*vsx_le_undo_permute_<mode>): Likewise.
16700 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16701 emit the split sequence.
16702 (*vsx_le_perm_store_<mode>): Likewise.
16704 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16706 PR tree-optimization/81555
16707 PR tree-optimization/81556
16708 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16709 if true, force CHANGED for the recursive invocation.
16710 (reassociate_bb): Remember original length of ops array, pass
16711 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16713 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16714 attributes for noipa attribute. For naked attribute use
16715 lookup_attribute first before lookup_attribute_spec.
16716 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16718 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16719 for functions with noipa attribute.
16720 (cgraph_externally_visible_p): Return true for functions with noipa
16722 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16723 for functions with noipa attribute.
16724 * doc/extend.texi: Document noipa function attribute.
16725 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16726 also for functions with noipa attribute.
16727 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16729 2017-07-26 Andrew Pinski <apinski@cavium.com>
16731 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16732 vec_unalign_load_cost and vec_unalign_store_cost.
16734 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16736 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16737 -mvsx-small-integer option.
16738 (ISA_3_0_MASKS_IEEE): Likewise.
16739 (OTHER_VSX_VECTOR_MASKS): Likewise.
16740 (POWERPC_MASKS): Likewise.
16741 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16742 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16743 code, only testing for DImode being allowed in non-VSX floating
16745 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16746 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16748 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16749 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16750 TARGET_P8_VECTOR test.
16751 (rs6000_secondary_reload_simple_move): Likewise.
16752 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16753 since TARGET_P9_VECTOR was already tested.
16754 (rs6000_opt_masks): Remove -mvsx-small-integer.
16755 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16756 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16758 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16759 test for TARGET_VEXTRACTUB was used, and that uses
16761 (p9 extract splitter): Likewise.
16762 (vsx_extract_<mode>_di_p9): Likewise.
16763 (vsx_extract_<mode>_store_p9): Likewise.
16764 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16765 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16766 the elimination of TARGET_VSX_SMALL_INTEGER.
16767 (vsx_extract_<mode>_p8): Likewise.
16768 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16769 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16770 (vsx_set_<mode>_p9): Likewise.
16771 (vsx_set_v4sf_p9): Likewise.
16772 (vsx_set_v4sf_p9_zero): Likewise.
16773 (vsx_insert_extract_v4sf_p9): Likewise.
16774 (vsx_insert_extract_v4sf_p9_2): Likewise.
16775 * config/rs6000/rs6000.md (sign extend splitter): Change
16776 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16777 (floatsi<mode>2_lfiwax_mem): Likewise.
16778 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16779 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16780 since a test for TARGET_P9_VECTOR was used.
16781 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16782 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16783 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16784 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16785 TARGET_P8_VECTOR test.
16786 (fix_trunc<mode>si2_stfiwx): Likewise.
16787 (fix_trunc<mode>si2_internal): Likewise.
16788 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16789 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16791 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16792 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16793 TARGET_P8_VECTOR test.
16794 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16795 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16796 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16798 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16799 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16800 since a test for TARGET_P9_VECTOR was used.
16801 (splitter for loading small constants): Likewise.
16803 2017-07-26 Andrew Pinski <apinski@cavium.com>
16805 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16808 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16811 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16812 (fp_valid_at): Likewise.
16814 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16816 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16817 (qdf24xx_addrcost_table): Likewise.
16818 (cortexa57_tunings): Update to use generic_branch_cost.
16819 (cortexa72_tunings): Likewise.
16820 (cortexa73_tunings): Likewise.
16821 (qdf24xx_tunings): Likewise.
16823 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16825 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16826 (thunderx2t99_branch_cost): Likewise.
16827 (cortexa35_tunings): Update to use generic_branch_cost.
16828 (cortexa53_tunings): Likewise.
16829 (cortexa57_tunings): Likewise.
16830 (cortexa72_tunings): Likewise.
16831 (cortexa73_tunings): Likewise.
16832 (thunderx2t99_tunings): Likewise.
16834 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16836 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16837 (sparc_option_override): Honour MASK_FSMULD.
16838 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16839 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16840 * config/sparc/sparc.opt (mfsmuld): New option.
16841 * doc/invoke.texi (mfsmuld): Document option.
16843 2017-07-26 Marek Polacek <polacek@redhat.com>
16845 PR middle-end/70992
16846 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16848 2017-07-26 Richard Biener <rguenther@suse.de>
16850 * gimple-match-head.c (do_valueize): Return OP if valueize
16852 (get_def): New helper to get at the def stmt of a SSA name
16853 if valueize allows.
16854 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16855 do_valueize to get at the def stmt.
16856 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16858 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16860 PR middle-end/46932
16861 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16863 2017-07-26 Martin Liska <mliska@suse.cz>
16866 * function.c (expand_function_start): Make expansion of
16867 nonlocal_goto_save_area after parm_birth_insn.
16869 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16871 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16872 from all CPU target flags enable members.
16874 2017-07-26 Richard Biener <rguenther@suse.de>
16876 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16877 (decision_tree::gen): Make 'type' const.
16878 (write_predicate): Likewise.
16880 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16882 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16883 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16884 (rs6000_option_override_internal): Likewise.
16885 (rs6000_expand_vector_set): Likewise.
16886 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16887 (TARGET_UPPER_REGS_SF): Likewise.
16888 (TARGET_UPPER_REGS_DI): Likewise.
16889 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16890 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16891 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16892 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16893 (Splitters for DI constants in Altivec registers): Likewise.
16894 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16895 (vsx_set_v4sf_p9): Likewise.
16896 (vsx_set_v4sf_p9_zero): Likewise.
16897 (vsx_insert_extract_v4sf_p9): Likewise.
16898 (vsx_insert_extract_v4sf_p9_2): Likewise.
16900 2017-07-25 Carl Love <cel@us.ibm.com>
16902 * doc/extend.texi: Update the built-in documentation file for the
16903 existing built-in functions
16904 vector signed char vec_cnttz (vector signed char);
16905 vector unsigned char vec_cnttz (vector unsigned char);
16906 vector signed short vec_cnttz (vector signed short);
16907 vector unsigned short vec_cnttz (vector unsigned short);
16908 vector signed int vec_cnttz (vector signed int);
16909 vector unsigned int vec_cnttz (vector unsigned int);
16910 vector signed long long vec_cnttz (vector signed long long);
16911 vector unsigned long long vec_cnttz (vector unsigned long long);
16913 2017-07-25 Andrew Pinski <apinski@cavium.com>
16915 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16916 accesses where the use is for the first operand of a BIT_INSERT.
16918 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16921 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16922 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16924 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16926 * config/i386/gstabs.h: Delete.
16927 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16929 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16931 * config/i386/i386.c (ix86_decompose_address): Do not check for
16932 register RTX when looking at index_reg or base_reg.
16933 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16935 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16937 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16938 to update EH info here.
16940 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16942 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16944 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16946 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16948 2017-07-25 Torsten Duwe <duwe@suse.de>
16950 * common.opt: Introduce -fpatchable-function-entry
16951 command line option, and its variables function_entry_patch_area_size
16952 and function_entry_patch_area_start.
16953 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16954 including a two-value parser.
16955 * target.def (print_patchable_function_entry): New target hook.
16956 * targhooks.h (default_print_patchable_function_entry): New function.
16957 * targhooks.c (default_print_patchable_function_entry): Likewise.
16958 * toplev.c (process_options): Switch off IPA-RA if
16959 patchable function entries are being generated.
16960 * varasm.c (assemble_start_function): Look at the
16961 patchable-function-entry command line switch and current
16962 function attributes and maybe generate NOP instructions by
16963 calling the print_patchable_function_entry hook.
16964 * doc/extend.texi: Document patchable_function_entry attribute.
16965 * doc/invoke.texi: Document -fpatchable_function_entry
16966 command line option.
16967 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16969 * doc/tm.texi: Re-generate.
16971 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16974 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16975 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16977 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16979 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16981 2017-07-25 Richard Biener <rguenther@suse.de>
16983 PR tree-optimization/81455
16984 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16985 not walk in cycles when looking for guards.
16987 2017-07-25 Richard Biener <rguenther@suse.de>
16989 PR tree-optimization/81529
16990 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16991 when optimizing backedge uses.
16993 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16995 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16997 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16998 to dl_section_ref. On AIX, append an expression to subtract
16999 the size of the section length to dl_section_ref.
17001 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17003 * configure.ac: If any of the config.* scripts fail, exit 1.
17004 * configure: Regenerate.
17006 2017-07-25 Richard Biener <rguenther@suse.de>
17008 PR middle-end/81546
17009 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17010 of immediate uses, be more verbose on errors.
17012 2017-07-25 Richard Biener <rguenther@suse.de>
17014 PR tree-optimization/81510
17015 * tree-vect-loop.c (vect_is_simple_reduction): When the
17016 reduction stmt is not inside the loop bail out.
17018 2017-07-25 Richard Biener <rguenther@suse.de>
17020 PR tree-optimization/81303
17021 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17022 profitability check against LOOP_VINFO_NITERSM1.
17024 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17026 * domwalk.c (cmp_bb_postorder): Simplify.
17027 (sort_bbs_postorder): New function. Use it...
17028 (dom_walker::walk): ...here to optimize common cases.
17030 2017-07-25 Martin Liska <mliska@suse.cz>
17033 * ipa-visibility.c (function_and_variable_visibility): Make the
17034 redirection just on target that supports aliasing.
17035 Fix GNU coding style.
17037 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17040 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17042 * config/arm/rtems.h: Likewise.
17043 * config/bfin/rtems.h: Likewise.
17044 * config/i386/rtemself.h: Likewise.
17045 * config/lm32/rtems.h: Likewise.
17046 * config/m32c/rtems.h: Likewise.
17047 * config/m68k/rtemself.h: Likewise.
17048 * config/microblaze/rtems.h: Likewise.
17049 * config/mips/rtems.h: Likewise.
17050 * config/moxie/rtems.h: Likewise.
17051 * config/nios2/rtems.h: Likewise.
17052 * config/powerpcspe/rtems.h: Likewise.
17053 * config/rs6000/rtems.h: Likewise.
17054 * config/rtems.h: Likewise.
17055 * config/sh/rtems.h: Likewise.
17056 * config/sh/rtemself.h: Likewise.
17057 * config/sparc/rtemself.h: Likewise.
17059 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17062 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17063 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17064 * tree-ssa-structalias.c (alias_get_name): Same.
17066 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17069 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17070 instructions if no du chain is found.
17072 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17074 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17076 2017-07-25 Richard Biener <rguenther@suse.de>
17078 PR middle-end/81505
17079 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17082 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17084 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17085 upper-regs options.
17086 (ISA_2_7_MASKS_SERVER): Likewise.
17087 (ISA_3_0_MASKS_IEEE): Likewise.
17088 (OTHER_P8_VECTOR_MASKS): Likewise.
17089 (OTHER_VSX_VECTOR_MASKS): Likewise.
17090 (POWERPC_MASKS): Likewise.
17091 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17092 duplicate list of options.
17093 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17094 explicit -mupper-regs options.
17095 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17096 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17097 alias for -mupper-regs-df.
17098 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17099 (rs6000_init_hard_regno_mode_ok): Likewise.
17100 (rs6000_option_override_internal): Likewise.
17101 (rs6000_opt_masks): Likewise.
17102 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17103 options in terms of whether -mvsx or -mpower8-vector was used.
17104 (TARGET_UPPER_REGS_DI): Likewise.
17105 (TARGET_UPPER_REGS_SF): Likewise.
17106 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17107 -mupper-regs-* options.
17109 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17111 * passes.c (emergency_dump_function): Print some empty lines and a
17112 header before the RTL dump.
17114 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17116 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17118 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17121 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17122 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17124 2017-07-24 Carl Love <cel@us.ibm.com>
17126 * config/rs6000/rs6000-c.c: Add support for built-in functions
17127 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17128 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17129 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17130 vec_extract_fp_from_shortl): Add defines for the two builtins.
17131 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17132 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17134 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17135 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17136 * doc/extend.texi: Update the built-in documentation file for the
17137 new built-in function.
17139 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17142 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17144 * doc/generic.texi: Likewise.
17145 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17146 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17148 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17150 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17151 (aarch64_mls_elt_merge<mode>): Likewise.
17153 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17155 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17156 having __cxa_atexit.
17158 2017-07-23 Michael Collison <michael.collison@arm.com>
17160 * config/arm/arm.c (arm_option_override): Deprecate
17161 use of -mstructure-size-boundary.
17162 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17163 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17165 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17168 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17169 Reduce cost estimate for direct moves.
17171 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17174 * config/i386/i386.c (ix86_option_override_internal): Disable
17175 BMI, BMI2 and TBM instructions for -m16.
17177 2017-07-21 Carl Love <cel@us.ibm.com>
17179 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17180 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17181 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17182 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17183 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17184 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17185 VMULOSW): New enum "unspec" values.
17186 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17187 altivec_vmulosw): New patterns.
17188 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17189 VMULOSW): Add definitions.
17191 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17193 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17194 (qdf24xx): Likewise.
17195 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17196 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17197 (AARCH64_FL_V8_1): Renumber.
17198 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17199 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17200 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17201 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17202 rdma to feature modifiers list.
17204 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17206 PR middle-end/56727
17207 * ipa-visibility (function_and_variable_visibility): Convert
17208 recursive PLT call to direct call if appropriate.
17210 2017-07-21 Andrew Pinski <apinski@cavium.com>
17212 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17213 operand 1 to see if the types precision matches.
17214 * fold-const.c (operand_equal_p): Likewise.
17216 2017-07-21 Richard Biener <rguenther@suse.de>
17218 PR tree-optimization/81303
17219 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17220 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17221 (vect_peeling_hash_get_lowest_cost): Adjust.
17222 (vect_enhance_data_refs_alignment): Likewise. Use
17223 vect_get_peeling_costs_all_drs to compute the penalty for no
17224 peeling to match up costs.
17226 2017-07-21 Richard Biener <rguenther@suse.de>
17228 PR tree-optimization/81500
17229 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17230 we didn't identify a reduction path.
17232 2017-07-21 Tom de Vries <tom@codesourcery.com>
17233 Cesar Philippidis <cesar@codesourcery.com>
17235 PR gcov-profile/81442
17236 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17239 2017-07-21 Tom de Vries <tom@codesourcery.com>
17242 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17244 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17245 nvptx_override_options_after_change.
17247 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17249 * dwarf2out.c (output_file_names): Avoid double testing for
17250 dwarf_version >= 5.
17252 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17254 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17256 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17258 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17260 (try_crossjump_to_edge): Do not punt on partitioned functions.
17262 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17264 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17265 Put all BBs reachable only via paths crossing cold region to cold
17267 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17269 2016-07-21 Richard Biener <rguenther@suse.de>
17271 PR tree-optimization/81303
17272 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17273 into account prologue and epilogue iterations when raising
17274 min_profitable_iters to sth at least covering one vector iteration.
17276 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17278 * config/arm/arm.c (arm_test_cpu_arch_dat):
17281 2017-07-20 Nathan Sidwell <nathan@acm.org>
17283 Remove TYPE_METHODS.
17284 * tree.h (TYPE_METHODS): Delete.
17285 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17286 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17287 (dbxout_type_methods): Scan TYPE_FIELDS.
17288 (dbxout_type): Don't check TYPE_METHODS here.
17289 * function.c (use_register_for_decl): Always ignore register for
17290 class types when not optimizing.
17291 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17292 * tree.c (free_lang_data_in_type): Stitch out member functions and
17293 templates from TYPE_FIELDS.
17294 (build_distinct_type_copy, verify_type_variant,
17295 verify_type): Member fns are on TYPE_FIELDS.
17296 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17297 * tree-pretty-print.c (dump_generic_node): Likewise.
17299 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17302 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17303 V2TImode and V4TImode.
17304 (ix86_expand_vector_extract): Likewise.
17305 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17306 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17307 (ssescalarmode): Handle V4TImode and V2TImode.
17308 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17309 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17310 (VEXTRACTI128_MODE): New mode iterator.
17311 (splitter for *vec_extractv?ti first element): New.
17312 (VEC_INIT_MODE): New mode iterator.
17313 (vec_init<mode>): Consolidate 3 expanders into one using
17314 VEC_INIT_MODE mode iterator.
17316 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17318 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17319 non_spilled_static_chain_regno_p.
17321 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17323 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17325 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17327 * bb-reorder.c (connect_traces): Allow copying of blocks within
17330 2017-07-20 Richard Biener <rguenther@suse.de>
17332 * gimple.h (gimple_phi_result): Add gphi * overload.
17333 (gimple_phi_result_ptr): Likewise.
17334 (gimple_phi_arg): Likewise. Adjust index assert to only
17335 allow actual argument accesses rather than all slots available
17337 (gimple_phi_arg_def): Add gphi * overload.
17338 * tree-phinodes.c (make_phi_node): Initialize only actual
17340 (resize_phi_node): Clear memory not covered by old node,
17341 do not initialize excess argument slots.
17342 (reserve_phi_args_for_new_edge): Initialize new argument slot
17345 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17347 PR tree-optimization/81388
17349 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17351 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17352 by removing computation of may_be_zero.
17354 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17355 Tom de Vries <tom@codesourcery.com>
17357 PR middle-end/81030
17358 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17359 when gimple level profile disagrees with what RTL expander did.
17361 2017-07-20 Richard Biener <rguenther@suse.de>
17363 PR tree-optimization/61171
17364 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17365 (vect_analyze_stmt): Add slp instance parameter.
17366 (vectorizable_reduction): Likewise.
17367 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17368 (vect_is_simple_reduction): Deal with chains not detected
17369 as SLP reduction chain, specifically not properly associated
17370 chains containing a mix of plus/minus.
17371 (get_reduction_op): Remove.
17372 (get_initial_defs_for_reduction): Simplify, pass in whether
17373 this is a reduction chain, pass in the SLP node for the PHIs.
17374 (vect_create_epilog_for_reduction): Get the SLP instance as
17376 (vectorizable_reduction): Get the SLP instance as arg.
17377 During analysis remember the SLP node with the PHIs in the
17378 instance. Simplify getting at the vectorized reduction PHIs.
17379 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17380 through SLP instance.
17381 (vect_slp_analyze_operations): Likewise.
17382 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17383 (vect_transform_stmt): Likewise.
17385 2017-07-20 Tom de Vries <tom@codesourcery.com>
17387 PR tree-optimization/81489
17388 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17389 read of phi arg location to before loop that modifies phi.
17391 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17393 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17396 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17398 PR middle-end/81331
17399 * except.c (execute): Fix ordering issue.
17401 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17403 PR rtl-optimization/81423
17404 * combine.c (make_compound_operation_int): Don't try to optimize
17405 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17407 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17409 PR rtl-optimization/81423
17410 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17411 with a constant that is -1 in the truncated to mode.
17413 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17415 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17416 (determine_unlikely_bbs): ... here.
17417 * predict.h (propagate_unlikely_bbs_forward): Declare.
17418 * cfgexpand.c (pass_expand::execute): Use it.
17419 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17421 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17422 propagate_unlikely_bbs_forward.
17424 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17426 PR middle-end/81331
17427 * except.c (maybe_add_nop_after_section_switch): New function.
17430 2017-07-19 Tom de Vries <tom@codesourcery.com>
17432 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17434 2017-07-19 Tom de Vries <tom@codesourcery.com>
17436 * gimple.h (gimple_phi_arg): Make assert more strict.
17438 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17440 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17441 * config/rs6000/mmintrin.h: New file.
17442 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17444 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17446 PR tree-optimization/81346
17447 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17449 2017-07-19 Tom de Vries <tom@codesourcery.com>
17451 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17453 2017-07-19 Tom de Vries <tom@codesourcery.com>
17455 * config/nvptx/nvptx-modes.def: Add V2DImode.
17456 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17457 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17458 (nvptx_output_mov_insn): Handle lack of mov.b128.
17459 (nvptx_print_operand): Handle 'H' and 'L' codes.
17460 (nvptx_vector_mode_supported): Allow V2DImode.
17461 (nvptx_preferred_simd_mode): New function.
17462 (nvptx_data_alignment): New function.
17463 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17464 nvptx_preferred_simd_mode.
17465 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17467 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17469 2017-07-19 Tom de Vries <tom@codesourcery.com>
17471 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17472 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17473 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17474 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17475 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17476 (mov<VECIM>_insn): New define_insn.
17477 (define_expand "mov<VECIM>): New define_expand.
17479 2017-07-19 Tom de Vries <tom@codesourcery.com>
17481 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17483 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17485 PR tree-optimization/81346
17486 * fold-const.h (fold_div_compare, range_check_type): Declare.
17487 * fold-const.c (range_check_type): New function.
17488 (build_range_check): Use range_check_type.
17489 (fold_div_compare): No longer static, rewritten into
17490 a match.pd helper function.
17491 (fold_comparison): Don't call fold_div_compare here.
17492 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17493 as helper function.
17495 2017-07-19 Nathan Sidwell <nathan@acm.org>
17497 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17498 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17499 * tree.c (find_decls_types_r, verify_type): Use
17500 TYPE_{MIN,MAX}_VALUE_RAW.
17501 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17502 (hash_tree): Likewise.
17503 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17505 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17508 2017-07-18 Tom de Vries <tom@codesourcery.com>
17510 PR middle-end/81464
17511 * omp-expand.c (expand_omp_for_static_chunk): Handle
17512 equal-argument loop exit phi.
17514 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17517 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17518 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17519 operand 2 predicate.
17520 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17521 operand 2 predicate.
17522 (ror,rol -> rorx splitters): Use const_int_operand as
17523 operand 2 predicate.
17525 2017-06-18 Richard Biener <rguenther@suse.de>
17527 PR tree-optimization/81410
17528 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17529 the gap in the ! slp_perm SLP case after each group.
17531 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17533 PR middle-end/81463
17534 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17537 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17539 PR middle-end/81462
17540 * predict.c (set_even_probabilities): Cleanup; do not affect
17541 probabilities that are already known.
17542 (combine_predictions_for_bb): Call even when count is set.
17544 2017-07-18 Nathan Sidwell <nathan@acm.org>
17546 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17549 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17552 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17553 optimization for loop niter analysis.
17555 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17558 * config/avr/avr.c (avr_optimize_casesi): Don't use
17559 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17561 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17563 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17564 body_cost_vec from _vect_peel_extended_info.
17565 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17566 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17569 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17571 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17573 2017-07-18 Richard Biener <rguenther@suse.de>
17575 PR tree-optimization/80620
17576 PR tree-optimization/81403
17577 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17578 info when re-using a VN table entry.
17580 2017-07-18 Richard Biener <rguenther@suse.de>
17582 PR tree-optimization/81418
17583 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17584 vectype_in. Verify that with lane-reducing reduction operations
17585 we have a single def-use cycle.
17587 2017-07-17 Carl Love <cel@us.ibm.com>
17589 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17591 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17592 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17593 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17594 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17595 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17596 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17597 VMULOSW): New enum "unspec" values.
17598 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17599 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17600 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17601 altivec_vmulosw): New patterns.
17602 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17603 VMULOSW): Add definitions.
17604 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17606 * config/alpha/alpha.c: Include predict.h.
17608 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17610 * tree-vrp.c (compare_assert_loc): Fix comparison function
17611 to return predictable results.
17613 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17615 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17617 (subdi3): Likewise.
17618 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17619 * doc/invoke.texi (mexpand-adddi): Update text.
17621 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17623 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17624 that also clobbers the CC register. The old expand code is moved
17626 (*arc_clzsi2): ... here.
17627 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17628 the CC register. The old expand code is moved to ...
17629 (arc_ctzsi2): ... here.
17631 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17633 * config/arc/arc.opt (mindexed-loads): Use initial value
17634 TARGET_INDEXED_LOADS_DEFAULT.
17635 (mauto-modify-reg): Use initial value
17636 TARGET_AUTO_MODIFY_REG_DEFAULT.
17637 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17638 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17639 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17640 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17642 2017-07-17 Martin Liska <mliska@suse.cz>
17645 * opts.c (finish_options): Do not allow -fgnu-tm
17646 w/ -fsanitize={kernel-,}address. Say sorry.
17648 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17651 * tree-loop-distribution.c (classify_partition): Only assert on
17652 numer of iterations.
17653 (merge_dep_scc_partitions): Delete prameter. Update function call.
17654 (distribute_loop): Remove code handling loop with unknown niters.
17655 (pass_loop_distribution::execute): Skip loop with unknown niters.
17657 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17660 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17661 function sort_partitions_by_post_order.
17663 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17665 PR tree-optimization/81374
17666 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17667 the max index of basic blocks, rather than number of basic blocks.
17669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17671 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17673 (arc_legitimate_pic_operand_p): Likewise.
17674 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17676 (arc_needs_pcl_p): Likewise.
17677 (arc_legitimate_pc_offset_p): Likewise.
17678 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17679 function is also used in constrains.md.
17680 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17681 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17682 PLUS. Only return true/false in known cases, otherwise assert.
17683 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17684 is already called in arc_legitimate_constant_p.
17685 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17687 (LEGITIMATE_PIC_OPERAND_P): Use
17688 arc_raw_symbolic_reference_mentioned_p function.
17689 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17694 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17695 Andrew Burgess <andrew.burgess@embecosm.com>
17697 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17698 (arc_return_address_register): New function.
17699 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17700 (arc_handle_fndecl_attribute): Add naked attribute.
17701 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17702 (TARGET_WARN_FUNC_RETURN): Likewise.
17703 (arc_allocate_stack_slots_for_args): New function.
17704 (arc_warn_func_return): Likewise.
17705 (machine_function): Change type fn_type.
17706 (arc_compute_function_type): Consider new naked function type,
17707 change function return type.
17708 (arc_must_save_register): Adapt to handle new
17709 arc_compute_function_type's return type.
17710 (arc_expand_prologue): Likewise.
17711 (arc_expand_epilogue): Likewise.
17712 (arc_return_address_regs): Delete.
17713 (arc_return_address_register): New function.
17714 (arc_epilogue_uses): Use above function.
17715 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17716 (arc_function_type): Change encoding, add naked type.
17717 (ARC_INTERRUPT_P): Change to handle the new encoding.
17718 (ARC_FAST_INTERRUPT_P): Likewise.
17719 (ARC_NORMAL_P): Define.
17720 (ARC_NAKED_P): Likewise.
17721 (arc_compute_function_type): Delete prototype.
17722 * config/arc/arc.md (in_ret_delay_slot): Use
17723 arc_return_address_register function.
17724 (simple_return): Likewise.
17725 (p_return_i): Likewise.
17727 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17729 PR tree-optimization/81428
17730 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17731 can't be built for those types.
17733 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17735 Remove stuff dead since r239246.
17737 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17738 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17739 (avr_inform_devices): Remove dead stuff.
17741 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17743 * config/arm/arm_neon.h: Fix softp typo.
17745 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17747 PR tree-optimization/81365
17748 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17749 aggregate moves onto bb predecessor edges, make sure there are no
17750 loads that could alias the lhs in between the start of bb and the
17753 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17756 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17757 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17758 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17760 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17762 PR tree-optimization/81396
17763 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17764 (init_symbolic_number): Initialize it to 1.
17765 (perform_symbolic_merge): Add n_ops from both operands into the new
17767 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17768 without base_addr as useless if they need more than one operation.
17769 (bswap_replace): Handle !bswap case for NULL base_addr.
17771 2017-07-17 Tom de Vries <tom@codesourcery.com>
17774 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17777 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17779 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17780 conditional builtin define __FIX_LEON3FT_B2BST.
17782 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17784 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17785 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17788 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17790 PR rtl-optimization/81424
17791 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17792 to remove potential trapping from operands if -fnon-call-exceptions.
17794 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17796 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17797 profile_proability for scalling.
17798 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17800 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17802 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17804 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17806 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17807 fixpoint arithmetics.
17809 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17811 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17812 fixpoint arithmetics.
17814 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17816 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17817 fixpoint arithmetics.
17819 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17821 * profile-count.h (profile_probability::from_reg_br_prob_note,
17822 profile_probability::to_reg_br_prob_note): New functions.
17823 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17824 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17825 * predict.c (probability_reliable_p): Update.
17826 (edge_probability_reliable_p): Update.
17827 (br_prob_note_reliable_p): Update.
17828 (invert_br_probabilities): Update.
17829 (add_reg_br_prob_note): New function.
17830 (combine_predictions_for_insn): Update.
17831 * asan.c (asan_clear_shadow): Update.
17832 * cfgbuild.c (compute_outgoing_frequencies): Update.
17833 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17834 (update_br_prob_note): Update.
17835 (rtl_verify_edges): Update.
17836 (purge_dead_edges): Update.
17837 (fixup_reorder_chain): Update.
17838 * emit-rtl.c (try_split): Update.
17839 * ifcvt.c (cond_exec_process_insns): Update.
17840 (cond_exec_process_if_block): Update.
17841 (dead_or_predicable): Update.
17842 * internal-fn.c (expand_addsub_overflow): Update.
17843 (expand_neg_overflow): Update.
17844 (expand_mul_overflow): Update.
17845 * loop-doloop.c (doloop_modify): Update.
17846 * loop-unroll.c (compare_and_jump_seq): Update.
17847 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17848 * predict.h: Update.
17849 * reorg.c (mostly_true_jump): Update.
17851 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17852 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17853 * config/arc/arc.c: (emit_unlikely_jump): Update.
17854 * config/arm/arm.c: (emit_unlikely_jump): Update.
17855 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17856 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17857 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17858 (ix86_print_operand): Update.
17859 (ix86_split_fp_branch): Update.
17860 (predict_jump): Update.
17861 * config/ia64/ia64.c (ia64_print_operand): Update.
17862 * config/mmix/mmix.c (mmix_print_operand): Update.
17863 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17864 (rs6000_expand_split_stack_prologue): Update.
17865 * config/rs6000/rs6000.c: Update.
17866 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17867 (s390_expand_vec_movstr): Update.
17868 (s390_expand_cs_tdsi): Update.
17869 (s390_expand_split_stack_prologue): Update.
17870 * config/sh/sh.c (sh_print_operand): Update.
17871 (expand_cbranchsi4): Update.
17872 (expand_cbranchdi4): Update.
17873 * config/sparc/sparc.c (output_v9branch): Update.
17874 * config/spu/spu.c (get_branch_target): Update.
17875 (ea_load_store_inline): Update.
17876 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17877 * config/tilepro/tilepro.c: Update.
17879 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17881 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17882 (gimplify_save_expr): Likewise.
17884 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17886 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17888 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17890 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17892 * timevar.def (TV_IPA_FNSUMMARY): Define.
17894 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17896 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17897 to back store errata sensitive sequence from being generated.
17898 (sqrtdf2_fix): Likewise.
17900 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17902 * tree-ssa-threadupdate.c (compute_path_counts,
17903 update_joiner_offpath_counts): Use profile_probability.
17905 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17908 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17910 * config/arm/arm-c.c (arm_cpu_builtins): Define
17911 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17913 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17915 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17916 array entries to represent __ieee128 versions of the
17917 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17918 scalar_extract_sig, and scalar_insert_exp built-in functions.
17919 (altivec_resolve_overloaded_builtin): Add special case handling
17920 for the __builtin_scalar_insert_exp function, as represented by
17921 the P9V_BUILTIN_VEC_VSIEDP constant.
17922 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17923 exponent support for __ieee128 argument.
17924 (VSESQP): Add scalar extract signature support for __ieee128
17926 (VSTDCNQP): Add scalar test negative support for __ieee128
17928 (VSIEQP): Add scalar insert exponent support for __int128 argument
17929 with __ieee128 result.
17930 (VSIEQPF): Add scalar insert exponent support for __ieee128
17931 argument with __ieee128 result.
17932 (VSTDCQP): Add scalar test data class support for __ieee128
17934 (VSTDCNQP): Add overload support for scalar test negative with
17935 __ieee128 argument.
17936 (VSTDCQP): Add overload support for scalar test data class
17937 __ieee128 argument.
17938 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17939 UNSPEC_VSX_SXSIGDP.
17940 (UNSPEC_VSX_SIEXPQP): New constant.
17941 (xsxexpqp): New insn for VSX scalar extract exponent quad
17943 (xsxsigqp): New insn for VSX scalar extract significand quad
17945 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17946 precision with floating point argument.
17947 (xststdcqp): New expand for VSX scalar test data class quad
17949 (xststdcnegqp): New expand for VSX scalar test negative quad
17951 (xststdcqp): New insn to match expansions for VSX scalar test data
17952 class quad precision and VSX scalar test negative quad precision.
17953 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17954 special case operand checking to enforce that second operand of
17955 VSX scalar test data class with quad precision argument is a 7-bit
17957 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17958 prototypes and descriptions of __ieee128 versions of
17959 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17960 scalar_test_data_class, and scalar_test_neg built-in functions.
17962 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17964 PR tree-optimization/81162
17965 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17966 replace a negate with an add.
17968 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17970 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17972 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17974 * config/arm/arm-c.c (arm_cpu_builtins): Define
17975 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17977 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17979 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17980 (armv8-r): Set ARM Cortex-R52 as default CPU.
17981 * config/arm/arm-tables.opt: Regenerate.
17982 * config/arm/arm-tune.md: Regenerate.
17983 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17985 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17986 extension for -mcpu=cortex-r52.
17988 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17990 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17991 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17992 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17993 (fp-armv8): Define it as FP_ARMv8 only.
17994 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17995 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17997 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17998 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17999 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18000 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18001 than TARGET_FPU_ARMV8.
18002 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18003 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18004 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18006 * config/arm/neon.md (neon_vrint): Likewise.
18007 (neon_vcvt): Likewise.
18008 (neon_<fmaxmin_op><mode>): Likewise.
18009 (<fmaxmin><mode>3): Likewise.
18010 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18011 * config/arm/predicates.md (arm_cond_move_operator): Check against
18012 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18014 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18016 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18017 to top of function.
18019 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18021 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18022 loop in comment with memset.
18024 2017-07-14 Martin Liska <mliska@suse.cz>
18026 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18027 * dwarf2out.c (is_java): Remove the function.
18028 (output_pubname): Remove usage of the function.
18029 (lower_bound_default): Remove usage of DW_LANG_Java.
18030 (gen_compile_unit_die): Likewise.
18031 * gcc.c: Remove compiler defaults for .java and .zip files.
18032 * gimple-expr.c (remove_suffix): Change as there's no longer
18033 extension than 4-letter one.
18034 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18035 (gimplify_save_expr): Likewise.
18036 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18037 as it's possible even for other languages than Java.
18038 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18039 * lto-opts.c (lto_write_options): Remove reference to Java.
18040 * opts.c (strip_off_ending): Update file extension handling.
18041 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18042 * tree-eh.c (lower_resx): Likewise.
18043 * tree.c (free_lang_data_in_type): Remove dead code.
18044 (find_decls_types_r): Likewise.
18045 (build_common_builtin_nodes): Remove Java from a comment.
18046 (verify_type): Remove dead code.
18047 * varasm.c (assemble_external): Remove Java from a comment.
18049 2017-07-14 Martin Liska <mliska@suse.cz>
18051 * opts.c (finish_options): Add quotes.
18052 (common_handle_option): Likewise.
18054 2017-07-14 Martin Liska <mliska@suse.cz>
18056 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18057 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18058 Remove N_SO_PASCAL.
18059 * dwarf2out.c (lower_bound_default): Do not handle
18061 (gen_compile_unit_die): Likewise.
18062 * gcc.c: Remove default extension binding for GNU Pascal.
18063 * stmt.c: Remove Pascal language from a comment.
18064 * xcoffout.c: Likewise.
18066 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18069 * diagnostic-show-locus.c (fixit_cmp): New function.
18070 (layout::layout): Sort m_fixit_hints.
18071 (column_range::column_range): Assert that the values are valid.
18072 (struct char_span): New struct.
18073 (correction::overwrite): New method.
18074 (struct source_line): New struct.
18075 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18076 calls in terms of classes source_line and char_span, and
18077 correction::overwrite.
18078 (selftest::test_overlapped_fixit_printing_2): New function.
18079 (selftest::diagnostic_show_locus_c_tests): Call it.
18081 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18083 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18084 early if there is no lhs.
18086 2017-07-13 Martin Liska <mliska@suse.cz>
18088 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18089 (gen_reference_type_die): Likewise.
18090 * stor-layout.c: Remove Pascal-related comment.
18092 2017-07-13 Martin Liska <mliska@suse.cz>
18094 * opts.c (finish_options): Add quotes to error messages.
18095 (parse_sanitizer_options): Likewise.
18097 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18099 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18101 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18103 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18105 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18107 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18109 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18111 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18114 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18115 provides the hardware capability bits, define the macro
18116 __BUILTIN_CPU_SUPPORTS__.
18117 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18118 if GLIBC does not provide the hardware capability bits. Add a
18119 gcc_unreachable call if the built-in cpu function is neither
18120 __builtin_cpu_is nor __builtin_cpu_supports.
18121 (rs6000_get_function_versions_dispatcher): Change the warning
18122 that an old GLIBC is used which does not export the capability
18123 bits to be an error.
18124 * doc/extend.texi (target_clones attribute): Document the
18125 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18126 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18127 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18128 the macros defined by GCC if the newer GLIBC is available.
18130 2017-07-12 Jeff Law <law@redhat.com>
18132 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18133 remaining includes slightly.
18134 * config/riscv/riscv-builtins.c: Include profile-count.h.
18136 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18139 * config/avr/avr.c (avr_set_current_function): In diagnostic
18140 messages: Quote keywords and (parts of) identifiers.
18141 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18144 2017-07-12 Carl Love <cel@us.ibm.com>
18146 * config/rs6000/rs6000-c.c: Add support for built-in functions
18147 vector bool char vec_revb (vector bool char);
18148 vector bool short vec_revb (vector short char);
18149 vector bool int vec_revb (vector bool int);
18150 vector bool long long vec_revb (vector bool long long);
18151 * doc/extend.texi: Update the built-in documentation file for the
18152 new built-in functions.
18154 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18156 * config/s390/s390.md: Remove movcc splitter.
18158 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18160 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18161 load/store on condition.
18163 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18166 * config/avr/avr.c (avr_encode_section_info)
18167 [progmem && !TREE_READONLY]: Error if progmem object needs
18170 2017-07-11 Michael Collison <michael.collison@arm.com>
18172 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18175 2017-07-11 Carl Love <cel@us.ibm.com>
18177 * config/rs6000/rs6000-c.c: Add support for builtins
18178 vector unsigned int vec_parity_lsbb (vector signed int);
18179 vector unsigned int vec_parity_lsbb (vector unsigned int);
18180 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18181 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18182 vector unsigned long long vec_parity_lsbb (vector signed long long);
18183 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18184 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18185 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18186 * doc/extend.texi: Update the built-in documentation file for the
18187 new built-in functions.
18189 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18191 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18192 (layout::m_primary_loc): New field.
18193 (layout::layout): Initialize new field. Move location filtering
18194 logic from here to...
18195 (layout::maybe_add_location_range): ...this new method. Add
18196 support for filtering to just the lines already specified by other
18198 (layout::will_show_line_p): New method.
18199 (gcc_rich_location::add_location_if_nearby): New method.
18200 (selftest::test_add_location_if_nearby): New test function.
18201 (selftest::diagnostic_show_locus_c_tests): Call it.
18202 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18205 2017-07-11 Tom de Vries <tom@codesourcery.com>
18207 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18208 (bb_first_real_insn): New function.
18209 (nvptx_single): Add extra initialization of broadcasted condition
18212 2017-07-11 Nathan Sidwell <nathan@acm.org>
18214 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18216 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18218 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18219 Binutils doc as TEXI will mess them up.
18220 * doc/invoke.texi (AVR Options): Same here.
18222 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18224 * config/sparc/sparc.opt (mfix-ut700): New option.
18225 (mfix-gr712rc): Likewise.
18226 (sparc_fix_b2bst): New variable.
18227 * doc/invoke.texi (SPARC options): Document them.
18228 (ARM options): Fix warnings.
18229 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18230 instructions to prevent sequences that can trigger the store-store
18231 errata for certain LEON3FT processors.
18232 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18233 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18234 * config/sparc/sparc.md (fix_b2bst): New attribute.
18235 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18237 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18240 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18242 (*rsqrtsf2_sse): Ditto.
18244 (div<mode>3): Macroize insn from divdf3 and divsf3
18245 using MODEF mode iterator.
18247 2017-07-10 Martin Sebor <msebor@redhat.com>
18249 PR tree-optimization/80397
18250 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18251 instead of testing for equality to INTEGER_TYPE.
18253 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18255 * config.gcc: Remove uclibc from arc target spec.
18257 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18259 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18261 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18264 * lto-wrapper.c (remove_option): New function.
18265 (merge_and_complain): Merge PIC/PIE options more realistically.
18267 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18269 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18273 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18274 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18275 * config.in: Regenerate.
18276 * configure: Regenerate.
18277 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18278 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18279 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18280 (TARGET_GASISR_PROLOGUES): ...target mask.
18281 * common/config/avr/avr-common.c
18282 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18283 Set -mgas-isr-prologues.
18284 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18285 INSERT_PASS_BEFORE for it.
18286 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18287 * config/avr/avr.c (avr_option_override)
18288 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18289 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18290 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18291 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18292 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18294 (make_avr_pass_pre_proep): New function.
18295 (emit_push_sfr) <treg>: Add argument to function and use it
18296 instead of TMP_REG.
18297 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18298 and set machine->gasisr.yes.
18299 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18300 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18301 __gcc_isr.n_pushed to .L__stack_usage.
18302 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18303 (avr_asm_final_postscan_insn): ...this new static function.
18304 * config/avr/avr.h (machine_function)
18305 <is_no_gccisr, use_L__stack_usage>: New fields.
18306 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18307 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18308 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18309 (gasisr, *gasisr): New expander and insn.
18310 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18311 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18312 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18314 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18316 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18319 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18321 Move jump-tables out of .text again.
18324 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18325 (ASM_OUTPUT_ADDR_VEC): New function.
18326 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18327 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18328 INSN_ADDRESSes as asm comment.
18329 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18330 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18331 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18332 * config/avr/avr.md (*tablejump): Adjust comment.
18333 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18334 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18336 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18337 (avr_output_addr_vec): New proto.
18338 (avr_log_t) <insn_addresses>: New field.
18340 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18343 * config/i386/i386.c (ix86_function_arg_advance): Set
18344 outgoing_args_on_stack to true if there are outgoing arguments
18346 (ix86_function_arg): Likewise.
18347 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18348 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18349 * config/i386/i386.h (machine_function): Add
18350 outgoing_args_on_stack.
18352 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18354 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18355 supporting pthreds.
18356 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18358 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18360 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18361 (REAL_H): Remove $(MACHMODE_H).
18362 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18364 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18365 $(MACHMODE_H) and double-int.h.
18366 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18368 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18371 2017-07-07 Andrew Pinski <apinski@cavium.com>
18373 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18374 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18376 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18378 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18379 Add warning if GCC was not configured to link against a GLIBC that
18380 exports the hardware capability bits.
18381 (make_resolver_func): Make resolver function private and not a
18382 COMDAT function. Create the name with clone_function_name instead
18383 of make_unique_name.
18386 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18387 correct operand in doing the split.
18389 2017-07-07 Carl Love <cel@us.ibm.com>
18391 * config/rs6000/rs6000-c: Add support for built-in function
18392 vector unsigned short vec_pack_to_short_fp32 (vector float,
18394 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18395 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18396 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18397 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18398 (convert_4f32_8i16): Add define_expand.
18399 * doc/extend.texi: Update the built-in documentation file for the
18400 new built-in function.
18402 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18404 * config/sparc/m8.md: New file.
18405 * config/sparc/sparc.md: Include m8.md.
18407 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18409 * config/sparc/sparc.opt: New option -mvis4b.
18410 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18411 (sparc_option_override): Handle VIS4B.
18412 (enum sparc_builtins): Define
18413 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18414 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18415 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18416 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18417 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18418 (check_constant_argument): New function.
18419 (sparc_vis_init_builtins): Define builtins
18420 __builtin_vis_dictunpack{8,16,32},
18421 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18422 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18423 __builtin_vis_fpcmpde{8,16,32}shl and
18424 __builtin_vis_fpcmpur{8,16,32}shl.
18425 (sparc_expand_builtin): Check that the constant operands to
18426 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18427 constant and in range.
18428 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18430 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18431 (SPARC_IMM5_P): Likewise.
18432 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18433 (enabled): Handle vis4b.
18434 (UNSPEC_DICTUNPACK): New unspec.
18435 (UNSPEC_FPCMPSHL): Likewise.
18436 (UNSPEC_FPUCMPSHL): Likewise.
18437 (UNSPEC_FPCMPDESHL): Likewise.
18438 (UNSPEC_FPCMPURSHL): Likewise.
18439 (cpu_feature): New CPU feature `vis4b'.
18440 (dictunpack{8,16,32}): New insns.
18441 (FPCSMODE): New mode iterator.
18442 (fpcscond): New code iterator.
18443 (fpcsucond): Likewise.
18444 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18445 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18446 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18447 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18448 * config/sparc/constraints.md: Define constraints `q' for unsigned
18449 2-bit integer constants and `t' for unsigned 5-bit integer
18451 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18453 (imm5_operand_dictunpack16): Likewise.
18454 (imm5_operand_dictunpack32): Likewise.
18455 (imm2_operand): Likewise.
18456 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18457 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18458 ditunpack* and fpcmp*shl builtins.
18460 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18462 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18463 * config.in: Add HAVE_AS_SPARC6 define.
18464 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18466 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18468 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18469 (CPP_CPU_SPEC): Handle m8.
18470 (ASM_CPU_SPEC): Likewise.
18471 * config/sparc/sparc-opts.h (enum processor_type): Add
18473 * config/sparc/sparc.c (m8_costs): New struct.
18474 (sparc_option_override): Handle TARGET_CPU_m8.
18475 (sparc32_initialize_trampoline): Likewise.
18476 (sparc64_initialize_trampoline): Likewise.
18477 (sparc_issue_rate): Likewise.
18478 (sparc_register_move_cost): Likewise.
18479 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18480 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18481 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18482 (CPP_CPU_SPEC): Handle M8.
18483 (ASM_CPU_SPEC): Likewise.
18484 (AS_M8_FLAG): Define.
18485 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18486 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18487 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18489 * configure: Regenerate.
18490 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18493 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18495 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18497 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18498 ("*movdi_insn_sp32"): Do not set v3pipe.
18499 ("*movsi_insn"): Likewise.
18500 ("*movdi_insn_sp64"): Likewise.
18501 ("*movsf_insn"): Likewise.
18502 ("*movdf_insn_sp32"): Likewise.
18503 ("*movdf_insn_sp64"): Likewise.
18504 ("*zero_extendsidi2_insn_sp64"): Likewise.
18505 ("*sign_extendsidi2_insn"): Likewise.
18506 ("*mov<VM32:mode>_insn"): Likewise.
18507 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18508 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18509 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18510 ("<vlop:code><VL:mode>3"): Likewise.
18511 ("*not_<vlop:code><VL:mode>3"): Likewise.
18512 ("*nand<VL:mode>_vis"): Likewise.
18513 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18514 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18515 ("one_cmpl<VL:mode>2"): Likewise.
18516 ("faligndata<VM64:mode>_vis"): Likewise.
18517 ("alignaddrsi_vis"): Likewise.
18518 ("alignaddrdi_vis"): Likweise.
18519 ("alignaddrlsi_vis"): Likewise.
18520 ("alignaddrldi_vis"): Likewise.
18521 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18522 ("bmaskdi_vis"): Likewise.
18523 ("bmasksi_vis"): Likewise.
18524 ("bshuffle<VM64:mode>_vis"): Likewise.
18525 ("cmask8<P:mode>_vis"): Likewise.
18526 ("cmask16<P:mode>_vis"): Likewise.
18527 ("cmask32<P:mode>_vis"): Likewise.
18528 ("pdistn<P:mode>_vis"): Likewise.
18529 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18531 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18533 * config/sparc/sparc.md ("subtype"): New insn attribute.
18534 ("*wrgsr_sp64"): Set insn subtype.
18535 ("*rdgsr_sp64"): Likewise.
18536 ("alignaddrsi_vis"): Likewise.
18537 ("alignaddrdi_vis"): Likewise.
18538 ("alignaddrlsi_vis"): Likewise.
18539 ("alignaddrldi_vis"): Likewise.
18540 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18541 ("fexpand_vis"): Likewise.
18542 ("fpmerge_vis"): Likewise.
18543 ("faligndata<VM64:mode>_vis"): Likewise.
18544 ("bshuffle<VM64:mode>_vis"): Likewise.
18545 ("cmask8<P:mode>_vis"): Likewise.
18546 ("cmask16<P:mode>_vis"): Likewise.
18547 ("cmask32<P:mode>_vis"): Likewise.
18548 ("fchksm16_vis"): Likewise.
18549 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18550 ("fmean16_vis"): Likewise.
18551 ("fp<plusminus_insn>64_vis"): Likewise.
18552 ("<plusminus_insn>v8qi3"): Likewise.
18553 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18554 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18555 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18556 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18557 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18558 ("*movqi_insn"): Likewise.
18559 ("*movhi_insn"): Likewise.
18560 ("*movsi_insn"): Likewise.
18561 ("movsi_pic_gotdata_op"): Likewise.
18562 ("*movdi_insn_sp32"): Likewise.
18563 ("*movdi_insn_sp64"): Likewise.
18564 ("movdi_pic_gotdata_op"): Likewise.
18565 ("*movsf_insn"): Likewise.
18566 ("*movdf_insn_sp32"): Likewise.
18567 ("*movdf_insn_sp64"): Likewise.
18568 ("*zero_extendhisi2_insn"): Likewise.
18569 ("*zero_extendqihi2_insn"): Likewise.
18570 ("*zero_extendqisi2_insn"): Likewise.
18571 ("*zero_extendqidi2_insn"): Likewise.
18572 ("*zero_extendhidi2_insn"): Likewise.
18573 ("*zero_extendsidi2_insn_sp64"): Likewise.
18574 ("ldfsr"): Likewise.
18575 ("prefetch_64"): Likewise.
18576 ("prefetch_32"): Likewise.
18577 ("tie_ld32"): Likewise.
18578 ("tie_ld64"): Likewise.
18579 ("*tldo_ldub_sp32"): Likewise.
18580 ("*tldo_ldub1_sp32"): Likewise.
18581 ("*tldo_ldub2_sp32"): Likewise.
18582 ("*tldo_ldub_sp64"): Likewise.
18583 ("*tldo_ldub1_sp64"): Likewise.
18584 ("*tldo_ldub2_sp64"): Likewise.
18585 ("*tldo_ldub3_sp64"): Likewise.
18586 ("*tldo_lduh_sp32"): Likewise.
18587 ("*tldo_lduh1_sp32"): Likewise.
18588 ("*tldo_lduh_sp64"): Likewise.
18589 ("*tldo_lduh1_sp64"): Likewise.
18590 ("*tldo_lduh2_sp64"): Likewise.
18591 ("*tldo_lduw_sp32"): Likewise.
18592 ("*tldo_lduw_sp64"): Likewise.
18593 ("*tldo_lduw1_sp64"): Likewise.
18594 ("*tldo_ldx_sp64"): Likewise.
18595 ("*mov<VM32:mode>_insn"): Likewise.
18596 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18597 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18599 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18601 * config/sparc/sparc.md ("type"): New insn type viscmp.
18602 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18604 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18605 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18606 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18607 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18609 ("n7_vis_logical_11cycle"): Likewise.
18610 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18611 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18612 * config/sparc/niagara.md ("niag_vis"): Likewise.
18613 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18614 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18616 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18618 * config/sparc/sparc.md: New instruction type `bmask'.
18619 (bmaskdi_vis): Use the `bmask' type.
18620 (bmasksi_vis): Likewise.
18621 * config/sparc/ultra3.md (us3_array): Likewise.
18622 * config/sparc/niagara7.md (n7_array): Likewise.
18623 * config/sparc/niagara4.md (n4_array): Likewise.
18624 * config/sparc/niagara2.md (niag2_vis): Likewise.
18625 (niag3_vis): Likewise.
18626 * config/sparc/niagara.md (niag_vis): Likewise.
18628 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18630 * ipa-comdats.c: Remove optimize check from gate.
18631 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18632 for functions not optimized.
18633 (ipa_fn_summary_read): Skip optimize check.
18634 (ipa_fn_summary_write): Likewise.
18635 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18637 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18639 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18640 (check_callers): Give up on caller which is not optimized.
18641 (inline_small_functions): Likewise.
18642 (ipa_inline): Do not give up when not optimizing.
18643 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18644 away unoptimizes cdtors.
18645 (whole_program_function_and_variable_visibility): Do
18646 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18647 * ipa.c (process_references): Do not check optimize.
18648 (symbol_table::remove_unreachable_nodes): Update optimize check.
18649 (set_writeonly_bit): Update optimize check.
18650 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18651 (pass_ipa_single_use::gate): Remove.
18653 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18655 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18656 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18657 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18658 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18659 permute_load, permute_store, adjust_extract, adjust_splat,
18660 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18661 replace_swap_with_copy, dump_swap_insn_table,
18662 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18663 recombine_lvx_pattern, recombine_stvx_pattern,
18664 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18665 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18666 to file rs6000-p8swap.c.
18667 * config/rs6000/rs6000-p8swap.c: New file.
18668 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18669 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18670 and rs6000*-*-* targets.
18672 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18674 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18676 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18678 * lto-wrapper.c (merge_and_complain): Do not merge
18679 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18680 fsigned_zeros, ftrapping_math, fwrapv.
18681 (append_compiler_options): Do not track these options.
18682 (append_linker_options): Likewie
18684 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18686 * cgraphunit.c (cgraph_node::finalize_function): When
18687 !flag_toplevel_reorde set no_reorder flag.
18688 (varpool_node::finalize_decl): Likewise.
18689 (symbol_table::compile): Drop no toplevel reorder path.
18691 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18693 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18694 edges; zero probability is not better than uninitialized.
18696 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18698 * asan.h (asan_sanitize_allocas_p): Declare.
18699 * asan.c (asan_sanitize_allocas_p): New function.
18700 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18701 (handle_builtin_alloca): Likewise.
18702 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18703 if !asan_sanitize_allocas_p.
18704 * params.def (asan-instrument-allocas): Add new option.
18705 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18706 * opts.c (common_handle_option): Disable allocas sanitization for
18709 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18711 * asan.c: Include gimple-fold.h.
18712 (get_last_alloca_addr): New function.
18713 (handle_builtin_stackrestore): Likewise.
18714 (handle_builtin_alloca): Likewise.
18715 (asan_emit_allocas_unpoison): Likewise.
18716 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18717 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18718 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18719 (instrument_builtin_call): Pass gimple iterator to
18720 get_mem_refs_of_builtin_call.
18721 (last_alloca_addr): New global.
18722 * asan.h (asan_emit_allocas_unpoison): Declare.
18723 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18724 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18725 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18726 if function calls alloca.
18727 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18728 * gimple-fold.h (replace_call_with_value): Declare.
18729 * internal-fn.c: Include asan.h.
18730 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18731 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18733 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18735 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18736 (C_SELFTEST_FLAGS): New.
18737 (CPP_SELFTEST_FLAGS): New.
18738 (SELFTEST_DEPS): New, from deps of s-selftest.
18739 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18740 (CPP_SELFTEST_DEPS): New.
18741 (selftest): Add dependency on s-selftest-c++.
18742 (s-selftest): Rename to...
18743 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18744 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18745 than SELFTEST_FLAGS.
18746 (selftest-gdb): Rename to...
18747 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18749 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18750 (selftest-valgrind): Rename to...
18751 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18753 (selftest-valgrind): Reintroduce as an alias for
18754 selftest-c-valgrind.
18755 (s-selftest-c++): New.
18756 (selftest-c++-gdb): New.
18757 (selftest-c++-valgrind): New.
18759 2017-07-06 Olivier Hainque <hainque@adacore.com>
18761 * gcc.c (process_command): When deciding if undefined variables
18762 should be ignored when processing specs, accept "gcc -v" as well.
18764 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18766 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18767 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18769 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18771 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18772 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18773 * config/arm/arm-tables.opt: Regenerate.
18774 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18776 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18778 2017-07-06 Carl Love <cel@us.ibm.com>
18780 * ChangeLog: Clean up from mid air collision
18782 2017-07-06 Carl Love <cel@us.ibm.com>
18784 * config/rs6000/rs6000-c.c: Add support for built-in functions
18785 vector signed int vec_subc (vector signed int, vector signed int);
18786 vector signed __int128 vec_subc (vector signed __int128,
18787 vector signed __int128);
18788 vector unsigned __int128 vec_subc (vector unsigned __int128,
18789 vector unsigned __int128);
18790 vector signed int vec_sube (vector signed int, vector signed int,
18791 vector signed int);
18792 vector unsigned int vec_sube (vector unsigned int,
18793 vector unsigned int,
18794 vector unsigned int);
18795 vector signed __int128 vec_sube (vector signed __int128,
18796 vector signed __int128,
18797 vector signed__int128);
18798 vector unsigned __int128 vec_sube (vector unsigned __int128,
18799 vector unsigned __int128,
18800 vector unsigned __int128);
18801 vector signed int vec_subec (vector signed int, vector signed int,
18802 vector signed int);
18803 vector unsigned int vec_subec (vector unsigned int,
18804 vector unsigned int,
18805 vector unsigned int);
18806 vector signed __int128 vec_subec (vector signed __int128,
18807 vector signed __int128,
18808 vector signed__int128);
18809 vector unsigned __int128 vec_subec (vector unsigned __int128,
18810 vector unsigned __int128,
18811 vector unsigned __int128);
18812 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18813 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18814 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18815 BU_ALTIVEC_OVERLOAD_X definitions.
18816 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18817 * doc/extend.texi: Update the built-in documentation file for the new
18818 built-in functions.
18820 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18823 * diagnostic-show-locus.c (layout::layout): Use start and finish
18824 spelling location for the start and finish of each range.
18825 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18826 Add unused aspect param.
18827 * input.c (expand_location_1): Add "aspect" param, and use it
18828 to access the correct part of the location.
18829 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18831 (expand_location_to_spelling_point): Likewise.
18832 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18833 param, and pass it to expand_location_1.
18835 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18837 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18838 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18839 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18840 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18841 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18842 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18843 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18844 _mm_maskz_getmant_ss): New intrinsics.
18845 (__builtin_ia32_getexpss128_mask): Changed to ...
18846 __builtin_ia32_getexpss128_round ... this.
18847 (__builtin_ia32_getexpsd128_mask): Changed to ...
18848 __builtin_ia32_getexpsd128_round ... this.
18849 * config/i386/i386-builtin-types.def
18850 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18851 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18852 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18853 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18854 __builtin_ia32_getmantss_mask_round): New builtins.
18855 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18856 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18857 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18858 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18859 * config/i386/sse.md
18860 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18861 avx512f_sgetexp<mode><mask_scalar_name>
18862 <round_saeonly_scalar_name> ... this.
18863 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18864 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18865 vgetexp<ssescalarmodesuffix>
18866 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18867 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18868 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18869 avx512f_vgetmant<mode><mask_scalar_name>
18870 <round_saeonly_scalar_name> ... this.
18871 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18872 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18873 vgetmant<ssescalarmodesuffix>
18874 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18875 %0<mask_scalar_operand4>, %1, %2
18876 <round_saeonly_scalar_mask_op4>, %3} ... this.
18877 * config/i386/subst.md (mask_scalar_operand4,
18878 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18879 round_saeonly_scalar_nimm_predicate): New subst attributes.
18881 2017-07-06 Julia Koval <julia.koval@intel.com>
18883 * config/i386/i386.c (ix86_erase_embedded_rounding):
18884 Remove code for old rounding pattern.
18886 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18888 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18890 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18892 * doc/sourcebuild.texi (Test Directives, Variants of
18893 dg-require-support): Add documentation for dg-require-stack-check.
18895 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18897 * config/i386/subst.md (mask_scalar, round_scalar,
18898 round_saeonly_scalar): New meta-templates.
18899 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18900 round_scalar_mask_operand3, round_scalar_mask_op3,
18901 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18902 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18903 round_saeonly_scalar_constraint,
18904 round_saeonly_scalar_prefix): New subst attribute.
18905 * config/i386/sse.md
18906 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18907 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18908 <round_scalar_name> ... this.
18909 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18910 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18911 <round_scalar_name> ... this.
18912 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18913 <sse>_vm<code><mode>3<mask_scalar_name>
18914 <round_saeonly_scalar_name> ... this.
18915 (v<plusminus_mnemonic><ssescalarmodesuffix>
18916 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18917 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18918 v<plusminus_mnemonic><ssescalarmodesuffix>
18919 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18920 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18921 (v<multdiv_mnemonic><ssescalarmodesuffix>
18922 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18923 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18924 v<multdiv_mnemonic><ssescalarmodesuffix>
18925 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18926 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18927 (v<maxmin_float><ssescalarmodesuffix>
18928 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18929 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18930 v<maxmin_float><ssescalarmodesuffix>
18931 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18932 %0<mask_scalar_operand3>, %1, %<iptr>2
18933 <round_saeonly_scalar_mask_op3>} ... this.
18935 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18937 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18938 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18940 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18941 Alan Hayward <alan.hayward@arm.com>
18942 David Sherwood <david.sherwood@arm.com>
18944 * combine.c (simplify_if_then_else): Remove "enum" before
18946 * compare-elim.c (can_eliminate_compare): Likewise.
18947 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18949 (aarch64_lookup_simd_builtin_type): Likewise.
18950 (aarch64_simd_builtin_type): Likewise.
18951 (aarch64_init_simd_builtin_types): Likewise.
18952 (aarch64_simd_expand_args): Likewise.
18953 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18955 (aarch64_reverse_mask): Likewise.
18956 (aarch64_simd_emit_reg_reg_move): Likewise.
18957 (aarch64_gen_adjusted_ldpstp): Likewise.
18958 (aarch64_ccmp_mode_to_code): Likewise.
18959 (aarch64_operands_ok_for_ldpstp): Likewise.
18960 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18961 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18963 (aarch64_min_divisions_for_recip_mul): Likewise.
18964 (aarch64_reassociation_width): Likewise.
18965 (aarch64_get_condition_code_1): Likewise.
18966 (aarch64_simd_emit_reg_reg_move): Likewise.
18967 (aarch64_simd_attr_length_rglist): Likewise.
18968 (aarch64_reverse_mask): Likewise.
18969 (aarch64_operands_ok_for_ldpstp): Likewise.
18970 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18971 (aarch64_gen_adjusted_ldpstp): Likewise.
18972 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18974 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18975 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18976 (arm_lookup_simd_builtin_type): Likewise.
18977 (arm_simd_builtin_type): Likewise.
18978 (arm_init_simd_builtin_types): Likewise.
18979 (arm_expand_builtin_args): Likewise.
18980 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18981 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18982 (ft32_setup_incoming_varargs): Likewise.
18983 (ft32_function_arg): Likewise.
18984 (ft32_function_arg_advance): Likewise.
18985 (ft32_pass_by_reference): Likewise.
18986 (ft32_arg_partial_bytes): Likewise.
18987 (ft32_valid_pointer_mode): Likewise.
18988 (ft32_addr_space_pointer_mode): Likewise.
18989 (ft32_addr_space_legitimate_address_p): Likewise.
18990 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18992 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18993 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18994 (iamcu_alignment): Likewise.
18995 (canonicalize_vector_int_perm): Likewise.
18996 (ix86_noce_conversion_profitable_p): Likewise.
18997 (ix86_mpx_bound_mode): Likewise.
18998 (ix86_operands_ok_for_move_multiple): Likewise.
18999 * config/microblaze/microblaze-protos.h
19000 (microblaze_expand_conditional_branch_reg): Likewise.
19001 * config/microblaze/microblaze.c
19002 (microblaze_expand_conditional_branch_reg): Likewise.
19003 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19005 (rs6000_reassociation_width): Likewise.
19006 (rs6000_invalid_binary_op): Likewise.
19007 (fusion_p9_p): Likewise.
19008 (emit_fusion_p9_load): Likewise.
19009 (emit_fusion_p9_store): Likewise.
19010 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19012 (riscv_hard_regno_mode_ok_p): Likewise.
19013 (riscv_address_insns): Likewise.
19014 (riscv_split_symbol): Likewise.
19015 (riscv_legitimize_move): Likewise.
19016 (riscv_function_value): Likewise.
19017 (riscv_hard_regno_nregs): Likewise.
19018 (riscv_expand_builtin): Likewise.
19019 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19020 (riscv_build_integer): Likewise.
19021 (riscv_split_integer): Likewise.
19022 (riscv_legitimate_constant_p): Likewise.
19023 (riscv_cannot_force_const_mem): Likewise.
19024 (riscv_regno_mode_ok_for_base_p): Likewise.
19025 (riscv_valid_base_register_p): Likewise.
19026 (riscv_valid_offset_p): Likewise.
19027 (riscv_valid_lo_sum_p): Likewise.
19028 (riscv_classify_address): Likewise.
19029 (riscv_legitimate_address_p): Likewise.
19030 (riscv_address_insns): Likewise.
19031 (riscv_load_store_insns): Likewise.
19032 (riscv_force_binary): Likewise.
19033 (riscv_split_symbol): Likewise.
19034 (riscv_force_address): Likewise.
19035 (riscv_legitimize_address): Likewise.
19036 (riscv_move_integer): Likewise.
19037 (riscv_legitimize_const_move): Likewise.
19038 (riscv_legitimize_move): Likewise.
19039 (riscv_address_cost): Likewise.
19040 (riscv_subword): Likewise.
19041 (riscv_output_move): Likewise.
19042 (riscv_canonicalize_int_order_test): Likewise.
19043 (riscv_emit_int_order_test): Likewise.
19044 (riscv_function_arg_boundary): Likewise.
19045 (riscv_pass_mode_in_fpr_p): Likewise.
19046 (riscv_pass_fpr_single): Likewise.
19047 (riscv_pass_fpr_pair): Likewise.
19048 (riscv_get_arg_info): Likewise.
19049 (riscv_function_arg): Likewise.
19050 (riscv_function_arg_advance): Likewise.
19051 (riscv_arg_partial_bytes): Likewise.
19052 (riscv_function_value): Likewise.
19053 (riscv_pass_by_reference): Likewise.
19054 (riscv_setup_incoming_varargs): Likewise.
19055 (riscv_print_operand): Likewise.
19056 (riscv_elf_select_rtx_section): Likewise.
19057 (riscv_save_restore_reg): Likewise.
19058 (riscv_for_each_saved_reg): Likewise.
19059 (riscv_register_move_cost): Likewise.
19060 (riscv_hard_regno_mode_ok_p): Likewise.
19061 (riscv_hard_regno_nregs): Likewise.
19062 (riscv_class_max_nregs): Likewise.
19063 (riscv_memory_move_cost): Likewise.
19064 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19065 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19066 (rl78_addr_space_address_mode): Likewise.
19067 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19069 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19070 (rs6000_reassociation_width): Likewise.
19071 (rs6000_invalid_binary_op): Likewise.
19072 (fusion_p9_p): Likewise.
19073 (emit_fusion_p9_load): Likewise.
19074 (emit_fusion_p9_store): Likewise.
19075 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19076 (ok_for_simple_move_operands): Likewise.
19077 (ok_for_simple_move_strict_operands): Likewise.
19078 (ok_for_simple_arith_logic_operands): Likewise.
19079 (visium_legitimize_reload_address): Likewise.
19080 (visium_select_cc_mode): Likewise.
19081 (output_cbranch): Likewise.
19082 (visium_split_double_move): Likewise.
19083 (visium_expand_copysign): Likewise.
19084 (visium_expand_int_cstore): Likewise.
19085 (visium_expand_fp_cstore): Likewise.
19086 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19087 (visium_function_arg): Likewise.
19088 (visium_function_arg_advance): Likewise.
19089 (visium_libcall_value): Likewise.
19090 (visium_setup_incoming_varargs): Likewise.
19091 (visium_legitimate_constant_p): Likewise.
19092 (visium_legitimate_address_p): Likewise.
19093 (visium_legitimize_address): Likewise.
19094 (visium_secondary_reload): Likewise.
19095 (visium_register_move_cost): Likewise.
19096 (visium_memory_move_cost): Likewise.
19097 (prepare_move_operands): Likewise.
19098 (ok_for_simple_move_operands): Likewise.
19099 (ok_for_simple_move_strict_operands): Likewise.
19100 (ok_for_simple_arith_logic_operands): Likewise.
19101 (visium_function_value_1): Likewise.
19102 (rtx_ok_for_offset_p): Likewise.
19103 (visium_legitimize_reload_address): Likewise.
19104 (visium_split_double_move): Likewise.
19105 (visium_expand_copysign): Likewise.
19106 (visium_expand_int_cstore): Likewise.
19107 (visium_expand_fp_cstore): Likewise.
19108 (visium_split_cstore): Likewise.
19109 (visium_select_cc_mode): Likewise.
19110 (visium_split_cbranch): Likewise.
19111 (output_cbranch): Likewise.
19112 (visium_print_operand_address): Likewise.
19113 * expmed.c (flip_storage_order): Likewise.
19114 * expmed.h (emit_cstore): Likewise.
19115 (flip_storage_order): Likewise.
19116 * genrecog.c (validate_pattern): Likewise.
19117 * hsa-gen.c (gen_hsa_addr): Likewise.
19118 * internal-fn.c (expand_arith_overflow): Likewise.
19119 * ira-color.c (allocno_copy_cost_saving): Likewise.
19120 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19121 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19122 (process_invariant_for_inheritance): Likewise.
19123 * lra-eliminations.c (move_plus_up): Likewise.
19124 * omp-low.c (lower_oacc_reductions): Likewise.
19125 * simplify-rtx.c (simplify_subreg): Likewise.
19126 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19127 (TARGET_CHKP_BOUND_MODE): Likewise..
19128 * targhooks.c (default_chkp_bound_mode): Likewise.
19129 (default_setup_incoming_vararg_bounds): Likewise.
19130 * targhooks.h (default_chkp_bound_mode): Likewise.
19131 (default_setup_incoming_vararg_bounds): Likewise.
19132 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19133 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19134 (have_whole_vector_shift): Likewise.
19135 * tree-vect-stmts.c (vectorizable_load): Likewise.
19136 * doc/tm.texi: Regenerate.
19138 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19140 Graceful degrade if Binutils PR21472 is not available.
19143 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19144 .rodata in flash test fails.
19145 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19146 * confgure: Regenerate.
19147 * config.in: Regenerate.
19148 * config/avr/avr.c (avr_asm_named_section)
19149 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19150 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19151 (avr_asm_init_sections): Same.
19153 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19155 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19156 (fma<VH:mode>4_intrinsic): Likewise.
19157 (*fmsub<VCVTF:mode>4): Likewise.
19158 (*fmsub<VH:mode>4_intrinsic): Likewise.
19160 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19163 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19164 Don't depend on "optimize > 0".
19165 (out_movhi_r_mr, out_movqi_mr_r): Same.
19166 (out_movhi_mr_r, out_movqi_r_mr): Same.
19167 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19168 io_address_operand on "optimize > 0".
19170 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19172 * tree-loop-distribution.c: Add general explanantion on the pass.
19173 (generate_loops_for_partition): Mark distributed loop.
19174 (pg_add_dependence_edges): New parameter. Handle alias data
19175 dependence specially and record it in the parameter if asked.
19176 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19177 (init_partition_graph_vertices, add_partition_graph_edge): New.
19178 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19179 (free_partition_graph_vdata, build_partition_graph): New.
19180 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19181 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19182 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19183 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19184 (version_for_distribution_p, finalize_partitions): New.
19185 (distribute_loop): Handle alias data dependence specially. Factor
19186 out loop fusion code as functions and call these functions.
19188 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19190 * tree-loop-distribution.c (classify_partition): New parameter and
19191 better handle reduction statement.
19192 (rdg_build_partitions): Revise comment.
19193 (distribute_loop): Compute statements in all partitions and pass it
19194 to classify_partition.
19196 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19198 * tree-loop-distribution.c (enum partition_type): New.
19199 (struct partition): New field type.
19200 (partition_merge_into): Add parameter. Update partition type.
19201 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19202 (build_rdg_partition_for_vertex): Compute partition type.
19203 (rdg_build_partitions): Dump partition type.
19204 (distribute_loop): Update calls to partition_merge_into.
19206 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19208 * tree-loop-distribution.c (struct ddr_hasher): New.
19209 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19211 (classify_partition): Call get_data_dependence.
19212 (pg_add_dependence_edges): Ditto.
19213 (distribute_loop): Release data dependence hash table.
19215 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19217 * tree-loop-distribution.c (ref_base_address): Delete.
19218 (similar_memory_accesses): Rename ...
19219 (share_memory_accesses): ... to this. Check if partitions access
19220 the same memory reference.
19221 (distribute_loop): Call share_memory_accesses.
19223 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19225 * tree-loop-distribution.c (struct partition): New field recording
19226 its data reference.
19227 (partition_alloc, partition_free): Init and release data refs.
19228 (partition_merge_into): Merge data refs.
19229 (build_rdg_partition_for_vertex): Collect data refs for partition.
19230 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19232 (distribute_loop): Remve data refs from vertice data of partition
19235 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19237 * tree-loop-distribution.c (params.h): Include header file.
19238 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19239 (datarefs_vec): New global var.
19240 (create_rdg_vertices): Use datarefs_vec directly.
19241 (free_rdg): Don't free data references.
19242 (build_rdg): Update use. Don't free data references.
19243 (distribute_loop): Compute global variable for data references.
19244 Bail out if there are too many data references.
19246 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19248 * tree-loop-distribution.c (loop_nest): New global var.
19249 (build_rdg): Use loop directly, rather than loop nest.
19250 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19252 (distribute_loop): Compute global variable loop nest. Update use.
19254 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19256 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19257 (partition_merge_into): New parameter. Dump reason for fusion.
19258 (distribute_loop): Update use of partition_merge_into.
19260 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19262 * tree-loop-distribution.c (bb_top_order_index): New.
19263 (bb_top_order_index_size, bb_top_order_cmp): New.
19264 (stmts_from_loop): Use topological order.
19265 (pass_loop_distribution::execute): Compute and release topological
19266 order for basic blocks.
19268 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19270 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19273 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19275 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19276 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19277 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19278 * internal-fn.def (LOOP_DIST_ALIAS): New.
19279 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19280 (fold_loop_internal_call): ... this.
19281 (vect_loop_dist_alias_call): New function.
19282 (set_uid_loop_bbs): Call fold_loop_internal_call.
19283 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19286 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19289 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19290 Require dead FLAGS_REG at the beginning of a peephole.
19292 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19295 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19296 arguments in the call to __builtin_ia32_sbb_u32.
19297 (_subborrow_u64): Swap _X and _Y arguments in the call to
19298 __builtin_ia32_sbb_u64.
19300 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19303 * tree-vrp.c (compare_assert_loc): Turn into a function template
19304 with stable template parameter. Only test if a->e is NULL,
19305 !a->e == !b->e has been verified already. Use e == NULL or
19306 e != NULL instead of e or ! e tests. If stable is true, don't use
19307 iterative_hash_expr, on the other side allow a or b or both NULL
19308 and sort the NULLs last.
19309 (process_assert_insertions): Sort using compare_assert_loc<false>
19310 instead of compare_assert_loc, later sort using
19311 compare_assert_loc<true> before calling process_assert_insertions_for
19312 in a loop. Use break instead of continue once seen NULL pointer.
19314 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19316 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19317 Cortex-R7 and Cortex-R8 processors.
19319 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19321 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19322 uninitialized while src is not.
19324 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19326 * common/config/arm/arm-common.c: Adjust include path for
19328 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19329 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19330 (arm-cpu-data.h): Likewise.
19331 (arm-cpu-cdata.h): Likewise.
19332 * config/arm/arm-cpu.h: Delete.
19333 * config/arm/arm-cpu-cdata.h: Delete.
19334 * config/arm/arm-cpu-data.h: Delete.
19336 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19338 * config/arm/arm-cpus.in (cortex-a55): New.
19339 (cortex-a75): Likewise.
19340 (cortex-a75.cortex-a55): Likewise.
19341 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19343 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19344 * config/arm/arm-cpu-cdata.h: Regenerate.
19345 * config/arm/arm-cpu-data.h: Regenerate.
19346 * config/arm/arm-cpu.h: Regenerate.
19347 * config/arm/arm-tables.opt: Regenerate.
19348 * config/arm/arm-tune.md: Regenerate.
19350 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19352 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19354 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19356 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19359 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19361 PR tree-optimization/81292
19362 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19363 full_string_p, also call adjust_related_strinfos if the adjustment
19364 is simple, otherwise invalidate related strinfos.
19366 2017-07-04 Martin Liska <mliska@suse.cz>
19369 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19370 newly created variable as DECL_IGNORED_P.
19372 2017-07-04 Martin Liska <mliska@suse.cz>
19375 * ipa-inline.c (inline_small_functions):
19376 Use xstrdup_for_dump.
19378 2017-07-04 Tom de Vries <tom@codesourcery.com>
19380 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19382 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19385 * config/darwin.c (darwin_function_switched_text_sections):
19388 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19390 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19392 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19394 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19396 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19398 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19399 min_profitable_iters, and th as inclusive lower bounds.
19400 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19401 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19402 for min_profitable_iters and min_profitable_estimate.
19403 (vect_transform_loop): Treat th as an inclusive lower bound.
19404 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19406 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19409 * config/darwin.c (darwin_function_switched_text_sections):
19410 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19411 in two pieces, and suppress the use of buf.
19413 2017-07-03 Nathan Sidwell <nathan@acm.org>
19415 * hash-table.h (hash_table_mod1): Fix indentation.
19417 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19419 PR middle-end/81290
19420 * predict.c (force_edge_cold): Be more careful about propagation
19422 * profile-count.h (profile_probability::guessed,
19423 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19425 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19427 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19429 * doc/invoke.texi (rcpc architecture extension): Document it.
19431 2017-07-03 Richard Biener <rguenther@suse.de>
19433 PR tree-optimization/60510
19434 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19435 the scalar reduction PHI and use it.
19436 (vectorizable_reduction): Properly guard the single_defuse_cycle
19437 path for non-SLP reduction chains where we cannot use it.
19438 Rework reduc_def/index and vector type deduction. Rework
19439 vector operand gathering during reduction op code-gen.
19440 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19441 chains dissolve the chain and leave it to non-SLP reduction
19444 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19446 * tree-data-ref.h (dr_alignment): Declare.
19447 * tree-data-ref.c (dr_alignment): New function.
19448 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19449 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19451 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19453 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19455 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19456 and base_misalignment fields.
19457 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19458 * tree-data-ref.c: Include builtins.h.
19459 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19460 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19461 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19462 * tree-vect-data-refs.c: Include tree-cfg.h.
19463 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19464 fields instead of calculating an alignment here.
19465 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19466 innermost_loop_behavior fields.
19468 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19470 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19472 (DR_STEP_ALIGNMENT): New macro.
19473 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19474 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19475 (create_data_ref): Print it.
19476 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19477 to tell whether the step preserves vector (mis)alignment.
19478 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19479 Move the check for an integer step and generalise to all INTEGER_CST.
19480 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19481 Print the outer step alignment.
19483 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19485 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19486 with offset_alignment.
19487 (DR_ALIGNED_TO): Delete.
19488 (DR_OFFSET_ALIGNMENT): New macro.
19489 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19490 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19491 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19492 (create_data_ref): Likewise.
19493 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19494 (vect_analyze_data_refs): Likewise.
19495 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19496 creating dummy innermost behavior.
19498 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19500 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19501 with a "innermost_loop_behavior *" and refeence tree.
19502 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19503 (create_data_ref): Update call accordingly.
19504 * tree-predcom.c (find_looparound_phi): Likewise.
19506 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19508 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19509 fields with dr_wrt_vec_loop.
19510 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19511 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19512 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19513 (vect_dr_behavior): New function.
19514 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19515 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19516 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19517 track whether the step preserves the misalignment.
19518 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19519 Use vect_dr_behavior.
19520 (vect_setup_realignment): Update call accordingly.
19521 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19522 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19523 call to vect_create_addr_base_for_vector_ref.
19524 (vect_create_cond_for_align_checks): Likewise.
19525 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19526 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19527 (vect_recog_mask_conversion_pattern): Likewise.
19528 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19529 (new_stmt_vec_info): Remove redundant zeroing.
19531 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19533 * common/config/arm/arm-common.c (arm_be8_option): New function.
19534 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19535 (ISA_ARMv6): Add isa_bit_be8.
19536 * config/arm/arm.h (arm_be8_option): Add prototype.
19537 (BE8_SPEC_FUNCTION): New define.
19538 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19539 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19540 (mlittle-endian): Similarly.
19541 (mbe8, mbe32): New options.
19542 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19543 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19545 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19547 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19549 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19551 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19552 (cleanup_tree_cfg_bb): Use it.
19553 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19555 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19557 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19560 * loop-doloop.c (add_test): Update profile.
19562 2017-07-03 Martin Liska <mliska@suse.cz>
19565 * sanopt.c (rewrite_usage_of_param): New function.
19566 (sanitize_rewrite_addressable_params): Likewise.
19567 (pass_sanopt::execute): Call rewrite_usage_of_param.
19569 2017-07-03 Richard Biener <rguenther@suse.de>
19571 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19572 back to using VIEW_CONVERT_EXPR.
19574 2017-07-03 Martin Liska <mliska@suse.cz>
19577 * doc/extend.texi: Document when a resolver function is
19578 generated for target_clones.
19580 2017-07-03 Martin Liska <mliska@suse.cz>
19582 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19583 and shadow memory of auto variables which are subject of
19584 use-after-scope sanitization.
19585 (asan_expand_mark_ifn): Add do set only when is_poison.
19587 2016-07-03 Richard Biener <rguenther@suse.de>
19589 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19591 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19592 (vectorizable_reduction): Perform reduction PHI creation when
19593 visiting a reduction PHI and adjust and simplify code generation
19594 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19595 (vect_transform_loop): Visit reduction PHIs.
19596 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19597 defs into the SLP tree.
19598 (vect_build_slp_tree): Reduction defs terminate the recursion.
19599 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19601 (vect_get_vec_defs_for_stmt_copy): Export.
19602 (vect_get_vec_defs): Likewise.
19603 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19605 (vect_get_vec_defs_for_stmt_copy): Declare.
19606 (vect_get_vec_defs): Likewise.
19608 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19610 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19611 parameter with a "loop" parameter and use it instead of the
19612 loop containing DR_STMT. Don't check simple_iv when doing
19613 BB analysis. Describe the two analysis modes in the comment.
19615 2017-07-03 Tom de Vries <tom@codesourcery.com>
19617 PR tree-optimization/69468
19618 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19619 (find_same_succ_bb): Handle ignore_edge_flags.
19621 2017-07-03 Tom de Vries <tom@codesourcery.com>
19623 PR tree-optimization/81192
19624 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19626 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19628 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19630 2017-07-03 Tom de Vries <tom@codesourcery.com>
19632 PR tree-optimization/81192
19633 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19634 BB_SAME_SUCC (bb) == NULL.
19636 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19638 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19641 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19643 * dumpfile.c: Include profile-count.h
19644 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19646 (insert_cond_bb): Update profile.
19647 * tree-cfg.h (insert_cond_bb): Update prototype.
19648 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19649 * tree-dump.c: Do not include tree-cfg.
19651 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19653 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19655 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19657 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19660 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19662 * tree-complex.c (expand_complex_div_wide): update profile.
19664 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19665 Alan Hayward <alan.hayward@arm.com>
19666 David Sherwood <david.sherwood@arm.com>
19668 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19669 (CORETYPES_H): New define.
19670 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19671 (insn-modes-inline.h, s-modes-inline-h): New rules.
19672 (generated_files): Add insn-modes-inline.h.
19673 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19674 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19675 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19676 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19677 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19678 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19679 (build/gencodes.o, build/genconditions.o): Likewise.
19680 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19681 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19682 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19683 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19684 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19685 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19686 * coretypes.h: Include everything up to real.h for generators.
19687 Include insn-modes.h first. Include wide-int-print.h after
19688 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19689 * machmode.h: Don't include insn-modes.h here.
19690 * function-tests.c: Remove includes of signop.h, machmode.h,
19691 double-int.h and wide-int.h.
19693 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19695 * optc-save-gen.awk: Likewise.
19696 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19697 * godump.c: Remove include of wide-int-print.h.
19698 * pretty-print.h: Likewise.
19699 * wide-int-print.cc: Likewise.
19700 * wide-int.cc: Likewise.
19701 * hash-map-tests.c: Remove include of signop.h.
19702 * hash-set-tests.c: Likewise.
19703 * rtl-tests.c: Likewise.
19704 * mkconfig.sh: Remove include of machmode.h.
19705 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19707 (emit_insn_modes_inline_h): ...this new function. Emit the code
19708 into an insn-modes-inline.h header file, adding appropriate
19709 include guards and end comments.
19710 (emit_insn_modes_c_header): Remove include of machmode.h.
19711 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19713 (main): Handle -i flag and call emit_insn_modes_inline_h when
19716 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19718 * tree-ssa-strlen.c (strinfo): Rename the length field to
19719 nonzero_chars. Add a full_string_p field.
19720 (compare_nonzero_chars, zero_length_string_p): New functions.
19721 (get_addr_stridx): Add an offset_out parameter.
19722 Use compare_nonzero_chars.
19723 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19724 (new_strinfo): Update after above changes to strinfo.
19725 (set_endptr_and_length): Set full_string_p.
19726 (get_string_length): Update after above changes to strinfo.
19727 (unshare_strinfo): Update call to new_strinfo.
19728 (maybe_invalidate): Likewise.
19729 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19730 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19731 as a uhwi instead of an shwi. Update after above changes to
19732 strinfo and new_strinfo.
19733 (zero_length_string): Assert that chainsi contains full strings.
19734 Use zero_length_string_p. Update call to new_strinfo.
19735 (adjust_related_strinfos): Update after above changes to strinfo.
19736 Copy full_string_p from origsi.
19737 (adjust_last_stmt): Use zero_length_string_p.
19738 (handle_builtin_strlen): Update after above changes to strinfo and
19739 new_strinfo. Install the lhs as the string length if the previous
19740 entry didn't describe a full string.
19741 (handle_builtin_strchr): Update after above changes to strinfo
19743 (handle_builtin_strcpy): Likewise.
19744 (handle_builtin_strcat): Likewise.
19745 (handle_builtin_malloc): Likewise.
19746 (handle_pointer_plus): Likewise.
19747 (handle_builtin_memcpy): Likewise. Track nonzero characters
19748 that aren't necessarily followed by a nul terminator.
19749 (handle_char_store): Likewise.
19751 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19753 PR tree-optimization/80769
19754 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19755 for malloc and calloc. Document the new invariant that all related
19756 strinfos have delayed lengths or none do.
19757 (verify_related_strinfos): Move earlier in file.
19758 (set_endptr_and_length): New function, split out from...
19759 (get_string_length): ...here. Also set the lengths of related
19761 (zero_length_string): Assert that chainsi has known (rather than
19763 (adjust_related_strinfos): Likewise.
19765 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19767 PR tree-optimization/81136
19768 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19769 assert that two references with the same misalignment have the same
19770 compile-time misalignment if those compile-time misalignments
19773 2017-07-01 Andi Kleen <ak@linux.intel.com>
19775 * print-tree.c (print_node): Print all attributes.
19777 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19779 * cfg.c (scale_bbs_frequencies): New function.
19780 * cfg.h (scale_bbs_frequencies): Declare it.
19781 * cfgloopanal.c (single_likely_exit): Cleanup.
19782 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19784 (scale_loop_profile): Likewise.
19785 (loop_version): Likewise.
19786 (create_empty_loop_on_edge): Update.
19787 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19788 scale_loop_frequencies, scale_loop_profile, loopify,
19789 loop_version): Update prototypes.
19790 * modulo-sched.c (sms_schedule): Update.
19791 * predict.c (unlikely_executed_edge_p): Also check probability.
19792 (probably_never_executed_edge_p): Fix typo.
19793 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19794 * tree-parloops.c (gen_parallel_loop): Update.
19795 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19796 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19797 * tree-ssa-loop-split.c (split_loop): Update.
19798 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19799 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19800 (vect_loop_versioning): Update.
19801 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19803 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19805 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19807 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19809 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19810 to keep profile consistent.
19812 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19814 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19815 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19816 * profile-count.h (max_safe_multiplier): Make unsigned.
19817 (profile_count::guessed_zero): New.
19819 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19821 * bb-reorder.c (fix_up_crossing_landing_pad,
19822 fix_crossing_conditional_branches): Use make_single_succ_edge
19823 to keep profile consistent.
19825 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19827 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19830 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19833 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19834 the right scopes, make sure cond_jump isn't preserved between multiple
19835 iterations. Search for fallthru edge whenever there are 3+ edges and
19836 use find_fallthru_edge for it.
19838 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19840 Patch by Alexander Monakov <amonakov@ispras.ru>
19841 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19842 probabilities consistently.
19844 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19846 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19847 emit_cmp_and_jump_insns.
19849 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19852 * tree-inline.c (expand_call_inline): Combine profile statuses.
19854 2017-06-30 Andrew Pinski <apinski@cavium.com>
19856 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19857 fold_stmt returned true.
19859 2017-06-30 Nathan Sidwell <nathan@acm.org>
19861 * ggc.h (empty_string): Delete.
19862 * cfgexpand.c (expand_asm_stmt): Use plain "".
19863 * optabs.c (expand_asm_memory_barrier): Likewise.
19864 * stringpool.c (empty_string): Delete.
19865 (digit_vector, digit_string): Delete.
19866 (ggc_alloc_string): Use plain "", don't optimize single digit
19867 strings. Use ggc_alloc_atomic.
19869 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19871 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19872 comparison set and one other set, use the cost of the non-comparison
19875 2017-06-30 Nathan Sidwell <nathan@acm.org>
19877 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19880 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19882 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19883 loops. Remove now unneeded calls to gimple_switch_set_label() that
19884 just set removed labels to NULL_TREE.
19886 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19888 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19889 (set_range_info): ...here. Only call set_range_info_raw if domain
19891 (set_nonzero_bits): Call set_range_info_raw.
19892 * tree-ssanames.h (set_range_info_raw): New.
19894 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19897 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19898 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19899 of nonimmediate_operand and <store_mask_constraint> instead of m
19900 for the input operand. For V8FI iterator, always split if input
19901 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19902 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19903 <store_mask_predicate> instead of register_operand and
19904 <store_mask_constraint> instead of v for the input operand. Make
19905 sure both operands aren't MEMs for if not <mask_applied>.
19907 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19909 * lto-wrapper.c (copy_file) Close both file descriptors before
19912 2017-06-30 Martin Liska <mliska@suse.cz>
19915 * multiple_target.c (create_dispatcher_calls): Make ifunc
19916 also for function that don't have calls or are not referenced.
19918 2017-06-30 Richard Biener <rguenther@suse.de>
19920 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19921 analyze the first scalar stmt. Move vector type computation
19922 for the BB case here from ...
19923 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19924 live operation processing in the SLP case properly.
19926 2017-06-30 Richard Biener <rguenther@suse.de>
19928 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19930 2017-06-30 Martin Liska <mliska@suse.cz>
19933 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19934 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19936 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19938 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19941 2017-06-30 Richard Biener <rguenther@suse.de>
19943 PR tree-optimization/81249
19944 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19945 condition reduction result to original scalar type.
19947 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19949 * profile-count.h (enum profile_quality): Fix typos and whitespace
19952 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19954 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19955 type for branch probabilities.
19957 2017-06-29 Julian Brown <julian@codesourcery.com>
19958 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19960 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19961 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19962 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19963 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19965 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19967 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19968 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19969 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19970 CC usage from generic code to here.
19971 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19972 CC usage into the target macros.
19974 2017-06-29 Maya Rashish <coypu@sdf.org>
19976 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19979 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19981 * arm/arm-builtins.c: Include profile-count.h
19982 * except.c (sjlj_emit_function_enter): Use
19983 profile_probability::unlikely.
19985 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19987 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19988 and tocrel_offset be pointer args rather than implicitly using
19990 (legitimate_constant_pool_address_p, rs6000_emit_move,
19991 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19992 tocrel_offset and use in toc_relative_expr_p call.
19993 (print_operand, print_operand_address): Use static tocrel_base_oac
19994 and tocrel_offset_oac.
19995 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19998 2017-06-29 Maya Rashish <coypu@sdf.org>
20000 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20002 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20004 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20005 objects, take into account only the alignment of 'op0' and 'mode1' if
20008 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20010 * ccmp.c (ccmp_tree_comparison_p): New function.
20011 (ccmp_candidate_p): Update to use above function.
20012 (get_compare_parts): New function.
20013 (expand_ccmp_next): Update to use new functions.
20014 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20016 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20017 take mode as argument.
20018 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20019 * expr.c (expand_expr_real_1): Pass mode as argument.
20021 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20023 * combine.c (combine_instructions): Print insns to dump_file, together
20026 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20028 * asan.c (asan_emit_stack_protection): Update.
20029 (create_cond_insert_point): Update.
20030 * auto-profile.c (afdo_propagate_circuit): Update.
20031 * basic-block.h (struct edge_def): Turn probability to
20032 profile_probability.
20033 (EDGE_FREQUENCY): Update.
20034 * bb-reorder.c (find_traces_1_round): Update.
20035 (better_edge_p): Update.
20036 (sanitize_hot_paths): Update.
20037 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20038 (make_single_succ_edge): Update.
20039 (check_bb_profile): Update.
20040 (dump_edge_info): Update.
20041 (update_bb_profile_for_threading): Update.
20042 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20043 probabilitycount to 0.
20044 * cfgbuild.c (compute_outgoing_frequencies): Update.
20045 * cfgcleanup.c (try_forward_edges): Update.
20046 (outgoing_edges_match): Update.
20047 (try_crossjump_to_edge): Update.
20048 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20049 (expand_gimple_tailcall): Update.
20050 (construct_init_block): Use make_single_succ_edge.
20051 (construct_exit_block): Use make_single_succ_edge.
20052 * cfghooks.c (verify_flow_info): Update.
20053 (redirect_edge_succ_nodup): Update.
20054 (split_edge): Update.
20055 (account_profile_record): Update.
20056 * cfgloopanal.c (single_likely_exit): Update.
20057 * cfgloopmanip.c (scale_loop_profile): Update.
20058 (set_zero_probability): Remove.
20059 (duplicate_loop_to_header_edge): Update.
20060 * cfgloopmanip.h (loop_version): Update prototype.
20061 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20062 (force_nonfallthru_and_redirect): Update.
20063 (update_br_prob_note): Update.
20064 (rtl_verify_edges): Update.
20065 (purge_dead_edges): Update.
20066 (rtl_lv_add_condition_to_bb): Update.
20067 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20068 * cgraphunit.c (init_lowered_empty_function): Update.
20069 (cgraph_node::expand_thunk): Update.
20070 * cilk-common.c: Include profile-count.h
20071 * dojump.c (inv): Remove.
20072 (jumpifnot): Update.
20073 (jumpifnot_1): Update.
20074 (do_jump_1): Update.
20076 (do_jump_by_parts_greater_rtx): Update.
20077 (do_compare_rtx_and_jump): Update.
20078 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20079 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20080 * dwarf2cfi.c: Include profile-count.h
20081 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20082 (sjlj_emit_dispatch_table): Likewise.
20083 * explow.c: Include profile-count.h
20084 * expmed.c (emit_store_flag_force): Update.
20085 (do_cmp_and_jump): Update.
20086 * expr.c (compare_by_pieces_d::generate): Update.
20087 (compare_by_pieces_d::finish_mode): Update.
20088 (emit_block_move_via_loop): Update.
20089 (store_expr_with_bounds): Update.
20090 (store_constructor): Update.
20091 (expand_expr_real_2): Update.
20092 (expand_expr_real_1): Update.
20093 * expr.h (try_casesi, try_tablejump): Update prototypes.
20094 * gimple-pretty-print.c (dump_probability): Update.
20095 (dump_profile): New.
20096 (dump_gimple_label): Update.
20097 (dump_gimple_bb_header): Update.
20098 * graph.c (draw_cfg_node_succ_edges): Update.
20099 * hsa-gen.c (convert_switch_statements): Update.
20100 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20101 (find_if_case_1): Update.
20102 (find_if_case_2): Update.
20103 * internal-fn.c (expand_arith_overflow_result_store): Update.
20104 (expand_addsub_overflow): Update.
20105 (expand_neg_overflow): Update.
20106 (expand_mul_overflow): Update.
20107 (expand_vector_ubsan_overflow): Update.
20108 * ipa-cp.c (good_cloning_opportunity_p): Update.
20109 * ipa-split.c (split_function): Use make_single_succ_edge.
20110 * ipa-utils.c (ipa_merge_profiles): Update.
20111 * loop-doloop.c (add_test): Update.
20112 (doloop_modify): Update.
20113 * loop-unroll.c (compare_and_jump_seq): Update.
20114 (unroll_loop_runtime_iterations): Update.
20115 * lra-constraints.c (lra_inheritance): Update.
20116 * lto-streamer-in.c (input_cfg): Update.
20117 * lto-streamer-out.c (output_cfg): Update.
20118 * mcf.c (adjust_cfg_counts): Update.
20119 * modulo-sched.c (sms_schedule): Update.
20120 * omp-expand.c (expand_omp_for_init_counts): Update.
20121 (extract_omp_for_update_vars): Update.
20122 (expand_omp_ordered_sink): Update.
20123 (expand_omp_for_ordered_loops): Update.
20124 (expand_omp_for_generic): Update.
20125 (expand_omp_for_static_nochunk): Update.
20126 (expand_omp_for_static_chunk): Update.
20127 (expand_cilk_for): Update.
20128 (expand_omp_simd): Update.
20129 (expand_omp_taskloop_for_outer): Update.
20130 (expand_omp_taskloop_for_inner): Update.
20131 * omp-simd-clone.c (simd_clone_adjust): Update.
20132 * optabs.c (expand_doubleword_shift): Update.
20133 (expand_abs): Update.
20134 (emit_cmp_and_jump_insn_1): Update.
20135 (expand_compare_and_swap_loop): Update.
20136 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20137 * predict.c (predictable_edge_p): Update.
20138 (edge_probability_reliable_p): Update.
20139 (set_even_probabilities): Update.
20140 (combine_predictions_for_insn): Update.
20141 (combine_predictions_for_bb): Update.
20142 (propagate_freq): Update.
20143 (estimate_bb_frequencies): Update.
20144 (force_edge_cold): Update.
20145 * profile-count.c (profile_count::dump): Add missing space into dump.
20146 (profile_count::debug): Add newline.
20147 (profile_count::differs_from_p): Explicitly convert to unsigned.
20148 (profile_count::stream_in): Update.
20149 (profile_probability::dump): New member function.
20150 (profile_probability::debug): New member function.
20151 (profile_probability::differs_from_p): New member function.
20152 (profile_probability::differs_lot_from_p): New member function.
20153 (profile_probability::stream_in): New member function.
20154 (profile_probability::stream_out): New member function.
20155 * profile-count.h (profile_count_quality): Rename to ...
20156 (profile_quality): ... this one.
20157 (profile_probability): New.
20158 (profile_count): Update.
20159 * profile.c (compute_branch_probabilities): Update.
20160 * recog.c (peep2_attempt): Update.
20161 * sched-ebb.c (schedule_ebbs): Update.
20162 * sched-rgn.c (find_single_block_region): Update.
20163 (compute_dom_prob_ps): Update.
20164 (schedule_region): Update.
20165 * sel-sched-ir.c (compute_succs_info): Update.
20166 * stmt.c (struct case_node): Update.
20167 (do_jump_if_equal): Update.
20168 (get_outgoing_edge_probs): Update.
20169 (conditional_probability): Update.
20170 (emit_case_dispatch_table): Update.
20171 (expand_case): Update.
20172 (expand_sjlj_dispatch_table): Update.
20173 (emit_case_nodes): Update.
20174 * targhooks.c: Update.
20175 * tracer.c (better_p): Update.
20176 (find_best_successor): Update.
20177 * trans-mem.c (expand_transaction): Update.
20178 * tree-call-cdce.c: Update.
20179 * tree-cfg.c (gimple_split_edge): Upate.
20180 (move_sese_region_to_fn): Upate.
20181 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20182 * tree-eh.c (lower_resx): Upate.
20183 (cleanup_empty_eh_move_lp): Upate.
20184 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20185 * tree-inline.c (copy_edges_for_bb): Update.
20186 (copy_cfg_body): Update.
20187 * tree-parloops.c (gen_parallel_loop): Update.
20188 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20189 (gimple_gen_time_profiler): Update.
20190 * tree-ssa-dce.c (remove_dead_stmt): Update.
20191 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20192 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20193 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20194 (unloop_loops): Update.
20195 (try_peel_loop): Update.
20196 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20197 * tree-ssa-loop-split.c (connect_loops): Update.
20198 (split_loop): Update.
20199 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20200 (hoist_guard): Update.
20201 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20202 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20203 (value_replacement): Update.
20204 * tree-ssa-reassoc.c (branch_fixup): Update.
20205 * tree-ssa-tail-merge.c (replace_block_by): Update.
20206 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20207 (create_edge_and_update_destination_phis): Update.
20208 (compute_path_counts): Update.
20209 (recompute_probabilities): Update.
20210 (update_joiner_offpath_counts): Update.
20211 (freqs_to_counts_path): Update.
20212 (duplicate_thread_path): Update.
20213 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20214 (struct switch_conv_info): Update.
20215 (gen_inbound_check): Update.
20216 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20217 (vect_do_peeling): Update.
20218 (vect_loop_versioning): Update.
20219 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20220 (optimize_mask_stores): Update.
20221 * ubsan.c (ubsan_expand_null_ifn): Update.
20222 * value-prof.c (gimple_divmod_fixed_value): Update.
20223 (gimple_divmod_fixed_value_transform): Update.
20224 (gimple_mod_pow2): Update.
20225 (gimple_mod_pow2_value_transform): Update.
20226 (gimple_mod_subtract): Update.
20227 (gimple_mod_subtract_transform): Update.
20228 (gimple_ic): Update.
20229 (gimple_stringop_fixed_value): Update.
20230 (gimple_stringops_transform): Update.
20231 * value-prof.h: Update.
20233 2017-06-29 Carl Love <cel@us.ibm.com>
20235 * config/rs6000/rs6000-c.c: Add support for built-in functions
20236 vector signed int vec_signed (vector float);
20237 vector signed long long vec_signed (vector double);
20238 vector signed int vec_signed2 (vector double, vector double);
20239 vector signed int vec_signede (vector double);
20240 vector signed int vec_signedo (vector double);
20241 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20242 instruction generator.
20243 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20244 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20245 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20247 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20248 vunsignede_v2df): Add define_expands.
20249 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20250 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20251 VEC_UNSIGNEDO): Add definitions.
20252 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20253 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20254 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20255 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20256 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20257 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20258 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20259 vec_unsignede, vec_unsignedo): Add builtin defines.
20260 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20262 * doc/extend.texi: Update the built-in documentation file for the
20263 new built-in functions.
20265 2017-06-29 Richard Biener <rguenther@suse.de>
20267 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20268 reduction chains to LOOP_VINFO_REDUCTIONS.
20269 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20270 SLP reductions after processing reduction chains.
20272 2017-06-29 Nathan Sidwell <nathan@acm.org>
20274 * builtins.c (fold_builtin_FUNCTION): Use
20275 lang_hooks.decl_printable_name.
20277 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20279 PR middle-end/81194
20280 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20281 with only one label.
20282 * stmt.c (expand_case): Assert NCASES is greater than one.
20284 2017-06-29 Richard Biener <rguenther@suse.de>
20286 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20288 (group_case_labels): Likewise.
20289 (find_taken_edge): Push sanity checking on val to workers...
20290 (find_taken_edge_cond_expr): ... here
20291 (find_taken_edge_switch_expr): ... and here, handle cases
20292 with just a default label.
20293 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20294 (group_case_labels): Likewise.
20295 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20296 group_case_labels does anything cleanup the CFG again.
20298 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20300 PR tree-optimization/81196
20301 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20302 exit condition comparing two IVs.
20304 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20306 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20307 profile to the dummy entry at the end of the list of architectures.
20308 * config/arm/arm-cpu-cdata.h: Regenerated.
20310 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20311 Michael Collison <michael.collison@arm.com>
20314 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20316 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20317 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20318 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20319 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20320 of shift when the shift amount is masked with constant equal to
20321 the size of the mode.
20322 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20325 2017-06-29 Martin Liska <mliska@suse.cz>
20327 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20329 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20332 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20335 (edge_growth_cache_entry::edge_growth_cache_entry): New
20337 (reset_edge_growth_cache): Update to use constructor.
20339 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20341 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20342 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20343 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20345 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20347 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20348 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20350 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20352 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20353 (*-linux-uclibc*): Add t-uclibc tmake_file.
20354 * config/t-musl: New.
20355 * config/t-uclibc: New.
20357 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20359 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20361 (gen_comm_data): Emit architectural setting of arch_prof.
20362 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20364 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20365 (armv8-m.base, armv8-m.main): Likewise.
20366 * arm-protos.h (arm_build_target): Add profile field.
20367 (arch_option): Likewise.
20368 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20370 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20371 arm_active_target.profile.
20373 2017-06-28 Richard Biener <rguenther@suse.de>
20375 PR middle-end/81227
20376 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20377 TYPE_OVERFLOW_WRAPS.
20378 * match.pd (negate_expr_p): Likewise.
20379 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20380 fold_build2, not fold_binary.
20382 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20384 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20385 Convert memory address to Pmode.
20386 (aarch64_print_operand): Assert MEM operands are always Pmode.
20388 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20391 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20392 Remove redundant if.
20393 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20394 * config/arm/aarch-common-protos.h
20395 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20396 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20398 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20401 * multiple_target.c (create_dispatcher_calls): Set the default
20402 clone to be static, not public.
20404 2017-06-28 Richard Biener <rguenther@suse.de>
20406 * tree-vect-loop.c (vectorizable_reduction): Move special
20407 cond reduction IV var creation ...
20408 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20409 parameter. Use STMT_VINFO_VECTYPE.
20410 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20413 2017-06-28 Martin Liska <mliska@suse.cz>
20416 * ipa-visibility.c (non_local_p): Handle visibility.
20418 2017-06-28 Martin Liska <mliska@suse.cz>
20421 * common.opt: Add IntegerRange to various options.
20422 * opt-functions.awk (integer_range_info): New function.
20423 * optc-gen.awk: Add integer_range_info to cl_options struct.
20424 * opts-common.c (decode_cmdline_option): Handle
20425 CL_ERR_INT_RANGE_ARG.
20426 (cmdline_handle_error): Likewise.
20427 * opts.c (print_filtered_help): Show valid interval in
20428 when --help is provided.
20429 * opts.h (struct cl_option): Add range_min and range_max fields.
20430 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20432 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20434 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20435 (x * C EQ/NE y * C): New transformation.
20437 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20439 * genmultilib (combination_space): Accept '+' in option names.
20441 2017-06-28 Martin Liska <mliska@suse.cz>
20444 * asan.c (instrument_derefs): Bail out inner references
20445 that are hard register variables.
20447 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20450 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20451 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20453 2017-06-28 Richard Biener <rguenther@suse.de>
20455 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20456 (vect_get_slp_defs): Adjust.
20457 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20459 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20461 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20462 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20463 (vectorizable_reduction): Adjust.
20464 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20466 (vect_get_slp_defs): Likewise.
20467 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20468 (vectorizable_bswap): Adjust.
20469 (vectorizable_call): Likewise.
20470 (vectorizable_conversion): Likewise.
20471 (vectorizable_assignment): Likewise.
20472 (vectorizable_shift): Likewise.
20473 (vectorizable_operation): Likewise.
20474 (vectorizable_store): Likewise.
20475 (vectorizable_condition): Likewise.
20476 (vectorizable_comparison): Likewise.
20478 2017-06-28 Michael Collison <michael.collison@arm.com>
20481 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20483 (arm_gen_movmemqi): Removed unused variable 'i'.
20484 Convert 'for' loop into 'while' loop.
20485 (arm_expand_prologue): Remove last unnecessary set of insn.
20486 (thumb_pop): Remove unused variable 'pushed_words'.
20487 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20489 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20491 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20492 * config/s390/s390-protos.h: Add prototype of
20493 s390_rel_address_ok_p.
20494 * config/s390/s390.c (s390_got_symbol): New function.
20495 (s390_rel_address_ok_p): New function.
20496 (legitimize_pic_address): Use s390_rel_address_ok_p.
20497 (s390_load_got): Use s390_got_symbol.
20498 (s390_option_override): Issue error if
20499 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20500 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20502 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20504 2017-06-27 Andrew Pinski <apinski@cavium.com>
20506 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20507 (X * copysign (1.0, X)): New pattern.
20508 (X * copysign (1.0, -X)): New pattern.
20509 (copysign (-1.0, CST)): New pattern.
20511 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20513 * genmultilib (combination_space): Remove variable.
20514 Validate reuse rules against regular expression for any sequence
20515 of multilib options in any order.
20517 2017-06-27 Michael Collison <michael.collison@arm.com>
20519 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20520 call aarch64_split_simd_combine.
20521 * (aarch64_combine_internal<mode>): Delete pattern.
20522 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20523 Allow register and subreg operands.
20525 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20527 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20528 specific need, just fallback on defaults.
20529 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20531 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20532 Olivier Hainque <hainque@adacore.com>
20534 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20536 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20537 targets. Pick a default if no particular attempt applied.
20538 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20541 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20543 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20544 (x86_64-wrs-vxworks7): Likewise.
20546 2017-06-27 Marek Polacek <polacek@redhat.com>
20549 * ubsan.c (instrument_null): Check get_base_address's result for null.
20551 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20553 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20555 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20557 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20558 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20559 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20560 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20561 New function types.
20562 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20563 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20564 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20565 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20566 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20567 BUILT_IN_FEUPDATEENV): New builtins.
20568 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20569 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20570 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20571 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20573 (builtin_structptr_types): Adjust size.
20574 * tree.c (builtin_structptr_types): Add four entries.
20576 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20577 Olivier Hainque <hainque@adacore.com>
20579 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20580 (TLS_SYM): New local macro, forcing reference to __tls__ on
20581 link command lines for VxWorks 7 RTPs, triggering initialization
20583 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20584 OS features TLS support, true for RTPs on VxWorks 7.
20585 * config/vxworks.c (vxworks_override_options): Setup emutls
20588 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20590 * predict.c (test_prediction_value_range): Use -1U instead of -1
20591 to avoid narrowing conversion warning.
20592 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20593 to avoid narrowing conversion warning.
20594 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20596 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20598 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20600 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20601 64bit configurations.
20602 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20603 (SIZE_TYPE): Likewise.
20604 * config/vxworks.c (vxworks_emutls_var_fields): Use
20605 long_unsigned_type_node instead of unsigned_type_node as the offset
20606 field type, which is "pointer" mode in emutls.c.
20608 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20611 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20613 PR middle-end/81207
20614 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20615 gimple_vuse copying separately from gimple_vdef copying.
20617 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20619 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20622 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20623 Olivier Hainque <hainque@adacore.com>
20625 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20626 for all vxworks7 targets.
20627 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20628 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20629 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20630 variations for VX6/VX7 and 32/64bits later on in ...
20631 (VXWORKS_LIB_SPEC): Leverage new macros.
20632 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20633 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20635 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20637 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20638 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20640 2017-06-26 Carl Love <cel@us.ibm.com>
20642 * config/rs6000/rs6000-c.c: Add support for built-in functions
20643 vector bool char vec_reve (vector bool char);
20644 vector signed char vec_reve (vector signed char);
20645 vector unsigned char vec_reve (vector unsigned char);
20646 vector bool int vec_reve (vector bool int);
20647 vector signed int vec_reve (vector signed int);
20648 vector unsigned int vec_reve (vector unsigned int);
20649 vector bool long long vec_reve (vector bool long long);
20650 vector signed long long vec_reve (vector signed long long);
20651 vector unsigned long long vec_reve (vector unsigned long long);
20652 vector bool short vec_reve (vector bool short);
20653 vector signed short vec_reve (vector signed short);
20654 vector double vec_reve (vector double);
20655 vector float vec_reve (vector float);
20656 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20657 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20658 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20659 (altivec_vreve): New pattern.
20660 * config/rs6000/altivec.h (vec_reve): New define.
20661 * doc/extend.texi (vec_rev): Update the built-in documentation file
20662 for the new built-in functions.
20664 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20666 PR tree-optimization/71815
20667 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20669 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20671 (slsr_process_phi): Likewise.
20672 (replace_uncond_cands_and_profitable_phis): Don't replace a
20673 multiply candidate with a stride of 1 (copy or cast).
20674 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20676 (lowest_cost_path): Likewise.
20677 (total_savings): Likewise.
20679 2017-06-26 Richard Biener <rguenther@suse.de>
20682 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20683 Use def_builtin_pure for all gather builtins.
20685 2017-06-26 Richard Biener <rguenther@suse.de>
20687 PR tree-optimization/81203
20688 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20689 non-dominating BBs.
20691 2017-06-26 Marek Polacek <polacek@redhat.com>
20694 * doc/invoke.texi: Document -Wmultistatement-macros.
20696 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20698 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20699 arm_neon_ok_no_float_abi effective target.
20701 2017-06-26 Richard Biener <rguenther@suse.de>
20703 PR tree-optimization/80928
20704 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20705 (copy_bbs): Set BB_DUPLICATED flag early.
20706 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20708 (execute_on_shrinking_pred): Likewise.
20709 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20710 BB_DUPLICATED blocks.
20711 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20712 iterate over all PHIs considering removal of *gsi.
20714 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20716 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20719 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20721 * config/rs6000/rs6000-string.c: (expand_block_clear,
20722 do_load_for_compare, select_block_compare_mode,
20723 compute_current_alignment, expand_block_compare,
20724 expand_strncmp_align_check, expand_strn_compare,
20725 expand_block_move, rs6000_output_load_multiple)
20726 Move functions related to string/block move/compare
20727 to a separate file.
20728 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20729 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20730 for this function which is now used in two files.
20731 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20732 * config.gcc: Add rs6000-string.o to extra_objs for
20733 targets powerpc*-*-* and rs6000*-*-*.
20735 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20738 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20739 32-bit, since indexed is not valid for DImode.
20740 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20741 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20742 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20743 (define_peephole2 for Altivec d-form store): Likewise.
20746 * multiple_target.c (create_dispatcher_calls): Only create the
20747 dispatcher call if the function is the default clone of a
20748 versioned function.
20750 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20752 PR middle-end/80902
20753 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20754 a call, force the call to not be a tail call.
20756 2017-06-23 Jeff Law <law@redhat.com>
20758 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20761 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20763 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20764 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20765 handling for early expansion of vector shifts (sl,sr,sra,rl).
20766 (builtin_function_type): Add vector shift right instructions
20767 to the unsigned argument list.
20769 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20771 rtl-optimizatoin/79286
20772 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20773 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20774 trap. PIC register plus a const unspec without offset can never trap.
20776 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20778 * tree.h (builtin_structptr_type): New type.
20779 (builtin_structptr_types): Declare new array.
20780 * tree.c (builtin_structptr_types): New array.
20781 (free_lang_data, build_common_tree_nodes): Use it.
20783 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20786 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20789 2017-06-22 Matt Turner <mattst88@gmail.com>
20791 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20792 Lake models to skylake case. Assume skylake for unknown
20793 models with clflushopt.
20795 2017-06-22 Jeff Law <law@redhat.com>
20797 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20798 frame sizes that do not satisfy aarch64_uimm12_shift.
20800 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20802 * profile-count.h (apply_probability,
20803 apply_scale, probability_in): Fix checks for zero.
20805 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20807 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20808 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20810 2016-06-22 Richard Biener <rguenther@suse.de>
20812 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20813 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20814 REDUC_MAX_EXPR support.
20815 (vectorizable_reduction): Likewise.
20816 (vect_create_epilog_for_reduction): Likewise.
20818 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20820 * match.pd (A / (1 << B) -> A >> B): New.
20821 * generic-match-head.c: Include optabs-tree.h.
20822 * gimple-match-head.c: Likewise.
20823 * optabs-tree.h (target_supports_op_p): New.
20824 * optabs-tree.c (target_supports_op_p): New.
20826 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20828 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20829 $gcc_cv_ld --help output.
20830 (gcc_cv_ld_demangle): Likewise.
20831 (gcc_cv_ld_eh_frame_hdr): Likewise.
20832 (gcc_cv_ld_pie): Likewise.
20833 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20834 (gcc_cv_ld_buildid): Likewise.
20835 (gcc_cv_ld_sysroot): Likewise.
20836 (ld_bndplt_support): Likewise.
20837 (ld_pushpopstate_support): Likewise.
20838 * configure: Regenerate.
20839 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20841 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20844 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20845 operands indexes to avoid gap between operands and match_dups.
20847 2017-06-21 Andrew Pinski <apinski@cavium.com>
20849 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20850 Increment Arith_shift and Arith_shift_reg by 1.
20851 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20853 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20854 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20855 (aarch64_strip_extend): Add new argument and test for it.
20856 (aarch64_cheap_mult_shift_p): New function.
20857 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20858 add a cost if it is true.
20859 Update calls to aarch64_strip_extend.
20860 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20862 2017-06-21 Andrew Pinski <apinski@cavium.com>
20864 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20866 (thunderxt88): Likewise.
20867 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20868 (thunderx_prefetch_tune): New variable.
20869 (thunderx2t99_prefetch_tune): Update for the correct values.
20870 (thunderxt88_tunings): New variable.
20871 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20872 generic_prefetch_tune.
20873 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20875 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20877 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20878 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20879 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20880 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20881 (aarch64_atomic_cas<mode>, GPI): Likewise.
20883 2017-06-21 Martin Liska <mliska@suse.cz>
20885 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20886 statements on cold and hot labels.
20887 * predict.c (tree_estimate_probability_bb): Remove the
20888 prediction from this place.
20890 2017-06-21 Martin Liska <mliska@suse.cz>
20892 PR tree-optimization/79489
20893 * gimplify.c (maybe_add_early_return_predict_stmt): New
20895 (gimplify_return_expr): Call the function.
20896 * predict.c (tree_estimate_probability_bb): Remove handling
20898 * predict.def: Update comment about early return predictor.
20899 * gimple-predict.h (is_gimple_predict): New function.
20900 * predict.def: Change default value of early return to 66.
20901 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20903 * passes.def: Put pass_strip_predict_hints to the beginning of
20906 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20908 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20909 FUNCTION_DECL declarations.
20910 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20912 (dwaf2out_decl): Likewise.
20913 * godump.c (go_early_global_decl): Skip call to the real debug hook
20914 for FUNCTION_DECL declarations.
20915 * passes.c (rest_of_decl_compilation): Skip call to the
20916 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20917 -fdump-go-spec is passed.
20919 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20921 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20922 Reorder for compactness.
20923 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20924 (def_builtin_pure, def_builtin_pure2): New functions.
20925 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20927 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20929 * match.pd (nop_convert): New predicate.
20930 ((A +- CST1) +- CST2): Allow some NOP conversions.
20932 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20935 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20936 with ctors/dtors if GOVD_SHARED is set.
20938 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20940 * config/aarch64/aarch64.md (movti_aarch64):
20941 Emit mov rather than orr.
20942 (movtf_aarch64): Likewise.
20943 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20944 Emit mov rather than orr.
20946 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20948 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20949 Swap alternatives, make integer dup more expensive.
20951 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20953 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20954 Return true for non-tls symbols.
20956 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20958 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20959 (cortex-a75): Likewise.
20960 (cortex-a75.cortex-a55): Likewise.
20961 * config/aarch64/aarch64-tune.md: Regenerate.
20962 * doc/invoke.texi (-mtune): Document new values for -mtune.
20964 2017-06-21 Tom de Vries <tom@codesourcery.com>
20966 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20967 stack_size feature.
20968 (Effective-Target Keywords, Other attributes): Suggest using
20969 dg-add-options stack_size feature to get stack limit in stack_size
20970 effective target documentation.
20972 2017-06-21 Julian Brown <julian@codesourcery.com>
20973 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20975 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20976 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20977 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20979 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20980 attribute type list for neon_multiply.
20981 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20982 attribute type list for neon_multiply.
20983 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20984 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20985 attribute type list for neon_multiply.
20986 * config/arm/types.md (crypto_pmull): Add.
20987 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20988 attribute type list.
20990 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20992 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20995 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20997 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20998 to make sure not to dereference a NULL cost_classes_ptr pointer.
21000 2017-06-20 Carl Love <cel@us.ibm.com>
21002 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21003 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21004 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21005 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21006 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21007 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21008 VMULOSW): New enum "unspec" values.
21009 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21010 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21011 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21012 altivec_vmulosw): New patterns.
21013 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21014 VMULOSW): Add definitions.
21016 2017-06-20 Julia Koval <julia.koval@intel.com>
21018 * config/i386/i386.c: Fix rounding expand for new pattern.
21019 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21021 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21023 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21024 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21026 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21028 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21031 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21033 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21034 architecture, then by implementer ID.
21035 * config/aarch64/aarch64-tune.md: Regenerate.
21037 2017-06-20 Richard Biener <rguenther@suse.de>
21039 PR middle-end/81097
21040 * fold-const.c (split_tree): Fold to type before negating.
21042 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21044 * diagnostic-show-locus.c
21045 (selftest::test_fixit_deletion_affecting_newline): New function.
21046 (selftest::diagnostic_show_locus_c_tests): Call it.
21048 2017-06-20 Andreas Schwab <schwab@suse.de>
21051 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21054 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21056 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21057 __ARM_FEATURE_COPROC according to support.
21059 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21061 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21062 Rewritten to avoid overflow for > 32-bit pointers.
21065 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21066 by removing enum keyword.
21067 (ubsan_type_descriptor): Likewise. Formatting fix.
21070 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21071 splitter): Require TARGET_SSE2 in the condition.
21073 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21076 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21077 for doing vector set of SFmode on ISA 3.0.
21078 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21079 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21081 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21082 SFmode value into a V4SF variable that was extracted from another
21083 V4SF variable without converting the element to double precision
21084 and back to single precision vector format.
21085 (vsx_insert_extract_v4sf_p9_2): Likewise.
21087 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21089 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21090 in UWHI to avoid undefined overflow.
21093 * ubsan.h (enum ubsan_encode_value_phase): New.
21094 (ubsan_encode_value): Change second argument to
21095 enum ubsan_encode_value_phase with default value of
21096 UBSAN_ENCODE_VALUE_GENERIC.
21097 * ubsan.c (ubsan_encode_value): Change second argument to
21098 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21099 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21100 create_tmp_var_raw instead of create_tmp_var and use a
21102 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21103 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21104 ubsan_encode_value callers.
21107 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21108 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21109 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21111 2017-06-19 Richard Biener <rguenther@suse.de>
21113 PR middle-end/81118
21114 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21115 estimates if we changed anything.
21117 2017-06-19 Richard Biener <rguenther@suse.de>
21119 PR tree-optimization/80887
21120 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21121 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21122 simplified lookups, then reset mprts_hook.
21123 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21125 (try_to_simplify): Likewise.
21127 2017-06-19 Martin Liska <mliska@suse.cz>
21130 * gimplify.c (gimplify_switch_expr):
21131 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21133 2017-06-19 Martin Liska <mliska@suse.cz>
21135 * doc/install.texi: Document that PGO runs in 4 stages.
21137 2017-06-19 Martin Liska <mliska@suse.cz>
21140 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21141 to dispatcher function name.
21142 * multiple_target.c (replace_function_decl): New function.
21143 (create_dispatcher_calls): Redirect both edges and references.
21145 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21147 * profile-count.c (profile_count::dump): Dump quality.
21148 (profile_count::differs_from_p): Update for unsigned val.
21149 * profile-count.h (profile_count_quality): New enum.
21150 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21152 2017-06-19 Richard Biener <rguenther@suse.de>
21154 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21155 struct function as arg.
21156 (estimate_numbers_of_iterations): Export overload with loop arg.
21157 (free_numbers_of_iterations_estimates_loop): Use an overload of
21158 free_numbers_of_iterations_estimates instead.
21159 * tree-cfg.c (remove_bb): Adjust.
21160 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21161 * tree-parloops.c (gen_parallel_loop): Likewise.
21162 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21164 (tree_unroll_loops_completely): Likewise.
21165 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21166 Use an overload instead and export.
21167 (estimated_loop_iterations): Adjust.
21168 (max_loop_iterations): Likewise.
21169 (likely_max_loop_iterations): Likewise.
21170 (estimate_numbers_of_iterations): Take struct function as arg
21172 (loop_exits_before_overflow): Adjust.
21173 (free_numbers_of_iterations_estimates_loop): Use an overload.
21174 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21175 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21177 2017-06-19 Richard Biener <rguenther@suse.de>
21180 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21181 RANGE_EXPR conservatively.
21183 2017-06-16 Carl Love <cel@us.ibm.com>
21185 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21186 definitions for vec_float, vec_float2, vec_floato,
21187 vec_floate built-ins.
21188 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21189 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21191 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21192 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21193 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21194 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21195 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21196 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21197 vec_floato): Add builtin defines.
21198 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21199 Update the built-in documentation file for the new built-in
21202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21204 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21205 (mthumb): Mark as the negative of -marm.
21207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21209 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21211 (ARM Options, -mtune): Document that this accepts the same
21212 extension options as -mcpu.
21213 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21217 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21218 permitted extensions.
21220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21222 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21223 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21224 (armv8-m.main): Add option +nodsp.
21225 * config/arm/arm-cpu-cdata.h: Regenerated.
21227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21229 * config/arm/t-fuchsia: New file.
21230 * config.gcc (arm*-*-fuchsia*): Use it.
21232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21234 * config/arm/t-symbian: Rewrite for new option infrastructure.
21236 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21238 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21239 (MULTILIB_REQUIRED): Likewise.
21241 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21243 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21244 (MULTILIB_RESUE): Likewise.
21245 (MULTILIB_MATCHES): Likewise.
21246 (MULTLIB_REQUIRED): Likewise.
21248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21250 * config/arm/t-rtems: Rewrite for new option framework.
21252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21254 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21255 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21256 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21257 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21258 * config/arm/t-multilib: ... here.
21259 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21260 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21261 armv7-a and armv8*-a when A-profile libraries have not been built.
21262 * config/arm/t-rmprofile: Rewrite.
21264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21266 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21267 with a backslash. Remove the backslash after substituting unescaped
21269 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21273 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21274 the driver to pass through the default setting of -mfloat-abi.
21275 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21277 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21278 (all_feat_combs): New rule.
21279 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21281 * config/arm/t-aprofile: Rewrite.
21283 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21285 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21286 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21287 fpu is not specified by the user/command-line.
21288 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21289 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21290 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21291 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21292 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21293 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21297 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21298 * config/arm/t-arm-elf: Rewritten.
21300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21302 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21303 have some floating-point instructions.
21304 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21305 (TARGET_MAYBE_HARD_FLOAT): New macro.
21306 * config/arm/arm-builtins.c (arm_init_builtins): Use
21307 TARGET_MAYBE_HARD_FLOAT.
21308 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21312 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21313 (configargs.h): Include it.
21314 (arm_print_hint_for_fpu_option): New function.
21315 (arm_parse_fpu_option): New function.
21316 (candidate_extension): New class.
21317 (arm_canon_for_multilib): New function.
21318 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21319 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21320 (ARCH_CANONICAL_SPECS): New macro.
21321 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21325 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21326 are set after handling multilib fragments. Set target_cpu_default2
21329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21331 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21333 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21334 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21336 (check_arch): Likewise.
21337 * config/arm/arm.c (arm_configure_build_target): Handle
21338 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21339 options in the default.
21341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21343 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21344 when an option is an alias of another.
21345 * config/arm/parsecpu.awk (optalias): New parser token.
21346 (gen_comm_data): Mark non-alias options as such. Emit entries
21347 for extension aliases.
21348 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21349 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21350 (armv6kz, armv6zk, armv6t2): Likewise.
21351 (armv7): Make vfpv3-d16 an alias.
21352 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21354 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21355 Sort in canonical order.
21356 (armv8-a): Sort in canonical order.
21357 (armv8.1-a, armv8.2-a): Likewise.
21358 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21360 (cortex-a9): Sort in canonical order.
21361 * config/arm/arm.c (selftests.h): Include it.
21362 (arm_test_cpu_arch_data): New function.
21363 (arm_run_self_tests): New function.
21364 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21365 (targetm): Move declaration to the end of the file.
21366 * arm-cpu-cdata.h: Regenerated.
21368 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21370 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21371 call to target_mode_check describing the type of option passed.
21372 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21373 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21374 arm_parse_cpu_option_name to match parameters against list of
21376 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21377 arm_arch_core_flags data structure.
21378 * config/arm/arm-cpu_cdata.h: Regenerated.
21380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21382 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21384 (arm_print_hint_for_cpu_option): Likewise.
21385 (arm_print_hint_for_arch_option): Likewise.
21386 (arm_parse_cpu_option_name): Likewise.
21387 (arm_parse_arch_option_name): Likewise.
21388 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21389 of entries in the all_fpus list.
21390 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21391 (arm_parse_cpu_option_name): Declare.
21392 (arm_parse_arch_option_name): Declare.
21393 (arm_parse_option_features): Declare.
21394 (arm_intialize_isa): Declare.
21395 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21397 (gen_comm_data): ... here. Make definitions non-static.
21398 * config/arm/arm-cpu-data.h: Regenerated.
21399 * config/arm/arm-cpu-cdata.h: Regenerated.
21401 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21403 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21404 (cpu_arch_extension): New structure.
21405 (cpu_arch_option, arch_option, cpu_option): New structures.
21406 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21407 architecture types.
21408 (gen_data): Generate new format data tables.
21409 * config/arm/arm.c (cpu_tune): New structure.
21410 (cpu_option, processors): Delete.
21411 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21412 (arm_print_hint_for_cpu_option): ... this and ...
21413 (arm_print_hint_for_arch_option): ... this.
21414 (arm_parse_arch_cpu_name): Delete. Replace with ...
21415 (arm_parse_cpu_option_name): ... this and ...
21416 (arm_parse_arch_option_name): ... this.
21417 (arm_unrecognized_feature): Change type of target parameter to
21419 (arm_parse_arch_cpu_features): Delete. Replace with ...
21420 (arm_parse_option_features): ... this.
21421 (arm_configure_build_target): Rework to use new configuration data
21423 (arm_print_tune_info): Rework for new configuration data tables.
21424 * config/arm/arm-cpu-data.h: Regenerated.
21425 * config/arm/arm-cpu.h: Regenerated.
21427 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21429 * Makefile.in (OBJS): Move sbitmap.o from here ...
21430 (OBJS-libcommon): ... to here.
21432 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21434 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21435 (ISA_ALL_CRYPTO): New macro.
21436 (ISA_ALL_SIMD): New macro
21437 (ISA_ALL_FP): New macro.
21438 * config/arm/arm.c (fpu_bitlist): Update initializer.
21439 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21441 (arm9e): Add fpu. Add option for nofp
21442 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21443 (arm926ej-s, arm1026ej-s): Likewise.
21444 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21445 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21446 neon-fp16, neon-vfpv4, nofp and nosimd.
21447 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21448 (cortex-a8): Add fpu. Add option for nofp.
21449 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21450 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21451 (cortex-r4f): Add fpu.
21452 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21453 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21455 (cortex-r8): Likewise.
21456 (cortex-m4): Add fpu. Add option for nofp.
21457 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21458 (cortex-a17.cortex-a7): Likewise.
21459 (cortex-a32): Add fpu. Add options for crypto and nofp.
21460 (cortex-a35, cortex-a53): Likewise.
21461 (cortex-a57): Add fpu. Add option for crypto.
21462 (cortex-a72, cortex-a73): Likewise.
21463 (exynos-m1): Likewise.
21464 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21465 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21466 (cortex-m33): Add fpu. Add option for nofp.
21467 * config/arm/arm-cpu-cdata.h: Regenerated
21468 * config/arm/arm-cpu-data.h: Regenerated.
21470 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21472 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21473 (armv5te, armv5tej): Likewise.
21474 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21475 (armv7): Add options fp and vfpv3-d16.
21476 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21477 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21479 (armv7ve): Likewise.
21480 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21481 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21482 (armv8-a): Add nocrypto option.
21483 (armv8.1-a, armv8.2-a): Likewise.
21484 (armv8-m.main): add options fp, fp.dp and nofp.
21486 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21488 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21490 (armv8-a+crc): Delete.
21491 (armv8.1-a): Add options simd, crypto and nofp.
21492 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21493 (armv8.2-a+fp16): Delete.
21494 (armv8-m.main): Add option dsp.
21495 (armv8-m.main+dsp): Delete.
21496 (cortex-a8): Add fpu. Add nofp option.
21497 (cortex-a9): Add fpu. Add nofp and nosimd options.
21498 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21499 link to main cpu and architecture data structures.
21500 (gen_comm_data): Only put isa attributes from the main architecture
21502 (option): New statement for architecture and CPU entries.
21503 * arm.c (struct cpu_option): New structure.
21504 (struct processors): Add entry for options.
21505 (arm_unrecognized_feature): New function.
21506 (arm_parse_arch_cpu_name): Ignore any characters after the first
21508 (arm_parse_arch_cpu_feature): New function.
21509 (arm_configure_build_target): Separate out any CPU and architecture
21510 features and parse separately. Don't error out if -mfpu=auto is
21511 used with only an architecture string.
21512 (arm_print_asm_arch_directives): New function.
21513 (arm_file_start): Call it.
21514 * config/arm/arm-cpu-cdata.h: Regenerated.
21515 * config/arm/arm-cpu-data.h: Likewise.
21516 * config/arm/arm-tables.opt: Likewise.
21518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21520 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21521 assembler when it is not -mfpu=auto.
21523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21525 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21526 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21527 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21528 (ASM_CPU_SPEC): Rewrite.
21529 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21530 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21531 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21532 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21533 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21534 copied string is NUL-terminated. Also strip any characters prefixed
21536 (arm_rewrite_selected_arch): New function.
21537 (arm_rewrite_march): New function.
21539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21541 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21542 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21543 (march, mcpu, mtune): Convert to string-based options.
21544 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21545 (arm_parse_arch_cpu_name): New function.
21546 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21547 identify selected architecture or CPU.
21548 (arm_option_save): New function.
21549 (TARGET_OPTION_SAVE): Redefine.
21550 (arm_option_restore): Restore string options.
21551 (arm_option_print): Print string options.
21553 2017-06-16 Martin Sebor <msebor@redhat.com>
21555 PR tree-optimization/80933
21556 PR tree-optimization/80934
21557 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21558 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21559 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21560 (gimple_fold_builtin): Call them.
21562 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21564 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21565 as unlikely; update profile.
21567 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21569 * predict.c (force_edge_cold): Handle declaring edges impossible
21572 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21574 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21576 (try_unroll_loop_completely): Fix reporting.
21578 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21580 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21582 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21585 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21586 if given a non-constant argument for an intrinsic which requires a
21589 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21591 * profile.c (compare_freqs): New function.
21592 (branch_prob): Sort edge list.
21593 (find_spanning_tree): Assume that the list is priority sorted.
21595 2017-06-16 Richard Biener <rguenther@suse.de>
21597 PR tree-optimization/81090
21598 * passes.def (pass_record_bounds): Remove.
21599 * tree-pass.h (make_pass_record_bounds): Likewise.
21600 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21601 make_pass_record_bounds): Likewise.
21602 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21603 not free niter estimates at the beginning but at the end.
21604 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21606 2017-06-16 Richard Biener <rguenther@suse.de>
21608 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21609 initializer to workaround ICE in host GCC 4.8.
21611 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21613 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21615 (clone_inlined_nodes): Update.
21617 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21619 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21620 prefetch settings, and enable prefetching by default at -O3.
21622 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21624 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21625 Set flag_prefetch_loop_arrays according to tuning data.
21627 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21629 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21630 New tune structure.
21631 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21632 [Unrelated to main purpose of the patch] Place the pointer field last
21633 to enable type checking errors when tune structure are wrongly merged.
21634 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21635 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21636 (thunderx2t99_prefetch_tune): New tune constants.
21637 (tune_params *_tunings): Update all tunings (no functional change).
21638 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21639 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21640 from tunings structures.
21642 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21645 * ubsan.c (instrument_null): Add T argument, use it instead
21646 of computing it based on IS_LHS.
21647 (instrument_object_size): Likewise.
21648 (pass_ubsan::execute): Adjust instrument_null and
21649 instrument_object_size callers to pass gimple_get_lhs or
21650 gimple_assign_rhs1 result to it. Use instrument_null instead of
21651 calling get_base_address and instrument_mem_ref. Handle
21652 aggregate call arguments for object-size sanitization.
21654 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21656 PR tree-optimization/81089
21657 * tree-vrp.c (is_masked_range_test): Validate operands of
21660 2017-06-15 Martin Sebor <msebor@redhat.com>
21663 * dumpfile.c (dump_register): Avoid calling memset to initialize
21664 a class with a default ctor.
21665 * gcc.c (struct compiler): Remove const qualification.
21666 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21667 * hash-table.h: Ditto.
21668 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21670 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21671 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21673 * params.h (struct param_info): Make struct members non-const.
21674 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21675 with default initialization.
21676 * vec.h (vec_copy_construct, vec_default_construct): New helper
21678 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21679 with vec_copy_construct.
21680 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21681 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21682 * doc/invoke.texi (-Wclass-memaccess): Document.
21684 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21686 * emit-rtl.h (is_leaf): Update comment about local
21687 register allocator.
21689 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21692 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21693 for a variable to have a section before checking if the section has a
21695 Set section to.persistent if persistent attribute is set.
21696 Warn if .persistent attribute is used on an automatic variable.
21698 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21700 PR rtl-optimization/80474
21701 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21703 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21705 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21708 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21710 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21711 (casesi): Emit bounds checking as RTL.
21712 (casesi_internal_mips16_<mode>): Remove bounds checking.
21714 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21716 * config/xtensa/xtensa.c (xtensa_option_override): Append
21717 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21718 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21719 xtensa_doloop_hooks): Define unconditionally.
21720 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21722 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21723 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21724 for it in xtensa_option_override.
21725 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21726 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21728 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21730 * doc/cppopts.texi: Document '-' special value to -MF.
21732 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21734 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21735 (cortex_a53_fconst): Likewise.
21736 (cortex_a53_fpmul): Likewise.
21737 (cortex_a53_f_load_64): Likewise.
21738 (cortex_a53_f_load_many): Likewise.
21739 (cortex_a53_advsimd_alu): Likewise.
21740 (cortex_a53_advsimd_alu_q): Likewise.
21741 (cortex_a53_advsimd_mul): Likewise.
21742 (cortex_a53_advsimd_mul_q): Likewise.
21743 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21744 Add missing fmul, r2f_cvt and fconst cases.
21746 2017-06-14 Richard Biener <rguenther@suse.de>
21748 PR middle-end/81088
21749 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21751 (fold_binary_loc): When associating do not treat pre-existing
21752 TREE_OVERFLOW on literal constants as a reason to allow
21753 TREE_OVERFLOW on associated literal constants.
21755 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21757 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21758 (MASK_FEATURES): New macro.
21759 * config/sparc/sparc.c (sparc_option_override): Remove the special
21760 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21762 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21764 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21765 a division of 0 if non-call exceptions are enabled.
21767 2017-06-14 Andrew Pinski <apinski@cavium.com>
21768 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21771 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21772 Improve vector initialization code gen for only variable case.
21774 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21776 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21778 2017-06-14 Richard Biener <rguenther@suse.de>
21780 PR tree-optimization/81083
21781 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21784 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21786 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21787 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21788 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21789 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21790 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21791 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21793 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21795 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21796 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21798 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21800 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21802 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21804 * config/rs6000/t-rtems: Don't handle SPE.
21806 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21808 * config/rs6000/t-linux: Don't handle SPE.
21810 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21812 * config/rs6000/eabispe.h: Delete file.
21814 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21816 * config/rs6000/t-spe: Delete file.
21818 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21820 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21821 (rs6000_legitimate_offset_address_p): Return false for anything in
21822 V2SImode or V2SFmode.
21824 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21826 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21827 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21828 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21830 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21831 (rs6000_legitimate_offset_address_p): Ditto.
21832 (rs6000_emit_move): Ditto.
21833 (rs6000_init_builtins): Remove V4HI_type_node.
21835 2017-06-13 Martin Liska <mliska@suse.cz>
21838 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21839 (gate_asan): Likewise.
21840 * asan.h (asan_no_sanitize_address_p): Remove the function.
21841 (sanitize_flags_p): New function.
21842 * builtins.def: Fix coding style.
21843 * common.opt: Use renamed enum value.
21844 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21845 * doc/extend.texi: Document no_sanitize attribute.
21846 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21847 to SANITIZE_UNDEFINED_NONDEFAULT.
21848 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21849 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21850 Use sanitize_flags_p.
21851 * gimplify.c (gimplify_function_tree): Likewise.
21852 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21853 * opts.c (parse_no_sanitize_attribute): New function.
21854 (common_handle_option): Use renamed enum value.
21855 * opts.h (parse_no_sanitize_attribute): Declare.
21856 * tree.c (sanitize_flags_p): New function.
21857 * tree.h: Declared here.
21858 * tsan.c: Use sanitize_flags_p.
21859 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21860 (instrument_mem_ref): Likewise.
21861 (instrument_bool_enum_load): Likewise.
21862 (do_ubsan_in_current_function): Remove the function.
21863 (pass_ubsan::execute): Use sanitize_flags_p.
21864 * ubsan.h: Remove do_ubsan_in_current_function
21865 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21866 (dump_function_to_file): Use it here.
21868 2017-06-13 Martin Jambor <mjambor@suse.cz>
21870 PR tree-optimization/80803
21871 PR tree-optimization/81063
21872 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21873 (propagate_subaccesses_across_link): Enqueue subtree whenever
21874 necessary instead of relying on the caller.
21876 2017-06-13 Martin Jambor <mjambor@suse.cz>
21878 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21879 that have a first_link.
21880 (sort_and_splice_var_accesses): Do not check first_link before
21882 (subtree_mark_written_and_enqueue): Likewise.
21883 (propagate_all_subaccesses): Likewise and do not stop at first
21884 parent with a first_link.
21886 2017-06-13 Martin Jambor <mjambor@suse.cz>
21888 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21891 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21893 * match.pd: New pattern.
21895 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21897 * tree-vrp.c (is_masked_range_test): New function.
21898 (register_edge_assert_for): Determine ranges for
21901 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21903 PR tree-optimization/67328
21904 * fold-const.c (maskable_range_p): New function.
21905 (build_range_check): Generate bittests if possible.
21907 2017-06-13 Martin Liska <mliska@suse.cz>
21909 * gimple-pretty-print.c (dump_probability): Add new argument.
21910 (dump_edge_probability): Dump both probability and count.
21911 (dump_gimple_label): Likewise.
21912 (dump_gimple_bb_header): Likewise.
21914 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21917 * config/avr/avr-devices.c: Fix indentation.
21918 * config/avr/gen-avr-mmcu-specs.c: Dito.
21920 2017-06-13 Richard Biener <rguenther@suse.de>
21922 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21923 instead get vector type from stmt_info.
21924 (vectorizable_reduction): Adjust. Remove dead code.
21926 2017-06-13 Richard Biener <rguenther@suse.de>
21928 PR middle-end/81065
21929 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21930 case of C * (x * C2 + C3).
21931 (fold_addr_of_array_ref_difference): Properly fold index difference.
21933 2017-06-12 David S. Miller <davem@davemloft.net>
21936 * config/sparc/sparc.md (return expander): Emit frame blockage if
21937 function uses alloca.
21939 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21941 * combine.c (make_field_assignment): Check len rather than the mode
21942 precision when calling force_to_mode.
21944 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21946 Support multilibs and devices that see flash in RAM address range.
21949 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21950 (avr_mcu_t) <flash_pm_offset>: New field.
21951 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21952 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21953 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21954 (AVR_TINY_PM_OFFSET): Remove macro.
21955 * config/avr/avr.opt (-mshort-calls): New option.
21956 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21957 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21958 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21959 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21960 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21961 instead of avr_arch->have_jmp_call.
21962 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21963 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21964 avr_arch->flash_pm_offset to define.
21965 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21966 new field flash_pm_offset. Add entry for avrxmega3.
21967 (avr_texinfo): Add entry for avrxmega3.
21968 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21969 attiny212, attiny214,
21970 attiny412, attiny414, attiny416, attiny417,
21971 attiny814, attiny816, attiny817,
21972 attiny1614, attiny1616, attiny1617,
21973 attiny3214, attiny3216, attiny3217.
21974 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21975 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21976 (avr_print_operand_address) [AVR_TINY]: Same.
21977 (avr_asm_init_sections) <readonly_data_section>: Only patch
21978 callback if avr_arch->flash_pm_offset = 0.
21979 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21980 for rodata if avr_arch->flash_pm_offset != 0.
21981 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21982 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21983 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21984 (m_options): Append opt_rcall.
21985 (m_dirnames): Append dir_rcall.
21986 * config/avr/t-multilib: Regenerate.
21988 * configure.ac [target=avr]: Check whether avrxmega3 default
21989 linker description file works as needed.
21990 * configure: Regenerate.
21991 * doc/avr-mmcu.texi: Regenerate.
21992 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21993 <__AVR_ARCH__>: Document avrxmega3 and 103.
21994 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21995 <__AVR_SHORT_CALLS__>: Document it.
21996 <__AVR_PM_BASE_ADDRESS__>: Document it.
21997 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21998 (AVR Variable Attributes) <progmem>: Document this is
21999 not needed for avrxmega3.
22000 (AVR Named Address Spaces) <__flash>: Dito.
22002 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22004 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22006 2017-06-12 Doug Rupp <rupp@adacore.com>
22008 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22009 Append vxworks-stdint.h to the tm_file list.
22010 * config/vxworks-stdint.h: New file.
22012 2017-06-12 Martin Liska <mliska@suse.cz>
22014 PR tree-optimization/81041
22015 * tree-profile.c (gimple_gen_ic_func_profiler):
22016 Create an extra BB in profile-generate
22017 (gimple_gen_time_profiler): Likewise.
22019 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22021 PR tree-optimization/81003
22022 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22023 (update_range_test): Use it instead of force_gimple_operand_gsi.
22025 2017-06-12 Richard Biener <rguenther@suse.de>
22027 PR tree-optimization/81053
22028 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22029 with backedge value not defined in loop. Simplify def stmt
22032 2017-06-11 Tom de Vries <tom@codesourcery.com>
22035 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22037 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22038 nvptx_cannot_force_const_mem.
22040 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22042 * opts.c (finish_options): Move test for flag_split_stack after
22043 it has been initialized.
22045 2017-06-11 Jason Merrill <jason@redhat.com>
22047 * tree.h (id_equal): New.
22048 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22049 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22050 instead of strcmp of IDENTIFIER_POINTER.
22052 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22054 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22055 (mark_all_inlined_calls_cdtor): Fix formating.
22056 (inline_transform): Rescale profile before inlining.
22058 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22060 * cgraph.h (cgraph_edge::clone): Update prototype.
22061 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22062 (cgraph_node::create_clone): Update.
22063 (cgraph_node::create_version_clone): Update.
22064 * tree-inline.c (copy_bb): Update.
22065 (expand_call_inline): Update.
22067 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22069 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22070 factored out from ...
22071 (rs6000_emit_prologue): ... here.
22073 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22075 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22076 factored out from ...
22077 (rs6000_emit_prologue): ... here.
22079 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22081 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22083 (handle_missing_profiles): Fix computation of tp_first_run.
22084 (counts_to_freqs): Do not touch freqs when count is 0.
22086 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22088 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22091 2017-06-10 Tom de Vries <tom@codesourcery.com>
22093 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22094 attributes): Document signal effective target.
22096 2017-06-10 Tom de Vries <tom@codesourcery.com>
22098 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22099 Document effective target stack_size.
22101 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22103 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22104 to the edit_context if they can be auto-applied.
22106 2017-06-9 Ian Lance Taylor <iant@golang.org>
22108 * opts.c (finish_options): If -fsplit-stack, disable implicit
22109 -forder-blocks-and-partition.
22110 * doc/invoke.texi (Optimize Options): Document that when using
22111 -fsplit-stack -forder-blocks-and-partition is not implicitly
22114 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22116 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22117 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22118 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22119 * builtins.def (abort, trap, unreachable): Declare cold.
22120 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22121 * tree-core.h (ECF_COLD): New.
22122 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22123 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22125 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22127 * predict.c (unlikely_executed_stmt_p): Cleanup.
22129 2017-06-09 Richard Biener <rguenther@suse.de>
22131 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22132 model if the ref is always written to.
22134 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22136 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22138 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22140 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22143 2017-06-09 Tom de Vries <tom@codesourcery.com>
22146 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22147 "target cannot support label values" when encountering LABEL_REF.
22149 2017-06-09 Martin Liska <mliska@suse.cz>
22151 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22152 (gimple_gen_ic_func_profiler): Emit direct comparison
22153 of __gcov_indirect_call_callee with NULL.
22154 (gimple_gen_time_profiler): Change probability from
22155 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22157 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22159 * profile.c (edge_gcov_counts): Turn to pointer.
22160 (compute_branch_probabilities, compute_branch_probabilities): Update.
22161 (branch_prob): Do not clear edge_gcov_count.
22162 * profile.h (edge_gcov_counts): Turn to pointer.
22163 (edge_gcov_count): Update.
22165 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22167 * gimple.h (gimple_check_failed): Mark cold.
22169 2017-06-09 Richard Biener <rguenther@suse.de>
22171 PR tree-optimization/66623
22172 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22173 refactor check_reduction into two parts, properly computing
22174 whether we have to check reduction validity for outer loop
22177 2017-06-09 Richard Biener <rguenther@suse.de>
22179 PR tree-optimization/79483
22180 * graphite-scop-detection.c (order): New global.
22181 (get_order): Compute bb to order mapping that satisfies code
22182 generation constraints.
22183 (cmp_pbbs): New helper.
22184 (build_scops): Start domwalk at entry block, sort generated
22187 2017-06-09 Richard Biener <rguenther@suse.de>
22189 PR middle-end/81007
22190 * ipa-polymorphic-call.c
22191 (ipa_polymorphic_call_context::restrict_to_inner_class):
22192 Skip FIELD_DECLs with error_mark_node type.
22193 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22196 2017-06-09 Martin Liska <mliska@suse.cz>
22198 * predict.c (struct branch_predictor): New struct.
22199 (test_prediction_value_range): New test.
22200 (predict_c_tests): New function.
22201 * selftest-run-tests.c (selftest::run_tests): Run the function.
22202 * selftest.h: Declare new tests.
22204 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22207 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22208 gen_add3_insn did not fail.
22209 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22210 r0, construct that number in a temporary reg and add that reg to r0.
22211 If asked to put the result in r0 as well, fail.
22213 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22215 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22216 for early expansion of vec_eqv.
22218 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22220 PR middle-end/81005
22221 * ubsan.c (instrument_null): Avoid pointless code temporary.
22222 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22224 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22228 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22231 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22234 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22236 * predict.c (unlikely_executed_edge_p): Move ahead.
22237 (probably_never_executed_edge_p): Use it.
22239 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22241 PR middle-end/79988
22242 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22243 gimple_call_builtin_p call.
22245 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22247 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22248 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22249 rtl_check_failed_type2, rtl_check_failed_code1,
22250 rtl_check_failed_code2, rtl_check_failed_code_mode,
22251 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22252 rtvec_check_failed_bounds, rtl_check_failed_flag,
22253 _fatal_insn_not_found, _fatal_insn): Likewise.
22254 * tree.h (tree_contains_struct_check_failed,
22255 tree_check_failed, tree_not_check_failed,
22256 tree_class_check_failed, tree_range_check_failed,
22257 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22258 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22259 tree_operand_check_failed, omp_clause_check_failed,
22260 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22263 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22265 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22266 flag_branch_probabilities.
22267 * ipa-inline.c (edge_badness): Likewise.
22268 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22269 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22270 * predict.c (maybe_hot_frequency_p): Likewise.
22271 (probably_never_executed): Likewise.
22272 * sched-ebb.c (schedule_ebbs): Likewise.
22273 * sched-rgn.c (find_single_block_region): Likewise.
22274 * tracer.c (tail_duplicate): Likewise.
22276 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22278 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22279 longer requires x_flag_profile_use.
22281 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22283 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22284 instead of flag_reorder_blocks_and_partition.
22285 * dbxout.c (dbxout_function_end): Likewise.
22286 * dwarf2out.c (gen_subprogram_die): Likewise.
22287 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22288 * hw-doloop.c (reorg_loops): Likewise.
22289 * varasm.c (assemble_start_function,
22290 assemble_end_function): Likewise.
22291 (decide_function_section): Do not check for
22292 flag_reorder_blocks_and_partition.
22294 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22296 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22298 (chkp_get_hard_register_fake_addr_expr): Ditto.
22299 (chkp_build_addr_expr): Add check for hard reg case.
22300 (chkp_parse_array_and_component_ref): Ditto.
22301 (chkp_find_bounds_1): Ditto.
22302 (chkp_process_stmt): Don't generate bounds store for
22305 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22307 * predict.c (maybe_hot_bb_p): Do not check profile status.
22308 (maybe_hot_edge_p): Likewise.
22309 (probably_never_executed): Check for zero counts even if profile
22311 (unlikely_executed_edge_p): New function.
22312 (unlikely_executed_stmt_p): New function.
22313 (unlikely_executed_bb_p): New function.
22314 (set_even_probabilities): Use unlikely predicates.
22315 (combine_predictions_for_bb): Likewise.
22316 (predict_paths_for_bb): Likewise.
22317 (predict_paths_leading_to_edge): Likewise.
22318 (determine_unlikely_bbs): New function.
22319 (estimate_bb_frequencies): Use it.
22320 (compute_function_frequency): Use zero counts even if profile is
22322 * profile-count.h: Fix typo.
22324 2017-08-08 Julia Koval <julia.koval@intel.com>
22326 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22327 _mm512_mask_cvtsepi16_storeu_epi8,
22328 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22329 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22330 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22331 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22332 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22333 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22334 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22335 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22336 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22337 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22338 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22339 __builtin_ia32_pmovuswb256mem_mask,
22340 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22341 __builtin_ia32_pmovwb512mem_mask): New builtins.
22343 2017-08-08 Julia Koval <julia.koval@intel.com>
22345 PR target/73350,80862
22346 * config/i386/subst.md (round): Fix round pattern.
22347 * config/i386/i386.c (ix86_erase_embedded_rounding):
22348 Fix erasing rounding for the fixed pattern.
22350 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22352 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22354 2017-06-08 Martin Liska <mliska@suse.cz>
22356 PR gcov-profile/80911
22357 * gcov.c (block_info::block_info): New constructor.
22359 2017-06-07 Carl Love <cel@us.ibm.com>
22361 * config/rs6000/rs6000-c: The return type of the following
22362 built-in functions was implemented as int not long long. Fix sign
22363 of return value for the unsigned version of vec_mulo and vec_mule.
22364 vector unsigned long long vec_bperm (vector unsigned long long,
22365 vector unsigned char)
22366 vector signed long long vec_mule (vector signed int,
22368 vector unsigned long long vec_mule (vector unsigned int,
22369 vector unsigned int)
22370 vector signed long long vec_mulo (vector signed int,
22372 vector unsigned long long vec_mulo (vector unsigned int,
22373 vector unsigned int)
22374 * doc/extend.texi: Fix the documentation for the built-in
22377 2017-06-07 Carl Love <cel@us.ibm.com>
22380 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22383 2017-06-07 Carl Love <cel@us.ibm.com>
22385 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22386 support, Generate doublehv for signed int/float for BE case only.
22388 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22390 * doc/invoke.texi (mcx16): Rewrite.
22392 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22394 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22395 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22396 *mov<mode>_softfloat, and an anonymous splitter): Use
22397 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22399 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22401 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22403 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22404 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22405 (rs6000_debug_reg_global): Adjust.
22406 (rs6000_init_hard_regno_mode_ok): Adjust.
22407 (rs6000_dbx_register_number): Adjust.
22408 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22409 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22410 Remove SPE_ACC and SPEFSCR.
22411 (REG_ALLOC_ORDER): Ditto.
22412 (FRAME_POINTER_REGNUM): Change to 111.
22413 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22414 (REG_CLASS_NAMES): Ditto.
22415 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22416 (REGISTER_NAMES): Ditto.
22417 (ADDITIONAL_REG_NAMES): Ditto.
22418 (rs6000_reg_names): Ditto.
22419 * config/rs6000/rs6000.md: Renumber some register number
22422 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22424 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22426 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22427 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22429 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22430 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22431 Delete the SPE high registers.
22432 (REG_ALLOC_ORDER): Ditto.
22433 (enum reg_class): Remove SPE_HIGH_REGS.
22434 (REG_CLASS_NAMES): Ditto.
22435 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22436 (REGISTER_NAMES): Ditto.
22437 (rs6000_reg_names): Ditto.
22438 * doc/tm.texi.in: Remove SPE as example.
22439 * doc/tm.texi: Regenerate.
22441 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22443 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22444 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22445 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22446 * config/rs6000/rs6000.md (type): Remove "brinc".
22448 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22450 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22451 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22452 * config/rs6000/linuxspe.h: Delete file.
22453 * config/rs6000/rs6000.md: Don't include spe.md.
22454 * config/rs6000/spe.h: Delete file.
22455 * config/rs6000/spe.md: Delete file.
22456 * config/rs6000/t-rs6000: Remove spe.md.
22458 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22460 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22461 (reg_or_none500mem_operand): Delete.
22462 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22463 instead of reg_or_none500mem_operand.
22465 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22467 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22468 handling of SPE flags.
22469 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22471 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22473 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22475 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22476 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22477 paired_divv2sf3): Similar.
22478 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22479 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22480 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22482 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22483 Rename the paired_* instruction patterns.
22484 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22486 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22487 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22488 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22489 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22490 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22491 PAIRED_VECTOR_MODE.
22492 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22493 (spe_func_has_64bit_regs_p): Delete.
22494 (spe_expand_predicate_builtin): Delete.
22495 (spe_expand_evsel_builtin): Delete.
22496 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22497 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22498 (invalid_e500_subreg): Delete.
22499 (rs6000_legitimize_address): Always force_reg op2 as well, for
22500 paired single memory accesses.
22501 (rs6000_member_type_forces_blk): Delete.
22502 (rs6000_spe_function_arg): Delete.
22503 (rs6000_expand_unop_builtin): Delete SPE handling.
22504 (rs6000_expand_binop_builtin): Ditto.
22505 (spe_expand_stv_builtin): Delete.
22506 (bdesc_2arg_spe): Delete.
22507 (spe_expand_builtin): Delete.
22508 (spe_expand_predicate_builtin): Delete.
22509 (spe_expand_evsel_builtin): Delete.
22510 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22511 (spe_init_builtins): Delete.
22512 (spe_func_has_64bit_regs_p): Delete.
22513 (savres_routine_name): Delete "info" parameter. Adjust callers.
22514 (rs6000_emit_stack_reset): Ditto.
22515 (rs6000_dwarf_register_span): Delete.
22516 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22517 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22518 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22520 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22522 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22523 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22524 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22525 mulv2sf3, divv2sf3): Delete expanders.
22527 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22529 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22531 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22533 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22534 * config/rs6000/rs6000.c: Ditto.
22536 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22538 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22539 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22540 comparison_operator.
22542 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22544 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22545 * config/rs6000/rs6000.opt: Ditto.
22546 * config/rs6000/t-rtems: Ditto.
22548 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22550 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22551 TARGET_E500_SINGLE by 0, simplify.
22552 * config/rs6000/rs6000.c: Ditto.
22553 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22554 (spe_build_register_parallel): Delete.
22555 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22556 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22557 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22558 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22559 (E500_CONVERT): Delete.
22560 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22562 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22564 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22565 * config/rs6000/dfp.md: Ditto.
22566 (negdd2, *negdd2_fpr): Merge.
22567 (absdd2, *absdd2_fpr): Merge.
22568 (negtd2, *negtd2_fpr): Merge.
22569 (abstd2, *abstd2_fpr): Merge.
22570 * config/rs6000/e500.h: Delete file.
22571 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22572 TARGET_FPRS by 1 and simplify.
22573 * config/rs6000/rs6000-c.c: Ditto.
22574 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22575 TARGET_DF_SPE by 0.
22576 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22578 * config/rs6000/rs6000.md: Ditto.
22579 (floatdidf2, *floatdidf2_fpr): Merge.
22580 (move_from_CR_gt_bit): Delete.
22581 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22582 (E500_CR_IOR_COMPARE): Delete.
22583 (All patterns that require !TARGET_FPRS): Delete.
22584 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22586 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22588 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22590 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22592 * graphds.c (add_edge): Intitialize edge's attached data.
22593 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22594 pointer parameter. Call pointed function on each edge during
22595 graph traversing. Skip traversing the edge when the function
22597 (graphds_dfs, graphds_scc): Ditto.
22598 (for_each_edge): New parameter. Pass the new parameter to callback
22600 * graphds.h (skip_edge_callback): New function pointer type.
22601 (graphds_dfs, graphds_scc): New function pointer parameter.
22602 (graphds_edge_callback, for_each_edge): New parameter.
22604 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22606 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22607 out code checking if runtime alias check is possible to below ...
22608 Call the new function.
22609 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22610 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22612 2017-06-07 Marek Polacek <polacek@redhat.com>
22615 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22616 TYPE_OVERFLOW_WRAPS check.
22618 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22620 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22621 if versioning is required.
22622 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22623 peeling with the check for versioning.
22625 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22627 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22628 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22629 Set true to new parameter if new ssa variable is defined.
22630 (vect_gen_vector_loop_niters): Refactor. Set range information
22631 for the new vector loop bound variable.
22632 (vect_do_peeling): Ditto.
22634 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22636 * tree-affine.c (ssa.h): Include header file.
22637 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22638 has wrapping overflow behavior.
22640 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22642 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22644 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22646 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22647 (tree_to_aff_combination): ... here.
22649 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22651 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22652 reg_pressure model function.
22653 (ivopts_global_cost_for_size): Delete.
22654 (determine_set_costs, iv_ca_recount_cost): Call new model function
22655 ivopts_estimate_reg_pressure.
22657 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22659 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22660 expensive than udiv. Remove floating point cases from mod.
22662 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22664 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22665 Increase idiv cost.
22667 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22669 * config/aarch64/aarch64.md
22670 (copysignsf3): Fix mask generation.
22672 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22674 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22676 (class dump_manager): Add register_dumps method.
22677 * dumpfile.c: Include langhooks.h.
22678 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22679 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22680 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22681 (dump_manager::dump_register): Start with 512 entries instead of 32.
22682 (dump_manager::register_dumps): New method.
22683 * toplev.c (general_init): Instead of invoking register_dumps
22684 langhook, invoke register_dumps method on the dump manager.
22685 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22688 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22690 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22691 Say that # requires an associated define_split to exist, and that
22692 the define_split must be suitable for use after register allocation.
22694 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22696 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22697 (compute_outgoing_frequencies): Also initialize zero counts.
22698 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22699 around loops; preserve more of profile when nothing changes.
22701 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22703 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22705 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22706 * config/arm/arm-cpu-cdata.h: Regenerate.
22707 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22708 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22709 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22710 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22711 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22713 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22715 * config/arm/t-rmprofile: Likewise.
22716 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22718 2017-06-06 David S. Miller <davem@davemloft.net>
22721 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22722 blockage if function uses alloca.
22724 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22726 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22727 New "uid" fields to hold pretty-print IDs of group and ref.
22728 Memory references are now identified as <group_id>:<ref_id>
22729 instead of using [random] addresses.
22730 (dump_mem_details): Simplify, no functional change.
22731 (dump_mem_ref): Simplify and make output more concise.
22732 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22733 (find_or_create_group): Initialize group uid.
22734 (record_ref): Initialize ref uid. Improve debug output.
22735 (prune_group_by_reuse, should_issue_prefetch_p,)
22736 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22737 (mark_nontemporal_store, determine_loop_nest_reuse):
22738 Improve debug output.
22740 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22742 * dbgcnt.def (prefetch): New debug counter.
22743 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22744 (schedule_prefetches): Stop issueing prefetches if debug counter
22747 2017-06-06 Tom de Vries <tom@codesourcery.com>
22749 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22750 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22752 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22754 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22755 Use aarch64_reg_or_zero predicate for operand 4.
22756 (aarch64_compare_and_swap<mode> define_insn_and_split):
22757 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22758 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22760 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22762 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22763 (arm_compute_save_core_reg_mask): This.
22764 (thumb1_compute_save_reg_mask): Rename into ...
22765 (thumb1_compute_save_core_reg_mask): This.
22766 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22767 (arm_compute_frame_layout): Likewise.
22769 2017-06-06 Richard Biener <rguenther@suse.de>
22771 PR tree-optimization/80974
22772 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22773 keep or clear leaders SSA info.
22775 2017-06-06 Tom de Vries <tom@codesourcery.com>
22777 * config/nvptx/nvptx.c (split_mode_p): New function.
22778 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22780 2017-06-06 Tom de Vries <tom@codesourcery.com>
22782 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22784 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22787 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22790 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22792 * shrink-wrap.c (handle_simple_exit): Update profile.
22793 (try_shrink_wrapping): Upate profile.
22795 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22797 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22798 (tree_guess_outgoing_edge_probabilities): New.
22799 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22800 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22802 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22804 * ipa-split.c (split_function): Initialize return bb profile.
22806 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22808 * profile.c (compute_branch_probabilities): Also initialize
22809 EXIT_BLOCK profile.
22811 2017-06-06 Richard Biener <rguenther@suse.de>
22813 PR tree-optimization/80928
22814 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22815 (vect_analyze_loop_operations): Properly guard analysis for
22817 (vect_transform_loop): Likewise.
22818 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22819 (vect_model_induction_cost): Do not cost for pure SLP.
22820 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22821 of induction in inner loop vectorization.
22822 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22823 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22824 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22826 (vect_analyze_slp_cost_1): Cost induction.
22827 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22828 (vect_get_slp_vect_defs): Likewise.
22829 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22830 (vect_transform_stmt): Handle SLP reductions.
22831 * tree-vectorizer.h (vectorizable_induction): Adjust.
22833 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22835 * config/rs6000/rs6000.c (make_resolver_func): Update
22836 init_lowered_empty_function call.
22838 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22840 * doc/invoke.texi: Document the -fprofile-abs-path option.
22841 * common.opt (fprofile-abs-path): New option.
22842 * gcov-io.h (gcov_write_filename): Declare.
22843 * gcov-io.c (gcov_write_filename): New function.
22844 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22845 * profile.c (output_location): Likewise.
22847 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22849 * shring-wrap.c: Revert accidental commit.
22851 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22853 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22855 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22857 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22859 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22860 profile in callgraph edge.
22861 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22862 (apply_scale): Likewise.
22863 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22864 Also scale profile when inlining function with zero profile.
22865 (initialize_cfun): Update exit block profile even when it is zero.
22866 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22867 when profile is read.
22869 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22871 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22872 (CLONE_*): New constants to define the processors we can generate
22873 code for with the target_clone attribute.
22874 (rs6000_clone_map): New array to identify which clone processors
22875 the current program is running on.
22876 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22877 target_clone attribute.
22878 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22879 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22880 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22881 (cpu_expand_builtin): Add support for target_clone attribute.
22882 (rs6000_valid_attribute_p): Allow "default" attribute.
22883 (get_decl_name): New debug function to simplify printing the
22884 current function name in debugging statements.
22885 (rs6000_clone_priority): New functions to support the target_clone
22886 attribute, and be able to generate code to switch between ISA 2.05
22887 through ISA 3.0 (power6 through power9).
22888 (rs6000_compare_version_priority): Likewise.
22889 (rs6000_get_function_versions_dispatcher): Likewise.
22890 (make_resolver_func): Likewise.
22891 (add_condition_to_bb): Likewise.
22892 (dispatch_function_versions): Likewise.
22893 (rs6000_generate_version_dispatcher_body): Likewise.
22894 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22895 (fusion_gpr_load_p): Fix a spacing issue.
22896 * doc/extend.texi (Common Function Attributes): Document that the
22897 PowerPC supports the target_clone attribute.
22899 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22901 * config/arm/arm.h: explain F symbol found in description of ARM
22902 register allocation in its legend.
22904 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22906 * config/mips/frame-header-opt.c: Include profile-count.h.
22907 * config/riscv/riscv.c: Include profile-count.h
22909 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22911 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22913 (sm_set_flag_if_changed): Add bbs field.
22914 (execute_sm_if_changed_flag_set): Pass BBS.
22915 (execute_sm): Update.
22917 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22919 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22922 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22924 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22925 (peephole2): New peephole2 to emit the above.
22926 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22928 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22930 * config/aarch64/aarch64.c (define_peephole2 above
22931 *sub_<shift>_<mode>): New peephole.
22933 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22935 * config/i386/i386.c (make_resolver_func): Update.
22936 * Makefile.in: Add profile-count.h and profile-count.o
22937 * auto-profile.c (afdo_indirect_call): Update to new API.
22938 (afdo_set_bb_count): Update.
22939 (afdo_propagate_edge): Update.
22940 (afdo_propagate_circuit): Update.
22941 (afdo_calculate_branch_prob): Update.
22942 (afdo_annotate_cfg): Update.
22943 * basic-block.h: Include profile-count.h
22944 (struct edge_def): Turn count to profile_count.
22945 (struct basic_block_def): Likewie.
22946 (REG_BR_PROB_BASE): Move to profile-count.h
22947 (RDIV): Move to profile-count.h
22948 * bb-reorder.c (max_entry_count): Turn to profile_count.
22949 (find_traces): Update.
22950 (rotate_loop):Update.
22951 (connect_traces):Update.
22952 (sanitize_hot_paths):Update.
22953 * bt-load.c (migrate_btr_defs): Update.
22954 * cfg.c (RDIV): Remove.
22955 (init_flow): Use alloc_block.
22956 (alloc_block): Uninitialize count.
22957 (unchecked_make_edge): Uninitialize count.
22958 (check_bb_profile): Update.
22959 (dump_edge_info): Update.
22960 (dump_bb_info): Update.
22961 (update_bb_profile_for_threading): Update.
22962 (scale_bbs_frequencies_int): Update.
22963 (scale_bbs_frequencies_gcov_type): Update.
22964 (scale_bbs_frequencies_profile_count): New.
22965 * cfg.h (update_bb_profile_for_threading): Update.
22966 (scale_bbs_frequencies_profile_count): Declare.
22967 * cfgbuild.c (compute_outgoing_frequencies): Update.
22968 (find_many_sub_basic_blocks): Update.
22969 * cfgcleanup.c (try_forward_edges): Update.
22970 (try_crossjump_to_edge): Update.
22971 * cfgexpand.c (expand_gimple_tailcall): Update.
22972 (construct_exit_block): Update.
22973 * cfghooks.c (verify_flow_info): Update.
22974 (dump_bb_for_graph): Update.
22975 (split_edge): Update.
22976 (make_forwarder_block): Update.
22977 (duplicate_block): Update.
22978 (account_profile_record): Update.
22979 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22980 (get_estimated_loop_iterations): Update.
22981 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22982 (single_likely_exit): Update.
22983 * cfgloopmanip.c (scale_loop_profile): Update.
22985 (set_zero_probability): Update.
22986 (lv_adjust_loop_entry_edge): Update.
22987 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22988 (purge_dead_edges): Update.
22989 (rtl_account_profile_record): Update.
22990 * cgraph.c (cgraph_node::create): Uninitialize count.
22991 (symbol_table::create_edge): Uninitialize count.
22992 (cgraph_update_edges_for_call_stmt_node): Update.
22993 (cgraph_edge::dump_edge_flags): Update.
22994 (cgraph_node::dump): Update.
22995 (cgraph_edge::maybe_hot_p): Update.
22996 * cgraph.h: Include profile-count.h
22997 (create_clone), create_edge, create_indirect_edge): Update.
22998 (cgraph_node): Turn count to profile_count.
22999 (cgraph_edge0: Likewise.
23000 (make_speculative, clone): Update.
23001 (create_edge): Update.
23002 (init_lowered_empty_function): Update.
23003 * cgraphclones.c (cgraph_edge::clone): Update.
23004 (duplicate_thunk_for_node): Update.
23005 (cgraph_node::create_clone): Update.
23006 * cgraphunit.c (cgraph_node::analyze): Update.
23007 (cgraph_node::expand_thunk): Update.
23008 * final.c (dump_basic_block_info): Update.
23009 * gimple-streamer-in.c (input_bb): Update.
23010 * gimple-streamer-out.c (output_bb): Update.
23011 * graphite.c (print_global_statistics): Update.
23012 (print_graphite_scop_statistics): Update.
23013 * hsa-brig.c: Include basic-block.h.
23014 * hsa-dump.c: Include basic-block.h.
23015 * hsa-gen.c (T sum_slice): Update.
23016 (convert_switch_statements):Update.
23017 * hsa-regalloc.c: Include basic-block.h.
23018 * ipa-chkp.c (chkp_produce_thunks): Update.
23019 * ipa-cp.c (struct caller_statistics): Update.
23020 (init_caller_stats): Update.
23021 (gather_caller_stats): Update.
23022 (ipcp_cloning_candidate_p): Update.
23023 (good_cloning_opportunity_p): Update.
23024 (get_info_about_necessary_edges): Update.
23025 (dump_profile_updates): Update.
23026 (update_profiling_info): Update.
23027 (update_specialized_profile): Update.
23028 (perhaps_add_new_callers): Update.
23029 (decide_about_value): Update.
23030 (ipa_cp_c_finalize): Update.
23031 * ipa-devirt.c (struct odr_type_warn_count): Update.
23032 (struct decl_warn_count): Update.
23033 (struct final_warning_record): Update.
23034 (possible_polymorphic_call_targets): Update.
23035 (ipa_devirt): Update.
23036 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23037 * ipa-icf.c (sem_function::merge): Update.
23038 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23039 * ipa-inline.c (compute_uninlined_call_time): Update.
23040 (compute_inlined_call_time): Update.
23041 (want_inline_small_function_p): Update.
23042 (want_inline_self_recursive_call_p): Update.
23043 (edge_badness): Update.
23044 (lookup_recursive_calls): Update.
23045 (recursive_inlining): Update.
23046 (inline_small_functions): Update.
23047 (dump_overall_stats): Update.
23048 (dump_inline_stats): Update.
23049 * ipa-profile.c (ipa_profile_generate_summary): Update.
23050 (ipa_propagate_frequency): Update.
23051 (ipa_profile): Update.
23052 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23053 * ipa-utils.c (ipa_merge_profiles): Update.
23054 * loop-doloop.c (doloop_modify): Update.
23055 * loop-unroll.c (report_unroll): Update.
23056 (unroll_loop_runtime_iterations): Update.
23057 * lto-cgraph.c (lto_output_edge): Update.
23058 (lto_output_node): Update.
23059 (input_node): Update.
23060 (input_edge): Update.
23061 (merge_profile_summaries): Update.
23062 * lto-streamer-in.c (input_cfg): Update.
23063 * lto-streamer-out.c (output_cfg): Update.
23064 * mcf.c (create_fixup_graph): Update.
23065 (adjust_cfg_counts): Update.
23066 (sum_edge_counts): Update.
23067 * modulo-sched.c (sms_schedule): Update.
23068 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23069 * predict.c (maybe_hot_count_p): Update.
23070 (probably_never_executed): Update.
23071 (dump_prediction): Update.
23072 (combine_predictions_for_bb): Update.
23073 (propagate_freq): Update.
23074 (handle_missing_profiles): Update.
23075 (counts_to_freqs): Update.
23076 (rebuild_frequencies): Update.
23077 (force_edge_cold): Update.
23078 * predict.h: Include profile-count.h
23079 (maybe_hot_count_p, counts_to_freqs): UPdate.
23080 * print-rtl-function.c: Do not include cfg.h
23081 * print-rtl.c: Include basic-block.h
23082 * profile-count.c: New file.
23083 * profile-count.h: New file.
23084 * profile.c (is_edge_inconsistent): Update.
23085 (correct_negative_edge_counts): Update.
23086 (is_inconsistent): Update.
23087 (set_bb_counts): Update.
23088 (read_profile_edge_counts): Update.
23089 (compute_frequency_overlap): Update.
23090 (compute_branch_probabilities): Update; Initialize and deinitialize
23092 (branch_prob): Update.
23093 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23094 (edge_gcov_count): New.
23095 (bb_gcov_count): New.
23096 * shrink-wrap.c (try_shrink_wrapping): Update.
23097 * tracer.c (better_p): Update.
23098 * trans-mem.c (expand_transaction): Update.
23099 (ipa_tm_insert_irr_call): Update.
23100 (ipa_tm_insert_gettmclone_call): Update.
23101 * tree-call-cdce.c: Update.
23102 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23103 (gimple_duplicate_sese_tail): Update.
23104 (gimple_account_profile_record): Update.
23105 (execute_fixup_cfg): Update.
23106 * tree-inline.c (copy_bb): Update.
23107 (copy_edges_for_bb): Update.
23108 (initialize_cfun): Update.
23109 (freqs_to_counts): Update.
23110 (copy_cfg_body): Update.
23111 (expand_call_inline): Update.
23112 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23113 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23114 (try_unroll_loop_completely): Update.
23115 (try_peel_loop): Update.
23116 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23117 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23118 * tree-ssa-loop-split.c (connect_loops): Update.
23119 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23120 * tree-ssa-reassoc.c (branch_fixup): Update.
23121 * tree-ssa-tail-merge.c (replace_block_by): Update.
23122 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23123 (compute_path_counts): Update.
23124 (update_profile): Update.
23125 (recompute_probabilities): Update.
23126 (update_joiner_offpath_counts): Update.
23127 (estimated_freqs_path): Update.
23128 (freqs_to_counts_path): Update.
23129 (clear_counts_path): Update.
23130 (ssa_fix_duplicate_block_edges): Update.
23131 (duplicate_thread_path): Update.
23132 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23133 (struct switch_conv_info): Update.
23134 * tree-tailcall.c (decrease_profile): Update.
23135 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23136 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23137 * value-prof.c (check_counter): Update.
23138 (gimple_divmod_fixed_value): Update.
23139 (gimple_mod_pow2): Update.
23140 (gimple_mod_subtract): Update.
23141 (gimple_ic_transform): Update.
23142 (gimple_stringop_fixed_value): Update.
23143 * value-prof.h (gimple_ic): Update.
23145 2017-06-02 Carl Love <cel@us.ibm.com>
23147 * config/rs6000/rs6000-c: Add support for built-in functions
23148 vector double vec_doublee (vector signed int);
23149 vector double vec_doublee (vector unsigned int);
23150 vector double vec_doublee (vector float);
23151 vector double vec_doubleh (vector signed int);
23152 vector double vec_doubleh (vector unsigned int);
23153 vector double vec_doubleh (vector float);
23154 vector double vec_doublel (vector signed int);
23155 vector double vec_doublel (vector unsigned int);
23156 vector double vec_doublel (vector float);
23157 vector double vec_doubleo (vector signed int);
23158 vector double vec_doubleo (vector unsigned int);
23159 vector double vec_doubleo (vector float);.
23160 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23161 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23163 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23164 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23165 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23167 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23168 vec_doublel, vec_doubleh.
23169 * doc/extend.texi: Update the built-in documentation file for the
23170 new built-in functions.
23172 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23175 * ipa-inline-analysis.c (free_growth_caches): Set
23176 edge_removal_hook_holder to NULL after removing it.
23178 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23180 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23181 comparision with zero.
23183 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23184 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23185 for early expansion of vec_min and vec_max builtins.
23186 (builtin_function_type): Add min/max unsigned variants to those
23187 identified as having unsigned arguments.
23189 2017-06-02 Olivier Hainque <hainque@adacore.com>
23191 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23193 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23195 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23196 Use VALL_F16 iterator rather than VALL.
23198 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23200 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23201 Emit CBNZ inside loop when doing a strong exchange and comparing
23202 against zero. Generate the CC flags after the loop.
23204 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23206 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23207 (dl_section_ref): New.
23208 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23209 On AIX, append an expression to subtract the size of the
23210 section length to dl_section_ref.
23212 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23214 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23215 for early expansion of vector absolute builtins.
23217 2017-06-02 Richard Biener <rguenther@suse.de>
23219 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23220 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23222 2017-06-02 Richard Biener <rguenther@suse.de>
23224 PR tree-optimization/80948
23225 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23226 stmt order as well.
23228 2017-06-02 Richard Biener <rguenther@suse.de>
23230 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23232 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23233 for inductions as relevant.
23235 2017-06-02 Richard Biener <rguenther@suse.de>
23237 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23238 (vectorizable_induction): ... this. Remove dead code.
23240 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23242 * builtins. (expand_builtin_alloca): Remove second parameter and
23243 infer its value from the first parameter instead.
23244 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23246 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23248 PR rtl-optimization/80903
23249 * loop-doloop.c (add_test): Unshare sequence.
23251 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23253 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23255 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23257 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23259 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23260 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23261 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23262 (xlogue_layout::get_stub_rtx): Make static.
23263 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23264 (xlogue_layout::compute_stub_managed_regs): Rename to...
23265 (xlogue_layout::count_stub_managed_regs): ...this.
23266 (xlogue_layout::is_stub_managed_reg): New function.
23267 (xlogue_layout::m_stub_names): Rename to...
23268 (xlogue_layout::s_stub_names): ...this, make static.
23269 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23270 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23271 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23272 xlogue_layout::s_stub_names): Instantiate statics.
23273 (stub_managed_regs): Remove.
23274 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23275 (disable_call_ms2sysv_xlogues): Rename to...
23276 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23277 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23279 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23280 change after reload_completed.
23281 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23283 (ix86_expand_prologue): Likewise.
23284 (ix86_expand_epilogue): Likewise.
23285 (ix86_expand_split_stack_prologue): Likewise.
23286 (ix86_compute_frame_layout): Remove frame parameter ...
23287 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23288 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23290 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23291 (ix86_frame): Move from here ...
23292 * config/i386/i386.h (ix86_frame): ... to here.
23293 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23294 complete ix86_frame data structure instead. Remove some_ld_name.
23296 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23298 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23299 symbols that hold a DECL_VALUE_EXPR.
23301 2017-06-01 Martin Jambor <mjambor@suse.cz>
23303 PR tree-optimization/80898
23304 * tree-sra.c (process_subtree_disqualification): Removed.
23305 (disqualify_candidate): Do not acll
23306 process_subtree_disqualification.
23307 (subtree_mark_written_and_enqueue): New function.
23308 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23309 RHS has been disqualified and re-queue LHS if necessary. Apart
23310 from that, ignore disqualified RHS.
23312 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23314 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23315 address fetch for z10 or later.
23317 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23319 * config/arc/arc.md (tst_movb): Add guard when splitting.
23321 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23323 * config/arc/arc.c (arc_can_eliminate): Test against
23324 arc_frame_pointer_needed.
23326 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23328 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23329 to prevent store reordering.
23330 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23331 (type): Add block type.
23332 (stack_tie): Define special instruction to be used in
23335 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23337 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23338 constraint. It is not valid for the pattern.
23339 (noncommutative_binary_comparison): Likewise.
23341 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23343 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23346 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23348 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23349 be used by the reg-alloc.
23351 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23353 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23354 reg-alloc when having mul64 or mul32x16 instructions.
23355 (mulsidi3): Likewise.
23356 (umulsidi3): Likewise.
23357 (mulsi32x16): New pattern.
23358 (mulsi64): Likewise.
23359 (mulsidi64): Likewise.
23360 (umulsidi64): Likewise.
23361 (MUL32x16_REG): Define.
23362 (mul64_600): Use MUL32x16_REG.
23363 (mac64_600): Likewise.
23364 (umul64_600): Likewise.
23365 (umac64_600): Likewise.
23367 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23369 * config/arc/arc.md (mulsi3_700): Make it commutative.
23371 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23373 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23375 (*sign_extendsidi2_insn): Likewise for movstosw.
23377 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23379 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23380 the type of the input discriminant value. Convert the
23381 discriminant value of signedness vary.
23383 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23385 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23386 Add to -Wall section.
23388 2017-06-01 Richard Biener <rguenther@suse.de>
23390 PR middle-end/66313
23391 * fold-const.c (fold_plusminus_mult_expr): If the factored
23392 factor may be zero use a wrapping type for the inner operation.
23393 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23394 and handle moved defs.
23395 (process_assignment): Properly guard the unary op case. Return a
23396 tri-state indicating that moving the stmt before the call may allow
23397 to continue. Pass through to_move.
23398 (find_tail_calls): Handle moving unrelated defs before
23401 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23404 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23405 splitter result in the canonical way.
23407 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23409 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23410 also for 32bit target. Update insn attributes.
23411 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23413 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23415 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23416 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23417 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23418 (_mm_maskz_min_ss): New intrinsics.
23420 2017-05-31 Martin Liska <mliska@suse.cz>
23422 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23423 Change comment style to one we normally use.
23424 (vectorizable_reduction): Likewise.
23425 (vectorizable_induction): Likewise.
23426 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23427 (vectorizable_call): Likewise.
23428 (vectorizable_simd_clone_call): Likewise.
23429 (vectorizable_conversion): Likewise.
23430 (vectorizable_assignment): Likewise.
23431 (vectorizable_shift): Likewise.
23432 (vectorizable_operation): Likewise.
23433 (vectorizable_store): Likewise.
23434 (vectorizable_load): Likewise.
23435 * tree-vectorizer.h: Likewise.
23437 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23439 * passes.c (emergency_dump_function): New.
23440 * tree-pass.h (emergency_dump_function): Declare.
23441 * plugin.c (plugins_internal_error_function): Remove.
23442 * plugin.h (plugins_internal_error_function): Remove declaration.
23443 * toplev.c (internal_error_function): New static function. Use it...
23444 (general_init): ...here.
23446 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23448 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23449 (arc_rtx_costs): Add costs for new patterns.
23450 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23451 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23453 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23455 * tree-ssa-strlen.c (get_next_strinfo): New function.
23456 (get_stridx_plus_constant): Use it.
23457 (zero_length_string): Likewise.
23458 (adjust_related_strinfos): Likewise.
23459 (adjust_last_stmt): Likewise.
23461 2017-05-31 Richard Biener <rguenther@suse.de>
23464 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23465 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23467 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23469 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23470 loop_vinfo argument and use of dependence distance vectors.
23471 Check instead whether the two references differ only in their
23472 initial value and assume that they have the same alignment if the
23473 difference is a multiple of the vector alignment.
23474 (vect_analyze_data_refs_alignment): Update call accordingly.
23476 2017-05-31 Martin Liska <mliska@suse.cz>
23479 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23481 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23483 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23484 (create_intersect_range_checks): Move from ...
23485 * tree-data-ref.c (create_intersect_range_checks_index)
23486 (create_intersect_range_checks): ... to here.
23487 (create_runtime_alias_checks): New function factored from ...
23488 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23489 here. Call above function.
23490 * tree-data-ref.h (create_runtime_alias_checks): New function.
23492 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23494 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23495 segment length for dr_b and compute it in wide_int.
23497 2017-05-31 Richard Biener <rguenther@suse.de>
23499 PR tree-optimization/80906
23500 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23501 and pass through iv_map.
23502 (copy_bb_and_scalar_dependences): Adjust.
23503 (translate_pending_phi_nodes): Likewise.
23504 (copy_loop_close_phi_args): Handle code-generating IVs instead
23507 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23509 * diagnostic-color.c (color_dict): Add "type-diff".
23510 (parse_gcc_colors): Update comment.
23511 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23512 -fdiagnostics-show-template-tree and -fno-elide-type.
23513 (GCC_COLORS): Add type-diff to example.
23515 (-fdiagnostics-show-template-tree): New.
23516 (-fno-elide-type): New.
23517 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23518 the pp_format_decoder callback. Call any m_format_postprocessor's
23520 (pretty_printer::pretty_printer): Initialize
23521 m_format_postprocessor.
23522 (pretty_printer::~pretty_printer): Delete any
23523 m_format_postprocessor.
23524 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23525 (class format_postprocessor): New class.
23526 (struct pretty_printer::format_decoder): Document the new parameters.
23527 (struct pretty_printer::m_format_postprocessor): New field.
23528 * tree-diagnostic.c (default_tree_printer): Update for new
23529 bool and const char ** params.
23530 * tree-diagnostic.h (default_tree_printer): Likewise.
23532 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23534 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23535 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23536 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23537 rs6000_gen_cell_microcode code.
23538 (rs6000_final_prescan_insn): Delete.
23539 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23540 "warn-cell-microcode" entries.
23541 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23542 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23543 throughout. Change cc_reg_not_micro_cr0_operand to
23544 cc_reg_not_cr0_operand throughout.
23545 (*extendhi<mode>2_noload): Delete.
23546 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23547 (mwarn-cell-microcode): Delete.
23548 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23549 -mgen-cell-microcode and -mwarn-cell-microcode.
23551 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23554 * config/i386/constraints.md (Yd): New constraint.
23556 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23557 and (?Yd, r) alternatives. Update insn attributes.
23558 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23559 and (?*Yd, r) alternatives. Update insn attributes.
23560 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23562 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23564 * gimplify.c (gimplify_modify_expr): Don't create a
23565 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23568 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23570 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23572 2017-05-30 Richard Biener <rguenther@suse.de>
23574 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23575 and reduc_def fields.
23576 (STMT_VINFO_REDUC_TYPE): New define.
23577 (STMT_VINFO_REDUC_DEF): Likewise.
23578 (vect_force_simple_reduction): Adjust prototype.
23579 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23580 (vect_is_simple_reduction): Remove check_reduction argument.
23581 (vect_force_simple_reduction): Adjust and set
23582 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23583 (vectorizable_reduction): Do not re-do reduction analysis
23584 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23585 * tree-parloops.c (gather_scalar_reductions): Adjust.
23587 2017-05-30 Richard Biener <rguenther@suse.de>
23589 PR middle-end/80901
23590 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23593 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23595 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23596 Introduce unknown_misalignment parameter and remove vf.
23597 (vect_peeling_hash_get_lowest_cost):
23598 Pass unknown_misalignment parameter.
23599 (vect_enhance_data_refs_alignment):
23600 Fix unsupportable data ref treatment.
23602 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23604 * tree-vect-data-refs.c (vect_get_data_access_cost):
23605 Workaround for SLP handling.
23606 (vect_enhance_data_refs_alignment):
23607 Compute costs for doing no peeling at all, compare to the best
23608 peeling costs so far and avoid peeling if cheaper.
23610 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23612 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23613 Return peeling info and set costs to zero for unlimited cost
23615 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23616 with unknown misalignment. Compute and costs for unknown
23617 misalignment, compare them to the costs for known misalignment
23618 and choose the cheapest for peeling.
23620 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23622 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23623 (vect_get_peeling_costs_all_drs): Create function.
23624 (vect_peeling_hash_get_lowest_cost):
23625 Use vect_get_peeling_costs_all_drs.
23626 (vect_peeling_supportable): Create function.
23627 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23629 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23631 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23632 DR_HAS_NEGATIVE_STEP.
23633 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23634 (vect_enhance_data_refs_alignment): Use.
23635 (vect_duplicate_ssa_name_ptr_info): Use.
23636 * tree-vectorizer.h (dr_misalignment): Use.
23637 (known_alignment_for_access_p): Use.
23639 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23642 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23644 (has_section_name): New function.
23646 2017-05-30 Martin Liska <mliska@suse.cz>
23649 * auto-profile.c (get_function_decl_from_block): Fix
23652 2017-05-30 Richard Biener <rguenther@suse.de>
23654 PR middle-end/80876
23655 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23657 2017-05-30 Martin Liska <mliska@suse.cz>
23659 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23660 * dumpfile.h (struct dump_file_info): Remove ctors.
23662 2017-05-30 Martin Liska <mliska@suse.cz>
23664 * predict.def: Fix GNU coding style.
23666 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23668 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23669 Mark 'to' argument with ATTRIBUTE_UNUSED.
23671 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23673 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23674 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23675 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23678 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23680 * doc/install.texi (Options specification): Restore entry of
23681 --enable-sjlj-exceptions.
23683 2017-05-27 Michael Eager <eager@eagercon.com>
23686 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23688 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23690 * config/microblaze/microblaze.h
23691 (FIXED_REGISTERS): Update in macro.
23692 (CALL_USED_REGISTERS): Update in macro.
23694 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23696 * doc/install.texi: Add links to macOS binary distributions.
23698 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23702 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23704 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23706 2017-05-26 Martin Liska <mliska@suse.cz>
23708 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23710 2017-05-26 Martin Liska <mliska@suse.cz>
23712 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23713 always leading ';; '.
23714 (dump_bb_info): Likewise.
23715 (brief_dump_cfg): Likewise.
23716 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23717 * dumpfile.c: Remove usage of TDF_VERBOSE.
23718 * dumpfile.h (enum dump_kind): Likewise.
23719 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23720 * print-tree.c (debug_verbose): Remove.
23721 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23722 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23723 * tree-diagnostic.c (default_tree_printer): Replace
23724 TDF_DIAGNOSTIC with TDF_SLIM.
23726 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23728 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23729 in parameter loop, rather than loop_vinfo.
23730 (create_intersect_range_checks): Ditto.
23731 (vect_create_cond_for_alias_checks): Update call to above functions.
23733 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23735 PR tree-optimization/80815
23736 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23737 for merging runtime alias checks. Handle negative DR_STEPs.
23739 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23741 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23743 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23744 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23745 out code pruning runtime alias checks.
23746 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23747 factored out from above.
23748 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23750 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23752 (prune_runtime_alias_test_list): New decalaration.
23754 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23756 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23757 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23758 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23759 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23760 (operator==, comp_dr_with_seg_len_pair): Ditto.
23761 (vect_prune_runtime_alias_test_list): Ditto.
23763 2017-05-26 Martin Liska <mliska@suse.cz>
23766 * params.def: Bound partial-inlining-entry-probability param.
23768 2017-05-26 Marek Polacek <polacek@redhat.com>
23771 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23774 2017-05-26 Richard Biener <rguenther@suse.de>
23776 PR tree-optimization/80842
23777 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23780 2017-05-26 Richard Biener <rguenther@suse.de>
23782 PR tree-optimization/80844
23783 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23785 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23787 * doc/md.texi (Machine Constraints): Update x86 family
23788 machine constraints section to match 'config/i386/constraints.md'.
23790 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23792 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23794 2017-05-25 Nathan Sidwell <nathan@acm.org>
23796 * doc/invoke.texi (--enable-languages): Update documentation.
23798 2017-05-25 Martin Liska <mliska@suse.cz>
23800 * dumpfile.c: Add TDF_FOLDING.
23801 * dumpfile.h (enum dump_kind): Likewise.
23802 * genmatch.c (dt_simplify::gen_1): Use it.
23804 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23806 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23808 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23810 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23811 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23813 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23815 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23816 * match.pd (X == C): Rewrite it here.
23817 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23818 with_certain_nonzero_bits2): New predicates.
23819 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23821 2017-05-24 Nathan Sidwell <nathan@acm.org>
23823 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23826 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23829 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23831 * config/powerpcspe: New port. Files are copied from the rs6000
23832 port, with "rs6000" in filenames replaced by "powerpcspe".
23834 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23836 PR rtl-optimization/80754
23837 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23839 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23841 * config/sparc/sparc.md (length): Return the correct value for -mflat
23842 sibcalls to match output_sibcall.
23844 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23848 * config/rs6000/rs6000.c (struct machine_function): Add new field
23850 (rs6000_get_separate_components): Init that field, use it.
23851 (rs6000_components_for_bb): Use the field.
23853 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23855 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23857 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23859 PR middle-end/80823
23860 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23862 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23865 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23866 address against address_operand predicate.
23867 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23869 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23871 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23872 parameters passed indirectly.
23874 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23876 * config/i386/i386.md (*movdi_internal): Remove SSE4
23877 alternative 18 (?r, *v). Update insn attributes.
23878 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23879 Update insn attributes.
23880 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23881 Update insn attributes.
23882 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23883 alternative 1 (r, v). Remove isa attribute.
23884 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23885 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23886 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23888 2017-05-23 Tom de Vries <tom@codesourcery.com>
23890 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23893 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23895 * cgraphunit.c (symbol_table::process_new_functions): Update.
23896 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23897 (inline_generate_summary): Rename to ...
23898 (ipa_fn_summary_generate): ... this one.
23899 (inline_read_summary): Rename to ...
23900 (ipa_fn_summary_read): ... this one.
23901 (inline_write_summary): Rename to ...
23902 (ipa_fn_summary_write): ... this one.
23903 (inline_free_summary): Rename to ...
23904 (ipa_free_fn_summary): ... this one.
23905 (pass_data_local_fn_summary, pass_local_fn_summary,
23906 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23907 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23908 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23909 make_pass_ipa_fn_summary): New.
23910 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23911 inline_write_summary, inline_free_summary): Remove.
23912 (ipa_free_fn_summary) : New.
23913 * ipa-inline.c (ipa_inline): Update.
23914 (pass_ipa_inline): Do not generate summaries.
23915 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23917 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23918 and add pass_ipa_fn_summary.
23919 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23921 (make_pass_inline_parameters): Remove.
23923 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23925 * omp-low.c (struct omp_context): Remove "default_kind" member.
23928 * omp-offload.c (execute_oacc_device_lower): Remove the
23929 parallelism dimensions function attributes for unparallelized
23930 OpenACC kernels constructs.
23932 2017-05-23 Martin Liska <mliska@suse.cz>
23934 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23936 (cgraph_edge::make_speculative): Likewise.
23937 (cgraph_edge::resolve_speculation): Likewise.
23938 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23939 (cgraph_node::dump): Likewise.
23940 * cgraph.h: Likewise.
23941 * cgraphunit.c (analyze_functions): Likewise.
23942 (symbol_table::compile): Likewise.
23943 * ipa-cp.c (print_all_lattices): Likewise.
23944 (determine_versionability): Likewise.
23945 (initialize_node_lattices): Likewise.
23946 (ipcp_verify_propagated_values): Likewise.
23947 (estimate_local_effects): Likewise.
23948 (update_profiling_info): Likewise.
23949 (create_specialized_node): Likewise.
23950 (perhaps_add_new_callers): Likewise.
23951 (decide_about_value): Likewise.
23952 (decide_whether_version_node): Likewise.
23953 (identify_dead_nodes): Likewise.
23954 (ipcp_store_bits_results): Likewise.
23955 * ipa-devirt.c (dump_targets): Likewise.
23956 (ipa_devirt): Likewise.
23957 * ipa-icf.c (sem_item::dump): Likewise.
23958 (sem_function::equals): Likewise.
23959 (sem_variable::equals): Likewise.
23960 (sem_item_optimizer::read_section): Likewise.
23961 (sem_item_optimizer::execute): Likewise.
23962 (congruence_class::dump): Likewise.
23963 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23964 (dump_inline_summary): Likewise.
23965 (estimate_node_size_and_time): Likewise.
23966 (inline_analyze_function): Likewise.
23967 * ipa-inline-transform.c (inline_call): Likewise.
23968 * ipa-inline.c (report_inline_failed_reason): Likewise.
23969 (want_early_inline_function_p): Likewise.
23970 (edge_badness): Likewise.
23971 (update_edge_key): Likewise.
23972 (inline_small_functions): Likewise.
23973 * ipa-profile.c (ipa_profile): Likewise.
23974 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23975 (ipa_make_edge_direct_to_target): Likewise.
23976 (remove_described_reference): Likewise.
23977 (ipa_impossible_devirt_target): Likewise.
23978 (propagate_controlled_uses): Likewise.
23979 (ipa_print_node_params): Likewise.
23980 (ipcp_transform_function): Likewise.
23981 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23982 (propagate_pure_const): Likewise.
23983 * ipa-reference.c (generate_summary): Likewise.
23984 (read_write_all_from_decl): Likewise.
23985 (propagate): Likewise.
23986 (ipa_reference_read_optimization_summary): Likewise.
23987 * ipa-utils.c (ipa_merge_profiles): Likewise.
23988 * ipa.c (walk_polymorphic_call_targets): Likewise.
23989 (symbol_table::remove_unreachable_nodes): Likewise.
23990 (ipa_single_use): Likewise.
23991 * passes.c (execute_todo): Likewise.
23992 * predict.c (drop_profile): Likewise.
23993 * symtab.c (symtab_node::get_dump_name): New function.
23994 (symtab_node::dump_name): Likewise.
23995 (symtab_node::dump_asm_name): Likewise.
23996 (symtab_node::dump_references): Likewise.
23997 (symtab_node::dump_referring): Likewise.
23998 (symtab_node::dump_base): Likewise.
23999 (symtab_node::debug_symtab): Likewise.
24000 * tree-sra.c (convert_callers_for_node): Likewise.
24001 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24002 * value-prof.c (init_node_map): Likewise.
24004 2017-05-23 Martin Liska <mliska@suse.cz>
24006 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24007 and symtab_node::debug_symtab to symbol_table::debug.
24008 * cgraphunit.c (analyze_functions): Use the renamed function.
24009 (symbol_table::compile): Likewise.
24010 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24011 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24012 * passes.c (execute_todo): Likewise.
24013 * symtab.c (symbol_table::dump): New function.
24014 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24016 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24018 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24019 that nonconst implies exec.
24021 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24023 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24024 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24025 (inline_edge_summary_vec): Turn into ...
24026 (ipa_call_summaries): ... this one.
24027 (redirect_to_unreachable, edge_set_predicate,
24028 evaluate_properties_for_edge, inline_summary_alloc,
24029 reset_ipa_call_summary, reset_inline_summary,
24030 inline_summary_t::duplicate): Update.
24031 (inline_edge_duplication_hook): Turn to ...
24032 (ipa_call_summary_t::duplicate): ... this one.
24033 (inline_edge_removal_hook): Turn to ...
24034 (ipa_call_summary_t::remove): ... this one.
24035 (dump_inline_edge_summary): Turn to ...
24036 (dump_ipa_call_summary): ... this one.
24037 (estimate_function_body_sizes): Update.
24038 (inline_update_callee_summaries): Update.
24039 (remap_edge_change_prob): Update.
24040 (remap_edge_summaries): Update.
24041 (inline_merge_summary): Update.
24042 (do_estimate_edge_time): Update.
24043 (inline_generate_summary): Update.
24044 (inline_read_section): Update.
24045 (inline_read_summary): Update.
24046 (inline_free_summary): Update.
24047 * ipa-inline.c (can_inline_edge_p): Update.
24048 (compute_inlined_call_time): Update.
24049 (want_inline_small_function_p): Update.
24050 (edge_badness): Update.
24051 (early_inliner): Update.
24052 * ipa-inline.h (inline_edge_summary): Turn to ...
24053 (ipa_call_summary): ... this one.
24054 (ipa_call_summary_t): New class.
24055 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24056 (ipa_call_summaries): New.
24057 (inline_edge_summary): Remove.
24058 (estimate_edge_growth): Update.
24059 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24060 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24061 * ipa-split.c (execute_split_functions): Update.
24062 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24064 2017-05-23 Tom de Vries <tom@codesourcery.com>
24066 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24067 attributes): Document rdrand effective target.
24069 2017-05-23 Tom de Vries <tom@codesourcery.com>
24071 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24072 attributes): Sort alphabetically.
24074 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24076 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24078 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24081 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24082 V2DF/V2DI splat into two separate patterns, one that handles
24083 registers, and the other that only handles memory. Drop support
24084 for splatting from a GPR on ISA 2.07 and then splitting the
24085 splat into direct move and splat.
24086 (vsx_splat_<mode>_reg): Likewise.
24087 (vsx_splat_<mode>_mem): Likewise.
24089 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24091 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24093 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24095 PR middle-end/80809
24096 * omp-low.c (finish_taskreg_remap): New function.
24097 (finish_taskreg_scan): If unit size of ctx->record_type
24098 is non-constant, unshare the size expression and replace
24099 decls in it with possible outer var refs.
24101 PR middle-end/80809
24102 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24103 GOVD_SHARED rather than GOVD_PRIVATE with it.
24104 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24105 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24107 PR middle-end/80853
24108 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24109 as last argument to build_outer_var_ref for pointer bases of array
24110 section reductions.
24112 2017-05-19 Martin Sebor <msebor@redhat.com>
24114 * print-tree.c (print_node): Print DECL_READ_P flag.
24116 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24118 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24119 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24120 * cgraph.c: Likewise.
24121 * cgraphunit.c: Likewise.
24122 * gengtype.c: Likewise.
24123 * ipa-cp.c: Likewise.
24124 * ipa-devirt.c: Likewise.
24125 * ipa-icf.c: Likewise.
24126 * ipa-predicate.c: Likewise.
24127 * ipa-profile.c: Likewise.
24128 * ipa-prop.c: Likewise.
24129 * ipa-split.c: Likewise.
24131 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24132 edge_predicate_pool, dump_inline_hints,
24133 inline_summary::account_size_time, redirect_to_unreachable,
24134 edge_set_predicate, set_hint_predicate,
24135 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24136 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24137 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24138 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24139 ipa_call_summary_t::remove, initialize_growth_caches,
24140 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24141 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24142 mark_modified, unmodified_parm_1, unmodified_parm,
24143 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24144 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24145 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24146 will_be_nonconstant_predicate, record_modified_bb_info,
24147 get_minimal_bb, record_modified, param_change_prob,
24148 phi_result_unknown_predicate, predicate_for_phi_result,
24149 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24150 estimate_function_body_sizes, compute_inline_parameters,
24151 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24152 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24153 inline_update_callee_summaries, remap_edge_change_prob,
24154 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24155 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24156 inline_analyze_function, inline_summary_t::insert,
24157 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24158 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24159 inline_free_summary): Move to ipa-fnsummary.h
24160 (predicate_t): Remove.
24161 * ipa-fnsummary.c: New file.
24162 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24163 (enum inline_hints_vals, inline_hints, agg_position_info,
24164 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24165 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24166 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24167 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24168 inline_read_summary, inline_write_summary, inline_free_summary,
24169 inline_analyze_function, initialize_inline_failed,
24170 inline_merge_summary, inline_update_overall_summary,
24171 compute_inline_parameters): Move to ipa-fnsummary.h
24172 * ipa-fnsummary.h: New file.
24173 * ipa-inline-transform.h: Include ipa-inline.h.
24174 * ipa-inline.c: LIkewise.
24176 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24178 * ipa-inline.c (edge_badness): Use inlined_time instead of
24179 inline_summaries->get.
24181 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24183 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24185 2017-05-22 Nathan Sidwell <nathan@acm.org>
24187 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24188 (fdump-lang): Document 'raw' option.
24189 * dumpfile.h (TDI_tu): Delete.
24190 * dumpfile.c (dump_files): Remove translation-unit.
24191 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24193 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24195 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24196 command option from $(AWK) call.
24197 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24198 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24199 [FORMAT]: Remove handling of variable.
24200 * config/avr/t-multilib: Regenerate.
24202 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24204 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24206 (dump_inline_summary): Do not print self_time.
24207 (estimate_function_body_sizes): Do not set self_time.
24208 (compute_inline_parameters): Likewise.
24209 (inline_read_section, inline_write_summary): Do not stream self_time.
24210 * ipa-inline.h (inline_summary): Drop self_time.
24212 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24214 * ipa-inline-analysis.c (account_size_time): Rename to ...
24215 (inline_summary::account_size_time): ... this one.
24216 (reset_ipa_call_summary): Turn to ...
24217 (ipa_call_summary::reset): ... this one.
24218 (reset_inline_summary): Turn to ...
24219 (inline_summary::reset): ... this one.
24220 (inline_summary_t::remove): Update.
24221 (inline_summary_t::duplicate): Update.
24222 (ipa_call_summary_t::remove): Update.
24223 (dump_inline_summary): Update.
24224 (estimate_function_body_sizes): Update.
24225 (compute_inline_parameters): Update.
24226 (estimate_node_size_and_time): Update.
24227 (inline_merge_summary): Update.
24228 (inline_update_overall_summary): Update.
24229 (inline_read_section): Update.
24230 (inline_write_summary): Update.
24231 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24232 add account_size_time and reset member functions.
24233 (ipa_call_summary): Add reset function.
24234 * ipa-predicate.h (predicate::operator &): Constify.
24236 2017-05-22 Richard Biener <rguenther@suse.de>
24238 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24240 2017-05-19 Jason Merrill <jason@redhat.com>
24242 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24244 2017-05-19 Marek Polacek <polacek@redhat.com>
24247 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24248 TYPE_OVERFLOW_WRAPS checks.
24250 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24252 * tree-core.h (enum omp_clause_default_kind): Add
24253 "OMP_CLAUSE_DEFAULT_PRESENT".
24254 * tree-pretty-print.c (dump_omp_clause): Handle it.
24255 * gimplify.c (enum gimplify_omp_var_data): Add
24256 "GOVD_MAP_FORCE_PRESENT".
24257 (gimplify_adjust_omp_clauses_1): Map it to
24258 "GOMP_MAP_FORCE_PRESENT".
24259 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24261 * gimplify.c (oacc_default_clause): Clarify.
24263 2017-05-19 Nathan Sidwell <nathan@acm.org>
24265 LANG_HOOK_REGISTER_DUMPS
24266 * toplev.c (general_init): Call register dump lang hook.
24267 * doc/invoke.texi: Document -fdump-lang option family.
24268 * dumpfile.c (dump_files): Remove class dump here.
24269 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24270 * dumpfile.h (tree_dump_index): Remove TDI_class.
24271 * langhooks-def.h (lhd_register_dumps): Declare.
24272 (LANG_HOOKS_REGISTER_DUMPS): Define.
24273 (LANG_HOOKS_INITIALIZER): Add it.
24274 * langhooks.c (lhd_register_dumps): Define.
24275 * langhooks.h (struct lang_hooks): Add register_dumps.
24277 2017-05-19 Nathan Sidwell <nathan@acm.org>
24279 * context.h (context::set_passes): New.
24280 * context.c (context::context): Do not create pass manager.
24281 * toplev.c (general_init): Create pass manager here.
24283 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24285 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24286 use this splitter if two add or or instructions would also work for
24287 the constant we want to generate.
24289 2017-05-19 Richard Biener <rguenther@suse.de>
24292 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24293 predicate evaluation.
24295 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24297 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24299 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24302 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24304 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24305 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24306 (inline_edge_summary_vec): Turn into ...
24307 (ipa_call_summaries): ... this one.
24308 (redirect_to_unreachable, edge_set_predicate,
24309 evaluate_properties_for_edge, inline_summary_alloc,
24310 reset_ipa_call_summary, reset_inline_summary,
24311 inline_summary_t::duplicate): Update.
24312 (inline_edge_duplication_hook): Turn to ...
24313 (ipa_call_summary_t::duplicate): ... this one.
24314 (inline_edge_removal_hook): Turn to ...
24315 (ipa_call_summary_t::remove): ... this one.
24316 (dump_inline_edge_summary): Turn to ...
24317 (dump_ipa_call_summary): ... this one.
24318 (estimate_function_body_sizes): Update.
24319 (inline_update_callee_summaries): Update.
24320 (remap_edge_change_prob): Update.
24321 (remap_edge_summaries): Update.
24322 (inline_merge_summary): Update.
24323 (do_estimate_edge_time): Update.
24324 (inline_generate_summary): Update.
24325 (inline_read_section): Update.
24326 (inline_read_summary): Update.
24327 (inline_free_summary): Update.
24328 * ipa-inline.c (can_inline_edge_p): Update.
24329 (compute_inlined_call_time): Update.
24330 (want_inline_small_function_p): Update.
24331 (edge_badness): Update.
24332 (early_inliner): Update.
24333 * ipa-inline.h (inline_edge_summary): Turn to ...
24334 (ipa_call_summary): ... this one.
24335 (ipa_call_summary_t): New class.
24336 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24337 (ipa_call_summaries): New.
24338 (inline_edge_summary): Remove.
24339 (estimate_edge_growth): Update.
24340 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24341 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24342 * ipa-split.c (execute_split_functions): Update.
24343 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24345 2017-05-19 Richard Biener <rguenther@suse.de>
24347 PR middle-end/80764
24348 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24350 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24352 * config/rs6000/rs6000.c (struct machine_function): Add field
24353 fpr_is_wrapped_separately.
24354 (rs6000_get_separate_components): Use 64 components. Handle the
24355 new FPR components.
24356 (rs6000_components_for_bb): Handle the FPR components.
24357 (rs6000_emit_prologue_components): Handle the FPR components.
24358 (rs6000_emit_epilogue_components): Handle the FPR components.
24359 (rs6000_set_handled_components): Handle the FPR components.
24360 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24361 that are already separately shrink-wrapped.
24362 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24363 that are already separately shrink-wrapped.
24365 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24368 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24371 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24372 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24373 cases where the register allocator uses a move and an offsettable
24374 memory operation to/from a FPR register on ISA 2.06/2.07.
24375 (define_peephole2 for Altivec d-form store): Likewise.
24377 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24380 * config/i386/mmx.md (*mov<mode>_internal): Enable
24381 alternatives 11, 12, 13 and 14 also for 32bit targets.
24382 Remove alternatives 15, 16, 17 and 18.
24383 * config/i386/sse.md (vec_concatv2di): Change
24384 alternative (!x, *y) to (x, ?!*Yn).
24386 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24388 * dumpfile.h (enum dump_kind): Remove stray comma.
24390 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24392 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24393 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24394 predicate::num_conditions
24395 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24396 (CHANGED): turn into predicate::changed.
24397 (agg_position_info): Move to ipa-predicate.h
24398 (add_condition, predicate::add_clause, predicate::operator &=,
24399 predicate::or_with, predicate::evaluate, predicate::probability,
24400 dump_condition, dump_clause, predicate::dump,
24401 predicate::remap_after_duplication, predicate::remap_after_inlining,
24402 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24403 (evaluate_conditions_for_known_args): Update.
24404 (set_cond_stmt_execution_predicate): Update.
24405 * ipa-inline.h: Include ipa-predicate.h
24406 (condition, inline_param_summary, conditions, agg_position_info,
24407 predicate): Move to ipa-predicate.h
24408 * ipa-predicate.c: New file.
24409 * ipa-predicate.h: New file.
24411 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24413 * final.c (leaf_function_p): Check we are not in a sequence.
24415 2017-05-18 Martin Liska <mliska@suse.cz>
24417 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24418 * dumpfile.c (dump_register): Use new enum dump_kind.
24419 (get_dump_file_name): Likewise.
24420 (dump_enable_all): Likewise.
24421 (dump_switch_p_1): Likewise.
24422 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24423 * dumpfile.h (enum dump_kind): New enum type.
24424 (struct dump_file_info): Create constructor and
24425 format fields and comments.
24426 * passes.c (pass_manager::register_one_dump_file):
24428 * statistics.c (statistics_early_init): Likewise.
24429 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24430 TDF_TREE with TDF_SLIM.
24431 (gather_memory_references_ref): Likewise.
24433 2017-05-18 Martin Liska <mliska@suse.cz>
24435 * vec.h (struct vnull): Use it.
24437 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24439 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24440 (true_predicate, false_predicate, true_predicate_p,
24441 false_predicate_p): Remove.
24442 (single_cond_predicate, not_inlined_predicate): Turn to member function
24444 (add_condition): Update.
24445 (add_clause): Turn to...
24446 (predicate::add_clause): ... this one; update; allow passing NULL
24448 (and_predicates): Turn to ...
24449 (predicate::operator &=): ... this one.
24450 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24451 (or_predicates): Turn to ...
24452 (predicate::or_with): ... this one.
24453 (evaluate_predicate): Turn to ...
24454 (predicate::evaluate): ... this one.
24455 (predicate_probability): Turn to ...
24456 (predicate::probability): ... this one.
24457 (dump_condition): Update.
24458 (dump_predicate): Turn to ...
24459 (predicate::dump): ... this one.
24460 (account_size_time): Update.
24461 (edge_set_predicate): Update.
24462 (set_hint_predicate): UPdate.
24463 (evaluate_conditions_for_known_args): Update.
24464 (evaluate_properties_for_edge): Update.
24465 (remap_predicate_after_duplication): Turn to...
24466 (predicate::remap_after_duplication): ... this one.
24467 (remap_hint_predicate_after_duplication): Update.
24468 (inline_summary_t::duplicate): UPdate.
24469 (dump_inline_edge_summary): Update.
24470 (dump_inline_summary): Update.
24471 (set_cond_stmt_execution_predicate): Update.
24472 (set_switch_stmt_execution_predicate): Update.
24473 (compute_bb_predicates): Update.
24474 (will_be_nonconstant_expr_predicate): Update.
24475 (will_be_nonconstant_predicate): Update.
24476 (phi_result_unknown_predicate): Update.
24477 (predicate_for_phi_result): Update.
24478 (array_index_predicate): Update.
24479 (estimate_function_body_sizes): Update.
24480 (estimate_node_size_and_time): Update.
24481 (estimate_ipcp_clone_size_and_time): Update.
24482 (remap_predicate): Rename to ...
24483 (predicate::remap_after_inlining): ... this one.
24484 (remap_hint_predicate): Update.
24485 (inline_merge_summary): Update.
24486 (inline_update_overall_summary): Update.
24487 (estimate_size_after_inlining): Update.
24488 (read_predicate): Rename to ...
24489 (predicate::stream_in): ... this one.
24490 (read_inline_edge_summary): Update.
24491 (write_predicate): Rename to ...
24492 (predicate::stream_out): ... this one.
24493 (write_inline_edge_summary): Update.
24494 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24495 (clause_t): Turn to uint32_t
24496 (predicate): Turn to class; implement constructor and operators
24498 (size_time_entry): Update.
24499 (inline_summary): Update.
24500 (inline_edge_summary): Update.
24502 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24504 * fold-const.c (fold_binary_loc): Move transformation...
24505 * match.pd (C - X CMP X): ... here.
24507 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24509 * config/sparc/sparc.c (sparc_option_override): Set function
24510 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24511 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24513 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24515 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24517 2017-05-18 Marek Polacek <polacek@redhat.com>
24520 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24521 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24522 gimple_assign_load_p.
24524 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24526 PR middle-end/80692
24527 * real.c (do_compare): Give decimal_do_compare preference over
24528 comparing just the signs.
24530 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24532 * doc/md.texi (Canonicalization of Instructions): Describe the
24533 canonical form of instructions that inherently set a condition
24536 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24538 PR middle-end/80775
24539 * tree-cfg.c: Move deletion of unreachable case statements to after
24540 the merging of consecutive case labels.
24542 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24544 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24545 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24546 restoring of callee-saved registers.
24548 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24550 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24551 * config/visium/visium.c (single_set_and_flags): Likewise.
24552 * config/visium/visium.md (Substitutions): Likewise.
24554 2017-05-17 Martin Liska <mliska@suse.cz>
24556 * cfg.c: Introduce dump_flags_t type and
24557 use it instead of int type.
24559 * cfghooks.c: Likewise.
24560 * cfghooks.h (struct cfg_hooks): Likewise.
24561 * cfgrtl.c: Likewise.
24562 * cfgrtl.h: Likewise.
24563 * cgraph.c (cgraph_node::get_body): Likewise.
24564 * coretypes.h: Likewise.
24565 * domwalk.c: Likewise.
24566 * domwalk.h: Likewise.
24567 * dumpfile.c (struct dump_option_value_info): Likewise.
24568 (dump_enable_all): Likewise.
24569 (dump_switch_p_1): Likewise.
24570 (opt_info_switch_p): Likewise.
24571 * dumpfile.h (enum tree_dump_index): Likewise.
24572 (struct dump_file_info): Likewise.
24573 * genemit.c: Likewise.
24574 * generic-match-head.c: Likewise.
24575 * gengtype.c (open_base_files): Likewise.
24576 * gimple-pretty-print.c: Likewise.
24577 * gimple-pretty-print.h: Likewise.
24578 * graph.c (print_graph_cfg): Likewise.
24579 * graphite-scop-detection.c (dot_all_sese): Likewise.
24580 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24581 * loop-unroll.c (report_unroll): Likewise.
24582 * passes.c (pass_manager::register_one_dump_file): Likewise.
24583 * print-tree.c: Likewise.
24584 * statistics.c: Likewise.
24585 * tree-cfg.c: Likewise.
24586 * tree-cfg.h: Likewise.
24587 * tree-dfa.c: Likewise.
24588 * tree-dfa.h: Likewise.
24589 * tree-dump.c (dump_function): Likewise.
24590 * tree-dump.h (struct dump_info): Likewise.
24591 * tree-pretty-print.c: Likewise.
24592 * tree-pretty-print.h: Likewise.
24593 * tree-ssa-live.c: Likewise.
24594 * tree-ssa-live.h: Likewise.
24595 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24596 * tree-vect-loop.c: Likewise.
24597 * tree-vect-slp.c: Likewise.
24599 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24600 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24602 PR tree-optimization/80457
24603 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24604 of all arguments to a statement as scalar_to_vec operations.
24605 (vectorizable_call): Adjust call to vect_model_simple_cost for
24607 (vectorizable_conversion): Likewise.
24608 (vectorizable_assignment): Likewise.
24609 (vectorizable_shift): Likewise.
24610 (vectorizable_operation): Likewise.
24611 (vectorizable_comparison): Likewise.
24612 (vect_is_simple_cond): Record the def types for operands.
24613 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24614 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24615 for statement argument count.
24617 2017-05-16 Carl Love <cel@us.ibm.com>
24619 * config/rs6000/rs6000-c: Add support for built-in functions
24620 vector unsigned long long vec_bperm (vector unsigned long long,
24621 vector unsigned char)
24622 vector signed long long vec_mule (vector signed int,
24624 vector unsigned long long vec_mule (vector unsigned int,
24625 vector unsigned int)
24626 vector signed long long vec_mulo (vector signed int,
24628 vector unsigned long long vec_mulo (vector unsigned int,
24629 vector unsigned int)
24630 vector signed char vec_sldw (vector signed char,
24631 vector signed char,
24633 vector unsigned char vec_sldw (vector unsigned char,
24634 vector unsigned char,
24636 vector signed short vec_sldw (vector signed short,
24637 vector signed short,
24639 vector unsigned short vec_sldw (vector unsigned short,
24640 vector unsigned short,
24642 vector signed int vec_sldw (vector signed int,
24645 vector unsigned int vec_sldw (vector unsigned int,
24646 vector unsigned int,
24648 vector signed long long vec_sldw (vector signed long long,
24649 vector signed long long,
24651 vector unsigned long long vec_sldw (vector unsigned long long,
24652 vector unsigned long long,
24654 * config/rs6000/rs6000-c: Add support for built-in functions
24655 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24656 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24657 * doc/extend.texi: Update the built-in documentation for the
24658 new built-in functions.
24660 2017-05-16 Marek Polacek <polacek@redhat.com>
24664 * tree.c (save_expr): Don't fold the expression.
24666 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24668 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24669 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24670 and (?*y,m). Update insn attributes.
24672 2017-05-16 Martin Liska <mliska@suse.cz>
24674 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24675 flags argument of print_gimple_stmt, print_gimple_expr,
24676 print_generic_stmt and print_generic_expr.
24677 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24678 * coretypes.h: Likewise.
24679 * except.c (dump_eh_tree): Likewise.
24680 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24681 * gimple-pretty-print.h: Likewise.
24682 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24683 (backprop::push_to_worklist): Likewise.
24684 (backprop::pop_from_worklist): Likewise.
24685 (backprop::process_use): Likewise.
24686 (backprop::intersect_uses): Likewise.
24687 (note_replacement): Likewise.
24688 * gimple-ssa-store-merging.c
24689 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24690 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24691 (pass_store_merging::execute): Likewise.
24692 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24693 (ssa_base_cand_dump_callback): Likewise.
24694 (dump_incr_vec): Likewise.
24695 (replace_refs): Likewise.
24696 (replace_mult_candidate): Likewise.
24697 (create_add_on_incoming_edge): Likewise.
24698 (create_phi_basis): Likewise.
24699 (insert_initializers): Likewise.
24700 (all_phi_incrs_profitable): Likewise.
24701 (introduce_cast_before_cand): Likewise.
24702 (replace_one_candidate): Likewise.
24703 * gimplify.c (gimplify_expr): Likewise.
24704 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24705 (set_rename): Likewise.
24706 (rename_uses): Likewise.
24707 (copy_loop_phi_nodes): Likewise.
24708 (add_close_phis_to_merge_points): Likewise.
24709 (copy_loop_close_phi_args): Likewise.
24710 (copy_cond_phi_args): Likewise.
24711 (graphite_copy_stmts_from_block): Likewise.
24712 (translate_pending_phi_nodes): Likewise.
24713 * graphite-poly.c (print_pdr): Likewise.
24714 (dump_gbb_cases): Likewise.
24715 (dump_gbb_conditions): Likewise.
24716 (print_scop_params): Likewise.
24717 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24718 (build_cross_bb_scalars_use): Likewise.
24719 (gather_bbs::before_dom_children): Likewise.
24720 * hsa-dump.c (dump_hsa_immed): Likewise.
24721 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24722 (get_replacement_map): Likewise.
24723 * ipa-inline-analysis.c (dump_condition): Likewise.
24724 (estimate_function_body_sizes): Likewise.
24725 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24726 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24727 * ipa-prop.c (ipa_dump_param): Likewise.
24728 (ipa_print_node_jump_functions_for_edge): Likewise.
24729 (ipa_modify_call_arguments): Likewise.
24730 (ipa_modify_expr): Likewise.
24731 (ipa_dump_param_adjustments): Likewise.
24732 (ipa_dump_agg_replacement_values): Likewise.
24733 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24734 * ipa-pure-const.c (check_stmt): Likewise.
24735 (pass_nothrow::execute): Likewise.
24736 * ipa-split.c (execute_split_functions): Likewise.
24737 * omp-offload.c (dump_oacc_loop_part): Likewise.
24738 (dump_oacc_loop): Likewise.
24739 * trans-mem.c (tm_log_emit): Likewise.
24740 (tm_memopt_accumulate_memops): Likewise.
24741 (dump_tm_memopt_set): Likewise.
24742 (dump_tm_memopt_transform): Likewise.
24743 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24744 (print_loop): Likewise.
24745 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24746 (chkp_gather_checks_info): Likewise.
24747 (chkp_get_check_result): Likewise.
24748 (chkp_remove_check_if_pass): Likewise.
24749 (chkp_use_outer_bounds_if_possible): Likewise.
24750 (chkp_reduce_bounds_lifetime): Likewise.
24751 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24752 (chkp_mark_completed_bounds): Likewise.
24753 (chkp_register_incomplete_bounds): Likewise.
24754 (chkp_mark_invalid_bounds): Likewise.
24755 (chkp_maybe_copy_and_register_bounds): Likewise.
24756 (chkp_build_returned_bound): Likewise.
24757 (chkp_get_bound_for_parm): Likewise.
24758 (chkp_build_bndldx): Likewise.
24759 (chkp_get_bounds_by_definition): Likewise.
24760 (chkp_generate_extern_var_bounds): Likewise.
24761 (chkp_get_bounds_for_decl_addr): Likewise.
24762 * tree-chrec.c (chrec_apply): Likewise.
24763 * tree-data-ref.c (dump_data_reference): Likewise.
24764 (dump_subscript): Likewise.
24765 (dump_data_dependence_relation): Likewise.
24766 (analyze_overlapping_iterations): Likewise.
24767 * tree-inline.c (expand_call_inline): Likewise.
24768 (tree_function_versioning): Likewise.
24769 * tree-into-ssa.c (dump_defs_stack): Likewise.
24770 (dump_currdefs): Likewise.
24771 (dump_names_replaced_by): Likewise.
24772 (dump_update_ssa): Likewise.
24773 (update_ssa): Likewise.
24774 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24775 * tree-parloops.c (build_new_reduction): Likewise.
24776 (try_create_reduction_list): Likewise.
24777 (ref_conflicts_with_region): Likewise.
24778 (oacc_entry_exit_ok_1): Likewise.
24779 (oacc_entry_exit_single_gang): Likewise.
24780 * tree-pretty-print.h: Likewise.
24781 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24782 (get_scalar_evolution): Likewise.
24783 (add_to_evolution): Likewise.
24784 (get_loop_exit_condition): Likewise.
24785 (analyze_evolution_in_loop): Likewise.
24786 (analyze_initial_condition): Likewise.
24787 (analyze_scalar_evolution): Likewise.
24788 (instantiate_scev): Likewise.
24789 (number_of_latch_executions): Likewise.
24790 (gather_chrec_stats): Likewise.
24791 (final_value_replacement_loop): Likewise.
24792 (scev_const_prop): Likewise.
24793 * tree-sra.c (dump_access): Likewise.
24794 (disqualify_candidate): Likewise.
24795 (create_access): Likewise.
24796 (reject): Likewise.
24797 (maybe_add_sra_candidate): Likewise.
24798 (create_access_replacement): Likewise.
24799 (analyze_access_subtree): Likewise.
24800 (analyze_all_variable_accesses): Likewise.
24801 (sra_modify_assign): Likewise.
24802 (initialize_constant_pool_replacements): Likewise.
24803 (find_param_candidates): Likewise.
24804 (decide_one_param_reduction): Likewise.
24805 (replace_removed_params_ssa_names): Likewise.
24806 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24807 * tree-ssa-copy.c (dump_copy_of): Likewise.
24808 (copy_prop_visit_cond_stmt): Likewise.
24809 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24810 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24811 (record_equivalences_from_stmt): Likewise.
24812 * tree-ssa-dse.c (compute_trims): Likewise.
24813 (delete_dead_call): Likewise.
24814 (delete_dead_assignment): Likewise.
24815 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24816 (forward_propagate_into_cond): Likewise.
24817 (pass_forwprop::execute): Likewise.
24818 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24819 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24821 (move_computations_worker): Likewise.
24822 (execute_sm): Likewise.
24823 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24824 (remove_exits_and_undefined_stmts): Likewise.
24825 (remove_redundant_iv_tests): Likewise.
24826 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24827 (adjust_iv_update_pos): Likewise.
24828 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24829 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24830 (value_replacement): Likewise.
24831 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24832 * tree-ssa-pre.c (print_pre_expr): Likewise.
24833 (get_representative_for): Likewise.
24834 (create_expression_by_pieces): Likewise.
24835 (insert_into_preds_of_block): Likewise.
24836 (eliminate_insert): Likewise.
24837 (eliminate_dom_walker::before_dom_children): Likewise.
24838 (eliminate): Likewise.
24839 (remove_dead_inserted_code): Likewise.
24840 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24841 * tree-ssa-reassoc.c (get_rank): Likewise.
24842 (eliminate_duplicate_pair): Likewise.
24843 (eliminate_plus_minus_pair): Likewise.
24844 (eliminate_not_pairs): Likewise.
24845 (undistribute_ops_list): Likewise.
24846 (eliminate_redundant_comparison): Likewise.
24847 (update_range_test): Likewise.
24848 (optimize_range_tests_var_bound): Likewise.
24849 (optimize_vec_cond_expr): Likewise.
24850 (rewrite_expr_tree): Likewise.
24851 (rewrite_expr_tree_parallel): Likewise.
24852 (linearize_expr): Likewise.
24853 (break_up_subtract): Likewise.
24854 (linearize_expr_tree): Likewise.
24855 (attempt_builtin_powi): Likewise.
24856 (attempt_builtin_copysign): Likewise.
24857 (transform_stmt_to_copy): Likewise.
24858 (transform_stmt_to_multiply): Likewise.
24859 (dump_ops_vector): Likewise.
24860 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24861 (print_scc): Likewise.
24862 (set_ssa_val_to): Likewise.
24863 (visit_reference_op_store): Likewise.
24864 (visit_use): Likewise.
24865 (sccvn_dom_walker::before_dom_children): Likewise.
24866 (run_scc_vn): Likewise.
24867 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24869 (expr_hash_elt::print): Likewise.
24870 (const_and_copies::pop_to_marker): Likewise.
24871 (const_and_copies::record_const_or_copy_raw): Likewise.
24872 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24873 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24874 (dump_predicates): Likewise.
24875 (find_uninit_use): Likewise.
24876 (warn_uninitialized_phi): Likewise.
24877 (pass_late_warn_uninitialized::execute): Likewise.
24878 * tree-ssa.c (verify_vssa): Likewise.
24879 (verify_ssa): Likewise.
24880 (maybe_optimize_var): Likewise.
24881 * tree-vrp.c (dump_value_range): Likewise.
24882 (dump_all_value_ranges): Likewise.
24883 (dump_asserts_for): Likewise.
24884 (register_edge_assert_for_2): Likewise.
24885 (vrp_visit_cond_stmt): Likewise.
24886 (vrp_visit_switch_stmt): Likewise.
24887 (vrp_visit_stmt): Likewise.
24888 (vrp_visit_phi_node): Likewise.
24889 (simplify_cond_using_ranges_1): Likewise.
24890 (fold_predicate_in): Likewise.
24891 (evrp_dom_walker::before_dom_children): Likewise.
24892 (evrp_dom_walker::push_value_range): Likewise.
24893 (evrp_dom_walker::pop_value_range): Likewise.
24894 (execute_early_vrp): Likewise.
24896 2017-05-16 Richard Biener <rguenther@suse.de>
24898 * dwarf2out.c (loc_list_from_tree_1): Do not create
24899 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24901 2017-05-16 Richard Biener <rguenther@suse.de>
24903 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24905 (note_variable_value_in_expr): If we resolved the decl ref
24906 do not push to the stack.
24908 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24910 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24911 operations in fast-math mode.
24912 (vaddq_f16): Likewise.
24913 (vmul_f16): Likewise.
24914 (vmulq_f16): Likewise.
24915 (vsub_f16): Likewise.
24916 (vsubq_f16): Likewise.
24917 * config/arm/neon.md (add<mode>3): New.
24919 (fma:<VH:mode>3): New. Also remove outdated comment.
24922 2017-05-16 Martin Liska <mliska@suse.cz>
24926 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24927 (odr_types_equivalent_p): Likewise.
24929 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24931 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24933 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24936 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24937 non-interunit SSE move alternatives with '?'.
24938 (zero-extendsidi peephole2): New peephole to skip intermediate
24939 general register in SSE zero-extend sequence.
24941 2017-05-15 Jeff Law <law@redhat.com>
24943 * reorg.c (relax_delay_slots): Create a new variable to hold
24944 the temporary target rather than clobbering TARGET_LABEL.
24946 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24947 missing argument to extract_bit_field call.
24948 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24950 2017-05-15 Martin Liska <mliska@suse.cz>
24953 * gcc.c (process_command): Do not allow empty argument of -o option.
24955 2017-05-15 Renlin Li <renlin.li@arm.com>
24957 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24958 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24959 * config/aarch64/constraints.md (Usf): Add long call check.
24960 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24961 (call_value): Likewise.
24962 (sibcall): Likewise.
24963 (sibcall_value): Likewise.
24965 (call_value_insn): New.
24966 (sibcall_insn): Update rtx pattern.
24967 (sibcall_value_insn): Likewise.
24968 (call_internal): Remove.
24969 (call_value_internal): Likewise.
24970 (sibcall_internal): Likewise.
24971 (sibcall_value_internal): Likewise.
24972 (call_reg): Likewise.
24973 (call_symbol): Likewise.
24974 (call_value_reg): Likewise.
24975 (call_value_symbol): Likewise.
24977 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24980 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24982 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24984 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24985 compatible with CCGOCmode and with CCZmode.
24987 2017-05-14 Martin Sebor <msebor@redhat.com>
24989 PR middle-end/77671
24990 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24991 (gimple_fold_builtin_snprintf): Same.
24992 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24993 (gimple_fold_builtin_snprintf): Same.
24994 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24995 of character types.
24996 (is_call_safe): New function.
24997 (try_substitute_return_value): Call it.
24998 (try_simplify_call): New function.
24999 (pass_sprintf_length::handle_gimple_call): Call it.
25001 2017-05-14 Martin Sebor <msebor@redhat.com>
25003 PR middle-end/80669
25004 * builtins.c (expand_builtin_stpncpy): Simplify.
25006 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25008 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25009 * config/i386/i386.h
25010 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25011 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25012 (struct machine_function): Add new members call_ms2sysv,
25013 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25014 (struct machine_frame_state): New fields sp_realigned and
25015 sp_realigned_offset.
25016 * config/i386/i386.c
25017 (enum xlogue_stub): New enum.
25018 (enum xlogue_stub_sets): New enum.
25019 (class xlogue_layout): New class.
25020 (struct ix86_frame): New fields stack_realign_allocate_offset,
25021 stack_realign_offset and outlined_save_offset. Modify comments to
25022 detail stack layout when using out-of-line stubs.
25023 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25024 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25025 -mcall-ms2sysv-xlogues.
25026 (stub_managed_regs): New static variable.
25027 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25028 registers managed by out-of-line stub.
25029 (disable_call_ms2sysv_xlogues): New function.
25030 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25031 m->call_ms2sysv when appropriate and compute frame layout for
25033 (sp_valid_at, fp_valid_at): New inline functions.
25034 (choose_basereg): New function.
25035 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25037 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25038 Use align parameter of choose_baseaddr to generated aligned SSE movs
25040 (pro_epilogue_adjust_stack): Modify to track
25041 machine_frame_state::sp_realigned.
25042 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25043 (ix86_nsaved_sseregs): Likewise.
25044 (ix86_emit_save_regs): Likewise.
25045 (ix86_emit_save_regs_using_mov): Likewise.
25046 (ix86_emit_save_sse_regs_using_mov): Likewise.
25047 (get_scratch_register_on_entry): Likewise.
25048 (gen_frame_set): New function.
25049 (gen_frame_load): Likewise.
25050 (gen_frame_store): Likewise.
25051 (emit_outlined_ms2sysv_save): Likewise.
25052 (emit_outlined_ms2sysv_restore): Likewise.
25053 (ix86_expand_prologue): Modify stack re-alignment code and call
25054 emit_outlined_ms2sysv_save when appropriate.
25055 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25056 parameter rtx_insn *insn, which allows the function to be used to only
25057 generate the notes.
25058 (ix86_expand_epilogue): Modify validity checks of frame and stack
25059 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25060 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25061 * config/i386/predicates.md
25062 (save_multiple): New predicate.
25063 (restore_multiple): Likewise.
25064 * config/i386/sse.md
25065 (save_multiple<mode>): New pattern.
25066 (save_multiple_realign<mode>): Likewise.
25067 (restore_multiple<mode>): Likewise.
25068 (restore_multiple_and_return<mode>): Likewise.
25069 (restore_multiple_leave_return<mode>): Likewise.
25070 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25072 2017-05-14 Julia Koval <julia.koval@intel.com>
25074 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25075 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25076 (__builtin_ia32_xsetbv): New builtins.
25077 * config/i386/i386.c (ix86_expand_special_args_builtin):
25079 (ix86_expand_builtin): Special expand for new intrinsics.
25080 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25081 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25082 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25084 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25086 * cfganal.c (inverted_post_order_compute): Change argument type
25088 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25089 * df-core.c (rest_of_handle_df_initialize): Adjust.
25090 (rest_of_handle_df_finish): Likewise.
25091 (df_analyze_1): Likewise.
25092 (df_analyze): Likewise.
25093 (loop_inverted_post_order_compute): Change argument to be a vec *.
25094 (df_analyze_loop): Adjust.
25095 (df_get_n_blocks): Likewise.
25096 (df_get_postorder): Likewise.
25097 * df.h (struct df_d): Change field to be a vec.
25098 * lcm.c (compute_laterin): Adjust.
25099 (compute_available): Likewise.
25100 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25101 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25102 * tree-ssa-pre.c (compute_antic): Likewise.
25104 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25106 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25107 (depth_first_search::depth_first_search): Change structure init
25108 function to this constructor.
25109 (depth_first_search::add_bb): Rename function to this member.
25110 (depth_first_search::execute): Likewise.
25111 (flow_dfs_compute_reverse_finish): Adjust.
25113 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25115 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25116 (longest_simple_path): Likewise.
25117 * shrink-wrap.c (spread_components): Likewise.
25118 (disqualify_problematic_components): Likewise.
25119 (emit_common_heads_for_components): Likewise.
25120 (emit_common_tails_for_components): Likewise.
25121 (insert_prologue_epilogue_for_components): Likewise.
25123 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25125 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25128 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25130 * df-core.c (df_set_blocks): Start using auto_bitmap.
25131 (df_compact_blocks): Likewise.
25132 * df-problems.c (df_rd_confluence_n): Likewise.
25133 * df-scan.c (df_insn_rescan_all): Likewise.
25134 (df_process_deferred_rescans): Likewise.
25135 (df_update_entry_block_defs): Likewise.
25136 (df_update_exit_block_uses): Likewise.
25137 (df_entry_block_bitmap_verify): Likewise.
25138 (df_exit_block_bitmap_verify): Likewise.
25139 (df_scan_verify): Likewise.
25140 * lra-constraints.c (lra_constraints): Likewise.
25141 (undo_optional_reloads): Likewise.
25142 (lra_undo_inheritance): Likewise.
25143 * lra-remat.c (calculate_gen_cands): Likewise.
25144 (do_remat): Likewise.
25145 * lra-spills.c (assign_spill_hard_regs): Likewise.
25146 (spill_pseudos): Likewise.
25147 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25148 (bitmap_set_subtract_values): Likewise.
25150 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25152 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25153 management with auto_bitmap.
25154 (fix_inter_tick): Likewise.
25155 (fix_recovery_deps): Likewise.
25156 * ira.c (add_store_equivs): Likewise.
25157 (find_moveable_pseudos): Likewise.
25158 (split_live_ranges_for_shrink_wrap): Likewise.
25159 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25160 (rtx_reuse_manager::seen_def_p): Likewise.
25161 (rtx_reuse_manager::set_seen_def): Likewise.
25162 * print-rtl.h (class rtx_reuse_manager): Likewise.
25164 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25166 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25168 (migrate_btr_def): Likewise.
25169 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25170 * df-core.c (loop_post_order_compute): Likewise.
25171 (loop_inverted_post_order_compute): Likewise.
25172 * hsa-common.h: Likewise.
25173 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25174 * init-regs.c (initialize_uninitialized_regs): Likewise.
25175 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25176 (inline_small_functions): Likewise.
25177 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25178 * ira.c (combine_and_move_insns): Likewise.
25179 (build_insn_chain): Likewise.
25180 * loop-invariant.c (find_invariants): Likewise.
25181 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25182 * predict.c (tree_predict_by_opcode): Likewise.
25183 (predict_paths_leading_to): Likewise.
25184 (predict_paths_leading_to_edge): Likewise.
25185 (estimate_loops_at_level): Likewise.
25186 (estimate_loops): Likewise.
25187 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25188 (spread_components): Likewise.
25189 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25190 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25191 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25192 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25193 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25194 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25195 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25196 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25197 (mark_threaded_blocks): Likewise.
25198 (thread_through_all_blocks): Likewise.
25199 * tree-ssa.c (verify_ssa): Likewise.
25200 (execute_update_addresses_taken): Likewise.
25201 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25203 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25205 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25207 (post_order_compute): Likewise.
25208 (inverted_post_order_compute): Likewise.
25209 (pre_and_rev_post_order_compute_fn): Likewise.
25211 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25213 * genrecog.c (int_set::int_set): Explicitly construct our
25214 auto_vec base class.
25215 * vec.h (auto_vec::auto_vec): New constructor.
25217 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25219 * bitmap.h (class auto_bitmap): New constructor taking
25220 bitmap_obstack * argument.
25222 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25224 * bitmap.h (class auto_bitmap): Change type of m_bits to
25225 bitmap_head, and adjust ctor / dtor and member operators.
25227 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25229 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25230 when returned register mode doesn't match original mode.
25232 2017-05-12 Jeff Law <law@redhat.com>
25233 Jakub Jelinek <jakub@redhat.com>
25235 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25236 we look for cc setter after the compare-elim changes.
25237 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25238 within the vector to match what compare-elim now expects.
25239 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25240 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25242 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25243 after the compare-elim changes.
25244 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25245 the vector to match what compare-elim now expects.
25246 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25247 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25248 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25249 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25250 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25252 * config/visium/visium.c (single_set_and_flags): Fix where
25253 we look for cc setter after the compare-elim changes.
25254 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25255 with the vector to match what compare-elim now expects.
25256 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25257 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25258 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25259 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25260 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25261 (neg<mode>2_insn_set_overflow): Likewise.
25263 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25265 PR middle-end/79794
25266 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25267 maybe_expand_insn call, set ops[0].target. If still set after call,
25268 set alt_rtl. Add extra arg to recursive calls.
25269 (extract_bit_field): Add alt_rtl argument. Pass to
25271 * expmed.h (extract_bit_field): Fix prototype.
25272 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25273 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25274 to extract_bit_field_calls.
25275 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25276 Pass alt_rtl to extract_bit_field calls.
25277 * calls.c (store_unaligned_arguments_into_psuedos)
25278 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25279 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25281 * optabs.h (struct expand_operand): Add target bitfield.
25283 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25285 * compare-elim.c (try_eliminate_compare): Canonicalize
25286 operation with embedded compare to
25287 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25288 (set (reg) (operation)].
25290 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25292 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25295 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25296 cost of adding a carry flag for ADC instruction.
25297 [case MINUS]: Ignore the cost of subtracting a carry flag
25298 for SBB instruction.
25300 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25302 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25304 * config/rs6000/bmiintrin.h: New file.
25305 * config/rs6000/bmi2intrin.h: New file.
25306 * config/rs6000/x86intrin.h: New file.
25308 2017-05-12 Jeff Law <law@redhat.com>
25310 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25313 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25315 PR middle-end/80707
25316 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25318 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25320 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25321 early expansion of vector divide builtins.
25322 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25323 builtins identified as having unsigned arguments.
25325 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25327 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25328 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25329 expansion of vector logical operations (and, andc, or, xor,
25332 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25334 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25335 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25337 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25339 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25340 early GIMPLE expansion of vector multiplies.
25342 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25344 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25345 TARGET_HAVE_MOVT conditional.
25346 (movt splitter): Likewise.
25348 2017-05-12 Richard Biener <rguenther@suse.de>
25350 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25351 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25352 Fold all stmts not inplace.
25354 2017-05-12 Richard Biener <rguenther@suse.de>
25356 PR tree-optimization/80713
25357 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25358 inserted_exprs bit for not removed stmts.
25360 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25362 PR middle-end/69921
25363 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25364 parallelized" attribute for parallelized OpenACC kernels.
25365 * omp-offload.c (execute_oacc_device_lower): Use it.
25367 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25368 Set "oacc kernels" attribute.
25369 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25370 parameter. Adjust all users.
25371 (oacc_fn_attrib_kernels_p): Remove function.
25372 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25373 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25374 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25375 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25376 assert "oacc kernels" attribute is set.
25378 2017-05-11 Carl Love <cel@us.ibm.com>
25380 * config/rs6000/rs6000-c: Add support for built-in functions
25381 vector unsigned char vec_popcnt (vector signed char)
25382 vector unsigned char vec_popcnt (vector unsigned char)
25383 vector unsigned short vec_popcnt (vector signed short)
25384 vector unsigned short vec_popcnt (vector unsigned short)
25385 vector unsigned int vec_popcnt (vector signed int)
25386 vector unsigned int vec_popcnt (vector unsigned int)
25387 vector unsigned long long vec_popcnt (vector signed long long)
25388 vector unsigned long long vec_popcnt (vector unsigned long long)
25389 vector signed long long vec_slo (vector signed long long,
25390 vector signed char)
25391 vector signed long long vec_slo (vector signed long long,
25392 vector unsigned char)
25393 vector unsigned long long vec_slo (vector unsigned long long,
25394 vector signed char)
25395 vector unsigned long long vec_slo (vector unsigned long long,
25396 vector unsigned char)
25397 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25398 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25399 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25400 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25401 * doc/extend.texi: Update the built-in documentation file for the
25402 new built-in functions.
25404 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25406 * attribs.h (sorted_attr_string): Move machine independent
25407 functions for target clone support from the i386 port to common
25408 code. Rename ix86_function_versions to common_function_versions.
25409 Rename make_name to make_unique_name.
25410 (common_function_versions): Likewise.
25411 (make_unique_name): Likewise.
25412 (make_dispatcher_decl): Likewise.
25413 (is_function_default_version): Likewise.
25414 * attribs.c (attr_strcmp): Likewise.
25415 (sorted_attr_string): Likewise.
25416 (common_function_versions): Likewise.
25417 (make_unique_name): Likewise.
25418 (make_dispatcher_decl): Likewise.
25419 (is_function_default_version): Likewise.
25420 * config/i386/i386.c (attr_strcmp): Likewise.
25421 (sorted_attr_string): Likewise.
25422 (ix86_function_versions): Likewise.
25423 (make_name): Likewise.
25424 (make_dispatcher_decl): Likewise.
25425 (is_function_default_version): Likewise.
25426 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25428 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25431 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25432 Account for direct move costs for vec_construct of integer
25435 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25438 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25439 (UNSPEC_STX_ATOMIC): Ditto.
25440 (loaddi_via_sse): New insn.
25441 (storedi_via_sse): Ditto.
25442 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25443 Update corresponding peephole2 patterns.
25444 (atomic_storedi_fpu): Ditto.
25446 2017-05-11 Julia Koval <julia.koval@intel.com>
25448 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25449 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25451 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25452 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25453 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25455 2017-05-11 Nathan Sidwell <nathan@acm.org>
25457 * graphite-poly.c: Include dumpfile.h.
25459 * dumpfle.h (dump_function): Declare here ...
25460 * tree-dump.h (dump_function): ... not here.
25461 * dumpfile.c: #include tree-cfg.h.
25462 (dump_function): Move here from ...
25463 * tree-dump.c (dump_function): ... here.
25464 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25465 * graphite-poly.c: Don't include tree-dump.h.
25466 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25467 * print-tree.c: Likewise.
25468 * stor-layout.c: Likewise.
25469 * tree-nested.c: Likewise.
25471 * dumpfile.c (dump_start): Use TDF_FLAGS.
25472 (dump_enable_all): Fix TDF_KIND check thinko.
25474 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25476 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25477 array entries to represent two legal parameterizations of the
25478 overloaded __builtin_cmpb function, as represented by the
25479 P6_OV_BUILTIN_CMPB constant.
25480 (altivec_resolve_overloaded_builtin): Add special case handling
25481 for the __builtin_cmpb function, as represented by the
25482 P6_OV_BUILTIN_CMPB constant.
25483 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25484 (BU_P6_64BIT_2): New macro.
25485 (BU_P6_OVERLOAD_2): New macro
25486 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25487 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25488 (CMPB): Add overload support to represent both 32-bit and 64-bit
25489 compare-bytes function.
25490 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25491 support for TARGET_CMPB.
25492 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25493 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25494 documentation of the __builtin_cmpb overloaded built-in function.
25496 2017-05-11 Richard Biener <rguenther@suse.de>
25498 PR tree-optimization/80705
25499 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25500 bases are not vectorizable.
25502 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25504 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25505 when counting register pressure.
25507 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25509 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25510 (iv_ca_more_deps): Renamed to ...
25511 (iv_ca_compare_deps): ... this.
25512 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25514 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25516 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25518 (determine_group_iv_costs): ... here.
25519 (find_inv_vars_cb): Record inv var if it's not recorded before.
25521 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25523 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25524 (get_shiftadd_cost): Ditto.
25526 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25528 * tree-ssa-address.c: Include header file.
25529 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25531 (add_to_parts): Refactor.
25532 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25533 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25536 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25538 PR tree-optimization/53090
25539 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25541 (extract_cond_operands): Detect condition with IV on both sides
25542 and return COMP_IV_EXPR_2.
25543 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25544 (rewrite_use_compare): Simplify by removing call to function
25545 extract_cond_operands.
25547 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25549 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25550 (extract_cond_operands): Detect condition comparing against non-
25551 invariant bound and return appropriate enum value.
25552 (find_interesting_uses_cond): Update use of extract_cond_operands.
25553 Handle its return value accordingly.
25554 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25556 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25558 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25559 nonlinear iv_use computation in loop invariant sensitive way.
25561 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25563 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25564 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25566 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25568 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25569 (dump_cand): Support iv_cand.inv_exprs.
25570 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25572 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25575 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25577 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25579 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25580 as local function. Include necessary header files.
25581 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25583 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25585 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25589 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25590 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25591 RSHIFT_EXPR and BIT_NOT_EXPR.
25593 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25595 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25596 (adjust_setup_cost): New parameter supporting round up adjustment.
25597 (struct address_cost_data): Delete.
25598 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25599 (split_address_cost, ptr_difference_cost): Delete.
25600 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25601 (struct ainc_cost_data): New struct.
25602 (get_address_cost_ainc): New function.
25603 (get_address_cost, get_computation_cost): Reimplement.
25604 (determine_group_iv_cost_address): Record inv_expr for all uses of
25606 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25607 (iv_ca_has_deps): Reimplemented to ...
25608 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25610 (iv_ca_extend): Call iv_ca_more_deps.
25612 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25614 * tree-ssa-address.c (struct mem_address): Move to header file.
25615 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25616 * tree-ssa-address.h (struct mem_address): Move from C file.
25617 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25619 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25621 * tree-affine.h (aff_combination_type): New interface.
25622 (aff_combination_zero_p): Remove static.
25623 (aff_combination_const_p): New interface.
25624 (aff_combination_singleton_var_p): New interfaces.
25626 2017-05-11 Richard Biener <rguenther@suse.de>
25628 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25629 Skip unreachable blocks and destinations.
25630 (eliminate): Move stmt removal and fixup ...
25631 (fini_eliminate): ... here. Skip inserted exprs.
25632 (pass_pre::execute): Move fini_pre after fini_eliminate.
25633 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25634 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25635 PRE to get rid of dead code that has invalid SSA form and
25636 split critical edges again.
25638 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25640 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25642 2017-05-11 Richard Biener <rguenther@suse.de>
25644 * passes.c (execute_function_todo): Verify loops if they are
25645 said to be up-to-date.
25646 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25647 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25649 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25652 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25653 handle calling assemble_external ourself.
25656 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25657 modes with zero size. Enhance comment.
25659 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25661 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25662 built-ins for vec_xl and vec_xst with short and char pointer
25665 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25667 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25668 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25669 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25670 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25671 (_mm_maskz_min_round_ss): New intrinsics.
25672 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25673 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25674 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25675 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25676 (__builtin_ia32_minss_mask_round): New builtins.
25677 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25678 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25679 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25681 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25682 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25684 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25687 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25689 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25690 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25691 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25692 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25693 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25694 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25695 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25696 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25697 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25698 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25699 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25700 (__builtin_ia32_mulss_mask_round): New builtins.
25701 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25702 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25703 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25705 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25706 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25708 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25711 2017-05-10 Julia Koval <julia.koval@intel.com>
25713 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25714 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25715 (_mm256_setr_m128i): New intrinsics.
25717 2017-05-10 Julia Koval <julia.koval@intel.com>
25719 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25720 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25721 (_mm_maskz_rcp14_ss): New intrinsics.
25722 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25723 (__builtin_ia32_rcp14ss_mask): New builtins.
25724 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25726 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25728 PR tree-optimization/51513
25729 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25730 (assert_unreachable_fallthru_edge_p): Use it.
25731 (group_case_labels_stmt): Likewise.
25732 * tree-cfg.h: Prototype it.
25733 * stmt.c: Include cfghooks.h and tree-cfg.h.
25734 (emit_case_dispatch_table) <gap_label>: New local variable.
25735 Use it to fill dispatch table gaps.
25736 Test for default_label before updating probabilities.
25737 (expand_case) <default_label>: Remove unneeded initialization.
25738 Test for unreachable default case statement and remove its edge.
25739 Set default_label accordingly.
25740 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25742 2017-05-10 Carl Love <cel@us.ibm.com>
25744 * config/rs6000/rs6000-c: Add support for built-in functions
25745 vector signed char vec_neg (vector signed char)
25746 vector signed short int vec_neg (vector short int)
25747 vector signed int vec_neg (vector signed int)
25748 vector signed long long vec_neg (vector signed long long)
25749 vector float vec_neg (vector float)
25750 vector double vec_neg (vector double)
25751 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25753 * config/rs6000/altivec.h: Add define for vec_neg
25754 * doc/extend.texi: Update the built-in documentation for the
25755 new built-in functions.
25757 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25759 PR tree-optimization/77644
25760 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25762 2017-05-10 Nathan Sidwell <nathan@acm.org>
25764 * dumpfile.h (TDI_lang_all): New.
25765 (TDF_KIND): New. Renumber others
25766 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25768 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25770 (get_dump_file_name): Adjust suffix generation.
25771 (dump_enable_all): Use TDF_KIND.
25772 * doc/invoke.texi (-fdump-lang-all): Document.
25774 * dumpfile.h: Tabify.
25776 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25779 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25780 Move member access before delete.
25782 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25784 * tree-inline.c (expand_call_inline): Split block at stmt
25787 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25790 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25791 are now unused after splitting mov{sf,sd}_hardfloat.
25792 (f32_lr2): Likewise.
25793 (f32_lm): Likewise.
25794 (f32_lm2): Likewise.
25795 (f32_li): Likewise.
25796 (f32_li2): Likewise.
25797 (f32_lv): Likewise.
25798 (f32_sr): Likewise.
25799 (f32_sr2): Likewise.
25800 (f32_sm): Likewise.
25801 (f32_sm2): Likewise.
25802 (f32_si): Likewise.
25803 (f32_si2): Likewise.
25804 (f32_sv): Likewise.
25805 (f32_dm): Likewise.
25806 (f32_vsx): Likewise.
25807 (f32_av): Likewise.
25808 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25809 For movsf, order stores so the VSX stores occur before the GPR
25810 store which encourages the register allocator to use a traditional
25811 FPR instead of a GPR. For movsd, order the stores so that the GPR
25812 store comes before the VSX stores to allow the power6 to work.
25813 This is due to the power6 not having a 32-bit integer store
25814 instruction from a FPR.
25815 (movsf_hardfloat): Likewise.
25816 (movsd_hardfloat): Likewise.
25818 2017-05-09 Martin Sebor <msebor@redhat.com>
25820 PR translation/80280
25821 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25824 PR translation/80280
25825 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25826 data member added in r247778.
25827 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25829 2017-05-09 Nathan Sidwell <nathan@acm.org>
25831 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25833 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25836 2017-05-09 Marek Polacek <polacek@redhat.com>
25838 * doc/invoke.texi: Fix typo.
25840 2017-05-09 Richard Biener <rguenther@suse.de>
25842 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25843 (vrp_val_is_min): Likewise.
25844 (set_value_range_to_value): Likewise.
25845 (set_value_range_to_nonnegative): Likewise.
25846 (gimple_assign_nonzero_p): Likewise.
25847 (gimple_stmt_nonzero_p): Likewise.
25848 (vrp_int_const_binop): Likewise. Remove unreachable case.
25849 (adjust_range_with_scev): Adjust comments.
25850 (compare_range_with_value): Likewise.
25851 (extract_range_from_phi_node): Likewise.
25852 (test_for_singularity): Likewise.
25854 2017-05-09 Richard Biener <rguenther@suse.de>
25856 * tree-vrp.c (get_single_symbol): Add assert that we don't
25857 get overflowed constants as invariant part.
25858 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25859 checks. Use wi::cmp instead of recursing for integer constants.
25860 (compare_values): Just ignore whether we assumed undefined
25861 overflow instead of failing the compare.
25862 (extract_range_for_var_from_comparison_expr): Add comment before the
25863 TREE_NO_WARNING sets.
25864 (test_for_singularity): Likewise.
25865 (extract_range_from_comparison): Do not disable optimization
25866 when we assumed undefined overflow.
25867 (extract_range_basic): Remove init of unused var.
25869 2017-05-09 Richard Biener <rguenther@suse.de>
25871 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25872 (extract_range_from_multiplicative_op_1): Adjust.
25873 (extract_range_from_binary_expr_1): Use int_const_binop.
25875 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25878 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25879 rs6000_store_data_bypass_p in seven define_bypass directives and
25880 in several comments.
25881 * config/rs6000/rs6000-protos.h: Add prototype for
25882 rs6000_store_data_bypass_p function.
25883 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25884 function implements slightly different (rs6000-specific) semantics
25885 than store_data_bypass_p, returning false rather than aborting
25886 with assertion error when arguments do not satisfy the
25887 requirements of store data bypass.
25888 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25889 rs6000_store_data_bypass_p.
25891 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25893 * config/xtensa/xtensa-protos.h
25894 (xtensa_initial_elimination_offset): New declaration.
25895 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25896 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25897 macro definition, add case for FRAME_POINTER_REGNUM when
25898 FRAME_GROWS_DOWNWARD.
25899 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25900 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25901 xtensa_initial_elimination_offset.
25903 2017-05-08 Nathan Sidwell <nathan@acm.org>
25905 * doc/invoke.texi: Alphabetize -fdump options.
25907 2017-05-08 Martin Sebor <msebor@redhat.com>
25909 PR translation/80280
25910 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25912 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25914 * target.def (compute_frame_layout): New optional target hook.
25915 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25916 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25917 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25919 * reload1.c (verify_initial_elim_offsets): Likewise.
25920 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25921 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25922 (arm_get_frame_offsets): Split up into this ...
25923 (arm_compute_frame_layout): ... and this function.
25925 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25927 * config/aarch64/constraints.md (Usa): New constraint.
25928 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25930 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25932 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25933 with_multilib_list after it has been checked.
25935 2017-05-08 Richard Biener <rguenther@suse.de>
25937 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25938 (bitmap_set_subtract_values): Likewise.
25940 2017-05-08 Richard Biener <rguenther@suse.de>
25942 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25943 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25945 (gimple_stmt_nonzero_warnv_p): Rename to ...
25946 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25948 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25949 (extract_range_basic): Adjust, do not disable propagation on
25950 strict overflow sensitive simplification.
25951 (vrp_visit_cond_stmt): Likewise.
25953 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25955 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25956 body size unconditionally.
25958 2017-05-07 Jeff Law <law@redhat.com>
25961 2017-05-06 Jeff Law <law@redhat.com>
25962 PR tree-optimization/78496
25963 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25966 PR tree-optimization/78496
25967 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25968 (simplify_stmt_using_ranges): Call it.
25969 (vrp_dom_walker::before_dom_children): Extract equivalences
25970 from an ASSERT_EXPR with an equality comparison against a
25973 2017-05-06 Jeff Law <law@redhat.com>
25975 PR tree-optimization/78496
25976 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25979 PR tree-optimization/78496
25980 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25981 (simplify_stmt_using_ranges): Call it.
25982 (vrp_dom_walker::before_dom_children): Extract equivalences
25983 from an ASSERT_EXPR with an equality comparison against a
25986 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25988 * lra-constraints.c (lra_copy_reg_equiv): New function.
25989 (split_reg): Use it to copy equivalence information from the
25990 original register to the spill register.
25992 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25994 PR rtl-optimization/75964
25995 * simplify-rtx.c (simplify_const_relational_operation): Remove
25996 invalid handling of comparisons of integer ABS.
25998 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26000 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26001 initialize to zero.
26002 (init_regs): Remove declaration.
26003 (function_arg_advance_32): Initialize error_p as boolean variable.
26005 2017-05-05 Nathan Sidwell <nathan@acm.org>
26007 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26008 lines. Use for (;;).
26010 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26012 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26013 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26014 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26015 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26016 VF=2 that require versioning.
26018 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26020 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26023 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26025 * diagnostic.h (diagnostic_override_option_index): Convert from
26026 macro to inline function.
26028 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26030 * diagnostic.c (last_module_changed_p): New function.
26031 (set_last_module): New function.
26032 (diagnostic_report_current_module): Convert macro usage to
26033 the above functions.
26034 * diagnostic.h (diagnostic_context::last_module): Strengthen
26035 from const line_map * to const line_map_ordinary *.
26036 (diagnostic_last_module_changed): Delete macro.
26037 (diagnostic_set_last_module): Delete macro.
26039 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26041 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26042 with diagnostic_report_diagnostic.
26043 (diagnostic_n_impl_richloc): Likewise.
26044 * diagnostic.h (report_diagnostic): Delete macro.
26045 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26046 with diagnostic_report_diagnostic.
26047 * substring-locations.c (format_warning_va): Likewise.
26049 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26051 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26052 save/restor of format_spec. Move option-printing code to...
26053 (print_option_information): ...this new function, and
26054 reimplement by simply printing to the pretty_printer,
26055 rather than appending to the format string.
26057 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26059 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26060 handling logic into...
26061 (update_effective_level_from_pragmas): ...this new function.
26063 2017-05-04 Andrew Waterman <andrew@sifive.com>
26065 * config/riscv/riscv.opt (mstrict-align): New option.
26066 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26067 (SLOW_UNALIGNED_ACCESS): Define.
26068 (riscv_slow_unaligned_access): Declare.
26069 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26071 (riscv_slow_unaligned_access): New variable.
26072 (rocket_tune_info): Set slow_unaligned_access to true.
26073 (optimize_size_tune_info): Set slow_unaligned_access to false.
26074 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26075 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26076 (riscv_option_override): Set riscv_slow_unaligned_access.
26077 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26079 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26081 * config/riscv/riscv.md: Unify indentation.
26083 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26088 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26090 (extendsi<mode>2): Add support for doing sign extension via
26091 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26092 don't have ISA 3.0 instructions.
26093 (extendsi<mode>2 splitter): Likewise.
26094 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26095 generate the normal insns since SImode can now go in vector
26096 registers. Disallow the special UNSPECs needed for previous
26097 machines to hide SImode being used. Add new insns
26098 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26099 (fix_trunc<mode>si2_stfiwx): Likewise.
26100 (fix_trunc<mode>si2_internal): Likewise.
26101 (fixuns_trunc<mode>si2): Likewise.
26102 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26103 (fctiw<u>z_<mode>_smallint): Likewise.
26104 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26105 of floating point to 32-bit integer from doing a direct move to
26106 the GPR registers to do a store.
26107 (fctiwz_<mode>): Break long line.
26109 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26111 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26112 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26113 (addr_list, addr_offset_valid_p): New.
26114 (split_address_groups): Check offset validity with above function.
26115 (gt-tree-ssa-loop-ivopts.h): Include header file.
26117 2017-05-05 Nathan Sidwell <nathan@acm.org>
26119 * config.gcc (arm*-*-*): Add missing 'fi'.
26121 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26123 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26124 included in -fopt-info does not matter.
26125 * doc/optinfo.texi (-fopt-info): Fix description of default
26126 behavour. Explicitly say order of options included in -fopt-info
26129 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26131 * config.gcc: Allow combinations of aprofile and rmprofile values for
26132 --with-multilib-list.
26133 * config/arm/t-multilib: New file.
26134 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26135 variables. Remove setting of ISA and floating-point ABI in
26136 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26137 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26138 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26139 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26141 * config/arm/t-rmprofile: Likewise except for the matches changes.
26142 * doc/install.texi (--with-multilib-list): Document the combination of
26143 aprofile and rmprofile values and warn about pitfalls in doing that.
26145 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26147 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26148 (movdi_aarch64): Likewise.
26150 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26152 PR tree-optimization/80632
26153 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26155 (build_arrays): Initialize it for virtual phis.
26156 (fix_phi_nodes): Use it for virtual phis.
26158 PR tree-optimization/80558
26159 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26160 [x, y] op z into [x op, y op z] for op & or | if conditions
26163 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26164 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26167 * config/arm/arm.md (use_literal_pool): Remove.
26168 (64-bit immediate split): No longer takes cost into consideration
26169 if arm_disable_literal_pool is enabled.
26170 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26171 used when arm_disable_literal_pool is enabled.
26172 (arm_max_const_double_inline_cost): Remove use of
26173 arm_disable_literal_pool.
26174 (push_minipool_fix): Add assert.
26175 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26176 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26177 (no_literal_pool_sf_immediate): New.
26179 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26181 PR tree-optimization/80613
26182 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26183 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26185 2017-05-05 Richard Biener <rguenther@suse.de>
26187 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26189 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26191 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26192 of this flag from insn conditions due to removal from r247495.
26194 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26196 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26198 (arm_early_store_addr_dep_ptr): Likewise.
26199 * config/arm/aarch-common-protos.h
26200 (arm_early_load_addr_dep_ptr): Add prototype.
26201 (arm_early_store_addr_dep_ptr): Likewise.
26202 * config/arm/cortex-a53.md: Add new bypasses.
26204 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26206 * tree.c (next_type_uid): Change type to unsigned.
26207 (type_hash_canon): Decrement back next_type_uid if
26208 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26209 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26212 2017-05-04 Martin Sebor <msebor@redhat.com>
26214 * builtins.c: Fix a trivial typo in a comment.
26216 PR middle-end/79234
26217 * builtins.c (check_sizes): Adjust to handle reading past the end.
26218 Avoid printing excessive upper bound of ranges. Use %E to print
26219 tree nodes instead of converting them to %wu.
26220 (expand_builtin_memchr): New function.
26221 (compute_dest_size): Rename...
26222 (compute_objsize): ...to this.
26223 (expand_builtin_memcpy): Adjust.
26224 (expand_builtin_mempcpy): Adjust.
26225 (expand_builtin_strcat): Adjust.
26226 (expand_builtin_strcpy): Adjust.
26227 (check_strncat_sizes): Adjust.
26228 (expand_builtin_strncat): Adjust.
26229 (expand_builtin_strncpy): Adjust and simplify.
26230 (expand_builtin_memset): Adjust.
26231 (expand_builtin_bzero): Adjust.
26232 (expand_builtin_memcmp): Adjust.
26233 (expand_builtin): Handle memcmp.
26234 (maybe_emit_chk_warning): Check strncat just once.
26236 2017-05-04 Martin Sebor <msebor@redhat.com>
26238 PR preprocessor/79214
26239 PR middle-end/79222
26240 PR middle-end/79223
26241 * builtins.c (check_sizes): Add inlining context and issue
26242 warnings even when -Wno-system-headers is set.
26243 (check_strncat_sizes): Same.
26244 (expand_builtin_strncat): Same.
26245 (expand_builtin_memmove): New function.
26246 (expand_builtin_stpncpy): Same.
26247 (expand_builtin): Handle memmove and stpncpy.
26249 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26251 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26252 which is not used any more.
26254 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26256 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26258 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26260 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26261 (cortexa53_tunings): Likewise.
26262 (cortexa57_tunings): Likewise.
26263 (cortexa72_tunings): Likewise.
26264 (cortexa73_tunings): Likewise.
26266 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26268 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26269 Set loop alignment to 8.
26271 2017-05-04 Martin Sebor <msebor@redhat.com>
26273 PR translation/80280
26274 * builtins.c (expand_builtin_object_size): Add missing quoting to
26275 %D and like directives.
26276 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26277 (hsa_type_for_tree_type): Same.
26278 (verify_function_arguments): Same.
26279 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26280 * varasm.c (get_section): Same.
26283 2017-05-04 Martin Sebor <msebor@redhat.com>
26285 PR translation/80280
26286 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26288 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26290 * config/aarch64/aarch64.c (generic_addrcost_table):
26291 Change HI/TI mode setting.
26293 2017-05-04 Martin Jambor <mjambor@suse.cz>
26295 PR tree-optimization/80622
26296 * tree-sra.c (comes_initialized_p): New function.
26297 (build_accesses_from_assign): Only set write lazily when
26298 comes_initialized_p is false.
26299 (analyze_access_subtree): Use comes_initialized_p.
26300 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26301 instead of testing for PARM_DECL.
26303 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26305 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26306 constraint on operand 0 to allow more general addressing modes.
26307 Adjust output template.
26308 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26310 * config/aarch64/aarch64-protos.h
26311 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26312 * config/aarch64/constraints.md (Dp): New address constraint.
26313 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26316 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26318 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26319 update use of estimate_ipcp_clone_size_and_time.
26320 (estimate_local_effects): Update use of
26321 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26322 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26323 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26324 Return nonspecialized time.
26326 2017-05-04 Richard Biener <rguenther@suse.de>
26328 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26329 for the last VUSE which def dominates the PHI. Directly call
26331 (get_continuation_for_phi_1): Remove.
26333 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26335 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26336 to explain the use of truncating division. Cap the number of
26337 iterations to the maximum given by nb_iterations_upper_bound,
26340 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26342 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26343 * configure: Regenerate.
26344 * config.in: Regenerate.
26345 * config/i386/driver-mingw32.c: new file.
26346 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26347 * config.host: Link driver-mingw32.o on MinGW host.
26348 * doc/install.texi: Document new --enable-mingw-wildcard configure
26351 2017-05-04 Marek Polacek <polacek@redhat.com>
26353 PR tree-optimization/80612
26354 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26356 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26357 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26359 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26360 (movt splitter): Likewise.
26361 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26362 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26363 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26364 block for Thumb-1 with MOVT.
26365 (thumb2_legitimate_address_p): Move code block ...
26366 (can_avoid_literal_pool_for_label_p): ... into this new function.
26367 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26369 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26370 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26371 "M-profile targets with the MOVT instruction".
26373 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26375 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26376 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26377 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26379 2017-05-04 Martin Liska <mliska@suse.cz>
26381 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26382 variable cond_code.
26384 2017-05-04 Richard Biener <rguenther@suse.de>
26386 * tree.c (array_at_struct_end_p): Handle arrays at struct
26387 end with flexarrays more conservatively. Refactor and treat
26388 arrays of arrays or aggregates more strict. Fix
26389 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26390 * tree.c (array_at_struct_end_p): Adjust prototype.
26391 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26392 * gimple-fold.c (get_range_strlen): Likewise.
26393 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26395 2017-05-04 Richard Biener <rguenther@suse.de>
26397 PR tree-optimization/31130
26398 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26400 (supports_overflow_infinity): Likewise.
26401 (is_negative_overflow_infinity): Likewise.
26402 (is_positive_overflow_infinity): Likewise.
26403 (is_overflow_infinity): Likewise.
26404 (stmt_overflow_infinity): Likewise.
26405 (overflow_infinity_range_p): Likewise.
26406 (usable_range_p): Remove as always returning true.
26407 (make_overflow_infinity): Remove.
26408 (negative_overflow_infinity): Likewise.
26409 (positive_overflow_infinity): Likewise.
26410 (avoid_overflow_infinity): Likewise.
26411 (set_value_range): Adjust accordingly.
26412 (set_value_range_to_nonnegative): Likewise, remove now unused
26413 overflow_infinity arg.
26414 (vrp_operand_equal_p): Adjust.
26415 (update_value_range): Likewise.
26416 (range_int_cst_singleton_p): Likewise.
26417 (operand_less_p): Likewise.
26418 (compare_values_warnv): Likewise.
26419 (extract_range_for_var_from_comparison_expr): Likewise.
26420 (vrp_int_const_binop): Likewise.
26421 (zero_nonzero_bits_from_vr): Likewise.
26422 (extract_range_from_multiplicative_op_1): Likewise.
26423 (extract_range_from_binary_expr_1): Likewise.
26424 (extract_range_from_unary_expr): Likewise.
26425 (extract_range_from_comparison): Likewise.
26426 (extract_range_basic): Likewise.
26427 (adjust_range_with_scev): Likewise.
26428 (compare_ranges): Likewise.
26429 (compare_range_with_value): Likewise.
26430 (dump_value_range): Likewise.
26431 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26433 (simplify_cond_using_ranges): Adjust.
26435 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26437 * brig-builtins.def: Added a builtin for class_f64.
26438 * builtin-types.def: Added a builtin type needed by class_f64.
26440 2017-05-03 Jason Merrill <jason@redhat.com>
26442 * timevar.def: Add TV_CONSTEXPR.
26444 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26446 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26448 2017-05-03 Martin Jambor <mjambor@suse.cz>
26450 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26451 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26452 * ipa-cp.c (ipcp_propagate_stage): Do not call
26453 ipa_update_after_lto_read.
26454 * ipa-inline.c (ipa_inline): Likewise.
26456 2017-05-03 Martin Jambor <mjambor@suse.cz>
26458 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26459 tag. Added a default constructor and a destructor.
26460 (ipa_edge_args_sum_t): New class;
26461 (ipa_edge_args_sum): Declare.
26462 (ipa_edge_args_vector): Remove declaration.
26463 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26464 (ipa_free_edge_args_substructures): Remove declaration.
26465 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26466 (ipa_edge_args_info_available_for_edge_p): Likewise.
26467 * ipa-prop.c (ipa_edge_args_vector): Removed.
26468 (edge_removal_hook_holder): Likewise.
26469 (edge_duplication_hook_holder): Likewise.
26470 (ipa_edge_args_sum): New variable.
26471 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26472 ipa_edge_args_vector.
26473 (ipa_free_edge_args_substructures): Likewise.
26474 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26475 ipa_edge_args_vector.
26476 (ipa_edge_removal_hook): Turned into method
26477 ipa_edge_args_sum_t::remove.
26478 (ipa_edge_duplication_hook): Turned into method
26479 ipa_edge_args_sum_t::duplicate.
26480 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26481 registering edge hooks.
26482 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26483 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26484 ipa_edge_args_sum instead of ipa_edge_args_vector.
26485 * ipa-profile.c (ipa_profile): Likewise.
26487 2017-05-03 Martin Jambor <mjambor@suse.cz>
26489 * symbol-summary.h (function_summary): New method exists.
26490 (function_summary::symtab_removal): Deallocate through release.
26491 (call_summary): New class.
26492 (gt_ggc_mx): New overload.
26493 (gt_pch_nx): Likewise.
26494 (gt_pch_nx): Likewise.
26496 2017-05-03 Jeff Law <law@redhat.com>
26498 PR tree-optimization/78496
26499 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26500 from simplify_cond_using_ranges. Split off code to walk
26501 backwards through casts into ...
26502 (simplify_cond_using_ranges_2): New function.
26503 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26504 (execute_vrp): After identifying jump threads, call
26505 simplify_cond_using_ranges_2.
26507 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26510 * ipa-inline.h (inline_summary): Add ctor.
26511 (create_ggc): Do not use ggc_cleared_alloc.
26513 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26514 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26516 * gcc.c (handle_braces): Support escaping in switch matching
26518 * doc/invoke.texi (Spec Files): Document it.
26519 Remove superfluous @code markup in items.
26521 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26523 * diagnostic-show-locus.c (struct column_range): New struct.
26524 (get_affected_columns): New function.
26525 (get_printed_columns): New function.
26526 (struct correction): New struct.
26527 (correction::ensure_capacity): New function.
26528 (correction::ensure_terminated): New function.
26529 (struct line_corrections): New struct.
26530 (line_corrections::~line_corrections): New dtor.
26531 (line_corrections::add_hint): New function.
26532 (layout::print_trailing_fixits): Reimplement in terms of the new
26534 (selftest::test_overlapped_fixit_printing): New function.
26535 (selftest::diagnostic_show_locus_c_tests): Call it.
26537 2017-05-03 Nathan Sidwell <nathan@acm.org>
26539 Canonicalize canonical type hashing
26540 * tree.h (type_hash_canon_hash): Declare.
26541 * tree.c (type_hash_list, attribute_hash_list): Move into
26542 type_hash_canon_hash.
26543 (build_type_attribute_qual_variant): Break out hash code calc into
26544 type_hash_canon_hash.
26545 (type_hash_canon_hash): New. Generic type hash computation.
26546 (build_range_type_1, build_array_type_1, build_function_type,
26547 build_method_type_directly, build_offset_type, build_complex_type,
26548 make_vector_type): Call it.
26550 2017-05-03 Richard Biener <rguenther@suse.de>
26552 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26553 When all DRs have unknown misaligned do not always peel
26554 when there is a store but apply the same costing model as if
26555 there were only loads.
26557 2017-05-03 Richard Biener <rguenther@suse.de>
26560 PR tree-optimization/80492
26561 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26562 compare_base_decls returning dont-know properly.
26564 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26566 * config/arm/iterators.md (CCSI): New mode iterator.
26567 (arch): New mode attribute.
26568 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26569 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26570 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26571 code iterator for success result mode.
26572 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26573 the corresponding new insn generators.
26575 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26578 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26579 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26581 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26583 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26584 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26585 (DDR_A): Wrap DDR argument in brackets.
26586 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26587 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26588 (DDR_REVERSED_P): Likewise.
26590 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26592 PR tree-optimization/79472
26593 * tree-switch-conversion.c (struct switch_conv_info): Add
26594 contiguous_range and default_case_nonstandard fields.
26595 (collect_switch_conv_info): Compute contiguous_range and
26596 default_case_nonstandard fields, don't clear final_bb if
26597 contiguous_range and only the default case doesn't have the required
26599 (check_all_empty_except_final): Set default_case_nonstandard instead
26600 of failing if contiguous_range and the default case doesn't have empty
26602 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26603 and only the default case doesn't have the required constants. Skip
26605 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26606 if default_case_nonstandard.
26607 (build_constructors): Build constant 1 just once. Assert that default
26608 values aren't inserted in between cases if contiguous_range. Skip
26610 (build_arrays): Skip virtual phis.
26611 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26612 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26613 Handle virtual phis.
26614 (gen_inbound_check): Handle default_case_nonstandard case.
26615 (process_switch): Adjust check_final_bb caller. Call
26616 gather_default_values with the first non-default case instead of
26617 default case if default_case_nonstandard.
26619 2017-05-02 Nathan Sidwell <nathan@acm.org>
26621 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26622 check. Fix formatting.
26624 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26626 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26627 errors when comparing specialized and unspecialized times.
26629 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26631 * diagnostic-show-locus.c
26632 (layout::should_print_annotation_line_p): Make private.
26633 (layout::print_annotation_line): Make private.
26634 (layout::annotation_line_showed_range_p): Make private.
26635 (layout::show_ruler): Make private.
26636 (layout::print_source_line): Make private. Pass in line and
26637 line_width, rather than calling location_get_source_line. Drop
26639 (layout::print_leading_fixits): New method.
26640 (layout::print_any_fixits): Rename to...
26641 (layout::print_trailing_fixits): ...this, and make private.
26642 Don't print newline fixits.
26643 (diagnostic_show_locus): Move logic for printing one row into...
26644 (layout::print_line): ...this new function. Move the
26645 location_get_source_line call and error-handling from
26646 print_source_line to here. Call print_leading_fixits, and rename
26647 print_any_fixits to print_trailing_fixits.
26648 (selftest::test_fixit_insert_containing_newline): Update now that
26649 newlines are partially supported.
26650 (selftest::test_fixit_insert_containing_newline_2): New test.
26651 (selftest::test_fixit_replace_containing_newline): Update comments.
26652 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26653 * edit-context.c (class added_line): New class.
26654 (class edited_line): Describe newline handling in comment.
26655 (edited_line::actually_edited_p): New method.
26656 (edited_line::print_content): Delete redundant decl.
26657 (edited_line::m_predecessors): New field.
26658 (edited_file::print_content): Call edited_line::print_content.
26659 (edited_file::print_diff): Update to support newlines.
26660 (edited_file::print_diff_hunk): Likewise.
26661 (edited_file::print_run_of_changed_lines): New function.
26662 (edited_file::print_diff_line): Convert to...
26663 (print_diff_line): ...this.
26664 (edited_file::get_effective_line_count): New function.
26665 (edited_line::edited_line): Initialize new field m_predecessors.
26666 (edited_line::~edited_line): Clean up m_predecessors.
26667 (edited_line::apply_fixit): Handle newlines.
26668 (edited_line::get_effective_line_count): New function.
26669 (edited_line::print_content): New function.
26670 (edited_line::print_diff_lines): New function.
26671 (selftest::test_applying_fixits_insert_containing_newline): New
26673 (selftest::test_applying_fixits_replace_containing_newline): New
26675 (selftest::insert_line): New function.
26676 (selftest::test_applying_fixits_multiple_lines): Add example of
26678 (selftest::edit_context_c_tests): Call the new tests.
26680 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26682 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26683 parameter cand. Update dump information.
26684 (get_computation_cost): Update uses.
26686 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26688 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26689 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26690 (get_computation_at, rewrite_use_address): Update use of
26691 get_computation_aff.
26693 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26695 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26696 (get_computation): Delete.
26697 (get_computation_cost): Implement like get_computation_cost_at.
26698 Use get_computation_at.
26699 (get_computation_cost_at): Delete.
26700 (rewrite_use_nonlinear_expr): Use get_computation_at.
26701 (rewrite_use_compare, remove_unused_ivs): Ditto.
26703 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26705 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26707 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26709 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26710 (ivopts_global_cost_for_size): Rename parameter and update uses.
26711 (iv_ca_recount_cost): Update uses.
26712 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26713 candidates seperately in n_invs and n_cands.
26714 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26716 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26718 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26719 (find_inv_vars_cb): New.
26720 (find_depends): Renamed to ...
26721 (find_inv_vars): ... this.
26722 (add_candidate_1, force_var_cost): Call find_inv_vars.
26723 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26725 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26727 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26728 inv_vars. Add inv_exprs.
26729 (struct iv_cand): Rename depends_on to inv_vars.
26730 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26731 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26732 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26733 (dump_cand): Dump inv_vars.
26734 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26735 (record_invariant, find_depends, add_candidate_1): Ditto.
26736 (set_group_iv_cost, force_var_cost): Ditto.
26737 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26738 (get_computation_cost_at, get_computation_cost): Ditto.
26739 (determine_group_iv_cost_generic): Ditto.
26740 (determine_group_iv_cost_address): Ditto.
26741 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26742 (determine_group_iv_costs): Ditto.
26743 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26744 (iv_ca_set_remove_invariants): Renamed to ...
26745 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26746 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26747 (iv_ca_set_add_invariants): Renamed to ...
26748 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26749 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26750 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26751 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26752 (create_new_ivs): Remove useless dump.
26754 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26756 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26758 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26759 (iv_ca_set_no_cp, create_new_iv): Ditto.
26761 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26763 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26765 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26767 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26768 function tree_check2.
26770 2017-05-02 Martin Liska <mliska@suse.cz>
26772 * doc/gcov.texi: Add missing preposition.
26773 * gcov.c (function_info::function_info): Properly fill up
26774 all member variables.
26776 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26778 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26780 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26782 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26784 2017-05-02 Martin Liska <mliska@suse.cz>
26787 * lto-streamer-in.c (lto_read_tree_1): Remove
26788 LTO_STREAMER_DEBUG.
26789 * lto-streamer.c (struct tree_hash_entry): Likewise.
26790 (struct tree_entry_hasher): Likewise.
26791 (tree_entry_hasher::hash): Likewise.
26792 (tree_entry_hasher::equal): Likewise.
26793 (lto_streamer_init): Likewise.
26794 (lto_orig_address_map): Likewise.
26795 (lto_orig_address_get): Likewise.
26796 (lto_orig_address_remove): Likewise.
26797 * lto-streamer.h: Likewise.
26798 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26799 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26801 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26803 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26804 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26805 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26806 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26807 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26808 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26809 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26810 (mm_maskz_sub_ss): New intrinsics.
26811 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26812 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26813 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26814 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26815 (__builtin_ia32_subss_mask_round): New builtins.
26816 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26817 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26818 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26820 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26821 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26823 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26826 2017-05-02 Martin Jambor <mjambor@suse.cz>
26828 PR tree-optimization/78687
26829 * tree-sra.c (access): New field parent.
26830 (process_subtree_disqualification): New function.
26831 (disqualify_candidate): Call it.
26832 (build_accesses_from_assign): Reset write flag if creating an
26834 (build_access_subtree): Fill in parent field and also prpagate
26835 down grp_write flag.
26836 (create_artificial_child_access): New parameter set_grp_write, set
26837 grp_write to its value.
26838 (propagate_subaccesses_across_link): Also propagate grp_write flag
26840 (propagate_all_subaccesses): Push the closest parent back to work
26841 queue if add_access_to_work_queue returned true.
26843 2017-05-02 Richard Biener <rguenther@suse.de>
26845 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26846 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26847 -fstrict-overflow documentation.
26848 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26849 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26850 flag_strict_overflow.
26851 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26852 * lto-opts.c (lto_write_options): Do not stream it.
26853 * lto-wrapper.c (merge_and_complain): Do not handle it.
26854 * opts.c (default_options_table): Do not set -fstrict-overflow.
26855 (finish_options): Likewise do not clear it when sanitizing.
26856 * simplify-rtx.c (simplify_const_relational_operation): Do not
26857 test flag_strict_overflow.
26859 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26861 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26862 using enabled attribute.
26863 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26864 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26865 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26866 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26867 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26868 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26869 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26870 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26871 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26872 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26874 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26876 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26878 2017-05-02 Richard Biener <rguenther@suse.de>
26880 PR tree-optimization/80591
26882 2017-04-10 Richard Biener <rguenther@suse.de>
26884 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26887 2017-05-02 Richard Biener <rguenther@suse.de>
26889 PR tree-optimization/80549
26890 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26891 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26892 headers if they do not have a preheader.
26894 2017-05-02 Martin Liska <mliska@suse.cz>
26897 * common.opt: Fix typo.
26898 * doc/invoke.texi: Likewise.
26900 2017-05-01 Jan Beulich <jbeulich@suse.com>
26902 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26903 swapping, add (x,x,m,x,n) alternative.
26905 2017-05-01 Nathan Sidwell <nathan@acm.org>
26907 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26908 unnecessary unadjusted_alignment check.
26910 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26913 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26915 * gimplify.c (gimplify_cilk_detach): New function.
26916 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26917 * tree-core.h: Document EXPR_CILK_SPAWN.
26918 * tree.h (EXPR_CILK_SPAWN): Define.
26920 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26922 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26923 to use new fixit_hint representation, using the "replace" logic.
26924 (get_line_span_for_fixit_hint): Likewise.
26925 (layout::print_any_fixits): Likewise.
26926 (selftest::test_one_liner_many_fixits): Rename to...
26927 (selftest::test_one_liner_many_fixits_1): ...this, and update
26928 comment and expected output to reflect that the multiple fix-it
26929 hints are now consolidated into one insertion.
26930 (selftest::test_one_liner_many_fixits_2): New test.
26931 (selftest::test_diagnostic_show_locus_one_liner): Update for
26933 (selftest::test_fixit_consolidation): Update for fix-it API
26935 * diagnostic.c (print_parseable_fixits): Likewise.
26936 * edit-context.c (edited_line::m_line_events): Convert from
26937 auto_vec <line_event *> to auto_vec <line_event>.
26938 (class line_event): Convert from abstract base class to a concrete
26939 class, taking over the role of replace_event.
26940 (class insert_event): Delete.
26941 (class replace_event): Rename to class line_event. Convert to
26943 (edit_context::add_fixits): Reimplement.
26944 (edit_context::apply_insert): Delete.
26945 (edit_context::apply_replace): Rename to...
26946 (edit_context::apply_fixit): ...this. Convert to half-open range.
26947 (edited_file::apply_insert): Delete.
26948 (edited_file::apply_replace): Rename to...
26949 (edited_file::apply_fixit): ...this.
26950 (edited_line::~edited_line): Drop deletion of events.
26951 (edited_line::apply_insert): Delete.
26952 (edited_line::apply_replace): Rename to...
26953 (edited_line::apply_fixit): ...this. Convert to half-open range.
26954 Update for change to type of m_line_events.
26955 * edit-context.h (edit_context::apply_insert): Delete.
26956 (edit_context::apply_replace): Rename to...
26957 (edit_context::apply_fixit): ...this.
26959 2017-05-01 Martin Sebor <msebor@redhat.com>
26961 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26964 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26967 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26968 __get_cpuid_max returns 0.
26969 (__get_cpuid_count): Ditto.
26971 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26973 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26974 replacement expression is another instance of one of its arguments.
26976 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26979 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26980 check for stack push/pop autoinc.
26981 * config/i386/i386.c (ix86_agi_dependent): Return false
26982 if the only reason why modified_in_p returned true is that
26983 addr is SP based and set_insn is a push or pop.
26985 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26987 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26990 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26993 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26994 (account_size_time): Use two predicates - exec_pred and
26996 (evaluate_conditions_for_known_args): Compute both clause and
26998 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26999 (inline_summary_t::duplicate): Update.
27000 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27002 (compute_inline_parameters): Likewise.
27003 (estimate_edge_size_and_time): Update caluclation of time.
27004 (estimate_node_size_and_time): Compute both time and nonspecialized
27006 (estimate_ipcp_clone_size_and_time): Update.
27007 (inline_merge_summary): Update.
27008 (do_estimate_edge_time): Update.
27009 (do_estimate_edge_size): Update.
27010 (do_estimate_edge_hints): Update.
27011 (inline_read_section, inline_write_summary): Stream both new predicates.
27012 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27014 (compute_inlined_call_time): Cleanup.
27015 (big_speedup_p): Update.
27016 (edge_badness): Update.
27017 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27018 (size_time_entry): Replace predicate by exec_predicate and
27019 nonconst_predicate.
27020 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27021 (estimate_edge_time): Return also nonspec_time.
27022 (reset_edge_growth_cache): Update.
27024 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27026 PR rtl-optimization/80491
27027 * ifcvt.c (noce_process_if_block): When looking for x setter
27028 with missing else_bb, don't check only the insn right before
27029 cond_earliest, but look for the last insn that x is modified in
27030 within the same bb.
27032 PR rtl-optimization/80491
27033 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27035 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27037 PR tree-optimization/80487
27038 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27040 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27042 PR tree-optimization/79697
27043 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27044 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27045 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27047 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27048 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27050 2017-04-28 Martin Sebor <msebor@redhat.com>
27052 PR tree-optimization/80523
27053 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27054 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27056 (maybe_warn, format_directive, parse_directive): Use new functions.
27057 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27059 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27061 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27063 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27065 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27066 target_header_dir): Set correctly.
27067 * configure: Regenerated.
27068 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27069 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27070 instead of SYSTEM_HEADER_DIR.
27072 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27074 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27075 (estimate_local_effects): Likewise.
27076 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27077 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27078 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27079 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27080 do_estimate_edge_time, estimate_edge_time): Likewise.
27081 * ipa-inline-analysis.c (estimate_node_size_and_time,
27082 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27083 (estimate_time_after_inlining): Remove.
27085 2017-04-28 Martin Liska <mliska@suse.cz>
27087 * doc/gcov.texi: Enhance documentation of gcov.
27089 2017-04-28 Martin Liska <mliska@suse.cz>
27091 * doc/gcov.texi: Sort options in alphabetic order.
27092 * doc/gcov-dump.texi: Likewise.
27093 * doc/gcov-tool.texi: Likewise.
27094 * gcov.c (print_usage): Likewise.
27095 * gcov-dump.c (print_usage): Likewise.
27096 * gcov-tool.c (print_merge_usage_message): Likewise.
27097 (print_rewrite_usage_message): Likewise.
27098 (print_overlap_usage_message): Likewise.
27100 2017-04-28 Martin Liska <mliska@suse.cz>
27102 PR gcov-profile/53915
27103 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27105 2017-04-28 Martin Liska <mliska@suse.cz>
27107 PR gcov-profile/79891
27108 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27109 is marked by compiler as living on a line.
27110 (get_cycles_count): Remove usage of the union.
27111 (output_intermediate_file): Likewise.
27112 (find_source): Fix GNU coding style.
27113 (accumulate_line_counts): Remove old non-all block mode.
27114 (output_lines): Remove usage of the union.
27115 * profile.c (output_location): Include all BBs, even if
27116 belonging to a same line (and file) as a previous BB.
27118 2017-04-28 Martin Liska <mliska@suse.cz>
27120 * gcov.c (process_args): Handle new argument 'w'.
27121 (read_graph_file): Assign ID to BBs.
27122 (output_branch_count): Display BB # if verbose flag is set.
27123 (output_lines): Likewise for arcs.
27124 (print_usage): Add '--verbose' option help.
27125 * doc/gcov.texi: Document --verbose (-w) option.
27127 2017-04-28 Martin Liska <mliska@suse.cz>
27129 * gcov.c (struct block_location_info): New struct.
27130 (process_file): Fill up the new structure.
27131 (read_graph_file): Replace usage of encoding by the newly added
27133 (add_line_counts): Likewise.
27134 (accumulate_line_counts): Remove usage of the union.
27135 (function_info::function_info): New function.
27136 (function_info::~function_info): Likewise.
27137 (process_file): Call delete instead of release_function.
27138 (release_function): Release the function.
27139 (release_structures): Call delete instead of release_function.
27140 (solve_flow_graph): Replace usage of num_blocks.
27141 (find_exception_blocks): Likewise.
27142 (output_lines): Fix GNU coding style.
27144 2017-04-28 Martin Liska <mliska@suse.cz>
27147 * coverage.c (coverage_remove_note_file): New function.
27148 * coverage.h: Declare the function.
27149 * toplev.c (finalize): Clean if an error has been seen.
27151 2017-04-28 Martin Liska <mliska@suse.cz>
27153 PR gcov-profile/80031
27154 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27155 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27156 * gcov.c (read_graph_file): Read just number of blocks.
27157 * profile.c (branch_prob): Do not stream 0 flags per a basic
27160 2017-04-28 Martin Liska <mliska@suse.cz>
27162 * gcov-dump.c (tag_*): Add new argument to declarations.
27163 (dump_gcov_file): Likewise.
27164 (tag_blocks): Add and use new argument depth.
27165 (tag_arcs): Likewise.
27166 (tag_lines): Likewise.
27167 (tag_counters): Likewise.
27168 (tag_summary): Likewise.
27169 (dump_working_sets): Use depth to do a proper indentation.
27171 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27174 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27175 * symtab.c (symtab_node::debug_symtab): Move definition here.
27177 2017-04-28 Richard Biener <rguenther@suse.de>
27179 * lto-streamer.h (LTO_major_version): Bump to 7.
27181 2017-04-28 Richard Biener <rguenther@suse.de>
27183 * tree-vrp.c (assert_info): New struct.
27184 (add_assert_info): New helper.
27185 (register_edge_assert_for_2): Refactor to add asserts to a vector
27187 (register_edge_assert_for_1): Likewise.
27188 (register_edge_assert_for): Likewise.
27189 (finish_register_edge_assert_for): New helper actually registering
27190 asserts where live on edge.
27191 (find_conditional_asserts): Adjust.
27192 (find_switch_asserts): Likewise.
27193 (evrp_dom_walker::try_find_new_range): Generalize.
27194 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27196 2017-04-27 Marek Polacek <polacek@redhat.com>
27199 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27200 arg10 and arg11 to itype.
27202 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27204 * doc/extend.texi (Object Size Checking): Improve grammar.
27206 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27209 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27210 that the logic for permitting reciprocal estimates matches that
27213 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27216 * tree.c (type_cache_hasher::equal): Only compare
27217 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27218 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27219 non-aggregate element types.
27220 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27221 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27223 2017-04-27 Richard Biener <rguenther@suse.de>
27225 PR middle-end/80533
27226 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27227 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27228 keeping a reference to a trailing array.
27230 2017-04-27 Richard Biener <rguenther@suse.de>
27232 PR middle-end/80539
27233 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27234 being in loop-closed SSA form conservatively.
27235 (chrec_fold_multiply_poly_poly): Likewise.
27237 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27239 PR middle-end/79665
27240 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27241 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27243 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27246 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27247 (aarch64_function_arg_alignment): Return unsigned int again, but still
27248 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27249 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27250 Don't emit -Wpsabi note.
27251 (aarch64_function_arg_boundary): Likewise.
27252 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27255 2017-04-26 Nathan Sidwell <nathan@acm.org>
27257 * tree.h (crc32_unsigned_n): Declare.
27258 (crc32_unsigned, crc32_unsigned): Make inline.
27259 * tree.c (crc32_unsigned_bits): Replace with ...
27260 (crc32_unsigned_n): ... this.
27261 (crc32_unsigned, crc32_byte): Remove.
27262 (crc32_string): Remove unnecessary braces.
27264 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27266 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27267 * ipa-inline-analysis.c (MAX_TIME): Remove.
27268 (account_size_time): Use sreal for time.
27269 (dump_inline_summary): Update.
27270 (estimate_function_body_sizes): Update.
27271 (estimate_edge_size_and_time): Update.
27272 (estimate_calls_size_and_time): Update.
27273 (estimate_node_size_and_time): Update.
27274 (inline_merge_summary): Update.
27275 (inline_update_overall_summary): Update.
27276 (estimate_time_after_inlining): Update.
27277 (inline_read_section): Update.
27278 (inline_write_summary): Update.
27279 * ipa-inline.c (compute_uninlined_call_time): Update.
27280 (compute_inlined_call_time): Update.
27281 (recursive_inlining): Update.
27282 (inline_small_functions): Update.
27283 (dump_overall_stats): Update.
27284 * ipa-inline.h: Include sreal.h.
27285 (size_time_entry): Turn time to sreal.
27286 (inline_summary): Turn self_time nad time to sreal.
27288 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27290 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27292 (sreal::stream_out, sreal::stream_in): New.
27293 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27295 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27297 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27300 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27303 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27305 (dimode_scalar_chain::compute_convert_gain): Ditto.
27306 (dimode_scalar_chain::make_vector_copies): Ditto.
27307 (dimode_scalar_chain::convert_reg): Ditto.
27308 (dimode_scalar_chain::convert_insn): Ditto.
27309 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27310 (VI248_AVX512BW_1): New mode iterator.
27311 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27312 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27315 2017-04-25 Martin Sebor <msebor@redhat.com>
27317 PR tree-optimization/80497
27318 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27319 constants are representable in HOST_WIDE_INT.
27320 (parse_directive): Ditto.
27322 2017-04-25 Martin Sebor <msebor@redhat.com>
27325 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27326 (new_zero_array): Adjust signature.
27327 (dom_info::dom_init): Used unsigned rather that size_t.
27328 (dom_info::dom_info): Same.
27330 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27331 Jakub Jelinek <jakub@redhat.com>
27334 * config/arm/arm.c: Include gimple.h.
27335 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27336 returns negative, increment ncrn only if it returned positive.
27337 (arm_needs_doubleword_align): Return int instead of bool,
27338 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27339 members, but if there is any such non-FIELD_DECL
27340 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27341 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27342 returns negative, increment nregs only if it returned positive.
27343 (arm_setup_incoming_varargs): Likewise.
27344 (arm_function_arg_boundary): Emit -Wpsabi note if
27345 arm_needs_doubleword_align returns negative, return
27346 DOUBLEWORD_ALIGNMENT only if it returned positive.
27348 2017-04-25 Marek Polacek <polacek@redhat.com>
27351 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27352 first argument to type.
27354 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27357 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27358 type checks to test for compatibility instead of equality.
27360 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27361 Jakub Jelinek <jakub@redhat.com>
27364 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27366 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27367 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27368 the alignment computation, but return their maximum in warn_alignment.
27369 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27370 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27372 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27373 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27376 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27378 * config/arc/simdext.md (dmpyh): Fix typo.
27380 2017-04-25 Richard Biener <rguenther@suse.de>
27382 PR tree-optimization/80492
27383 * alias.c (compare_base_decls): Handle registers with asm
27384 specification conservatively.
27385 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27386 compare_base_decls returning dont-know properly.
27388 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27390 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27391 (legitimate_offset_address_p): New function.
27392 (arc_legitimate_address_p): Use above function.
27394 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27396 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27398 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27400 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27401 ACCH registers whenever they are available.
27403 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27405 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27406 double regs fix when not used.
27408 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27410 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27412 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27413 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27415 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27417 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27418 when emitting short ADD instructions.
27420 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27422 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27424 (cmpsi_cc_c_insn): Likewise.
27425 (cbranchsi4_scratch): Compute proper instruction length using
27426 compact_hreg_operand.
27427 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27429 2017-04-25 Richard Biener <rguenther@suse.de>
27431 PR middle-end/80509
27432 * passes.c (pass_manager::pass_manager): Initialize
27433 m_name_to_pass_map.
27435 2017-04-25 Richard Biener <rguenther@suse.de>
27437 PR tree-optimization/79201
27438 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27440 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27443 * config/s390/vector.md: Split MEM->GPR vector moves for
27444 non-s_operand addresses.
27446 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27449 * config/s390/predicates.md (reload_const_wide_int_operand): New
27451 * config/s390/s390.md ("movti"): Remove d/P alternative.
27452 ("movti_bigconst"): New pattern definition.
27454 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27457 * s390-protos.h (s390_expand_cs_hqi): Removed.
27458 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27459 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27460 modes as well as CCZ1mode and CCZmode.
27461 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27462 signature of s390_emit_compare_and_swap.
27463 (s390_expand_cs_hqi): Likewise, make static.
27464 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27465 compare-and-swap, in some cases.
27466 (s390_expand_cs): Wrapper function.
27467 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27469 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27470 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27471 patterns for small and large integers. Forbid symref memory operands.
27472 Move expander to s390.c. Require cc register.
27473 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27474 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27475 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27476 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27477 symref memory operands. Remove CC mode and call s390_match_ccmode
27479 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27481 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27483 * config/s390/s390.md (define_peephole2): New peephole to help
27484 combining the load-and-test pattern with volatile memory.
27486 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27488 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27489 with CCZmode for TARGET_Z196.
27491 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27493 PR rtl-optimization/80501
27494 * combine.c (make_compound_operation_int): Set subreg_code to SET
27495 even for AND with mask of the sign bit of mode.
27497 PR rtl-optimization/80500
27498 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27499 sum's initial value.
27501 2017-04-25 Julian Brown <julian@codesourcery.com>
27502 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27504 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27506 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27508 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27510 2017-04-25 Julian Brown <julian@codesourcery.com>
27511 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27513 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27514 (thunderx2t99_sha): New Reservation.
27516 2017-04-25 Julian Brown <julian@codesourcery.com>
27517 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27519 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27520 type for 1-element load.
27522 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27524 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27526 2017-04-24 Martin Jambor <mjambor@suse.cz>
27528 PR tree-optimization/80293
27529 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27530 char arrays not totally scalarizable if it is false.
27531 (analyze_all_variable_accesses): Pass correct value in the new
27532 parameter. Add a statistics counter.
27534 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27536 PR middle-end/79931
27537 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27539 2017-04-24 Richard Biener <rguenther@suse.de>
27541 PR tree-optimization/80494
27542 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27543 out for complex types.
27545 2017-04-24 Richard Biener <rguenther@suse.de>
27547 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27548 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27549 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27550 (DFS): Adjust and never fail.
27551 (sccvn_dom_walker::fail): Remove.
27552 (sccvn_dom_walker::before_dom_children): Adjust.
27553 (run_scc_vn): Likewise and never fail.
27554 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27555 (pass_fre::execute): Likewise.
27557 2017-04-24 Richard Biener <rguenther@suse.de>
27559 PR tree-optimization/79725
27560 * tree-ssa-sink.c (statement_sink_location): Return whether
27561 failure reason was zero uses. Move that check later.
27562 (sink_code_in_bb): Deal with zero uses by removing the stmt
27565 2017-04-24 Richard Biener <rguenther@suse.de>
27568 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27569 pointer-based references.
27571 2017-04-24 Richard Biener <rguenther@suse.de>
27574 * pass_manager.h (pass_manager::operator new): Remove.
27575 (pass_manager::operator delete): Likewise.
27576 * passes.c (pass_manager::operator new): Remove.
27577 (pass_manager::operator delete): Likewise.
27578 (pass_manager::pass_manager): Zero individual pass members.
27580 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27583 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27584 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27585 Check "XEXP (src, 1)" operand here.
27586 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27587 Check "XEXP (src, 1)" operand here.
27588 (dimode_scalar_chain::make_vector_copies): Detect count register
27589 of a shift instruction. Zero extend count register from QImode
27590 to DImode to satisfy vector shift pattern count operand predicate.
27591 Substitute vector shift count operand with a DImode copy.
27592 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27595 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27597 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27598 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27599 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27600 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27601 (UNSPEC_NOREX_MEM): Remove definition.
27603 2017-04-21 Richard Biener <rguenther@suse.de>
27605 PR tree-optimization/79547
27606 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27607 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27608 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27609 without any constraints.
27611 2017-04-21 Richard Biener <rguenther@suse.de>
27613 PR tree-optimization/78847
27614 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27616 2017-04-21 Richard Biener <rguenther@suse.de>
27618 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27619 (build_distinct_type_copy): Likewise.
27620 (build_variant_type_copy): Likewise.
27621 * tree.c (build_qualified_type): Pass down mem-stat info.
27622 (build_distinct_type_copy): Likewise.
27623 (build_variant_type_copy): Likewise.
27625 2017-04-21 Richard Biener <rguenther@suse.de>
27627 PR tree-optimization/80237
27628 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27630 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27631 for a simplified result.
27633 2016-04-21 Richard Biener <rguenther@suse.de>
27635 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27636 sth as strict as a simple_iv but a chrec without symbols and an
27637 operand defined in the loop we are peeling (and not some subloop).
27638 (propagate_constants_for_unrolling): Propagate all constants.
27640 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27643 * config/i386/i386.c (print_reg): Remove assert for disalowed
27644 regno values, call output_operand_lossage instead.
27646 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27649 * config/i386/constraints.md (Yc): New register constraint.
27650 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27651 Use Yc constraint for alternative 2 of operand 0. Remove
27652 preferred_for_speed attribute.
27654 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27656 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27657 lastprivate clauses in SIMT case.
27659 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27661 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27663 2017-04-20 Richard Biener <rguenther@suse.de>
27665 PR tree-optimization/57796
27666 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27667 as N scalar stores.
27668 (vect_model_load_cost): Cost gathers as N scalar loads.
27670 2017-04-20 Richard Biener <rguenther@suse.de>
27672 * ggc-page.c (ggc_allocated_p): Rename to ...
27673 (safe_lookup_page_table_entry): ... this and return the lookup
27675 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27677 2017-04-20 Richard Biener <rguenther@suse.de>
27679 PR tree-optimization/80453
27680 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27681 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27682 from the conditions.
27683 (vn_phi_eq): Pass them down.
27684 (vn_phi_lookup): Record them.
27685 (vn_phi_insert): Likewise.
27687 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27689 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27690 uninitialized variable warning to avoid buffer overrun.
27692 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27695 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27696 is suppressed for '{ 0 }' in C.
27698 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27700 * BASE-VER: Set to 8.0.0.
27702 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27704 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27705 priority .init_array and .fini_array section with SECTION_NOTYPE
27708 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27710 PR middle-end/80423
27711 * tree.h (build_array_type): Add typeless_storage default argument.
27712 * tree.c (type_cache_hasher::equal): Also compare
27713 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27714 (build_array_type): Add typeless_storage argument, set
27715 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27717 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27718 (build_array_type): Likewise. Add typeless_storage argument.
27720 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27721 Jakub Jelinek <jakub@redhat.com>
27723 PR tree-optimization/80426
27724 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27725 operation on symbolic operands, also compute the overflow for the
27726 invariant part when the operation degenerates into a negation.
27728 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27731 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27732 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27735 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27737 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27740 * config/avr/avr.c (tree.h): Include it.
27741 (cgraph.h): Include it.
27742 (avr_encode_section_info): Don't warn for uninitialized progmem
27743 variable if it's just an alias.
27745 2017-04-19 Richard Biener <rguenther@suse.de>
27748 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27749 when needed by AutoPGO.
27751 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27754 * doc/lto.texi: Remove an extra 'that'.
27756 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27758 PR rtl-optimization/80429
27759 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27760 are only used in debug insns.
27762 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27763 Vladimir Makarov <vmakarov@redhat.com>
27765 * config/sparc/predicates.md (input_operand): Add comment. Return
27766 true for any memory operand when LRA is in progress.
27767 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27769 2017-04-18 Jeff Law <law@redhat.com>
27772 * mips.md ({return,simple_return}_internal): Do not overwrite
27775 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27777 PR tree-optimization/80443
27778 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27779 instead of adding 1, subtract -1 and similarly instead of subtracting
27782 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27784 PR rtl-optimization/80357
27785 * haifa-sched.c (tmp_bitmap): New variable.
27786 (model_recompute): Handle duplicate use records.
27787 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27788 (free_global_sched_pressure_data): Free it.
27790 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27793 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27794 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27795 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27796 instead of SYSTEM_HEADER_DIR.
27798 2017-04-18 Jeff Law <law@redhat.com>
27800 PR middle-end/80422
27801 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27802 predecessors after walking up the insn chain.
27804 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27807 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27808 sizetype type into debug info.
27810 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27813 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27814 unneeded test for TARGET_UPPER_REGS_SF.
27815 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27817 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27820 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27821 instead of gsi_after_labels.
27823 2017-04-18 Jeff Law <law@redhat.com>
27825 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27829 2017-04-13 Jeff Law <law@redhat.com>
27830 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27831 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27833 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27836 * config/avr/avr.c (intl.h): Include it.
27837 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27839 2017-04-18 Martin Liska <mliska@suse.cz>
27841 PR gcov-profile/78783
27842 * gcov-tool.c (gcov_output_files): Validate that destination
27843 file is either removed by the tool or by a user.
27845 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27846 Guy Benyei <guybe@mellanox.com>
27848 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27849 block, and do not negate it, the stored id is already negative.
27851 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27853 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27855 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27858 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27859 masks of options that should be turned off if the VSX vector
27860 options are turned off.
27861 (OTHER_P8_VECTOR_MASKS): Likewise.
27862 (OTHER_VSX_VECTOR_MASKS): Likewise.
27863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27864 rs6000_disable_incompatible_switches to validate no type switches
27866 (rs6000_incompatible_switch): New function to disallow turning on
27867 other vector options if -mno-vsx, -mno-power8-vector, or
27868 -mno-power9-vector are specified.
27870 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27872 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27874 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27876 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27877 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27878 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27879 (ARG_POINTER_CFA_OFFSET): Likewise.
27881 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27883 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27884 conditions to take advantage of various optimizations.
27886 2017-04-13 Jeff Law <law@redhat.com>
27888 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27889 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27890 (zero_extendsidi2_dext): Likewise.
27892 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27895 * fold-const.c (fold_ternary_loc): Revert
27896 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27899 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27901 PR rtl-optimization/80343
27902 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27903 new scratch pseudo.
27905 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27908 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27909 to ubsan_encode_value.
27911 2017-04-13 Jeff Law <law@redhat.com>
27913 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27914 appearing in DEBUG_INSNs.
27916 2017-04-13 Martin Liska <mliska@suse.cz>
27918 PR gcov-profile/80413
27919 * gcov-io.c (gcov_write_string): Copy to buffer just when
27920 allocated size is greater than zero.
27922 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27925 * dwarf2out.c (decls_for_scope): Ignore declarations of
27926 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27928 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27931 * ipa-visibility.c (non_local_p): Fix typos.
27932 (localize_node): When localizing symbol in same comdat group,
27933 dissolve the group only when we know external symbols are going
27935 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27937 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27939 PR tree-optimization/79390
27940 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27941 order does not result in usable sequence, retry with reversed operand
27947 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27948 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27949 op0 instead of fold_convert_loc (loc, type, arg0).
27951 2017-04-12 Jeff Law <law@redhat.com>
27953 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27954 has a delay slot in the generated code.
27956 * config/cris/cris.md (cris_preferred_reload_class): Return
27957 GENNONACR_REGS rather than GENERAL_REGS.
27959 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27962 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27963 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27964 signedness of the result type.
27966 2017-04-12 Richard Biener <rguenther@suse.de>
27967 Jeff Law <law@redhat.com>
27969 PR tree-optimization/80359
27970 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27971 trim stores to TARGET_MEM_REFs.
27973 2017-04-12 Richard Biener <rguenther@suse.de>
27975 PR tree-optimization/79390
27976 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27977 threading case even more.
27979 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27982 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27983 for quad_address_p for TImode, instead of just not indexed_address.
27985 2017-04-12 Richard Biener <rguenther@suse.de>
27986 Bernd Edlinger <bernd.edlinger@hotmail.de>
27988 PR middle-end/79671
27989 * alias.c (component_uses_parent_alias_set_from): Handle
27990 TYPE_TYPELESS_STORAGE.
27991 (get_alias_set): Likewise.
27992 * tree-core.h (tree_type_common): Add typeless_storage flag.
27993 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27994 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27995 for types containing members with TYPE_TYPELESS_STORAGE.
27996 (place_field): Likewise.
27997 (layout_type): Likewise for ARRAY_TYPE.
27998 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27999 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28000 TYPE_TYPELESS_STORAGE.
28001 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28003 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28006 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28007 first argument to type.
28009 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28013 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28014 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28015 (rs6000_expand_binop_builtin): Likewise.
28016 (rs6000_expand_ternop_builtin): Likewise; also add missing
28017 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28018 vshasigma built-ins.
28019 * doc/extend.texi: Document that vec_xxpermdi's third argument
28020 must be a constant.
28022 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28024 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28025 Use shift_const cost parameter when calculating gain of STV shifts.
28027 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28029 PR rtl-optimization/70478
28030 * lra-constraints.c (process_alt_operands): Check memory for
28031 disfavoring memory insn operand.
28033 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28035 PR middle-end/80100
28036 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28037 left shift in unsigned HOST_WIDE_INT type.
28039 PR rtl-optimization/80385
28040 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28041 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28044 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28045 if they have any depend clauses.
28047 2017-04-11 Martin Liska <mliska@suse.cz>
28050 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28051 * ipa-split.c (split_function): Create a local comdat symbol
28052 if caller is in a comdat group.
28054 2017-04-11 Martin Liska <mliska@suse.cz>
28057 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28060 2017-04-11 Martin Sebor <msebor@redhat.com>
28062 PR middle-end/80364
28063 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28064 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28066 (directive::set_width, directive::set_precision, format_character):
28068 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28071 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28074 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28075 conflict, set target->arch_name instead of target->cpu_name.
28077 2017-04-11 Richard Biener <rguenther@suse.de>
28079 PR tree-optimization/80374
28080 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28081 build_zero_cst, remove fold_convertible_p check again.
28083 2017-04-11 Martin Liska <mliska@suse.cz>
28086 * ubsan.c (instrument_object_size): Do not instrument register
28089 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28092 * config/i386/i386-builtin-types.def
28093 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28094 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28095 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28096 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28097 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28098 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28099 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28100 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28101 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28102 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28103 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28104 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28105 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28106 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28107 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28108 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28109 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28110 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28111 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28112 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28113 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28114 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28115 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28116 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28117 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28118 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28119 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28120 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28121 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28122 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28123 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28124 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28125 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28126 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28127 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28128 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28129 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28130 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28131 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28132 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28133 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28134 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28135 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28136 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28137 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28138 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28140 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28141 flag to second_arg_count, handle 4 argument function type _COUNT
28142 aliases, handle second_arg_count on second argument rather than last.
28144 2017-04-10 Jeff Law <law@redhat.com>
28146 PR tree-optimization/80374
28147 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28148 record anything if we can not convert integer_zero_node to the
28151 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28154 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28155 Enhance special handling given to the TARGET_P9_MINMAX option in
28156 relation to certain other options.
28158 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28160 PR tree-optimization/80153
28161 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28162 remove POINTER_PLUS_EXPR's base part directly, rather than through
28165 2017-04-10 Richard Biener <rguenther@suse.de>
28166 Bin Cheng <bin.cheng@arm.com>
28168 PR tree-optimization/80153
28169 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28170 the first element of pointer type aff_tree. Build result expr in
28172 (add_elt_to_tree): Convert to type unconditionally. Remove other
28173 fold_convert calls.
28174 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28175 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28177 2017-04-10 Richard Biener <rguenther@suse.de>
28179 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28182 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28184 PR rtl-optimization/70478
28185 * lra-constraints.c (curr_small_class_check): New.
28186 (update_and_check_small_class_inputs): New.
28187 (process_alt_operands): Update curr_small_class_check. Disfavor
28188 alternative insn memory operands. Check available regs for small
28191 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28194 * config/mips/mips.opt (-mvirt): Update description.
28195 * doc/invoke.texi (-mvirt): Likewise.
28197 2017-04-10 Richard Biener <rguenther@suse.de>
28199 PR middle-end/80362
28200 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28201 looking for NEGATE_EXPR in -A / -B to A / B folding.
28203 2017-04-10 Martin Liska <mliska@suse.cz>
28205 PR gcov-profile/80224
28206 * gcov.c (print_usage): Fix usage string.
28207 (get_gcov_intermediate_filename): Remove.
28208 (output_gcov_file): Use both for normal and intermediate format.
28209 (generate_results): Do not initialize special file for
28210 intermediate format.
28212 2017-04-10 Richard Biener <rguenther@suse.de>
28214 PR tree-optimization/80304
28215 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28218 2017-04-10 Nathan Sidwell <nathan@acm.org>
28221 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28222 (rs6000_init_builtins): Use it.
28224 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28226 * config/arm/arm.md (<mrc>): Add mode to SET source.
28227 (<mrrc>): Likewise.
28229 2017-04-10 Richard Biener <rguenther@suse.de>
28231 PR middle-end/80344
28232 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28234 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28237 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28238 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28239 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28240 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28241 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28242 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28243 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28244 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28245 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28246 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28247 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28248 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28249 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28250 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28251 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28252 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28253 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28254 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28255 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28256 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28257 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28258 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28259 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28261 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28263 PR rtl-optimization/70478
28264 * lra-constraints.c: Reverse the last patch.
28266 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28268 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28269 Add comment for WCHAR_T.
28271 2017-04-08 Martin Liska <mliska@suse.cz>
28274 2017-04-07 Martin Liska <mliska@suse.cz>
28277 * ipa-split.c (split_function): Add function part to a same comdat
28280 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28283 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28285 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28287 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28288 (vec_pairing): ...to this.
28289 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28290 (rs6000_sched_init): Adjust for name change.
28291 (struct rs6000_sched_context): Likewise.
28292 (rs6000_init_sched_context): Likewise.
28293 (rs6000_set_sched_context): Likewise.
28295 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28301 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28303 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28304 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28305 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28307 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28309 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28311 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28313 PR rtl-optimization/70703
28314 * ira-color.c (update_conflict_hard_regno_costs): Use
28315 int64_t instead of HOST_WIDE_INT.
28317 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28319 PR rtl-optimization/70478
28320 * lra-constraints.c (process_alt_operands): Disfavor alternative
28321 insn memory operands.
28323 2017-04-07 Jeff Law <law@redhat.com>
28325 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28326 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28328 2017-04-07 Martin Liska <mliska@suse.cz>
28331 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28332 Show error message instead of an ICE.
28334 2017-04-07 Martin Liska <mliska@suse.cz>
28337 * ipa-split.c (split_function): Add function part to a same comdat
28340 2017-04-07 Richard Biener <rguenther@suse.de>
28342 PR middle-end/80341
28343 * tree.c (get_unwidened): Also handle ! for_type case for
28345 * convert.c (do_narrow): Split out from ...
28346 (convert_to_integer_1): ... here. Do not pass final truncation
28347 type to get_unwidened for TRUNC_DIV_EXPR.
28349 2017-04-07 Richard Biener <rguenther@suse.de>
28351 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28353 (aff_combination_const): Adjust.
28354 (aff_combination_scale): Likewise.
28355 (aff_combination_add_elt): Likewise.
28356 (aff_combination_add_cst): Likewise.
28357 (aff_combination_convert): Likewise.
28358 (add_elt_to_tree): Likewise. Remove unused argument.
28359 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28361 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28363 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28365 * config/arm/arm.c (arm_default_short_enums): Use
28366 ARM_DEFAULT_SHORT_ENUMS.
28367 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28369 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28372 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28373 members with redundant out-of-class redeclaration.
28375 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28378 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28379 * config/i386/i386.md (*zero_extendsidi2):
28380 Add (?*x,*x) and (?*v,*v) alternatives.
28382 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28385 * config/i386/i386.c (ix86_expand_builtin)
28386 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28387 mode from insn data. Convert operands to insn operand mode.
28388 Copy operands that don't satisfy insn predicate to a register.
28390 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28392 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28395 2017-04-06 Richard Biener <rguenther@suse.de>
28397 PR tree-optimization/80334
28398 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28399 preserve alignment of accesses.
28401 2017-04-06 Richard Biener <rguenther@suse.de>
28403 PR tree-optimization/80262
28404 * tree-sra.c (build_ref_for_offset): Preserve address-space
28406 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28407 Drop useless address-space information on MEM_REF offsets.
28409 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28411 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28413 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28415 PR rtl-optimization/70703
28416 * ira-color.c (update_conflict_hard_regno_costs): Use
28417 HOST_WIDE_INT instead of long.
28419 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28422 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28423 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28425 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28426 for x86_64 target. Handle -m3dnowa option.
28428 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28430 PR rtl-optimization/70703
28431 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28432 (update_conflict_hard_regno_costs): Use long instead of unsigned
28433 arithmetic for cost calculation.
28435 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28436 Bernd Edlinger <bernd.edlinger@hotmail.de>
28439 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28442 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28445 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28446 ptr_mode with Pmode throughout.
28447 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28448 into probe_stack_range and use DImode.
28450 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28453 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28454 call_eh_return is true.
28456 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28458 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28459 Initialize last_match_fntype_index.
28461 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28464 * tree-nvr.c: Include internal-fn.h.
28465 (pass_return_slot::execute): Ignore internal calls without
28468 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28469 Richard Biener <rguenther@suse.de>
28472 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28473 captures used multiple times, except for the last use.
28474 * generic-match-head.c: Include gimplify.h.
28476 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28478 PR tree-optimization/79390
28479 * target.h (struct noce_if_info): Declare.
28480 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28481 * target.def (noce_conversion_profitable_p): New target hook.
28482 * ifcvt.h (struct noce_if_info): New type, moved from ...
28483 * ifcvt.c (struct noce_if_info): ... here.
28484 (noce_conversion_profitable_p): Renamed to ...
28485 (default_noce_conversion_profitable_p): ... this. No longer
28487 (noce_try_store_flag_constants, noce_try_addcc,
28488 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28489 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28490 instead of noce_conversion_profitable_p.
28491 * config/i386/i386.c: Include ifcvt.h.
28492 (ix86_option_override_internal): Don't override
28493 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28494 (ix86_noce_conversion_profitable_p): New function.
28495 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28496 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28497 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28498 * doc/tm.texi: Regenerated.
28500 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28502 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28505 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28508 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28509 instructions for small multiply cores.
28511 2017-04-04 Jeff Law <law@redhat.com>
28513 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28515 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28516 that are not set by the loop over the elements.
28518 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28521 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28522 int mode, convert_modes it to mode as unsigned, otherwise use
28523 lowpart_subreg to mode rather than SImode.
28524 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28525 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28526 Use DImode instead of SImode for the shift count operand.
28527 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28530 2017-04-04 Richard Biener <rguenther@suse.de>
28532 PR middle-end/80281
28533 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28534 arithmetic done for the negate or the plus. Simplify.
28535 (A - (-B) -> A + B): Likewise.
28536 * fold-const.c (split_tree): Make sure to not negate pointers.
28538 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28540 PR rtl-optimization/60818
28541 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28542 a compare of comparisons with the thing compared if this results
28543 in a different machine mode.
28545 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28547 * alias.c (base_alias_check): Fix typo in comment.
28548 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28549 * cgraphunit.c (symbol_table::compile): Likewise.
28550 * collect2.c (maybe_run_lto_and_relink): Likewise.
28551 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28552 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28553 * config/avr/avr.c (avr_map_op_t): Likewise.
28554 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28555 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28556 * config/epiphany/epiphany.md (movcc): Likewise.
28557 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28558 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28560 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28561 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28562 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28563 * config/sh/sh.c (sh_rtx_costs): Likewise.
28564 * fold-const.c (fold_truth_andor): Likewise.
28565 * genautomata.c (collapse_flag): Likewise.
28566 * gengtype.h (struct type::u::s): Likewise.
28567 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28568 * input.c (FORMAT_AMOUNT): Likewise.
28569 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28570 (known_aggs_to_agg_replacement_list): Likewise.
28571 * ipa-inline-analysis.c: Likewise.
28572 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28573 * ipa-polymorphic-call.c
28574 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28575 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28576 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28578 * modulo-sched.c (apply_reg_moves): Likewise.
28579 * omp-expand.c (build_omp_regions_1): Likewise.
28580 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28581 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28582 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28583 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28584 * value-prof.c: Likewise.
28585 * var-tracking.c (val_reset): Likewise.
28587 2017-04-03 Richard Biener <rguenther@suse.de>
28589 PR tree-optimization/80275
28590 * fold-const.c (split_address_to_core_and_offset): Handle
28593 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28595 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28596 descriptors is at least equal to that of functions.
28598 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28600 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28602 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28605 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28606 (mov<IMOD4:mode>): New expander.
28607 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28609 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28611 PR rtl-optimization/79405
28612 * fwprop.c (propagations_left): New variable.
28613 (forward_propagate_into): Decrement it.
28614 (fwprop_init): Initialize it.
28615 (fw_prop): If the variable has reached zero, stop propagating.
28616 (fwprop_addr): Ditto.
28618 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28621 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28622 a FUNCTION_DECL, pass it as decl instead of origin to
28625 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28627 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28630 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28633 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28634 TARGET_VSX_SMALL_INTEGER.
28636 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28638 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28639 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28641 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28643 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28644 extraction from odd-numbered MSA register.
28646 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28648 PR middle-end/80173
28649 * expmed.c (store_bit_field_1): Don't attempt to create
28650 a word subreg out of hard registers wider than word if they
28651 have HARD_REGNO_NREGS of 1 for their mode.
28653 PR middle-end/80163
28654 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28655 conversions to integer types wider than word and pointer.
28658 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28659 (rtx_equal_for_cselib_p): Pass 0 to it.
28660 * cselib.c (cselib_hasher::equal): Likewise.
28661 (rtx_equal_for_cselib_1): Add depth argument. If depth
28662 is 128, don't look up VALUE locs and punt. Increment
28663 depth in recursive calls when walking VALUE locs.
28665 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28667 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28668 (make_gcov_file_name): Use the canonical path name for generating
28670 (read_line): Fix handling of files with ascii null bytes.
28672 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28674 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28675 to initialise a vector register instead
28676 of using a const_int.
28678 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28680 PR translation/80189
28681 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28682 diagnostic messages.
28684 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28687 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28688 (dfp_diex_<mode>): Update mode of operand 1.
28689 * doc/extend.texi (dxex, dxexq): Document change to return type.
28690 (diex, diexq): Document change to argument type.
28692 2017-03-30 Martin Jambor <mjambor@suse.cz>
28695 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28696 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28697 it reflects the signature changes performed at the callee side.
28698 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28699 to cgraph_build_function_type_skip_args.
28700 (build_function_decl_skip_args): Adjust call to the above function.
28702 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28705 * config/i386/sse.md
28706 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28707 register as dest whenever it is a MEM not rtx_equal_p to the
28708 corresponding dup operand, and when forcing into reg move the
28709 reg into the memory afterwards.
28710 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28711 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28712 for the force_reg mode.
28713 (avx512vl_vextractf128<mode>): Use register as dest either
28714 always when a MEM, or when it is a MEM not rtx_equal_p to the
28715 corresponding dup operand, or even not when it is a CONST_VECTOR
28716 depending on the mode and lo vs. hi.
28717 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28719 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28720 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28721 Likewise. Require that operands[2] is even.
28722 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28723 Remove extraneous parens. Require that operands[2] is a multiple
28725 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28726 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28727 disallow memory then.
28729 2017-03-30 Richard Biener <rguenther@suse.de>
28731 PR tree-optimization/77498
28732 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28733 to non-constants over backedges.
28735 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28737 PR rtl-optimization/80233
28738 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28739 as last_combined_insn. Do not test for BARRIER_P separately.
28741 2017-03-29 Andreas Schwab <schwab@suse.de>
28744 * calls.c (prepare_call_address): Convert funexp to Pmode before
28745 copying to temp reg.
28747 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28749 PR tree-optimization/80158
28750 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28751 Handle possible future case of more than one alternate
28753 (replace_rhs_if_not_dup): Likewise.
28754 (replace_one_candidate): Likewise.
28756 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28758 PR rtl-optimization/80193
28759 * ira.c (ira): Do not check allocation for LRA.
28761 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28763 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28764 (nvptx_output_simt_exit): Declare.
28765 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28766 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28767 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28768 (nvptx_declare_function_name): ...here. Emit declaration of local
28769 memory space buffer for omp_simt_enter insn.
28770 (nvptx_output_unisimt_switch): New.
28771 (nvptx_output_softstack_switch): New.
28772 (nvptx_output_simt_enter): New.
28773 (nvptx_output_simt_exit): New.
28774 * config/nvptx/nvptx.h (struct machine_function): New fields
28775 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28776 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28777 (UNSPECV_SIMT_EXIT): Ditto.
28778 (omp_simt_enter_insn): New insn.
28779 (omp_simt_enter): New expansion.
28780 (omp_simt_exit): New insn.
28781 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28783 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28784 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28785 (expand_GOMP_SIMT_EXIT): New.
28786 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28787 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28788 (GOMP_SIMT_EXIT): Ditto.
28789 * target-insns.def (omp_simt_enter): New insn.
28790 (omp_simt_exit): Ditto.
28791 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28793 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28794 (lower_rec_input_clauses): Likewise.
28795 (lower_lastprivate_clauses): Handle SIMT privatization.
28797 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28798 (ompdevlow_adjust_simt_enter): New.
28799 (find_simtpriv_var_op): New.
28800 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28801 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28803 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28804 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28805 (copy_decl_for_dup_finish): Ditto.
28807 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28809 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28812 * config/i386/i386.c (ix86_option_override_internal): Always
28813 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28815 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28817 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28819 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28821 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28822 mark new edge's irreducible flag accordign to it.
28823 (vect_do_peeling): Check loop preheader edge's irreducible flag
28824 and pass it to function slpeel_add_loop_guard.
28826 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28828 PR tree-optimization/80218
28829 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28830 Update block frequencies and counts.
28832 2017-03-28 Richard Biener <rguenther@suse.de>
28834 PR tree-optimization/78644
28835 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28836 of a simplification result we may not use it at all.
28838 2017-03-28 Richard Biener <rguenther@suse.de>
28841 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28842 without arguments, generate default definition of a SSA name.
28844 2017-03-28 Richard Biener <rguenther@suse.de>
28846 PR middle-end/80222
28847 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28848 TYPE_REF_CAN_ALIAS_ALL references.
28849 * fold-const.c (fold_indirect_ref_1): Likewise.
28851 2017-03-28 Martin Liska <mliska@suse.cz>
28854 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28855 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28857 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28858 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28860 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28861 (EXTRA_SPECS): Define.
28862 (SUBTARGET_EXTRA_SPECS): Likewise.
28863 (SUBTARGET_CPP_SPEC): Likewise.
28864 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28865 SUBTARGET_EXTRA_SPECS.
28866 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28868 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28870 * config/arc/simdext.md (vst64_insn): Update pattern.
28871 (vld32wh_insn): Likewise.
28872 (vld32wl_insn): Likewise.
28873 (vld64_insn): Likewise.
28874 (vld32_insn): Likewise.
28876 2017-03-28 Marek Polacek <polacek@redhat.com>
28879 * fold-const.c (fold_comparison): Use protected_set_expr_location
28880 instead of SET_EXPR_LOCATION.
28882 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28884 * tree.c (add_expr): Avoid name lookup warning.
28886 2017-03-27 Jeff Law <law@redhat.com>
28888 PR tree-optimization/80216
28889 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28890 function name. Limit recursion depth.
28891 (record_temporary_equivalences): Corresponding changes.
28893 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28895 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28898 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28901 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28903 * cfgcleanup.c (reg_note_cfa_p): New array.
28904 (insns_have_identical_cfa_notes): New function.
28905 (old_insns_match_p): Don't cross-jump in between /f
28906 and non-/f instructions. If both i1 and i2 are frame related,
28907 verify all CFA notes, their order and content.
28909 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28912 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28913 HImode and SImode with zero extend to DImode to one insn.
28914 (bswap<mode>2_extenddi): Likewise.
28915 (bswapsi2_extenddi): Likewise.
28916 (bswaphi2_extendsi): Likewise.
28917 (bswaphi2): Combine bswap HImode and SImode into one insn.
28918 Separate memory insns from swapping register.
28919 (bswapsi2): Likewise.
28920 (bswap<mode>2): Likewise.
28921 (bswaphi2_internal): Delete, no longer used.
28922 (bswapsi2_internal): Likewise.
28923 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28924 store, and gpr<-gpr swap insns.
28925 (bswap<mode>2_store): Likewise.
28926 (bswaphi2_reg): Register only splitter, combine with the splitter.
28927 (bswaphi2 splitter): Likewise.
28928 (bswapsi2_reg): Likewise.
28929 (bswapsi2 splitter): Likewise.
28930 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28931 the insns into load, store, and register/register insns.
28932 (bswapdi2_ldbrx): Likewise.
28933 (bswapdi2_load): Likewise.
28934 (bswapdi2_store): Likewise.
28935 (bswapdi2_reg): Likewise.
28937 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28939 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28940 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28942 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28945 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28947 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28948 special handling for target option conflicts between dform
28949 options (-mpower9-dform, -mpower9-dform-vector,
28950 -mpower9-dform-scalar) and -mno-direct-move.
28952 2017-03-27 Richard Biener <rguenther@suse.de>
28954 PR tree-optimization/80181
28955 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28957 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28959 * config/arc/predicates.md (move_double_src_operand): Replace the
28960 call to move_double_src_operand with a call to address_operand.
28962 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28964 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28965 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28966 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28968 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28970 * config/arc/predicates.md (long_immediate_loadstore_operand):
28971 Consider scaled addresses cases.
28973 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28975 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28976 restored when in interrupt.
28977 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28978 doesn't have delay slot.
28980 2017-03-27 Richard Biener <rguenther@suse.de>
28983 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28984 inlined thunk clones.
28986 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28989 * asan.c (instrument_derefs): Copy over last operand from
28990 original COMPONENT_REF to the new COMPONENT_REF with
28991 DECL_BIT_FIELD_REPRESENTATIVE.
28992 * ubsan.c (instrument_object_size): Likewise.
28994 2017-03-27 Richard Biener <rguenther@suse.de>
28996 PR tree-optimization/80170
28997 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28998 sure DR/SCEV didnt fold in constants we do not see when looking
28999 at the reference base alignment.
29001 2017-03-27 Richard Biener <rguenther@suse.de>
29003 PR middle-end/80171
29004 * gimple-fold.c (fold_ctor_reference): Properly guard against
29005 NULL return value from canonicalize_constructor_val.
29007 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29010 * config/i386/i386.c (ix86_expand_builtin)
29011 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29012 flags reg setting and flags reg using instructions.
29013 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29014 clobbering instructions to zero extend op2.
29016 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29018 * doc/install.texi (Configuration) <--with-aix-soname>:
29019 Update link to AIX ld.
29021 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29023 PR rtl-optimization/80160
29024 PR rtl-optimization/80159
29025 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29026 reg_alternate_class into account.
29028 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29031 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29032 to consider in curr_insn_transform.
29034 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29036 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29037 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29038 and emit_mode_inner.
29040 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29042 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29043 argument to the overloaded builtin variants. Use the new flag to
29044 deprecate certain builtin variants.
29045 * config/s390/s390-builtin-types.def: Add new builtin types.
29046 * config/s390/s390-builtins.h: Support new flags field for
29047 overloaded builtins.
29048 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29049 (s390_macro_to_expand): Enable vector float data type.
29050 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29051 builtins by incrementing the __VEC__ version number.
29052 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29054 (s390_resolve_overloaded_builtin): Emit error messages depending
29055 on the builtin flags.
29056 * config/s390/s390.c (s390_expand_builtin): Support additional
29057 flags argument. Change error message to match the messages
29058 emitted in s390-c.c.
29059 * config/s390/s390.md: New UNSPEC_* constants.
29060 (op_type): Add new instruction types.
29061 * config/s390/vecintrin.h: Add new builtins and test data class
29063 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29064 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29065 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29066 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29067 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29068 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29070 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29071 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29072 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29073 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29075 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29076 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29077 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29078 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29080 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29081 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29082 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29083 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29084 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29085 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29086 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29088 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29089 ("vec_scatter_element<V_HW_4:mode>_DI")
29090 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29091 ("vec_fpint<mode>", "vflls")
29092 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29093 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29094 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29095 ("*vec_cmphe<mode>_cc"): ... these.
29097 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29098 mode constant instead of magic value.
29100 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29102 * config/s390/s390.c (s390_expand_vec_compare): Support other
29103 vector floating point modes than just V2DF.
29104 (s390_expand_vcond): Likewise.
29105 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29106 (s390_cannot_change_mode_class): Prevent mode changes between TF
29107 and V1TF in vector registers.
29108 * config/s390/s390.md (DF, SF): New mode attributes.
29109 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29110 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29111 SFmode support for VRs.
29112 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29114 (VFT, VF_HW): New mode iterators.
29115 (vw, sdx): New mode attributes.
29116 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29117 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29118 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29119 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29120 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29121 also the new vector floating point modes. Renaming to ...
29123 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29124 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29125 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29126 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29127 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29128 ("vec_unordered<mode>"): ... these.
29130 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29131 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29132 ("*vec_extendv2df"): New insn definitions.
29134 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29136 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29137 ("mulditi3_2", "*muldi3_sign"): New patterns.
29138 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29139 rename the pattern definition.
29141 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29143 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29145 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29147 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29149 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29150 instruction if possible.
29151 * config/s390/vector.md (vec_halfnumelts): New mode
29153 ("*vec_vllezlf<mode>"): New pattern.
29155 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29157 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29158 ("popcountv4si2", "popcountv2di2"): Rename to ...
29159 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29160 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29162 ("popcount<mode>2_vxe"): New pattern.
29164 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29166 * common/config/s390/s390-common.c (processor_flags_table): Add
29168 * config.gcc: Add arch12.
29169 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29170 Default to arch12 for unknown CPU model numbers.
29171 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29172 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29173 PROCESSOR_max sanity check.
29174 * config/s390/s390-opts.h (enum processor_type): Add
29176 * config/s390/s390.c (processor_table): Add arch12.
29177 (s390_expand_builtin): Add check for B_VXE flag.
29178 (s390_issue_rate): Add PROCESSOR_ARCH12.
29179 (s390_get_sched_attrmask): Likewise.
29180 (s390_get_unit_mask): Likewise.
29181 (s390_sched_score): Enable z13 scheduling for arch12.
29182 (s390_sched_reorder): Likewise.
29183 (s390_sched_variable_issue): Likewise.
29184 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29186 (s390_tune_attr): Use z13 scheduling also for arch12.
29187 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29188 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29189 (TARGET_VXE_P): New macros.
29190 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29191 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29192 * config/s390/s390.opt: Add arch12 as processor_type.
29194 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29196 * config/s390/s390.md
29197 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29198 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29199 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29201 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29202 Rename expanders to ...
29204 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29205 ("fixuns_truncdddi2_emu"): ... these.
29207 ("fixuns_trunc<mode>si2_emu"): New expander.
29209 ("*fixuns_truncdfdi2_z13"): Rename to ...
29210 ("*fixuns_truncdfdi2_vx"): ... this.
29212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29214 * config/s390/2964.md: Remove the single element vector compare
29215 instructions which are no longer used.
29216 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29218 (s390_canonicalize_comparison): Remove handling of DFmode
29220 (s390_expand_vec_compare_scalar): Remove function.
29221 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29222 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29224 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29228 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29230 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29231 will anyway by matched by mov<mode>_64dfp.
29233 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29235 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29236 vlef/vstef. Add missing operand to vleif.
29238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29240 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29241 pair for all vector types with 64 bit elements.
29242 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29243 * config/s390/vector.md (V_HW_64): ... here.
29244 (V_128_NOSINGLE): New mode iterator.
29245 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29246 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29247 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29248 ("*vec_load_pairv2di"): Change to ...
29249 ("*vec_load_pair<mode>"): ... this one.
29251 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29253 * config/s390/constraints.md: Add comments.
29254 (jKK): Reject element sizes > 8 bytes.
29255 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29257 * config/s390/s390.md: Add the s_operand checks formerly in
29258 s390_split_ok_p to various splitters where they are still
29260 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29261 for 128 bit vectors. Plus two splitters.
29263 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29265 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29268 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29271 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29272 error if the boundary argument is not constant.
29274 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29276 PR rtl-optimization/80112
29277 * loop-doloop.c (doloop_condition_get): Don't check condition
29278 if cmp isn't SET with IF_THEN_ELSE src.
29280 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29282 PR tree-optimization/80158
29283 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29284 replacing a candidate statement, also replace it for the
29285 candidate's alternate interpretation.
29286 (replace_rhs_if_not_dup): Likewise.
29287 (replace_one_candidate): Likewise.
29289 2017-03-24 Richard Biener <rguenther@suse.de>
29291 PR tree-optimization/80167
29292 * graphite-isl-ast-to-gimple.c
29293 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29295 (translate_isl_ast_to_gimple::get_rename): Likewise.
29297 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29299 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29300 handling of certain combinations of target options, including the
29301 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29302 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29304 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29307 * config/arm/arm.md (*load_multiple): Add reload_completed to
29308 matching condition.
29310 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29311 Richard Biener <rguenth@suse.de>
29313 PR tree-optimization/79908
29314 PR tree-optimization/80136
29315 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29316 been cast away, gimplify_and_add suffices.
29318 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29320 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29322 2017-03-23 Richard Biener <rguenther@suse.de>
29324 PR tree-optimization/80032
29325 * gimplify.c (gimple_push_cleanup): Forced unconditional
29326 cleanups still have to go to the conditional_cleanups
29329 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29331 PR tree-optimization/80072
29332 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29334 (next_operand_entry_id): Change type to unsigned int.
29335 (sort_by_operand_rank): Make sure to return the right return value
29336 even if unsigned fields are bigger than INT_MAX.
29337 (struct oecount): Change cnt and id type to unsigned int.
29338 (oecount_hasher::equal): Formatting fix.
29339 (oecount_cmp): Make sure to return the right return value
29340 even if unsigned fields are bigger than INT_MAX.
29341 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29344 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29345 TREE_READONLY on result if writing it more than once.
29348 * doc/invoke.texi (-fsanitize=thread): Document that with
29349 -fnon-call-exceptions atomics are not able to throw
29353 * tsan.c: Include tree-eh.h.
29354 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29355 maybe_clean_or_replace_eh_stmt where needed.
29356 (instrument_memory_accesses): Add cfg_changed argument.
29357 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29358 if it returned true.
29359 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29361 PR rtl-optimization/63191
29362 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29363 wrapper function, moved the whole old content into ...
29364 (ix86_delegitimize_address_1): ... this. New inline function.
29365 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29366 true as last argument instead of ix86_delegitimize_address.
29368 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29370 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29371 cortexa57_branch_cost.
29373 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29375 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29377 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29380 * doc/md.texi (Constraints): Document wA constraint.
29381 * config/rs6000/constraints.md (wA): New.
29382 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29383 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29384 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29385 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29387 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29390 * gimplify.c (is_oacc_declared): New function.
29391 (oacc_default_clause): Use it to set default flags for acc declared
29392 variables inside parallel regions.
29393 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29394 declared variables.
29395 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29396 declare attribute to any decl as necessary.
29398 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29401 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29402 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29403 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29404 (arm_arch_lpae): This.
29405 * config/arm/arm.c (arm_arch7ve): Rename into ...
29406 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29407 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29410 2017-03-22 Martin Liska <mliska@suse.cz>
29413 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29414 error message instead of an ICE.
29416 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29418 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29420 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29422 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29424 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29427 2017-03-21 Martin Sebor <msebor@redhat.com>
29429 * doc/extend.texi: Use "cannot" instead of "can't."
29430 * doc/hostconfig.texi: Same.
29431 * doc/install.texi: Same.
29432 * doc/invoke.texi: Same.
29433 * doc/loop.texi: Same.
29434 * doc/md.texi: Same.
29435 * doc/objc.texi: Same.
29436 * doc/rtl.texi: Same.
29437 * doc/tm.texi: Same.
29438 * doc/tm.texi.in: Same.
29439 * doc/trouble.texi: Same.
29441 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29444 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29445 (collect_checksum_attributes): Set it.
29446 (die_checksum_ordered): Use it.
29448 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29450 PR tree-optimization/79908
29451 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29452 change: For a VA_ARG whose LHS has been cast away, use
29453 force_gimple_operand to construct the side effects.
29455 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29457 PR translation/80001
29458 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29459 more amenable to translation.
29460 (oacc_loop_auto_partitions): Likewise.
29462 2017-03-21 Marek Polacek <polacek@redhat.com>
29463 Martin Sebor <msebor@redhat.com>
29465 PR tree-optimization/80109
29466 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29467 on INTEGRAL_TYPE_P.
29469 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29470 Segher Boessenkool <segher@kernel.crashing.org>
29473 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29474 check reg_used_between_p between insn and one of succ or succ2
29475 depending on if succ is artificial insn not inserted into insn
29478 2017-03-21 Martin Liska <mliska@suse.cz>
29480 PR gcov-profile/80081
29481 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29482 * doc/gcc.texi: Include gcov-dump stuff.
29483 * doc/gcov-dump.texi: New file.
29485 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29487 PR rtl-optimization/79150
29488 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29489 conditional jump, if the jump is the last insn of the loop.
29491 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29492 Richard Biener <rguenth@suse.de>
29494 PR tree-optimization/79908
29495 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29496 been cast away, use force_gimple_operand to construct the side
29499 2017-03-21 Martin Liska <mliska@suse.cz>
29501 PR libfortran/79956
29502 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29505 2017-03-21 Brad Spengler <spender@grsecurity.net>
29508 * plugin.c (htab_hash_plugin): New function.
29509 (add_new_plugin): Use it and adjust.
29510 (parse_plugin_arg_opt): Adjust.
29511 (init_one_plugin): Likewise.
29513 2017-03-21 Richard Biener <rguenther@suse.de>
29515 PR tree-optimization/80032
29516 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29517 if set force the cleanup to happen unconditionally.
29518 (gimplify_target_expr): Push inserted clobbers with force_uncond
29519 to avoid them being removed by control-dependent DCE.
29521 2017-03-21 Richard Biener <rguenther@suse.de>
29523 PR tree-optimization/80122
29524 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29525 va_arg_pack_len when the inlined call stmt requires pack
29527 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29529 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29532 * tsan.c (instrument_builtin_call): If the memory model argument
29533 is not a constant, assume it is valid.
29536 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29539 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29541 PR rtl-optimization/79910
29542 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29543 if its dest is used by an insn before I2 (other than the combined
29544 insns themselves, which are properly handled already).
29546 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29549 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29551 * combine.c (record_used_regs): New static function.
29552 (try_combine): Handle situations where there is an additional
29553 instruction between I2 and I3 which needs to have a LOG_LINK
29557 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29559 * combine.c (try_combine): Delete redundant i1 test. Call
29560 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29562 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29565 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29566 alternatives 13/14.
29568 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29570 PR tree-optimization/80054
29571 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29572 the optimization if a PHI or any of its arguments is not dominated
29573 by the candidate's basis. Use gphi* rather than gimple* as
29575 (replace_profitable_candidates): Clean up a gimple* variable that
29576 should be a gphi* variable.
29578 2017-03-20 Martin Sebor <msebor@redhat.com>
29581 * doc/extend.texi (attribute constructor): Document present limitation.
29583 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29586 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29587 __POWER9_VECTOR__ #ifdef control, change template definition to
29588 use Power9-specific built-in function.
29589 (vec_any_eq): Likewise.
29590 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29591 to control outcomes from this test.
29592 (vector_ae_<mode>p): For VEC_F modes, likewise.
29594 2017-03-20 Ian Lance Taylor <iant@google.com>
29596 * config/i386/i386.c (ix86_function_regparm): Save an extra
29597 register for -fsplit-stack with DECL_STATIC_CHAIN.
29599 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29602 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29603 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29605 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29607 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29609 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29612 2017-03-20 Marek Polacek <polacek@redhat.com>
29615 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29617 2017-03-20 Richard Biener <rguenther@suse.de>
29619 PR tree-optimization/80113
29620 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29621 allocate extra SSA name for PHI def.
29622 (add_close_phis_to_outer_loops): Likewise.
29623 (add_close_phis_to_merge_points): Likewise.
29624 (copy_loop_close_phi_args): Likewise.
29625 (copy_cond_phi_nodes): Likewise.
29627 2017-03-20 Martin Liska <mliska@suse.cz>
29629 PR middle-end/79753
29630 * tree-chkp.c (chkp_build_returned_bound): Do not build
29631 returned bounds for a LHS that's not a BOUNDED_P type.
29633 2017-03-20 Martin Liska <mliska@suse.cz>
29637 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29638 COMPLEX_CST and VECTOR_CST.
29640 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29643 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29644 target operand. A new splitter adds the clobber statement in case
29645 the target operand is dead anyway.
29647 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29649 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29650 to age-old versions of binutils and glibc.
29652 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29654 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29656 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29658 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29660 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29662 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29663 requirement for binutils 2.13.
29665 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29667 * combine.c (try_combine): Delete redundant i1 test. Call
29668 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29670 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29672 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29673 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29675 <riscv64-*-elf>: Re-arrange section
29676 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29677 <riscv32-*-linux>: Likewise.
29678 <riscv64-*-elf>: Likewise
29679 <riscv64-*-linux>: Likewise.
29681 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29684 * aarch64.opt(verbose-cost-dump): Fix typo.
29686 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29689 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29690 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29692 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29694 * reload.c (find_reloads): When reloading a nonoffsettable address,
29695 use RELOAD_OTHER for it and its address reloads.
29697 PR rtl-optimization/79910
29698 * combine.c (record_used_regs): New static function.
29699 (try_combine): Handle situations where there is an additional
29700 instruction between I2 and I3 which needs to have a LOG_LINK
29703 2017-03-17 Jeff Law <law@redhat.com>
29705 PR tree-optimization/71437
29706 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29707 conditional in the hash table first.
29708 (vrp_dom_walker::before_dom_children): Extract condition from
29709 ASSERT_EXPR. Record condition, its inverion and any implied
29710 conditions as well.
29712 2017-03-17 Marek Polacek <polacek@redhat.com>
29713 Markus Trippelsdorf <markus@trippelsdorf.de>
29715 PR tree-optimization/80079
29716 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29719 2017-03-17 Richard Biener <rguenther@suse.de>
29721 PR middle-end/80075
29722 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29723 Properly verify the LHS before the RHS possibly claims to be
29725 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29728 2017-03-17 Martin Jambor <mjambor@suse.cz>
29730 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29731 (List of -O2 options): Likewise.
29732 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29735 2017-03-17 Tom de Vries <tom@codesourcery.com>
29737 * gcov-dump.c (print_usage): Print bug_report_url.
29739 2017-03-17 Richard Biener <rguenther@suse.de>
29741 PR middle-end/80050
29742 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29743 (parser::peek): Likewise.
29745 2017-03-17 Richard Biener <rguenther@suse.de>
29747 PR tree-optimization/80048
29748 * sese.c (free_sese_info): Properly release rename_map and
29749 copied_bb_map elements.
29751 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29753 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29754 Add linked-list forward and backlinks. Insert on
29755 construction, remove on destruction.
29756 (class pass_store_merging): Add m_stores_head field.
29757 (pass_store_merging::terminate_and_process_all_chains):
29758 Iterate over m_stores_head list.
29759 (pass_store_merging::terminate_all_aliasing_chains):
29761 (pass_store_merging::execute): Check for debug stmts first.
29762 Push new chains onto the m_stores_head stack.
29764 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29767 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29768 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29769 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29771 2017-03-16 Jeff Law <law@redhat.com>
29773 PR tree-optimization/71437
29774 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29775 member function. Implementation moved into after_dom_children
29776 member function and into the threader's thread_outgoing_edges
29778 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29779 some code into new thread_outgoing_edges.
29780 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29781 definition. Simplify marker handling (do it here). Assume we always
29782 have the available expression and the const/copies tables.
29783 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29785 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29786 * tree-vrp.c (equiv_stack): No longer file scoped.
29787 (vrp_dom_walker): New class.
29788 (vrp_dom_walker::before_dom_children): New member function.
29789 (vrp_dom_walker::after_dom_children): Likewise.
29790 (identify_jump_threads): Setup domwalker. Use it rather than
29791 walking edges in a random order by hand. Simplify setup/finalization.
29792 (finalize_jump_threads): Remove.
29793 (vrp_finalize): Do not call identify_jump_threads here.
29794 (execute_vrp): Do it here instead and call thread_through_all_blocks
29797 PR tree-optimization/71437
29798 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29800 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29802 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29803 (dom_opt_dom_walker::thread_across_edge): Remove
29804 handle_dominating_asserts argument. All callers changed.
29805 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29806 changes. Remove calls to lhs_of_dominating_assert. Other
29807 uses of handle_dominating_asserts turn into unconditional code
29808 (simplify_control_stmt_condition_1): Likewise.
29809 (simplify_control_stmt_condition): Likewise.
29810 (thread_through_normal_block, thread_across_edge): Likewise.
29811 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29812 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29813 object if it is not an SSA_NAME.
29814 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29815 before calling into the VRP specific simplifiers.
29816 (identify_jump_threads): Remove handle_dominating_asserts
29819 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29822 * tree-diagnostic.c (default_tree_printer): No longer static.
29823 * tree-diagnostic.h (default_tree_printer): New prototype.
29825 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29827 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29828 Change ins into fmov.
29830 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29832 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29833 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29834 Use h_con constraint for operand 1.
29835 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29836 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29838 2017-03-15 Jeff Law <law@redhat.com>
29840 PR tree-optimization/71437
29841 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29842 (record_temporary_equivalences): Use it.
29844 PR tree-optimization/71437
29845 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29846 tree-ssa-scopedtables.
29847 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29848 (record_conditions, record_cond, vuse_eq): Likewise.
29849 (record_edge_info): Adjust to API tweak of record_conditions.
29850 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29851 (record_temporary_equivalences, optimize_stmt): Likewise.
29852 (eliminate_redundant_computations): Likewise.
29853 (record_equivalences_from_stmt): Likewise.
29854 * tree-ssa-scopedtables.c: Include options.h and params.h.
29855 (vuse_eq): New function, moved from tree-ssa-dom.c
29856 (build_and_record_new_cond): Likewise.
29857 (record_conditions): Likewise. Accept vector of conditions rather
29858 than edge_equivalence structure for first argument.
29859 for the first argument.
29860 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29861 from tree-ssa-dom.c.
29862 (avail_exprs_stack::record_cond): Likewise.
29863 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29864 from tree-ssa-dom.c.
29865 (avail_exprs_stack): Add new member functions lookup_avail_expr
29867 (record_conditions): Declare.
29869 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29872 * lra-constraints.c (process_alt_operands): Increase reject for
29873 reloading an input/output operand.
29875 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29878 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29879 insns to convert from signed/unsigned char/short to IEEE 128-bit
29881 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29883 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29886 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29887 subreg of inner mode for values already in registers.
29889 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29891 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29892 iteration reg is used after the loop.
29894 2017-03-14 Martin Sebor <msebor@redhat.com>
29896 PR tree-optimization/79800
29897 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29898 precision in negative-positive range.
29899 (format_floating): Call non-const overload with adjusted precision.
29901 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29904 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29907 2017-03-14 Martin Sebor <msebor@redhat.com>
29909 PR middle-end/80020
29910 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29911 * builtins.def (aligned_alloc): Use it.
29914 * Makefile.in (GTFILES): Add calls.c.
29915 * calls.c: Include "gt-calls.h".
29917 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29919 PR rtl-optimization/79728
29920 * regs.h (struct target_regs): New field
29921 x_contains_allocatable_regs_of_mode.
29922 (contains_allocatable_regs_of_mode): New macro.
29923 * reginfo.c (init_reg_sets_1): Initialize it, and change
29924 contains_reg_of_mode so it includes global regs as well.
29925 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29926 rather than contains_regs_of_mode.
29928 2017-03-14 Martin Liska <mliska@suse.cz>
29930 * doc/invoke.texi: Document options that can't be combined with
29931 -fcheck-pointer-bounds.
29933 2017-03-14 Martin Liska <mliska@suse.cz>
29935 PR middle-end/79831
29936 * doc/invoke.texi (-Wchkp): Document the option.
29938 2017-03-14 Martin Liska <mliska@suse.cz>
29940 * Makefile.in: Install gcov-dump.
29942 2017-03-14 Martin Liska <mliska@suse.cz>
29944 * multiple_target.c (expand_target_clones): Bail out for
29945 an invalid attribute.
29947 2017-03-14 Richard Biener <rguenther@suse.de>
29949 * alias.c (struct alias_set_entry): Pack properly.
29950 * cfgloop.h (struct loop): Likewise.
29951 * cse.c (struct set): Likewise.
29952 * ipa-utils.c (struct searchc_env): Likewise.
29953 * loop-invariant.c (struct invariant): Likewise.
29954 * lra-remat.c (struct cand): Likewise.
29955 * recog.c (struct change_t): Likewise.
29956 * rtl.h (struct address_info): Likewise.
29957 * symbol-summary.h (function_summary): Likewise.
29958 * tree-loop-distribution.c (struct partition): Likewise.
29959 * tree-object-size.c (struct object_size_info): Likewise.
29960 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29961 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29962 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29963 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29964 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29965 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29966 (struct _stmt_vec_info): Likewise.
29968 2017-03-14 Martin Liska <mliska@suse.cz>
29971 * multiple_target.c (create_dispatcher_calls): Check that
29972 a target can create a function dispatcher.
29974 2017-03-14 Martin Liska <mliska@suse.cz>
29977 * multiple_target.c (expand_target_clones): Drop local.local
29978 flag for default implementation.
29980 2017-03-14 Richard Biener <rguenther@suse.de>
29982 PR tree-optimization/80030
29983 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29985 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29987 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29988 gcc_fallthrough() instead of __attribute__((fallthrough));
29990 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29992 * doc/gcc.texi: Remove "up" link to (DIR).
29993 * doc/gccint.texi: Ditto.
29995 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29997 * doc/install.texi (Specific) <avr>: Remove reference to
30000 2017-03-13 Jeff Law <law@redhat.com>
30002 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30003 attribute rather than comments.
30005 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30006 match_scratch operand is highest.
30008 2017-03-13 Martin Liska <mliska@suse.cz>
30010 PR middle-end/78339
30011 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30012 is a CHKP clone, use original declaration.
30014 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30016 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30017 (arc_conditional_register_usage): Use a different allocation order
30018 when optimizing for size.
30019 * common/config/arc/arc-common.c (arc_option_optimization_table):
30020 Section anchors default on when optimizing for size.
30022 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30024 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30026 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30028 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30029 * config/arc/arc.md (cpu_facility): Add cd variant.
30030 (*movqi_insn): Add code density variant.
30031 (*movhi_insn): Likewise.
30032 (*movqi_insn): Likewise.
30033 (*addsi3_mixed): Likewise.
30034 (subsi3_insn): Likewise.
30036 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30038 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30040 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30042 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30043 expressions with MINUS and UNARY ops.
30045 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30048 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30050 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30051 between vec_select and vector argument.
30052 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30053 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30054 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30055 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30056 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30057 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30059 2017-03-13 Richard Biener <rguenther@suse.de>
30062 * params.def (vect-max-peeling-for-alignment): Fix typo.
30064 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30066 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30067 issue that only occurred with binutils below 2.18.
30069 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30071 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30072 refer to binutils 2.11/2.12 minimum.
30074 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30076 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30077 ftp.kernel.org and simplify binutils requirement.
30079 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30081 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30083 (Optimize Options): Ditto. Also remove redundancy.
30085 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30087 PR translation/79848
30088 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30090 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30091 to G_ to avoid double translation.
30093 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30095 PR translation/79923
30096 * auto-profile.c (get_combined_location): Convert leading
30097 character of diagnostics to lower case and remove trailing period.
30098 (read_profile): Likewise for various diagnostics.
30099 * config/arm/arm.c (arm_option_override): Remove trailing period
30100 from various diagnostics.
30101 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30102 (msp430_expand_delay_cycles): Likewise.
30104 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30107 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30108 full command-line argument, rather than just "str".
30109 (aarch64_validate_march): Likewise.
30110 (aarch64_validate_mtune): Likewise.
30112 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30114 PR rtl-optimization/78911
30115 * lra-assigns.c (must_not_spill_p): New function.
30116 (spill_for): Use it.
30118 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30120 PR tree-optimization/79981
30121 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30122 ATOMIC_COMPARE_EXCHANGE ifn result.
30123 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30124 IFN_ATOMIC_COMPARE_EXCHANGE.
30126 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30129 * opts.c (parse_sanitizer_options): Add missing question mark to
30130 "did you mean" message.
30132 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30134 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30136 (VMULEUH_UNS): Likewise.
30137 (VMULOUB_UNS): Likewise.
30138 (VMULOUH_UNS): Likewise.
30139 * config/rs6000/rs6000.c (builtin_function_type): Remove
30140 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30142 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30145 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30146 x with result of extra_parsing_for_operand_code_0.
30147 (function_reader::extra_parsing_for_operand_code_0): Convert
30148 return type from void to rtx, returning x. When reading
30149 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30150 larger size containing struct block_symbol.
30152 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30154 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30155 -mfloat128-hardware without -m64.
30157 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30160 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30161 entries to the case statement that marks unsigned arguments to
30162 overloaded functions.
30164 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30167 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30169 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30172 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30173 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30175 2017-03-10 Martin Liska <mliska@suse.cz>
30179 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30180 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30183 2017-03-10 Olivier Hainque <hainque@adacore.com>
30185 * tree-switch-conversion (array_value_type): Start by resetting
30186 candidate type to it's main variant.
30188 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30190 PR rtl-optimization/79909
30191 * combine.c (try_combine): Use simplify_replace_rtx on individual
30192 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30193 of the whole CALL_INSN_FUNCTION_USAGE.
30195 PR tree-optimization/79972
30196 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30197 get_range_info on SSA_NAMEs. Formatting fixes.
30199 2017-03-10 Richard Biener <rguenther@suse.de>
30200 Jakub Jelinek <jakub@redhat.com>
30202 PR tree-optimization/77975
30203 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30204 edge to be constant.
30205 (get_val_for): For constant x return it. Formatting fix.
30206 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30207 would use the same bases as the current one.
30209 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30211 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30212 instead of vec_select for V1TImode.
30213 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30215 (VSX_LE_128): Add V1TI to this mode iterator.
30216 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30217 (*vsx_le_perm_store_<mode>): Likewise.
30218 (pre-reload splitter for VSX stores): Likewise.
30219 (post-reload splitter for VSX stores): Likewise.
30220 (*vsx_xxpermdi2_le_<mode>): Likewise.
30221 (*vsx_lxvd2x2_le_<mode>): Likewise.
30222 (*vsx_stxvd2x2_le_<mode>): Likewise.
30224 2017-03-09 Michael Eager <eager@eagercon.com>
30226 Correct failures with --enable-checking=yes,rtl.
30228 * config/microblaze/microblaze.c (microblaze_expand_shift):
30229 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30230 test for const0_rtx.
30231 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30232 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30234 2017-03-09 Richard Biener <rguenther@suse.de>
30236 PR tree-optimization/79977
30237 * graphite-scop-detection.c (scop_detection::merge_sese):
30238 Handle the case of extra exits to blocks dominating the entry.
30240 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30242 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30245 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30247 PR rtl-optimization/79949
30248 * lra-constraints.c (process_alt_operands): Check memory when
30249 trying to predict a cycle. Print about the overall increase.
30251 2017-03-09 Richard Biener <rguenther@suse.de>
30253 PR middle-end/79971
30254 * gimple-expr.c (useless_type_conversion_p): Preserve
30255 TYPE_SATURATING for fixed-point types.
30257 2017-03-09 Richard Biener <rguenther@suse.de>
30260 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30261 alignment of BLKmode params.
30263 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30266 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30267 (VALL_NO_V2Q): Likewise.
30269 * config/aarch64/aarch64-simd.md
30270 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30272 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30273 VALL_NO_V2Q mode iterator.
30274 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30276 2017-03-09 Martin Liska <mliska@suse.cz>
30278 PR tree-optimization/79631
30279 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30280 tree_int_cst_sign_bit just for INTEGER constants.
30282 2017-03-09 Martin Liska <mliska@suse.cz>
30286 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30289 2017-03-09 Marek Polacek <polacek@redhat.com>
30292 * tree.c (inchash::add_expr): Handle TREE_VEC.
30294 2017-03-09 Martin Liska <mliska@suse.cz>
30297 (chkp_narrow_size_and_offset): New function.
30298 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30299 (void chkp_parse_bit_field_ref): New function.
30300 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30301 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30303 2017-03-09 Martin Liska <mliska@suse.cz>
30306 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30307 (chkp_find_bounds_1): Remove gcc_unreachable.
30309 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30312 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30313 BUILT_IN_SYNC*, determine the access type from the size suffix and
30314 always build a MEM_REF with that type. Handle forgotten
30315 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30318 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30319 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30320 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30321 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30322 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30323 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30324 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30325 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30326 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30327 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30328 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30329 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30330 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30331 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30332 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30333 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30334 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30335 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30336 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30337 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30338 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30339 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30340 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30341 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30342 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30343 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30344 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30345 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30346 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30347 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30348 definitions outside of __OPTIMIZE__ guarded section.
30351 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30352 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30353 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30354 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30357 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30359 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30360 ("vfenez<mode>"): Add missing constraints.
30362 2017-03-08 Martin Sebor <msebor@redhat.com>
30365 * config/nds32/nds32.c (nds32_option_override):
30366 Fix misspelled diagnostic.
30368 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30371 * gimplify.c (gimplify_omp_for): Replace index var in outer
30372 taskloop statement with an artificial variable and add
30373 OMP_CLAUSE_PRIVATE clause for it.
30375 2017-03-08 Richard Biener <rguenther@suse.de>
30377 PR tree-optimization/79955
30378 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30379 for accesses that are completely outside of the variable.
30381 2017-03-08 Andrew Haley <aph@redhat.com>
30383 PR tree-optimization/79943
30384 * tree-ssa-loop-split.c (compute_new_first_bound): When
30385 calculating the new upper bound, (END-BEG) should be added, not
30388 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30390 * config/avr/avr.md (setmemhi): Make sure match_dup
30391 operand number comes before match_scratch.
30393 2017-03-08 Richard Biener <rguenther@suse.de>
30395 PR tree-optimization/79920
30396 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30397 with ncopies == 1 to ...
30398 (vect_transform_slp_perm_load): ... here. Properly compute
30399 all element loads by iterating VF times over the group. Do
30400 not handle ncopies (computed in a broken way) in
30401 vect_create_mask_and_perm.
30403 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30406 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30407 is a uniform vector, use uniform_vector_p return value instead of
30408 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30410 2017-03-07 Marek Polacek <polacek@redhat.com>
30412 PR middle-end/79809
30413 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30414 (alloca_call_type): Likewise.
30416 2017-03-07 Martin Liska <mliska@suse.cz>
30418 * gcov.c (process_args): Put comment to correct location.
30420 2017-03-07 Martin Liska <mliska@suse.cz>
30422 PR middle-end/68270
30423 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30424 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30425 (chkp_narrow_bounds_for_field): Likewise.
30426 (chkp_parse_array_and_component_ref): Pass one more argument to
30429 2017-03-07 Richard Biener <rguenther@suse.de>
30431 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30434 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30436 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30437 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30439 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30442 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30443 to end of description.
30444 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30446 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30448 PR rtl-optimization/79901
30449 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30451 (*avx512f_<code><mode>3<mask_name>): ... this.
30452 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30453 iterator instead of VI8_AVX2_AVX512BW.
30455 PR rtl-optimization/79901
30456 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30457 min/max expander, expand it using expand_vec_cond_expr.
30460 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30463 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30466 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30467 to void * for PCH reasons.
30468 * dwarf2out.c (output_loc_operands, output_die): Cast
30469 v.val_vec.array to unsigned char *.
30471 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30474 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30477 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30479 PR rtl-optimization/79571
30480 * lra-constraints.c (process_alt_operands): Calculate static
30481 reject and subtract it from overall when only addresses will be
30484 2017-03-06 Julia Koval <julia.koval@intel.com>
30487 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30488 incoming stack boundary to 128 for 64-bit targets.
30490 2017-03-06 Richard Biener <rguenther@suse.de>
30492 PR tree-optimization/79894
30493 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30494 to NULL after folding it.
30496 2017-03-06 Richard Biener <rguenther@suse.de>
30498 PR tree-optimization/79824
30499 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30500 check disabling peeling for gaps.
30502 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30504 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30505 attributes): Document gettimeofday.
30507 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30509 * config/s390/s390.c (s390_option_override_internal): Set
30510 PARAM_MIN_VECT_LOOP_BOUND
30512 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30514 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30515 * config/s390/s390.md: Likewise.
30517 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30520 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30521 (<avx2_avx512>_perm<mode>): Rename to ...
30522 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30524 (<avx512>_perm<mode>_mask): Rename to ...
30525 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30527 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30528 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30529 instead of VI8F_256_512.
30530 (avx512f_perm<mode>): New define_expand.
30531 (avx512f_perm<mode>_mask): Likewise.
30532 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30533 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30535 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30537 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30538 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30540 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30542 2017-03-06 Martin Liska <mliska@suse.cz>
30545 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30546 when having a SSA NAME w/o VAR_DECL assigned to it.
30548 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30550 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30551 msa_dpsub_<su>_d): Fix MODE for vec_select.
30553 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30555 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30557 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30559 2017-03-06 Richard Biener <rguenther@suse.de>
30561 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30562 * plugin.c (register_plugin_info): Likewise.
30563 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30565 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30567 * config/i386/sse.md (sse_storehps, sse_storelps,
30568 avx_<castmode><avxsizesuffix>_<castmode>,
30569 avx512f_<castmode><avxsizesuffix>_<castmode>,
30570 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30571 in condition that at least one operand is not a MEM.
30573 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30575 PR middle-end/79805
30576 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30577 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30579 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30580 gimple_call_nothrow_p flag based on whether original builtin can throw.
30581 If it can, emit following stmts on the fallthrough edge.
30582 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30583 don't create new bb if inserting just debug stmts on the edge, try to
30584 insert them on the fallthru bb or just reset debug stmts.
30586 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30589 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30590 restore recog_data (including the operand rtxes inside it) around
30591 the call to get_insn_template.
30593 2017-03-03 Martin Sebor <msebor@redhat.com>
30595 PR tree-optimization/79699
30596 * context.c (context::~context): Free MPFR caches to avoid
30597 a memory leak on program exit.
30599 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30601 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30602 Use wide_int::ulow () instead of .elt (0).
30604 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30606 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30607 (*pushxf): Limit oF constraint to 32bit targets and add oC
30608 constraint for 64bit targets.
30609 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30610 (*pushdf): Change rmF constraint to rmC.
30612 2017-03-03 Martin Liska <mliska@suse.cz>
30614 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30615 Remove unused variable.
30617 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30620 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30621 is a memory operand, increase num_memory.
30622 (ix86_expand_args_builtin): Likewise.
30624 2017-03-03 Jan Hubicka <jh@suse.cz>
30627 * ipa-devirt.c (maybe_record_node): Properly handle
30628 __cxa_pure_virtual visibility.
30630 2017-03-03 Martin Liska <mliska@suse.cz>
30632 PR tree-optimization/79803
30633 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30635 (pass_loop_prefetch::execute): Disabled optimization if an
30636 assumption about L1 cache size is not met.
30638 2017-03-03 Martin Liska <mliska@suse.cz>
30640 PR rtl-optimization/79574
30641 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30642 (hash_scan_set): Likewise.
30643 (dump_hash_table): Likewise.
30644 (hoist_code): Likewise.
30646 2017-03-03 Richard Biener <rguenther@suse.de>
30648 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30649 in place of uhigh ().
30650 (fixed_convert_from_real): Likewise.
30652 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30655 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30657 2017-03-03 Richard Biener <rguenther@suse.de>
30659 PR middle-end/79818
30660 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30661 TYPE_OVERFLOW_UNDEFINED check.
30663 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30665 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30667 (vector_ae_<mode>_p): Likewise.
30668 (vector_nez_<mode>_p): Likewise.
30669 (vector_ne_v2di_p): Likewise.
30670 (vector_ae_v2di_p): Likewise.
30671 (vector_ne_<mode>_p): Likewise.
30672 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30674 (vsx_tsqrt<mode>2_fe): Likewise.
30676 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30679 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30681 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30683 PR rtl-optimization/79780
30684 * cprop.c (one_cprop_pass): When second and further conditional trap
30685 in a single basic block is turned into an unconditional trap, turn it
30686 into a deleted note to avoid RTL verification failures.
30688 2017-03-02 Richard Biener <rguenther@suse.de>
30690 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30692 2017-03-02 Richard Biener <rguenther@suse.de>
30694 PR tree-optimization/79345
30696 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30697 param and abort the walk, returning -1 if it is hit.
30698 (walk_aliased_vdefs): Take a limit param and pass it on.
30699 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30700 defaulting to 0 and return a signed int.
30701 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30702 (check_defs): New helper.
30703 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30704 about uninitialized memory.
30705 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30706 bogus uninitialized warning.
30707 (fixed_convert_from_real): Likewise.
30709 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30711 PR tree-optimization/66768
30712 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30713 iv_use if base object can't be determined.
30715 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30717 PR tree-optimization/79345
30718 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30719 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30720 (get_pattern_stats): Initialize it.
30721 * genemit.c (gen_expand): Verify match_scratch numbers come after
30722 match_operand/match_dup numbers.
30723 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30724 match_scratch numbers.
30725 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30727 * config/s390/s390.md (trunctdsd2): Likewise.
30729 2017-03-02 Richard Biener <rguenther@suse.de>
30731 * wide-int.h (wide_int_storage::operator=): Implement in terms
30734 2017-03-02 Richard Biener <rguenther@suse.de>
30736 PR tree-optimization/79777
30737 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30738 the to insert expression to sth existing.
30740 2017-03-01 Martin Sebor <msebor@redhat.com>
30742 PR middle-end/79692
30743 * gimple-ssa-sprintf.c
30744 (directive::known_width_and_precision): New function.
30745 (format_integer): Use it.
30746 (get_mpfr_format_length): Consider the full range of precision
30747 when computing %g output with the # flag. Set the likely byte
30748 count to 3 rather than 1 when precision is indeterminate.
30749 (format_floating): Correct the lower bound of precision.
30751 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30753 * doc/invoke.texi: Document default code model for 64-bit Linux.
30755 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30758 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30759 udiv rather than div since input pattern is unsigned.
30761 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30763 * config/i386/i386.c (print_reg): Warn for values of
30764 unsupported size in integer register.
30766 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30769 * config/rs6000/predicates.md (current_file_function_operand): Do
30770 not allow self calls to be local if the function is replaceable.
30772 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30775 * config/rs6000/altivec.h (vec_ctz and others): Change the
30776 preprocessor macro that controls conditional compilation from
30777 _ARCH_PWR9 to __POWER9_VECTOR__.
30778 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30779 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30780 control (instead of _ARCH_PWR9 control) so that template
30781 definition uses power9-specific function.
30782 (vec_any_eq): Likewise.
30783 (vec_all_ne): Change macro definition to use a power9-specific
30784 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30785 _ARCH_PWR9 control).
30786 (vec_any_eq) Likewise.
30787 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30788 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30789 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30790 support for xvcmpnedp instruction.
30791 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30792 macro expansion so that Power9 implementation of vec_all_ne does
30793 not use the AltiVec predicate framework.
30794 (VCMPNEH_P): Likewise.
30795 (VCMPNEW_P): Likewise.
30796 (VCMPNED_P): Likewise.
30797 (VCMPNEFP_P): Likewise.
30798 (VCMPNEDP_P): Likewise.
30799 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30800 implementation of vec_any_eq to not use AltiVec predicate
30802 (VCMPAEH_P): Likewise.
30803 (VCMPAEW_P): Likewise.
30804 (VCMPAED_P): Likewise.
30805 (VCMPAEFP_P): Likewise.
30806 (VCMPAEDP_P): Likewise.
30807 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30808 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30809 not use the AltiVec predicate framework.
30810 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30811 of vec_any_eq to not use AltiVec predicate framework.
30812 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30813 support for predefined __POWER9_VECTOR__ macro to indicate that
30814 Power9 instruction selection is enabled.
30815 (altivec_overloaded_builtins): Remove extraneous
30816 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30817 function argument types RS6000_BTI_bool_V16QI and
30818 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30819 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30820 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30821 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30822 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30823 Power9 for implementations of vec_cmpne. Change the signature for
30824 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30825 (representing vec_all_ne) to remove the previously described first
30826 argument of type RS6000_BTI_INTSI, as this was an artifact of
30827 reliance on the AltiVec predicate framework, which is no longer
30828 used in the implementation of these functions. Add
30829 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30830 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30831 since, unlike the AltiVec predicate framework implementation, we
30832 do not share function descriptors between vec_alle and vec_anyeq.
30833 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30834 set of modes that receive special treatment even when
30835 TARGET_P9_VECTOR is true. The special treatment emits code that
30836 does not depend on Power9 instructions.
30837 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30838 define_expand to not rely on AltiVec predicate framework.
30839 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30841 (vector_ne_v2di_p): Change this define_expand to not rely on
30842 AltiVec predicate framework.
30843 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30845 (vector_ne_<mode>_p): Change this define_expand to not rely on
30846 AltiVec predicate framework.
30847 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30849 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30850 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30851 define_insn pattern.
30852 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30853 define_insn pattern because the xvcmpne<VSs>. instruction is not
30855 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30856 instruction is not supported.
30858 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30860 * config/nvptx/nvptx.c: Include intl.h.
30862 2017-03-01 Martin Jambor <mjambor@suse.cz>
30865 * ipa-prop.h (ipa_bits): Removed field known.
30866 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30867 to pointers. Adjusted their comments to warn about their sharing.
30868 (ipcp_transformation_summary): Change bits to a vector of pointers.
30869 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30870 (ipa_get_ipa_bits_for_value): Declare.
30871 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30872 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30873 (ipa_bits_hash_table): Likewise.
30874 (ipa_vr_ggc_hash_traits): Likewise.
30875 (ipa_vr_hash_table): Likewise.
30876 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30877 being pointers and vr_known being removed.
30878 (ipa_set_jf_unknown): Likewise.
30879 (ipa_get_ipa_bits_for_value): New function.
30880 (ipa_set_jfunc_bits): Likewise.
30881 (ipa_get_value_range): New overloaded functions.
30882 (ipa_set_jfunc_vr): Likewise.
30883 (ipa_compute_jump_functions_for_edge): Use the above functions to
30884 construct bits and vr parts of jump functions.
30885 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30886 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30888 (ipcp_grow_transformations_if_necessary): Also allocate
30889 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30891 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30892 them. Fix too long lines.
30893 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30894 vr_known being removed.
30895 (ipa_read_jump_function): Use new setter functions to construct bits
30896 and vr parts of jump functions or set them to NULL.
30897 (write_ipcp_transformation_info): Adjust for bits being pointers.
30898 (read_ipcp_transformation_info): Likewise.
30899 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30901 Include gt-ipa-prop.h.
30902 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30904 (ipcp_store_bits_results): Likewise.
30905 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30906 Do not write to existing jump functions but use a temporary instead.
30908 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30911 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30912 attempt to use its first operand as BIT_FIELD_REF base.
30914 2017-03-01 Richard Biener <rguenther@suse.de>
30916 PR middle-end/79721
30917 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30918 interpolating formula in wrapping arithmetic.
30919 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30921 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30923 PR tree-optimization/79734
30924 * tree-vect-generic.c (expand_vector_condition): Optimize
30925 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30926 Handle VEC_COND_EXPR where comparison has different inner width from
30927 type's inner width.
30929 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30931 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30932 markup, and similar issues. Remove @opindex entries for things
30933 that aren't options. Add missing -mmpy-option entries.
30935 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30937 PR tree-optimization/79737
30938 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30939 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30940 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30941 instead of byte_size. Formatting fix.
30942 (shift_bytes_in_array_right): Formatting fix.
30944 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30947 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30948 condition on optimize for the leaf function test.
30950 2017-02-28 Martin Liska <mliska@suse.cz>
30953 * read-rtl-function.c (function_reader::handle_unknown_directive):
30954 Bail out when one uses -flto.
30956 2017-02-28 Martin Liska <mliska@suse.cz>
30958 * common.opt: Replace space with tabular for options of <number>
30960 * config/i386/i386.opt: Show <number> value for
30961 -mlarge-data-threshold.
30962 * opts.c (print_filtered_help): Do not display number in hexadecimal
30965 2017-02-28 Martin Liska <mliska@suse.cz>
30967 * common.opt: Fix --help=option -Q for options which are of
30970 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30972 * config/i386/i386.c (print_reg): Error out for values
30973 of 8-bit size in invalid integer register.
30975 2017-02-28 Martin Sebor <msebor@redhat.com>
30977 PR tree-optimization/79691
30978 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30980 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30983 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30984 gcc_unreachable with output_operand_lossage.
30986 2017-02-28 Richard Biener <rguenther@suse.de>
30988 PR tree-optimization/79740
30989 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30991 (visit_nary_op): Insert the nary into the hashtable if we
30992 pattern-matched sth.
30993 * tree-ssa-pre.c (eliminate_insert): Robustify.
30995 2017-02-28 Richard Biener <rguenther@suse.de>
30997 PR middle-end/79731
30998 * fold-const.c (decode_field_reference): Reject out-of-bound
31001 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31003 * config/i386/i386.c: Include intl.h.
31004 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31005 instead of just cond ? "..." : "...".
31006 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31007 * coverage.c (read_counts_file): Likewise.
31008 * omp-offload.c: Include intl.h.
31009 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31010 of just cond ? "..." : "...".
31011 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31012 of just cond ? "..." : "...".
31014 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31017 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31019 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31020 'tune for' CPU name.
31021 * config/arm/arm-cpu-data.h: Regenerated.
31023 2017-02-28 Richard Biener <rguenther@suse.de>
31025 PR tree-optimization/79732
31026 * tree-inline.c (expand_call_inline): Do not shadow var.
31028 2017-02-28 Richard Biener <rguenther@suse.de>
31030 PR tree-optimization/79723
31031 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31032 address-space properly.
31034 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31036 * doc/optinfo.texi (Optimization groups): Fix option used for
31038 * doc/invoke.texi (-fopt-info): Document "omp".
31039 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31040 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31041 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31042 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31043 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31045 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31047 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31048 OPTGROUP_OMP with "omp".
31050 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31053 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31054 for arithmetic shift of unsigned V2DI.
31056 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31058 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31059 arc/linux.h headers.
31060 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31061 (LINK_SPEC): Likewise.
31062 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31063 (EXTRA_SPECS): Likewise.
31064 (STARTFILE_SPEC): Likewise.
31065 (ENDFILE_SPEC): Likewise.
31066 (LIB_SPEC): Likewise.
31067 (TARGET_SDATA_DEFAULT): Likewise.
31068 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31069 (MULTILIB_DEFAULTS): Likewise.
31070 (DWARF2_UNWIND_INFO): Likewise.
31071 * config/arc/big.h: New file.
31072 * config/arc/elf.h: Likewise.
31073 * config/arc/linux.h: Likewise.
31074 * config/arc/t-uClibc: Remove.
31076 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31078 PR tree-optimization/77536
31079 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31080 (tree_transform_and_unroll_loop): Use above function to compute the
31081 estimated niter of unrolled loop and use it when scaling profile.
31082 Also use count info rather than frequency if it's non-zero.
31083 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31084 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31085 (vect_transform_loop): Call above function.
31087 2017-02-27 Richard Biener <rguenther@suse.de>
31089 PR tree-optimization/45397
31090 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31091 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31092 (visit_nary_op): Add pattern matching for CSEing sign-changed
31093 or truncated operations with wider ones.
31095 2017-02-27 Richard Biener <rguenther@suse.de>
31097 PR tree-optimization/79690
31098 * tree-vect-stmts.c (vectorizable_store): Use vector type
31099 built from the DR with address-space.
31101 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31103 * doc/invoke.texi (Optimize Options): Refine the description
31104 of asan-use-after-return.
31106 2017-02-25 Alan Modra <amodra@gmail.com>
31108 PR rtl-optimization/79584
31109 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31110 base, not ad->base_term, the reg within base. Remove assertion
31111 that ad->base == ad->base_term. Replace gen_int_mode using
31112 bogus mode with const0_rtx.
31114 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31116 PR middle-end/79396
31117 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31118 FMA_EXPR like tcc_binary or tcc_unary.
31120 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31123 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31125 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31126 (output_loc_operands): Handle DW_OP_call_ref and
31127 DW_OP_GNU_variable_value.
31128 (struct variable_value_struct): New type.
31129 (struct variable_value_hasher): Likewise.
31130 (variable_value_hash): New variable.
31131 (string_types): Remove.
31132 (copy_loc_descr): New function.
31133 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31134 (prepend_loc_descr_to_each): New function.
31135 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31136 instead of add_loc_descr_to_each if the first argument is single
31137 location list and the second has multiple.
31138 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31139 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31140 when looking for variable value which doesn't have other location info.
31141 (loc_list_from_tree): Formatting fix.
31142 (gen_array_type_die): Simplify DW_AT_string_length handling.
31143 (adjust_string_types): Remove.
31144 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31145 string_types. Call resolve_variable_values.
31146 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31147 (resolve_addr_in_expr): Likewise. Add A argument.
31148 (copy_deref_exprloc): Remove deref argument. Adjust for the
31149 original expression being DW_OP_GNU_variable_value with optionally
31150 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31151 optionally after it.
31152 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31153 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31154 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31155 (variable_value_hasher::hash, variable_value_hasher::equal): New
31157 (resolve_variable_value_in_expr, resolve_variable_value,
31158 resolve_variable_values, note_variable_value_in_expr,
31159 note_variable_value): New functions.
31160 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31163 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31166 * opts.h (handle_generated_option): Add GENERATED_P argument.
31167 * opts-common.c (handle_option): Adjust function comment.
31168 (handle_generated_option): Add GENERATED_P argument, pass it to
31170 (control_warning_option): Pass false to handle_generated_option
31172 * opts.c (maybe_default_option): Pass true to handle_generated_option
31174 * optc-gen.awk: Likewise.
31176 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31178 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31179 a REG, look at the REG it is a SUBREG of.
31180 (splitter for cmpeqsi_t): Ditto.
31182 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31184 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31185 the special USEs with the pattern of the insn, not the insn itself.
31187 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31190 * doc/invoke.texi: Document -mload-store-pairs.
31192 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31193 Sandra Loosemore <sandra@codesourcery.com>
31195 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31196 argument isn't a CONST_INT.
31197 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31198 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31199 (nios2_validate_compare): Bypass alternate compare logic if *op2
31200 is not a CONST_INT.
31201 (ldstwm_operation_p): Return false if first_base is not a REG or
31202 if first_offset is not a CONST_INT.
31204 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31206 * config/cris/cris.md: Use correct operand in a define_peephole2.
31208 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31210 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31212 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31214 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31215 this_insn if it is an INSN or JUMP_INSN.
31216 (force_offsettable): Look at base, not at addr.
31217 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31218 on things that aren't necessarily CONST_INTs.
31220 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31222 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31223 -mfpmath=sse is the default also for x86-32 targets with SSE2
31224 instruction set when @option{-ffast-math} is enabled
31226 2017-02-24 Jeff Law <law@redhat.com>
31228 PR rtl-optimizatoin/79286
31229 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31232 2017-02-24 Richard Biener <rguenther@suse.de>
31234 PR tree-optimization/79389
31235 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31238 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31240 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31241 function comment to reflect reality.
31242 (loop_exits_before_overflow): Fix typo in function description.
31244 2017-02-24 Richard Biener <rguenther@suse.de>
31246 PR tree-optimization/79389
31247 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31248 properly that a threading opportunity exists. Detect conditional
31249 copy/constant propagation opportunities.
31251 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31253 * config/visium/visium.md (type): Add trap.
31254 (b): New mode attribute.
31255 (*btst): Rename into...
31256 (*btst<mode>): ...this and adjust.
31257 (*cbranchsi4_btst_insn): Rename into...
31258 (*cbranch<mode>4_btst_insn): ...this and adjust.
31259 (trap): New define_insn.
31261 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31263 PR tree-optimization/79389
31264 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31265 (noce_reversed_cond_code): New function.
31266 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31267 reversed_comparison_code. Formatting fix.
31268 (noce_try_store_flag): Test rev_cond != NULL in addition to
31269 reversed_comparison_code.
31270 (noce_try_store_flag_constants): Likewise.
31271 (noce_try_store_flag_mask): Likewise.
31272 (noce_try_addcc): Use rev_cond if non-NULL instead of
31273 reversed_comparison_code.
31274 (noce_try_cmove_arith): Likewise. Formatting fixes.
31275 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31276 (noce_find_if_block): Initialize rev_cond.
31277 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31278 instead of false as last argument never attempt to reverse it
31281 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31283 PR tree-optimization/79663
31284 * tree-predcom.c (combine_chains): Process refs in reverse order
31285 only for ZERO length chains, and add explaining comment.
31287 2017-02-23 Jeff Law <law@redhat.com>
31289 PR tree-optimization/79578
31290 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31291 in call to operand_equal_p.
31293 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31296 * config/i386/cpuid.h: Fix another undefined behavior.
31298 2017-02-23 Richard Biener <rguenther@suse.de>
31300 PR tree-optimization/79683
31301 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31302 vector types for data-refs.
31304 2017-02-23 Martin Liska <mliska@suse.cz>
31306 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31308 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31310 PR middle-end/79665
31311 * internal-fn.c (get_range_pos_neg): Moved to ...
31312 * tree.c (get_range_pos_neg): ... here. No longer static.
31313 * tree.h (get_range_pos_neg): New prototype.
31314 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31315 are known to be in between 0 and signed maximum inclusive, try to
31316 expand both unsigned and signed divmod and use the cheaper one from
31319 2017-02-22 Jeff Law <law@redhat.com>
31321 PR tree-optimization/79578
31322 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31323 to compare base operands.
31325 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31328 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31329 gpc_reg_operand instead of fpr_reg_operand.
31331 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31333 * config/mips/mips.c (mips_return_in_memory): Force FP
31334 vector types to be returned in memory for o32 ABI.
31336 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31338 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31339 instead of DW_TAG_member for static data member declarations and don't
31340 set no_linkage_name for static inline data members.
31341 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31344 2017-02-22 Martin Liska <mliska@suse.cz>
31346 * doc/invoke.texi: Replace inequality signs with square brackets
31349 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31352 * lra-constraints.c (simplify_operand_subreg): Handle
31353 WORD_REGISTER_OPERATIONS targets.
31355 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31358 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31359 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31360 elimination by swapping fld*.
31362 2017-02-22 Richard Biener <rguenther@suse.de>
31364 PR tree-optimization/79673
31365 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31366 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31367 irrelevant address-space qualifiers and avoiding a
31368 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31370 2017-02-22 Richard Biener <rguenther@suse.de>
31372 PR tree-optimization/79666
31373 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31374 to not symbolically negate if that may introduce undefined
31377 2017-02-22 Martin Liska <mliska@suse.cz>
31380 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31381 * data-streamer-out.c (streamer_write_gcov_count_stream):
31383 * value-prof.c (stream_out_histogram_value): Make assert more
31384 precise based on type of counter.
31386 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31389 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31390 Use nonimmediate_operand instead of memory_operand for operand 1.
31391 (float-extend standard_x87sse_constant_load splitter): Ditto.
31393 2017-02-21 Jeff Law <law@redhat.com>
31395 PR tree-optimization/79621
31396 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31397 blocks with edges to themselves.
31399 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31402 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31403 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31404 Use gimple_call_builtin_p.
31407 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31408 on temporarily removed DEBUG_INSNs.
31410 PR tree-optimization/79649
31411 * tree-loop-distribution.c (classify_partition): Give up on
31412 non-generic address space loads/stores.
31414 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31416 * doc/loop.texi (Loop manipulation): Remove nonexistent
31417 tree_ssa_loop_version from the documentation.
31418 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31420 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31423 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31424 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31425 * config/rs6000/rs6000.c: Include except.h.
31426 (rs6000_expand_split_stack_prologue): Call
31427 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31429 2017-02-21 Martin Jambor <mjambor@suse.cz>
31432 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31433 have been analyzed.
31435 2017-02-21 Martin Jambor <mjambor@suse.cz>
31437 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31438 for backward compatibility only.
31439 * doc/invoke.texi (Option Summary): Remove all references to
31440 -fipa-cp-alignment.
31442 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31446 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31448 * lra-constraints.c (curr_insn_transform): Handle
31449 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31451 2017-02-21 Martin Liska <mliska@suse.cz>
31453 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31455 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31457 PR translation/79638
31458 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31460 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31463 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31464 (arm_function_ok_for_sibcall): Return false for an indirect call by
31465 descriptor if all the argument registers are used.
31466 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31467 alignment of the function.
31469 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31471 PR tree-optimization/61441
31472 * simplify-rtx.c (simplify_const_unary_operation): For
31473 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31474 the sNaN unmodified.
31476 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31478 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31479 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31480 instead of SYSTEM_HEADER_DIR.
31482 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31483 Martin Liška <mliska@suse.cz>
31485 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31486 Fix typos and grammar, use active voice, and clarify.
31488 2017-02-20 Marek Polacek <polacek@redhat.com>
31490 PR middle-end/79537
31491 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31494 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31496 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31499 * config/i386/i386.c (ix86_expand_builtin): Handle
31500 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31501 ix86_builtins_isa[fcode].isa as a requirement of those
31502 flags and any other flag in the bitmask.
31503 (ix86_init_mmx_sse_builtins): Use 0 instead of
31504 ~OPTION_MASK_ISA_64BIT as mask.
31505 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31506 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31507 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31508 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31510 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31513 * lra-constraints.c (split_reg): Check requested split mode
31514 is supported by the register.
31516 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31518 * lra-constraints.c (simplify_operand_subreg): Remove early
31521 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31524 * lra-constraints.c (curr_insn_transform): Tighten condition
31525 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31527 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31530 * lra-constraints.c (curr_insn_transform): Handle
31531 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31533 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31536 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31538 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31540 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31543 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31546 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31548 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31551 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31553 * final.c (last_columnnum, override_columnnum): New variables.
31554 (final_start_function): Set last_columnnum, pass it to begin_prologue
31555 hook and pass 0 to dwarf2out_begin_prologue.
31556 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31557 to source_line debug hook.
31558 (notice_source_line): Compute last_columnnum and for debug_column_info
31559 return true on column changes.
31560 * debug.h (struct gcc_debug_hooks): Add column argument to
31561 source_line and begin_prologue hooks.
31562 (debug_nothing_int_charstar_int_bool): Remove prototype.
31563 (debug_nothing_int_int_charstar,
31564 debug_nothing_int_int_charstar_int_bool): New prototypes.
31565 (dwarf2out_begin_prologue): Add column argument.
31566 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31567 begin_prologue hooks.
31568 (debug_nothing_int_charstar_int_bool): Remove.
31569 (debug_nothing_int_int_charstar,
31570 debug_nothing_int_int_charstar_int_bool): New functions.
31571 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31572 through to dwarf2out_source_line.
31573 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31574 (dwarf2out_source_line): Add column argument, emit it if requested.
31575 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31577 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31578 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31579 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31580 through to dwarf2out_begin_prologue.
31581 (vmsdbgout_source_line): Add column argument, pass it through to
31582 dwarf2out_source_line.
31583 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31584 dbxout_source_line caller.
31585 (dbxout_source_line): Add column argument.
31587 * common.opt (gno-column-info, gcolumn-info): New options.
31588 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31589 (check_die): Also test for multiple DW_AT_decl_column attributes.
31590 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31591 DW_AT_decl_column if requested.
31592 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31594 (gen_variable_die): Likewise.
31595 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31596 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31599 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31600 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31601 (ix86_handle_option): Handle OPT_m3dnowa.
31602 * doc/invoke.texi (-m3dnowa): Document.
31603 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31604 -m3dnowa instead of -m3dnow -march=athlon.
31607 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31608 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31610 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31613 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31614 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31615 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31616 generator for vsx_xxpermdi_<mode>_be.
31617 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31618 force big-endian semantics.
31619 (vsx_xxpermdi_<mode>_be): New define_expand with same
31620 implementation as previous version of vsx_xxpermdi_<mode>.
31622 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31624 PR tree-optimization/79327
31625 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31626 variable, its initialization and use.
31628 2017-02-17 Julia Koval <julia.koval@intel.com>
31630 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31631 (OPTION_MASK_ISA_PKU_UNSET): New.
31632 (ix86_handle_option): Handle -mrdpid.
31633 * config/i386/cpuid.h (bit_RDPID): New.
31634 * config/i386/driver-i386.c (host_detect_local_cpu):
31635 Detect RDPID feature.
31636 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31637 * config/i386/i386-c.c (ix86_target_macros_internal):
31639 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31640 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31641 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31642 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31643 * config/i386/i386.md (define_insn "rdpid"): New.
31644 * config/i386/i386.opt Add -mrdpid.
31645 * config/i386/immintrin.h (_rdpid_u32): New.
31647 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31649 PR rtl-optimization/79541
31650 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31651 instead of transforming it into USE.
31653 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31655 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31656 If HONOR_SNANS (SFmode) force the input to a register.
31657 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31658 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31659 an frsp or similar insn.
31661 2017-02-17 Martin Liska <mliska@suse.cz>
31663 PR rtl-optimization/79577
31664 * params.def (selsched-max-sched-times): Increase minimum to 1.
31666 2017-02-17 Martin Liska <mliska@suse.cz>
31668 PR rtl-optimization/79574
31669 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31671 2017-02-17 Martin Liska <mliska@suse.cz>
31673 PR tree-optimization/79529
31674 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31675 ssa_defined_default_def_p to handle cases which are implicitly
31677 * tree-ssa.c (ssa_defined_default_def_p): New function.
31678 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31679 which are implicitly defined.
31680 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31682 2017-02-17 Richard Biener <rguenther@suse.de>
31684 PR middle-end/79576
31685 * params.def (max-ssa-name-query-depth): Limit to 10.
31687 2017-02-17 Richard Biener <rguenther@suse.de>
31689 PR tree-optimization/79552
31690 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31693 2017-02-17 Richard Biener <rguenther@suse.de>
31696 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31698 2017-02-17 Marek Polacek <polacek@redhat.com>
31700 PR middle-end/79536
31701 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31702 (fold_negate_expr): New wrapper.
31704 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31706 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31707 Correct terminology and de-emphasize pre-standard behavior.
31709 2017-02-16 Alan Modra <amodra@gmail.com>
31711 PR rtl-optimization/79286
31712 * ira.c (def_dominates_uses): New function.
31713 (update_equiv_regs): Don't create an equivalence for insns that
31714 may trap where the register def does not dominate the use.
31716 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31718 PR rtl-optimization/78127
31719 * lra.c (lra): Call lra_eliminate before finish the loop after
31722 2017-02-16 Richard Biener <rguenther@suse.de>
31724 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31726 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31727 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31728 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31729 (isl_val_int_from_wi): New function.
31730 (extract_affine_gmp): Rename to ...
31731 (extract_affine_wi): ... this, take a widest_int.
31732 (extract_affine_int): Just wrap extract_affine_wi.
31733 (add_param_constraints): Use isl_val_int_from_wi.
31734 (add_loop_constraints): Likewise, and extract_affine_wi.
31736 2017-02-15 Jeff Law <law@redhat.com>
31738 PR middle-end/79521
31739 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31740 ira_init_register_move_cost_if_necessary.
31742 2017-02-15 Martin Sebor <msebor@redhat.com>
31744 PR middle-end/32003
31745 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31746 removed in a prior commit.
31748 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31750 PR tree-optimization/79347
31751 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31752 counters during peeling.
31754 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31756 * Makefile.in (site.exp): Remove "set ISLVER".
31758 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31761 * real.c (real_from_integer): Call real_convert even for decimal.
31763 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31766 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31768 2017-02-14 Andrew Pinski <apinski@cavium.com>
31770 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31771 cores and change the partno/implementer to be correct.
31772 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31773 the 'B" as the implementer.
31774 * config/aarch64/aarch64-tune.md: Regenerate.
31776 2017-02-14 Carl Love <cel@us.ibm.com>
31778 * config/rs6000/rs6000.c: Add case statement entry to make the
31779 xvcvuxdsp built-in argument unsigned.
31780 * config/rs6000/vsx.md: Fix the source and return operand types so they
31781 match the instruction definitions from the ISA document. Fix typo
31782 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31785 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31788 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31789 member early_clobber_alts.
31790 * lra-lives.c (reg_early_clobber_p): New.
31791 (process_bb_lives): Use it.
31792 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31793 (debug_operand_data): Initialize early_clobber_alts.
31794 (setup_operand_alternative): Set up early_clobber_alts.
31795 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31796 alternatives to new_insn_reg.
31797 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31799 (lra_update_insn_regno_info): Pass the new arg.
31801 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31803 PR middle-end/79505
31804 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31805 (new_oacc_loop_raw): Don't clear already cleared fields.
31808 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31809 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31810 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31812 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31815 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31817 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31820 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31821 the extra instruction to the right place to store 128-bit constant
31824 2017-02-14 Martin Sebor <msebor@redhat.com>
31826 PR middle-end/79448
31827 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31828 warning for strings of unknown length.
31830 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31832 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31834 2017-02-14 Jeff Law <law@redhat.com>
31837 * ira-costs.c (scan_one_insn): Initialize register move costs
31838 for pseudos seen in USE/CLOBBER insns.
31840 PR tree-optimization/79095
31841 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31842 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31843 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31844 if the operands are known to be not equal, then the resulting range
31846 (intersect_ranges): If the new range is ~[0,0] and the old range is
31847 wide, then prefer ~[0,0].
31848 * tree-vrp.c (overflow_comparison_p_1): New function.
31849 (overflow_comparison_p): New function.
31850 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31851 if NAME is used in an overflow test.
31852 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31853 overflow check that can be expressed as an equality test, then adjust
31854 ops to be that equality test.
31856 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31858 * config/s390/s390-builtin-types.def: Remove flags argument.
31859 * config/s390/s390.c (s390_init_builtins): Likewise.
31861 2017-02-14 Martin Liska <mliska@suse.cz>
31863 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31864 vector. Fix trailing white spaces.
31866 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31868 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31871 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31873 PR rtl-optimization/68664
31874 * config/arm/arm.c (arm_sched_can_speculate_insn):
31875 New function. Declare prototype.
31876 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31878 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31880 PR rtl-optimization/68664
31881 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31883 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31885 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31887 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31888 max skip bytes for function, loop and jump.
31890 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31892 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31893 ABS_EXPR for gimple dump.
31895 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31898 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31900 PR tree-optimization/79408
31901 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31902 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31903 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31904 also if rhs1 is INTEGER_CST.
31906 2017-02-14 Richard Biener <rguenther@suse.de>
31908 PR middle-end/79432
31909 * tree-into-ssa.c (insert_phi_nodes): When the function can
31910 have abnormal edges rewrite SSA names with broken use-def
31911 dominance out of SSA and register them for PHI insertion.
31913 2017-02-13 Martin Sebor <msebor@redhat.com>
31915 PR middle-end/79496
31916 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31917 clearing info.nowrite flag when snprintf size argument is a range.
31919 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31921 * cprop.c (cprop_jump): Add missing space in string literal.
31922 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31923 (get_constraint_for_component_ref): Likewise.
31924 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31925 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31926 * lra-constraints.c (process_alt_operands): Likewise.
31927 * ipa-inline.c (inline_small_functions): Likewise.
31928 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31929 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31930 * trans-mem.c (diagnose_tm_1_op): Likewise.
31931 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31932 (grid_parallel_clauses_gridifiable): Likewise.
31934 * config/nvptx/mkoffload.c (process): Add space in between
31937 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31938 "MOD4_SSE_REGS" and "ALL_REGS".
31940 * spellcheck.c (test_data): Add , in between "foo" and "food".
31942 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31945 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31946 boundary crossing check and subsequent code generation agree.
31948 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31950 * config/aarch64/aarch64.c (has_memory_op): Delete.
31951 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31954 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31956 PR rtl-optimization/79388
31957 PR rtl-optimization/79450
31958 * combine.c (distribute_notes): When removing TEM_INSN for which
31959 corresponding dest has last value recorded, invalidate that last
31962 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31964 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31965 of explicit '@'. Add missing assembly comment marker on branch costs
31968 2017-02-13 Nathan Sidwell <nathan@acm.org>
31970 * gengtype-lex.l (<in_struct>): Add '/'.
31972 2017-02-13 Martin Liska <mliska@suse.cz>
31975 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31977 2017-02-13 Richard Biener <rguenther@suse.de>
31979 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31981 * configure: Re-generate.
31982 * config.in: Likewise.
31983 * graphite-dependences.c: Simplify as if
31984 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31985 * graphite-isl-ast-to-gimple.c: Likewise.
31986 * graphite-optimize-isl.c: Likewise.
31987 * graphite-poly.c: Likewise.
31988 * graphite-sese-to-poly.c: Likewise.
31989 * graphite.h: Likewise.
31990 * toplev.c: Include isl/version.h and use isl_version () for
31991 printing the ISL version.
31992 * doc/install.texi: Update ISL requirement.
31994 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31996 * doc/standards.texi (Standards): Update reference to
31999 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32001 * doc/extend.texi (Named Address Spaces): sourceware.org now
32003 * doc/install.texi (Binaries): Ditto.
32006 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32008 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32009 standard terminology "stringize"/"stringizing" throughout.
32010 * doc/cppinternals.texi: Likewise.
32012 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32014 * doc/extend.texi: Fix some spelling mistakes and typos.
32015 * doc/invoke.texi: Likewise.
32017 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32020 * params.def (inline-min-speedup) Change from 10 to 8.
32022 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32024 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32027 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32030 * ipa-inline-analysis.c (get_minimal_bb): New function.
32031 (record_modified): Use it.
32032 (remap_edge_change_prob): Handle also ancestor functions.
32034 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32036 * doc/contrib.texi (Contributors): Remove broken link into
32037 the Mauve CVS repository.
32039 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32041 PR middle-end/79454
32042 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32043 result computation whenever lhs doesn't have vector mode, not
32044 just when it has BLKmode.
32046 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32048 * doc/makefile.texi (profiledbootstrap): Refer to the
32049 installation instructions only in textual form.
32051 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32054 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32056 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32058 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32059 (Specific): Update mingw-w64 reference.
32061 (Specific): Remove broken link to Renesas RX processor.
32063 2017-02-10 Richard Biener <rguenther@suse.de>
32065 * toplev.c (process_options): Do not mention obsolete graphite
32066 options when printing sorry message about missing graphite support.
32067 Mention -floop-nest-optimize.
32069 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32071 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32072 (vtst_p16): Likewise.
32073 (vtstq_p8): Likewise.
32074 (vtstq_p16): Likewise.
32076 (vtstq_p64): Likewise.
32077 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32078 (vset_lane_p64): New.
32079 (vsetq_lane_p64): New.
32081 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32083 PR tree-optimization/79411
32084 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32085 stmt operands are SSA_NAMEs used in abnormal phis.
32086 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32089 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32092 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32095 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32097 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32099 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32101 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32105 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32106 not arbitrary TREE_CONSTANT.
32109 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32110 "omp declare target link" attribute unless is_global_var.
32111 * omp-offload.c (find_link_var_op): Likewise.
32113 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32114 Chung-Lin Tang <cltang@codesourcery.com>
32116 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32118 (gimplify_adjust_omp_clauses): Don't delete TILE.
32119 (gimplify_omp_for): Deal with TILE.
32120 * internal-fn.c (expand_GOACC_TILE): New function.
32121 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32123 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32124 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32126 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32127 avoid DIV for outermost collapse var.
32128 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32129 Remove out of date comments, fix whitespace.
32130 * omp-general.c (omp_extract_for_data): Deal with tiling.
32131 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32132 adjust OLF_DIM_BASE value.
32133 (struct omp_for_data): Add tiling field.
32134 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32135 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32136 for auto loops. Remove default auto determining, moved to
32137 oacc_loop_fixed_partitions.
32138 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32139 stmts, add e_mask field.
32140 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32141 (oacc_thread_numbers): Use oacc_dim_call.
32142 (oacc_xform_tile): New.
32143 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32144 (finish_oacc_loop): Adjust for ifns vector.
32145 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32146 add case for GOACC_TILE fns.
32147 (oacc_loop_xform_loop): Delete.
32148 (oacc_loop_process): Iterate over call list directly, and add
32149 handling for GOACC_TILE fns.
32150 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32152 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32153 vector partitioning to outer loops. Assign 2 partitions to loops
32154 when available. Add TILE handling.
32155 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32156 (execite_oacc_device_lower): Process GOACC_TILE fns,
32157 ignore unknown specs.
32158 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32159 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32160 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32162 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32164 * configure.ac (ACX_BUGURL): Update.
32165 * configure: Regenerate.
32167 2017-02-09 Richard Biener <rguenther@suse.de>
32169 PR tree-optimization/69823
32170 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32171 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32173 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32175 * config/arc/arc-c.def: Add __NPS400__ definition.
32176 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32177 (TARGET_NPS400): Define.
32179 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32181 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32183 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32184 pointer, arch_info.
32185 (arc_cpu_types): Fill the arch_info field with a pointer into the
32186 arc_arch_types table.
32187 (arc_selected_cpu): Declare.
32188 * config/arc/arc.c (arc_selected_cpu): Make global.
32189 (arc_selected_arch): Delete.
32190 (arc_base_cpu): Delete.
32191 (arc_override_options): Remove references to deleted variables,
32192 update access to arch information.
32193 (ARC_OPT): Update access to arch information.
32194 (ARC_OPTX): Likewise.
32195 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32196 (TARGET_ARC600): Update access to arch information.
32197 (TARGET_ARC601): Likewise.
32198 (TARGET_ARC700): Likewise.
32199 (TARGET_EM): Likewise.
32200 (TARGET_HS): Likewise.
32201 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32204 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32207 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32208 condition/operands for integer GE/LE/GEU/LEU operations.
32210 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32212 PR translation/79397
32213 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32216 2017-02-08 Martin Jambor <mjambor@suse.cz>
32219 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32220 whether allocation happened.
32221 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32222 nothing was allocated.
32224 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32226 PR tree-optimization/79408
32227 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32228 constant, but SSA_NAME with a known integer range, use the minimum
32229 of that range instead of op1 to determine if modulo can be replaced
32230 with its first operand.
32232 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32234 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32236 2017-02-08 Richard Biener <rguenther@suse.de>
32238 PR tree-optimization/71824
32239 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32240 Check all loops contained in the merged region.
32242 2017-02-07 Andrew Pinski <apinski@cavium.com>
32244 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32246 2017-02-07 Andrew Pinski <apinski@cavium.com>
32248 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32249 (thunderxt88): Likewise.
32250 (thunderxt81): Disable LSE and change v8.1 to v8.
32251 (thunderxt83): Likewise.
32253 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32254 Richard Biener <rguenther@suse.de>
32256 PR middle-end/79399
32257 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32258 type from int to size_t.
32259 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32261 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32263 PR rtl-optimization/79386
32264 * cprop.c (bypass_conditional_jumps): Initialize
32265 bypass_last_basic_block already before splitting bbs after
32266 unconditional traps...
32267 (bypass_conditional_jumps): ... rather than here.
32270 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32271 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32272 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32273 fix -masm=intel patterns.
32275 2017-02-07 Richard Biener <rguenther@suse.de>
32277 PR tree-optimization/79256
32278 PR middle-end/79278
32279 * builtins.c (get_object_alignment_2): Use min_align_of_type
32280 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32281 and ADJUST_FIELD_ALIGN.
32283 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32285 * doc/tm.texi: Regenerate.
32286 * stor-layout.c (layout_decl): Adjust.
32287 (update_alignment_for_field): Likewise.
32288 (place_field): Likewise.
32289 (min_align_of_type): Likewise.
32290 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32291 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32292 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32293 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32294 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32295 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32296 * config/i386/i386.c (x86_field_alignment): Likewise.
32297 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32298 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32299 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32300 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32301 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32302 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32306 2017-01-30 Richard Biener <rguenther@suse.de>
32308 PR tree-optimization/79256
32309 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32310 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32313 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32315 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32316 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32317 builtins to SImode and emit a zero-extend, if necessary.
32319 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32321 * docs/invoke.texi (RISC-V Options): Alphabetize.
32323 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32325 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32328 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32330 * config/riscv/riscv.c: New file.
32331 * common/config/riscv/riscv-common.c: Likewise.
32332 * config.gcc: Likewise.
32333 * config/riscv/constraints.md: Likewise.
32334 * config/riscv/elf.h: Likewise.
32335 * config/riscv/generic.md: Likewise.
32336 * config/riscv/linux.h: Likewise.
32337 * config/riscv/multilib-generator: Likewise.
32338 * config/riscv/peephole.md: Likewise.
32339 * config/riscv/pic.md: Likewise.
32340 * config/riscv/predicates.md: Likewise.
32341 * config/riscv/riscv-builtins.c: Likewise.
32342 * config/riscv/riscv-c.c: Likewise.
32343 * config/riscv/riscv-ftypes.def: Likewise.
32344 * config/riscv/riscv-modes.def: Likewise.
32345 * config/riscv/riscv-opts.h: Likewise.
32346 * config/riscv/riscv-protos.h: Likewise.
32347 * config/riscv/riscv.h: Likewise.
32348 * config/riscv/riscv.md: Likewise.
32349 * config/riscv/riscv.opt: Likewise.
32350 * config/riscv/sync.md: Likewise.
32351 * config/riscv/t-elf-multilib: Likewise.
32352 * config/riscv/t-linux: Likewise.
32353 * config/riscv/t-linux-multilib: Likewise.
32354 * config/riscv/t-riscv: Likewise.
32355 * configure.ac: Likewise.
32356 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32357 Waterman as RISC-V maintainers.
32358 * doc/install.texi: Add RISC-V entries.
32359 * doc/invoke.texi: Add RISC-V options section.
32360 * doc/md.texi: Add RISC-V constraints section.
32361 * configure: Regenerated.
32363 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32366 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32367 false values to be constant vectors with all 0 or all 1 bits set.
32368 (vcondu<mode><mode>): Likewise.
32369 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32371 (fpmask_comparison_operator): Update comment.
32372 (vecint_comparison_operator): New predicate.
32373 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32374 vector conditionals when the true and false values are constant
32375 vectors with all 0 bits or all 1 bits set.
32377 2017-02-06 Martin Sebor <msebor@redhat.com>
32379 PR tree-optimization/79376
32380 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32382 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32384 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32385 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32386 to simplify split condition.
32388 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32390 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32391 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32394 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32396 PR rtl-optimization/68664
32397 * target.def (can_speculate_insn): New hook.
32398 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32399 * doc/tm.texi: Regenerate.
32400 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32401 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32402 (rs6000_sched_can_speculate_insn): New function.
32404 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32406 PR tree-optimization/79284
32407 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32408 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32409 vectorizable_mask_load_store, vectorizable_operation,
32410 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32411 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32412 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32413 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32414 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32415 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32416 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32417 is_gimple_assign (stmt). Replace another such test with
32418 is_gimple_assign (stmt).
32420 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32423 * config/avr/avr.c (rtl-iter.h): Include it.
32424 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32425 (avr_legitimate_combined_insn): ...and implementation.
32427 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32429 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32430 * config/s390/s390.c (s390_const_operand_ok)
32431 (s390_canonicalize_comparison, s390_extract_part)
32432 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32433 (s390_contiguous_bitmask_p, s390_rtx_costs)
32434 (legitimize_pic_address): Likewise.
32435 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32436 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32437 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32438 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32439 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32441 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32443 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32444 REGNO($0) == REGNO($1).
32446 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32448 * config/s390/linux.h(SIZE_TYPE): Add comment.
32450 2017-02-06 Julian Brown <julian@codesourcery.com>
32451 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32452 Virendra Pathak <virendra.pathak@broadcom.com>
32454 * config/aarch64/aarch64-cores.def: Change the scheduler
32456 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32457 * config/aarch64/thunderx2t99.md: New file.
32459 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32461 * doc/standards.texi (Go Language): Update link to language
32464 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32466 * tree-eh.c (lower_resx): Sanitize profile.
32467 (cleanup_empty_eh_move_lp): Likewise.
32469 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32472 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32474 * cfgloopmanip.h (loop_version): Update prototype.
32475 * modulo-sched.c (sms_schedule): Update call of loop_version.
32476 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32477 * tree-parloops.c (gen_parallel_loop): Likewise.
32478 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32479 * tree-ssa-loop-split.c (split_loop): Likewise.
32480 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32481 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32483 2017-02-05 Martin Liska <mliska@suse.cz>
32486 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32488 (print_operand_address): Initialize a struct to zero.
32490 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32492 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32493 garbage collector only in textual form.
32495 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32497 * doc/extend.texi (x86 specific memory model extensions for
32498 transactional memory): Simplify a phrase.
32500 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32503 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32504 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32505 (atomic_storedi_1): Likewise.
32507 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32509 PR tree-optimization/79338
32510 * tree-parloops.c (gather_scalar_reductions): Don't call
32511 vect_analyze_loop_form for loop->inner before destroying loop's
32514 2017-02-03 Martin Sebor <msebor@redhat.com>
32516 PR tree-optimization/79327
32517 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32518 when precision has resulted in leading zeros.
32519 (format_integer): Adjust the likely counter to assume an unknown
32520 argument that may be zero is non-zero.
32522 2017-02-03 Jason Merrill <jason@redhat.com>
32525 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32526 avoid copying non-taken branch.
32528 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32530 PR tree-optimization/79340
32531 * tree-vect-loop.c (vectorizable_reduction): Release
32532 vec_defs elements after safe_splicing them into other vectors.
32535 PR tree-optimization/79327
32536 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32537 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32539 (format_integer): Use wide_int_to_tree instead of build_int_cst
32540 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32541 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32542 of shortest and longest sequence.
32544 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32546 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32547 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32549 2017-02-03 Walter Lee <walt@tilera.com>
32552 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32553 after initial stackframe link reg save.
32554 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32556 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32559 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32560 wu for stxssp alternative.
32562 2017-02-03 Martin Sebor <msebor@redhat.com>
32564 PR tree-optimization/79352
32565 * gimple-fold.c (get_range_strlen): Add argument.
32566 (get_range_strlen): Change return type to bool.
32567 (get_maxval_strlen): Pass in a dummy argument.
32568 * gimple-fold.h (get_range_strlen): Change return type to bool.
32569 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32570 * tree.h (array_at_struct_end_p): Add argument.
32571 * tree.c (array_at_struct_end_p): Handle it.
32573 2017-02-03 Martin Liska <mliska@suse.cz>
32576 * multiple_target.c (create_dispatcher_calls): Redirect edge
32577 from a caller of a dispatcher.
32578 (expand_target_clones): Make the clones local.
32579 (ipa_target_clone): Do both target clones and resolvers.
32580 (ipa_dispatcher_calls): Remove the pass.
32581 (pass_dispatcher_calls::gate): Likewise.
32582 (make_pass_dispatcher_calls): Likewise.
32583 * passes.def (pass_target_clone): Put as very first IPA early
32586 2017-02-03 Martin Liska <mliska@suse.cz>
32588 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32589 in case of a function with ifunc attribute.
32591 2017-02-03 Martin Liska <mliska@suse.cz>
32593 * cgraph.c (cgraph_node::dump): Dump function version info.
32594 * symtab.c (symtab_node::dump_base): Add missing new line.
32596 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32598 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32599 (ifcombine_ifandif): Use it.
32601 2017-02-03 Martin Liska <mliska@suse.cz>
32603 * doc/invoke.texi: Document default value for
32604 use-after-scope-direct-emission-threshold.
32606 2017-02-03 Martin Liska <mliska@suse.cz>
32608 PR tree-optimization/79339
32609 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32610 (format_floating): Likewise.
32612 2017-02-03 Martin Liska <mliska@suse.cz>
32615 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32617 (ipa_node_params_t::remove): Likewise.
32618 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32619 initialization from removed ipa_node_params_t::insert.
32620 (ipa_node_params::~ipa_node_params): Move from removed
32621 ipa_node_params_t::release.
32622 * symbol-summary.h (symbol_summary::m_released): New member.
32623 Do not release a summary twice. Do not allow to call finalizer
32624 for types of a summary that live in GGC memory.
32626 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32628 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32631 2017-02-02 Martin Sebor <msebor@redhat.com>
32633 PR middle-end/79275
32634 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32635 (format_string): Tighten up the range of output for non-constant
32636 strings and correct the expected range for wide non-constant strings.
32638 2017-02-02 Martin Sebor <msebor@redhat.com>
32640 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32642 PR middle-end/32003
32643 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32645 (-fdump-tree-@var): Add to index and document how to come up
32646 with pass-specific option and dump file names.
32647 (-fdump-passes): Clarify where to look for output.
32649 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32651 PR middle-end/77445
32652 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32653 statistics of the analyzed path; allow threading for speed when
32654 any of BBs along the path are optimized for speed.
32656 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32658 PR middle-end/78468
32659 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32660 settings of the virtual registers.
32663 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32665 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32666 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32669 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32671 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32672 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32674 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32676 * config/s390/s390.md: Add missing comments with the expanded
32678 * config/s390/vector.md: Likewise.
32679 * config/s390/vx-builtins.md: Likewise.
32681 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32684 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32685 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32686 conditions on a single line.
32688 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32690 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32691 __S390_VX__ to __VX__.
32693 2017-02-01 Andrew Pinski <apinski@cavium.com>
32695 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32696 stmt_info to record_stmt_cost.
32697 (vect_get_known_peeling_cost): Pass stmt_info if known to
32699 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32700 cpu_vector_cost field into
32701 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32702 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32703 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32704 splitting of scalar_stmt_cost and vec_stmt_cost.
32705 (thunderx_vector_cost): Likewise.
32706 (cortexa57_vector_cost): LIkewise.
32707 (exynosm1_vector_cost): Likewise.
32708 (xgene1_vector_cost): Likewise.
32709 (thunderx2t99_vector_cost): Improve after the splitting of the two
32711 (aarch64_builtin_vectorization_cost): Update for the splitting of
32712 scalar_stmt_cost and vec_stmt_cost.
32714 2017-02-01 Torvald Riegel <triegel@redhat.com>
32715 Richard Henderson <rth@redhat.com>
32717 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32718 conditional on existance of a fast atomic load.
32719 * optabs-query.c (can_atomic_load_p): New function.
32720 * optabs-query.h (can_atomic_load_p): Declare it.
32721 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32722 no fast atomic load is available for the particular size of access.
32723 (expand_atomic_compare_and_swap): Likewise.
32724 (expand_atomic_load): Likewise.
32725 (expand_atomic_store): Likewise.
32726 (expand_atomic_fetch_op): Likewise.
32727 * testsuite/lib/target-supports.exp
32728 (check_effective_target_sync_int_128): Remove x86 because it provides
32729 no fast atomic load.
32730 (check_effective_target_sync_int_128_runtime): Likewise.
32732 2017-02-01 Richard Biener <rguenther@suse.de>
32734 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32735 (graphite_transform_loops): Provide opt-info for optimized nests.
32736 * tree-parloop.c (parallelize_loops): Provide opt-info for
32737 parallelized loops.
32739 2017-02-01 Richard Biener <rguenther@suse.de>
32741 PR middle-end/79315
32742 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32743 was not set before.
32745 2017-02-01 Richard Biener <rguenther@suse.de>
32747 PR tree-optimization/71824
32748 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32749 Verify the loops are valid in the merged SESE region.
32750 (scop_detection::can_represent_loop_1): Check analyzing the
32751 evolution of the number of iterations in the region succeeds.
32753 2017-01-31 Ian Lance Taylor <iant@golang.org>
32755 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32756 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32758 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32760 PR preprocessor/79210
32761 * input.c (get_substring_ranges_for_loc): Replace line_width
32762 assertion with error-handling.
32764 2017-01-31 Richard Biener <rguenther@suse.de>
32766 PR tree-optimization/77318
32767 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32768 (create_pw_aff_from_tree): Take loop parameter.
32769 (add_condition_to_pbb): Pass loop of the condition to
32770 create_pw_aff_from_tree.
32772 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32774 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32775 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32777 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32781 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32783 (convert_int_to_float128): Likewise.
32784 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32785 (convert_int_to_float128): Likewise.
32786 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32787 (UNSPEC_IEEE128_CONVERT): Likewise.
32788 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32789 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32790 Use local variables for IBM extended format.
32791 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32792 (fix_trunc<mode>si2_fprs): Likewise.
32793 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32794 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32795 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32796 to know that we can now have integers of all sizes in vector
32798 (fix<uns>_<mode>di2_hw): Likewise.
32799 (float<uns>_<mode>si2_hw): Likewise.
32800 (fix_<mode>si2_hw): Likewise.
32801 (fixuns_<mode>si2_hw): Likewise.
32802 (float<uns>_<mode>di2_hw): Likewise.
32803 (float_<mode>di2_hw): Likewise.
32804 (float_<mode>si2_hw): Likewise.
32805 (floatuns_<mode>di2_hw): Likewise.
32806 (floatuns_<mode>si2_hw): Likewise.
32807 (xscvqp<su>wz_<mode>): Delete, no longer used.
32808 (xscvqp<su>dz_<mode>): Likewise.
32809 (xscv<su>dqp_<mode>): Likewise.
32810 (ieee128_mfvsrd_64bit): Likewise.
32811 (ieee128_mfvsrd_32bit): Likewise.
32812 (ieee128_mfvsrwz): Likewise.
32813 (ieee128_mtvsrw): Likewise.
32814 (ieee128_mtvsrd_64bit): Likewise.
32815 (ieee128_mtvsrd_32bit): Likewise.
32817 2017-01-31 Martin Liska <mliska@suse.cz>
32820 * ipa-prop.c (ipa_free_all_node_params): Call release method
32821 instead of ~sumbol_summary to not to trigger double times
32824 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32826 PR tree-optimization/71691
32827 * bitmap.h (class auto_bitmap): New.
32828 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32829 is_maybe_undefined instead of ssa_undefined_value_p.
32831 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32833 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32834 __S390_ARCH_LEVEL__ to __ARCH__.
32836 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32838 PR tree-optimization/79267
32839 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32840 if should_remove_lhs_p is true.
32842 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32845 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32846 (add_alignment_attribute): New.
32847 (base_type_die): Add alignment attribute.
32848 (subrange_type_die): Likewise.
32849 (modified_type_die): Likewise.
32850 (gen_array_type_die): Likewise.
32851 (gen_descr_array_type_die: Likewise.
32852 (gen_enumeration_type_die): Likewise.
32853 (gen_subprogram_die): Likewise.
32854 (gen_variable_die): Likewise.
32855 (gen_field_die): Likewise.
32856 (gen_ptr_to_mbr_type_die): Likewise.
32857 (gen_struct_or_union_type_die): Likewise.
32858 (gen_subroutine_type_die): Likewise.
32859 (gen_typedef_die): Likewise.
32860 (base_type_cmp): Compare alignment attribute.
32862 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32865 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32866 (setb_unsigned) New pattern for setb with CCUNS.
32867 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32868 subfc./subfe sequence to avoid overflow problems. Generate a
32869 shorter sequence with cmpld/setb for power9.
32870 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32871 for generating subfc. instruction.
32872 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32873 now uses this instruction.
32875 2017-01-30 Ian Lance Taylor <iant@google.com>
32878 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32879 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32881 2017-01-30 Martin Sebor <msebor@redhat.com>
32883 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32884 Move constant to the right of a relational operator.
32885 (get_mpfr_format_length, format_character, format_string): Ditto.
32886 (should_warn_p, maybe_warn): Same.
32888 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32890 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32893 * asan.c (get_translation_unit_decl): Remove function.
32894 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32896 2017-01-30 Martin Liska <mliska@suse.cz>
32898 PR gcov-profile/79259
32899 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32900 -fprofile-generate.
32902 2017-01-30 Martin Liska <mliska@suse.cz>
32905 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32906 Initialize variables with NULL value.
32908 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32911 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32913 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32915 2017-01-30 Richard Biener <rguenther@suse.de>
32917 PR tree-optimization/79276
32918 * tree-vrp.c (process_assert_insertions): Properly adjust common
32919 when removing a duplicate.
32921 2017-01-30 Richard Biener <rguenther@suse.de>
32923 PR tree-optimization/79256
32924 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32925 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32927 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32929 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32932 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32933 ("*r<noxa>sbg_<mode>_sll_bitmask")
32934 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32935 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32936 Use contiguous_bitmask_nowrap_operand.
32938 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32941 * config/rs6000/altivec.h (vec_xl): Revise #define.
32942 (vec_xst): Likewise.
32944 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32946 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32948 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32950 PR rtl-optimization/79194
32951 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32952 traps before call to bypass_conditional_jumps.
32954 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32956 PR tree-optimization/71374
32957 * lra-constraints.c (check_conflict_input_operands): New.
32958 (match_reload): Use it.
32960 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32963 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32964 account to calculate conflict_set.
32966 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32968 PR rtl-optimization/78559
32969 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32970 other_insn in combine.
32972 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32974 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32975 uint16_type_node for BT_UINT16.
32977 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32979 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32980 "RTL Tests" to menu.
32981 (GIMPLE Tests): New node.
32982 (RTL Tests): New node.
32984 2017-01-27 Richard Biener <rguenther@suse.de>
32986 PR tree-optimization/79245
32987 * tree-loop-distribution.c (distribute_loop): Apply cost
32988 modeling also to detected patterns.
32990 2017-01-27 Richard Biener <rguenther@suse.de>
32992 PR tree-optimization/71433
32993 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32994 (compare_assert_loc): New function.
32995 (process_assert_insertions): Sort and optimize assert locations
32996 to remove duplicates and push down identical assertions on
32997 edges to their destination block.
32999 2017-01-27 Richard Biener <rguenther@suse.de>
33001 PR tree-optimization/79244
33002 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33003 out SSA names even if abnormal.
33005 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33007 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33008 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33009 instead of MPFR_RNDN.
33011 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33014 * arm.c (arm_option_override): Don't call build_target_option_node
33015 until after doing all option overrides.
33016 (arm_valid_target_attribute_tree): Likewise.
33018 2017-01-27 Martin Liska <mliska@suse.cz>
33020 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33021 for {cd}tors and C++ {cd}tors.
33023 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33025 * config/s390/s390.md ("*setmem_long_and")
33026 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33028 2017-01-26 Martin Sebor <msebor@redhat.com>
33030 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33033 2017-01-26 Martin Sebor <msebor@redhat.com>
33035 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33036 HAVE_DFmode before using XFmode or DFmode.
33037 (parse_directive): Avoid using the z length modifier to avoid
33038 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33040 PR middle-end/78703
33041 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33042 to accept adjustment as an array.
33043 (get_int_range): New function.
33044 (struct directive): Make width and prec arrays.
33045 (directive::set_width, directive::set_precision): Call get_int_range.
33046 (format_integer, format_floating): Handle width and precision ranges.
33047 (format_string, parse_directive): Same.
33049 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33052 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33053 comdat_type_p set, just clone them, but keep the children in the
33057 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33058 which have direct callers with -fvar-tracking-assignments enabled
33060 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33061 inside of type units.
33063 2017-01-26 Martin Sebor <msebor@redhat.com>
33065 PR middle-end/78703
33066 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33068 (struct format_result): Replace number_chars, number_chars_min,
33069 and number_chars_max with a single member of struct result_range.
33071 (format_result::operator+=): Adjust.
33072 (struct fmtresult): Remove bounded. Handle likely and unlikely
33074 (fmtresult::adjust_for_width_or_precision): New function.
33075 (fmtresult:type_max_digits): New function.
33076 (bytes_remaining): Handle likely and unlikely counters.
33077 (min_bytes_remaining): Remove.
33078 (format_percent): Simplify.
33079 (format_integer, format_floating): Set likely and unlikely counters.
33080 (get_string_length, format_character, format_string): Same.
33081 (format_plain, should_warn_p): New function.
33082 (maybe_warn): Call should_warn_p. Update diagnostic messages
33083 and handle those for all directives, including plain strings.
33084 (format_directive): Handle likely and unlikely counters.
33085 Remove unnecessary quoting from diagnostics. Add an informational
33087 (add_bytes): Remove.
33088 (pass_sprintf_length::compute_format_length): Simplify.
33089 (try_substitute_return_value): Handle likely and unlikely counters.
33091 2017-01-26 Carl Love <cel@us.ibm.com>
33093 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33094 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33096 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33099 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33100 endianess for subregs into account.
33101 * lra-constraints.c (lra_constraints): Do risky transformations
33102 always on the first iteration.
33103 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33104 last_call_used_reg_set.
33105 (process_bb_lives): Define and use last_call_used_reg_set.
33106 * lra.c (lra): Always continue after lra_constraints on the first
33109 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33111 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33113 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33115 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33117 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33118 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33119 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33120 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33121 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33122 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33123 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33124 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33125 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33127 2017-01-26 Marek Polacek <polacek@redhat.com>
33130 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33131 for the third operand.
33133 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33135 PR middle-end/79236
33136 * omp-low.c (struct omp_context): Add simt_stmt field.
33137 (scan_omp_for): Return omp_context *.
33138 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33139 context to the _simt_ SIMD stmt.
33140 (lower_omp_for): For combined SIMD with sibling _simt_
33141 SIMD, make sure to use the same decls in _looptemp_
33142 clauses as in the sibling.
33144 2017-01-26 David Sherwood <david.sherwood@arm.com>
33146 PR middle-end/79212
33147 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33150 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33153 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33154 emit fld b; fld a; if possible.
33156 * brig-builtins.def: Update copyright years.
33157 * config/arm/arm_acle_builtins.def: Update copyright years.
33159 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33162 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33163 constraint instead of o for the stxsd instruction.
33165 2017-01-25 Carl Love <cel@us.ibm.com>
33167 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33168 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33170 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33172 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33174 2017-01-25 Richard Biener <rguenther@suse.de>
33176 PR tree-optimization/69264
33177 * target.def (vector_alignment_reachable): Improve documentation.
33178 * doc/tm.texi: Regenerate.
33179 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33181 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33182 earlier changes with respect to TYPE_USER_ALIGN.
33183 (vector_alignment_reachable_p): Likewise. Improve dumping.
33185 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33188 * config/arm/arm.md (xordi3): Force constant operand into a register
33191 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33193 * doc/invoke.texi (-fstore-merging): Correct default optimization
33194 levels at which it is enabled.
33195 (-O): Move -fstore-merging from list to...
33198 2017-01-25 Richard Biener <rguenther@suse.de>
33201 * omp-expand.c: Include debug.h.
33202 (expand_omp_taskreg): Make sure to generate early debug before
33203 outlining anything from a function.
33204 (expand_omp_target): Likewise.
33205 (grid_expand_target_grid_body): Likewise.
33207 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33210 * asan.c (get_translation_unit_decl): New function.
33211 (asan_add_global): Extract modules file name from globals
33212 TRANSLATION_UNIT_DECL name.
33214 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33217 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33218 for long calls with APCS frame and VFP.
33220 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33222 * cfg.c (original_copy_tables_initialized_p): New function.
33223 * cfg.h (original_copy_tables_initialized_p): New decl.
33224 * cfgrtl.c (relink_block_chain): Guard the call to
33225 free_original_copy_tables with a call to
33226 original_copy_tables_initialized_p.
33227 * cgraph.h (symtab_node::native_rtl_p): New decl.
33228 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33229 (symtab_node::needed_p): Don't assert for early assembly output
33230 for __RTL functions.
33231 (cgraph_node::finalize_function): Set "force_output" for __RTL
33233 (cgraph_node::analyze): Bail out early for __RTL functions.
33234 (analyze_functions): Update assertion to support __RTL functions.
33235 (cgraph_node::expand): Bail out early for __RTL functions.
33236 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33238 * function.h (struct function): Update comment for field
33240 * gimple-expr.c: Include "tree-pass.h".
33241 (gimple_has_body_p): Return false for __RTL functions.
33242 * Makefile.in (OBJS): Add run-rtl-passes.o.
33243 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33245 (gcc::pass_manager::get_clean_slate): New accessor.
33246 * passes.c: Include "insn-addr.h".
33247 (should_skip_pass_p): Add logging. Update logic for running
33248 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33249 property-provider override so it is only done for gimple passes.
33251 (skip_pass): New function.
33252 (execute_one_pass): Call skip_pass when skipping passes.
33253 * read-md.c (md_reader::read_char): Support filtering
33254 the input to a subset of line numbers.
33255 (md_reader::md_reader): Initialize fields
33256 m_first_line and m_last_line.
33257 (md_reader::read_file_fragment): New function.
33258 * read-md.h (md_reader::read_file_fragment): New decl.
33259 (md_reader::m_first_line): New field.
33260 (md_reader::m_last_line): New field.
33261 * read-rtl-function.c (function_reader::create_function): Only
33262 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33263 curr_properties. Set DECL_INITIAL to a dummy block.
33264 (read_rtl_function_body_from_file_range): New function.
33265 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33267 * run-rtl-passes.c: New file.
33268 * run-rtl-passes.h: New file.
33270 2017-01-24 Jeff Law <law@redhat.com>
33272 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33275 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33277 PR tree-optimization/79159
33278 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33279 (record_nonwrapping_iv): Improve boundary using above function if no
33280 value range information.
33282 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33283 Martin Jambor <mjambor@suse.cz>
33285 * brig-builtins.def: New file.
33286 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33287 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33288 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33289 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33290 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33291 * builtin-types.def (BT_INT8): New.
33292 (BT_INT16): Likewise.
33293 (BT_UINT8): Likewise.
33294 (BT_UINT16): Likewise.
33295 (BT_FN_ULONG): Likewise.
33296 (BT_FN_UINT_INT): Likewise.
33297 (BT_FN_UINT_ULONG): Likewise.
33298 (BT_FN_UINT_LONG): Likewise.
33299 (BT_FN_UINT_PTR): Likewise.
33300 (BT_FN_ULONG_PTR): Likewise.
33301 (BT_FN_INT8_FLOAT): Likewise.
33302 (BT_FN_INT16_FLOAT): Likewise.
33303 (BT_FN_UINT32_FLOAT): Likewise.
33304 (BT_FN_UINT16_FLOAT): Likewise.
33305 (BT_FN_UINT8_FLOAT): Likewise.
33306 (BT_FN_UINT64_FLOAT): Likewise.
33307 (BT_FN_UINT16_UINT32): Likewise.
33308 (BT_FN_UINT32_UINT16): Likewise.
33309 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33310 (BT_FN_INT_PTR_INT): Likewise.
33311 (BT_FN_UINT_PTR_UINT): Likewise.
33312 (BT_FN_LONG_PTR_LONG): Likewise.
33313 (BT_FN_ULONG_PTR_ULONG): Likewise.
33314 (BT_FN_VOID_UINT64_UINT64): Likewise.
33315 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33316 (BT_FN_INT8_INT8_INT8): Likewise.
33317 (BT_FN_INT16_INT16_INT16): Likewise.
33318 (BT_FN_INT_INT_INT): Likewise.
33319 (BT_FN_UINT_FLOAT_UINT): Likewise.
33320 (BT_FN_FLOAT_UINT_UINT): Likewise.
33321 (BT_FN_ULONG_UINT_UINT): Likewise.
33322 (BT_FN_ULONG_UINT_PTR): Likewise.
33323 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33324 (BT_FN_UINT_UINT_UINT): Likewise.
33325 (BT_FN_VOID_UINT_PTR): Likewise.
33326 (BT_FN_UINT_UINT_PTR: Likewise.
33327 (BT_FN_UINT32_UINT64_PTR): Likewise.
33328 (BT_FN_INT_INT_UINT_UINT): Likewise.
33329 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33330 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33331 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33332 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33333 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33334 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33335 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33336 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33337 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33338 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33339 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33340 * doc/frontends.texi: List BRIG FE.
33341 * doc/install.texi (Testing): Add BRIG tesring requirements.
33342 * doc/invoke.texi (Overall Options): Mention BRIG.
33343 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33345 2017-01-24 Richard Biener <rguenther@suse.de>
33347 PR translation/79208
33348 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33350 2017-01-24 Martin Jambor <mjambor@suse.cz>
33353 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33354 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33355 and known_contexts.
33357 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33359 PR middle-end/79123
33360 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33361 casts from signed to unsigned really don't have a range.
33363 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33365 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33366 GMP_RNDx for compatiblity.
33368 2017-01-24 Martin Liska <mliska@suse.cz>
33371 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33372 that would prevent us to call alloca with -1 as argument.
33374 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33376 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33377 Avoid -Wformat-security warning.
33379 2017-01-23 Andrew Pinski <apinski@cavium.com>
33381 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33384 2017-01-23 Martin Sebor <msebor@redhat.com>
33386 PR middle-end/78703
33387 * gimple-ssa-sprintf.c (warn_level): New global.
33388 (format_integer): Use it here and throughout the rest of the file.
33389 Use the same switch to compute sign as base.
33390 (maybe_warn): New function.
33391 (format_directive): Factor out warnings into maybe_warn.
33392 Add debugging output. Use warn_level.
33393 (add_bytes): Use warn_level.
33394 (pass_sprintf_length::compute_format_length): Add debugging output.
33395 (try_substitute_return_value): Same.
33396 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33398 PR middle-end/78703
33399 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33400 (struct fmtresult, format_integer, format_floating): Adjust.
33401 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33402 (get_string_length, format_string,format_directive): Same.
33403 (pass_sprintf_length::compute_format_length): Same.
33404 (try_substitute_return_value): Simplify slightly.
33406 PR middle-end/78703
33407 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33408 (fmtresult::operator+=): Outlined.
33409 (struct fmtresult): Add ctors.
33410 (struct conversion_spec): Rename...
33411 (struct directive): ...to this. Add and remove data members.
33412 (directive::set_width, directive::set_precision): New functions.
33413 (format_percent): Use fmtresult ctor.
33414 (get_width_and_precision): Remove.
33415 (format_integer): Make naming changes. Avoid computing width and
33417 (format_floating): Same. Adjust indentation.
33418 (format_character, format_none): New functions.
33419 (format_string): Moved character handling to format_character.
33420 (format_directive): Remove arguments, change return type.
33421 (parse_directive): New function.
33422 (pass_sprintf_length::compute_format_length): Move directive
33423 parsing to parse_directive.
33425 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33427 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33428 (assign_assembler_name_if_needed): ... this.
33429 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33430 (assign_assembler_name_if_needed): ... this.
33431 (free_lang_data_in_cgraph): Adjust callers.
33432 * cgraphunit.c (cgraph_node::analyze): Likewise.
33433 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33436 2017-01-23 Richard Biener <rguenther@suse.de>
33438 PR tree-optimization/79088
33439 PR tree-optimization/79188
33440 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33441 resetting loop bounds after last path deletion. Reset loop
33442 bounds of the target loop, make code match the comments.
33443 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33444 Make sure loops need no fixups.
33446 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33448 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33449 exponent support with double type for first argument.
33450 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33451 type returned by __builtin_vec_extract_sig,
33452 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33453 functions from "vector int" to "vector unsigned int" or from
33454 "vector long long int" to "vector unsigned long long int".
33455 Changed type returned by __builtin_vec_extract_exp,
33456 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33457 functions from "vector int" to "vector unsigned int" or from
33458 "vector long long int" to "vector unsigned long long int".
33459 Changed return type of __builtin_vec_test_data_class,
33460 __builtin_vec_test_data_class_sp, and
33461 __builtin_vec_test_data_class_dp from "vector int" to
33462 "vector bool int" or from "vector long long int" to "vector bool
33463 long long int" and changed second argument type from "unsigned
33464 int" to "int". Added new overloaded function forms "vector float
33465 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33466 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33467 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33468 double, vector unsigned long long int)" and "vector double
33469 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33470 long int)". Changed return type of
33471 __builtin_scalar_test_data_class and
33472 __builtin_scalar_test_data_class_sp and
33473 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33474 int" and changed second argument from "unsigned int" to "int".
33475 Changed type returned by __builtin_scalar_test_neg,
33476 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33477 from "int" to "bool int". Added new overloaded function form
33478 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33479 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33480 exponent double-precision with floating point first argument.
33481 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33482 documentation of scalar_test_data_class, scalar_test_neg,
33483 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33484 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33485 vec_test_data_class built-in functions to reflect refinements in
33486 their type signatures.
33488 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33490 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33492 (aarch64_elf_asm_destructor): Likewise.
33494 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33496 PR rtl-optimization/78634
33497 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33498 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33499 * ifcvt.c (noce_try_cmove): Add missing cost check.
33501 PR rtl-optimization/71724
33502 * combine.c (if_then_else_cond): Look for situations where it is
33503 beneficial to undo the work of one of the recursive calls.
33505 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33507 PR tree-optimization/70754
33508 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33509 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33510 combined stmt before it if not NULL.
33511 (combine_chains): Process refs reversely and compute dominance point
33514 2017-01-23 Martin Liska <mliska@suse.cz>
33516 PR tree-optimization/79196
33517 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33518 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33520 (strlen_optimize_stmt): Call the renamed function.
33522 2017-01-23 Michael Matz <matz@suse.de>
33524 PR tree-optimization/78384
33525 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33527 2017-01-23 Richard Biener <rguenther@suse.de>
33529 PR tree-optimization/79186
33530 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33531 both incoming edges before moving an assert.
33533 2017-01-23 Martin Jambor <mjambor@suse.cz>
33535 * ipa-prop.c (load_from_param_1): Removed.
33536 (load_from_unmodified_param): Bits from load_from_param_1 put back
33538 (load_from_param): Removed.
33539 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33540 with stmt. Reverted back to use of load_from_unmodified_param.
33542 2017-01-23 Martin Jambor <mjambor@suse.cz>
33545 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33546 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33547 field a pointer to garbage collected vector, mark lattices and
33548 ipcp_orig_node with GTY((skip)).
33549 (ipa_get_param_count): Adjust to descriptors being a pointer.
33550 (ipa_get_param): Likewise.
33551 (ipa_get_type): Likewise.
33552 (ipa_get_param_move_cost): Likewise.
33553 (ipa_set_param_used): Likewise.
33554 (ipa_get_controlled_uses): Likewise.
33555 (ipa_set_controlled_uses): Likewise.
33556 (ipa_is_param_used): Likewise.
33557 (ipa_node_params_t): Move into garbage collector. New methods insert
33559 (ipa_node_params_sum): Annotate wth GTY(()).
33560 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33562 (ipa_load_from_parm_agg): Adjust declaration.
33563 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33564 * ipa-profile.c (ipa_profile): Likewise.
33565 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33566 (ipa_populate_param_decls): Make descriptors parameter garbage
33568 (ipa_dump_param): Adjust to descriptors being a pointer.
33569 (ipa_alloc_node_params): Likewise.
33570 (ipa_initialize_node_params): Likewise.
33571 (load_from_param_1): Make descriptors parameter garbage collected.
33572 (load_from_unmodified_param): Likewise.
33573 (load_from_param): Likewise.
33574 (ipa_load_from_parm_agg): Likewise.
33575 (ipa_node_params::~ipa_node_params): Removed.
33576 (ipa_free_all_node_params): Remove call to delete operator.
33577 (ipa_node_params_t::insert): New.
33578 (ipa_node_params_t::remove): Likewise.
33579 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33580 copy known_csts and known_contexts vectors.
33581 (ipa_read_node_info): Adjust to descriptors being a pointer.
33582 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33584 (ipcp_transform_function): Make descriptors variable garbage
33587 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33589 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33590 * config/i386/avx512dqintrin.h: Ditto.
33591 * config/i386/avx512fintrin.h: Ditto.
33592 * config/i386/i386.c: Handle new builtins.
33593 * config/i386/i386-builtin.def: Add new builtins.
33594 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33595 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33597 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33598 Martin Liska <mliska@suse.cz>
33600 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33601 * asan.c (asan_expand_poison_ifn): Support stores and use
33602 appropriate ASAN report function.
33603 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33604 * internal-fn.def (ASAN_POISON_USE): Declare.
33605 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33606 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33607 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33608 ASAN_POISON calls w/o LHS.
33609 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33610 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33611 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33612 * gimplify.c (asan_poison_variables): Add attribute
33613 use_after_scope_memory to variables that really needs to live
33615 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33616 having the attribute.
33618 2017-01-23 Martin Liska <mliska@suse.cz>
33620 * asan.c (create_asan_shadow_var): New function.
33621 (asan_expand_poison_ifn): Likewise.
33622 * asan.h (asan_expand_poison_ifn): New declaration.
33623 * internal-fn.c (expand_ASAN_POISON): Likewise.
33624 * internal-fn.def (ASAN_POISON): New builtin.
33625 * sanopt.c (pass_sanopt::execute): Expand
33626 asan_expand_poison_ifn.
33627 * tree-inline.c (copy_decl_for_dup_finish): Make function
33629 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33630 * tree-ssa.c (is_asan_mark_p): New function.
33631 (execute_update_addresses_taken): Rewrite local variables
33632 (identified just by use-after-scope as addressable) into SSA.
33634 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33636 * doc/install.texi (Specific): opensource.apple.com uses https
33637 now. Remove trailing slash.
33639 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33641 * README.Portability: Remove note on an Irix compatibility issue.
33643 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33645 * gcov.c (INCLUDE_ALGORITHM): Define.
33646 (INCLUDE_VECTOR): Define.
33647 No longer include <vector> and <algorithm> directly.
33649 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33651 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33653 * doc/invoke.texi (Code Gen Options): Ditto.
33655 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33658 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33660 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33662 rtl-optimization/79125
33663 * cprop.c (local_cprop_pass): Handle cases where we make an
33664 unconditional trap.
33666 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33670 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33671 read from, for big endian.
33673 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33675 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33676 register pauth builtins for LP64 only.
33678 2017-01-20 Marek Polacek <polacek@redhat.com>
33681 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33684 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33686 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33688 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33689 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33690 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33692 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33695 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33696 in big-endian mode when they are not a single duplicated value.
33698 2017-01-20 Richard Biener <rguenther@suse.de>
33700 * BASE-VER: Bump to 7.0.1.
33702 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33704 * omp-low.c (omplow_simd_context): New struct. Use it...
33705 (lower_rec_simd_input_clauses): ...here and...
33706 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33707 references to idx, lane, max_vf, is_simt.
33709 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33711 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33714 2017-01-20 Martin Jambor <mjambor@suse.cz>
33716 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33717 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33719 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33720 (GTFILES): Rename hsa.c to hsa-common.c.
33721 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33722 * hsa-dump.c: Likewise.
33723 * hsa-gen.c: Likewise.
33724 * hsa-regalloc.c: Likewise.
33725 * ipa-hsa.c: Likewise.
33726 * omp-expand.c: Likewise.
33727 * omp-low.c: Likewise.
33728 * toplev.c: Likewise.
33730 2017-01-20 Marek Polacek <polacek@redhat.com>
33733 * doc/invoke.texi: Document -Wduplicated-branches.
33734 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33735 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33736 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33737 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33738 return 0 only when not OEP_LEXICOGRAPHIC.
33739 (fold_build_cleanup_point_expr): Use the expression
33740 location when building CLEANUP_POINT_EXPR.
33741 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33742 * tree.c (add_expr): Handle error_mark_node.
33744 2017-01-20 Martin Liska <mliska@suse.cz>
33747 * tree-profile.c (init_ic_make_global_vars): Do not call
33749 (gimple_init_gcov_profiler): Likewise.
33751 2017-01-20 Martin Liska <mliska@suse.cz>
33754 * cgraph.h (maybe_create_reference): Remove argument and
33756 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33758 * ipa-cp.c (create_specialized_node): Likewise.
33759 * symtab.c (symtab_node::maybe_create_reference): Handle
33760 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33762 2017-01-20 Martin Liska <mliska@suse.cz>
33764 * read-rtl-function.c (function_reader::create_function): Use
33765 build_decl instread of build_decl_stat.
33767 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33769 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33770 * config/i386/avx512dqintrin.h: Ditto.
33771 * config/i386/avx512fintrin.h: Ditto.
33772 * config/i386/i386-builtin-types.def: Add new types.
33773 * config/i386/i386.c: Handle new types.
33774 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33775 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33776 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33777 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33778 (__builtin_ia32_kshiftridi): New.
33779 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33781 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33785 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33786 define to rs6000_init_stack_protect_guard.
33787 (rs6000_init_stack_protect_guard): New function.
33789 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33790 Yunqiang Su <yunqiang.su@imgtec.com>
33792 * config.gcc (supported_defaults): Add madd4.
33793 (with_madd4): Add validation.
33794 (all_defaults): Add madd4.
33795 * config/mips/mips.opt (mmadd4): New option.
33796 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33798 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33800 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33801 (ISA_HAS_FUSED_MADD4): Likewise.
33802 * doc/invoke.texi (-mmadd4): Document the new option.
33803 * doc/install.texi (--with-madd4): Document the new option.
33805 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33807 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33808 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33809 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33810 (aarch64_init_pauth_hint_builtins): New.
33811 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33812 (aarch64_expand_builtin): Expand new builtins.
33814 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33816 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33817 * combine-stack-adj.c (no_unhandled_cfa): Handle
33818 REG_CFA_TOGGLE_RA_MANGLE.
33819 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33820 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33821 info for return address signing.
33822 (aarch64_expand_epilogue): Likewise.
33824 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33826 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33827 * config/aarch64/aarch64-protos.h
33828 (aarch64_return_address_signing_enabled): New declaration.
33829 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33831 (aarch64_expand_prologue): Sign return address before it's pushed onto
33833 (aarch64_expand_epilogue): Authenticate return address fetched from
33835 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33836 (aarch64_attributes): New function attributes for "sign-return-address".
33837 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33838 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33839 ("*do_return"): Generate combined instructions according to key index.
33840 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33841 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33843 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33844 * config/aarch64/aarch64.opt (msign-return-address=): New.
33845 * doc/extend.texi (AArch64 Function Attributes): Documents
33846 "sign-return-address=".
33847 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33849 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33851 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33852 overall option summary.
33854 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33856 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33857 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33858 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33859 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33861 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33863 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33864 -mpower9-minmax by default for -mcpu=power9.
33865 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33866 128-bit floating point.
33868 2017-01-20 Alan Modra <amodra@gmail.com>
33870 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33871 optimizing for size.
33873 2017-01-20 Alan Modra <amodra@gmail.com>
33876 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33877 for strcmp and strncmp from corresponding builtin decl.
33879 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33881 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33882 instead of i386/rtems-64.h.
33883 * config/i386/rtems-64.h: Remove.
33885 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33889 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33891 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33893 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33895 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33896 Change int to HOST_WIDE_INT.
33897 * config/aarch64/aarch64-protos.h
33898 (aarch64_simd_gen_const_vector_dup): Likewise.
33899 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33901 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33903 * langhooks-def.h (lhd_type_for_size): New decl.
33904 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33905 * langhooks.c (lhd_type_for_size): New function, taken from
33908 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33910 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33911 define_bypass for CR latency.
33912 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33913 (power9-alu2): Add define_bypass for CR latency.
33915 (power9-mul): Update insn latency.
33916 (power9-mul-compare): Update insn latency, bypass latency and remove
33919 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33921 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33923 * config/aarch64/aarch64.md
33924 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33925 aarch64_nopcrelative_literal_loads.
33926 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33928 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33930 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33931 TARGET_LOONGSON_3A.
33932 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33934 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33937 * config.gcc (supported_defaults): Add lxc1-sxc1.
33938 (with_lxc1_sxc1): Add validation.
33939 (all_defaults): Add lxc1-sxc1.
33940 * config/mips/mips.opt (mlxc1-sxc1): New option.
33941 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33943 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33944 __mips_no_lxc1_sxc1.
33945 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33946 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33947 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33949 2017-01-19 Richard Biener <rguenther@suse.de>
33951 PR tree-optimization/72488
33952 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33953 sure to restore SSA info.
33954 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33956 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33958 PR rtl-optimization/79121
33959 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33960 of the inner type when shifting an extended value.
33962 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33965 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33966 interposable aliases.
33968 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33971 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33972 Use the evmergelohi instruction.
33973 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33974 (mov_sitf_e500_subreg8_2_be): Likewise.
33975 (mov_sitf_e500_subreg12_2_le): Likewise.
33976 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33977 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33978 (mov_sitf_e500_subreg8_2_le): Likewise.
33979 (mov_sitf_e500_subreg12_2_be): Likewise.
33981 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33983 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33984 attribute from vecsimple to vecperm.
33985 (altivec_vbpermq2): Likewise.
33987 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33990 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33992 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33993 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33994 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33995 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33996 case where N arg is SIZE_MAX.
33997 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33998 (cmpstrsi): Add pattern.
34000 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34002 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34003 __builtin_vec_revb builtins.
34004 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34005 built-in functions to support generation of the ISA 3.0 XXBR<x>
34006 vector byte reverse instructions.
34007 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34008 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34009 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34010 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34011 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34012 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34013 (P9V_BUILTIN_VEC_REVB): Likewise.
34014 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34015 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34016 (p9_xxbrq_v16qi): Likewise.
34017 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34018 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34019 (p9_xxbrh_v8hi): Likewise.
34020 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34021 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34022 vec_revb built-in functions.
34024 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34026 PR rtl-optimization/78952
34027 * config/i386/i386.md (any_extract): New code iterator.
34028 (*insvqi_2): Use any_extract for source operand.
34029 (*insvqi_3): Use any_shiftrt for source operand.
34031 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34033 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34035 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34037 2017-01-18 Matthias Klose <doko@ubuntu.com>
34039 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34041 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34043 * config/rs6000/altivec.h (vec_bperm): Change #define.
34044 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34045 (altivec_vbpermq2): New define_insn.
34046 (altivec_vbpermd): Likewise.
34047 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34048 function interface.
34049 (VBPERMD): Likewise.
34050 (VBPERM): New polymorphic function interface.
34051 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34052 Add entries for P9V_BUILTIN_VEC_VBPERM.
34053 * doc/extend.texi: Add interfaces for vec_bperm.
34055 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34057 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34058 first letter of error messages.
34059 (s390_resolve_overloaded_builtin): Likewise.
34060 * config/s390/s390.c (s390_expand_builtin): Likewise.
34061 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34062 (s390_valid_target_attribute_inner_p): Likewise.
34063 * config/s390/s390.md ("tabort"): Likewise.
34065 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34067 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34068 (ISA_AVOID_DIV_HILO): New macro.
34069 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34070 (ISA_HAS_DDIV): Likewise.
34072 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34074 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34076 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34078 * doc/invoke.texi (fabi-version): Spelling fix.
34080 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34083 * doc/invoke.texi (fabi-version): Mention mangling fix for
34086 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34089 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34091 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34094 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34095 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34097 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34099 (stack_protect_test): Ditto.
34100 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34101 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34103 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34104 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34105 -mstack-protector-guard-offset=.
34106 (RS/6000 and PowerPC Options): Ditto.
34108 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34110 * config/i386/i386.h (MASK_CLASS_P): New define.
34111 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34112 there are no registers from different register sets also when
34113 mask registers are used. Update function comment.
34114 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34115 to (*k/*r) and (*k/*km) alternatives.
34117 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34119 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34120 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34121 (EH_RETURN_HANDLER_RTX): New define.
34122 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34123 Force frame pointer in EH return functions.
34124 (aarch64_expand_epilogue): Add barrier for eh_return.
34125 (aarch64_final_eh_return_addr): Remove.
34126 (aarch64_eh_return_handler_rtx): New function.
34127 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34129 (aarch64_eh_return_handler_rtx): New prototype.
34131 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34133 * config/rs6000/altivec.h (vec_rlmi): New #define.
34134 (vec_vrlnm): Likewise.
34135 (vec_rlnm): Likewise.
34136 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34137 (UNSPEC_VRLNM): Likewise.
34138 (VIlong): New mode iterator.
34139 (altivec_vrl<VI_char>mi): New define_insn.
34140 (altivec_vrl<VI_char>nm): Likewise.
34141 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34143 (VRLDNM): Likewise.
34144 (RLNM): New polymorphic function entry.
34145 (VRLWMI): New monomorphic function entry.
34146 (VRLDMI): Likewise.
34147 (RLMI): New polymorphic function entry.
34148 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34149 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34150 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34153 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34156 * dwarf2out.c (field_byte_offset): Restore the
34157 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34158 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34159 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34162 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34165 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34167 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34170 * dwarf2out.c (add_data_member_location_attribute): For constant
34171 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34172 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34173 DW_AT_byte_size attributes.
34175 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34177 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34178 after forcing to constant memory when the code model is medium.
34180 2017-01-17 Julia Koval <julia.koval@intel.com>
34183 * config/i386/avx512fintrin.h
34184 (_mm512_i32gather_ps): Change __addr type to void const*.
34185 (_mm512_mask_i32gather_ps): Ditto.
34186 (_mm512_i32gather_pd): Ditto.
34187 (_mm512_mask_i32gather_pd): Ditto.
34188 (_mm512_i64gather_ps): Ditto.
34189 (_mm512_mask_i64gather_ps): Ditto.
34190 (_mm512_i64gather_pd): Ditto.
34191 (_mm512_mask_i64gather_pd): Ditto.
34192 (_mm512_i32gather_epi32): Ditto.
34193 (_mm512_mask_i32gather_epi32): Ditto.
34194 (_mm512_i32gather_epi64): Ditto.
34195 (_mm512_mask_i32gather_epi64): Ditto.
34196 (_mm512_i64gather_epi32): Ditto.
34197 (_mm512_mask_i64gather_epi32): Ditto.
34198 (_mm512_i64gather_epi64): Ditto.
34199 (_mm512_mask_i64gather_epi64): Ditto.
34200 (_mm512_i32scatter_ps): Change __addr type to void*.
34201 (_mm512_mask_i32scatter_ps): Ditto.
34202 (_mm512_i32scatter_pd): Ditto.
34203 (_mm512_mask_i32scatter_pd): Ditto.
34204 (_mm512_i64scatter_ps): Ditto.
34205 (_mm512_mask_i64scatter_ps): Ditto.
34206 (_mm512_i64scatter_pd): Ditto.
34207 (_mm512_mask_i64scatter_pd): Ditto.
34208 (_mm512_i32scatter_epi32): Ditto.
34209 (_mm512_mask_i32scatter_epi32): Ditto.
34210 (_mm512_i32scatter_epi64): Ditto.
34211 (_mm512_mask_i32scatter_epi64): Ditto.
34212 (_mm512_i64scatter_epi32): Ditto.
34213 (_mm512_mask_i64scatter_epi32): Ditto.
34214 (_mm512_i64scatter_epi64): Ditto.
34215 (_mm512_mask_i64scatter_epi64): Ditto.
34216 * config/i386/avx512pfintrin.h
34217 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34218 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34219 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34220 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34221 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34222 (_mm512_prefetch_i32scatter_ps): Ditto.
34223 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34224 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34225 (_mm512_prefetch_i64scatter_pd): Ditto.
34226 (_mm512_prefetch_i64scatter_ps): Ditto.
34227 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34228 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34229 * config/i386/avx512vlintrin.h
34230 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34231 (_mm_mmask_i32gather_ps): Ditto.
34232 (_mm256_mmask_i32gather_pd): Ditto.
34233 (_mm_mmask_i32gather_pd): Ditto.
34234 (_mm256_mmask_i64gather_ps): Ditto.
34235 (_mm_mmask_i64gather_ps): Ditto.
34236 (_mm256_mmask_i64gather_pd): Ditto.
34237 (_mm_mmask_i64gather_pd): Ditto.
34238 (_mm256_mmask_i32gather_epi32): Ditto.
34239 (_mm_mmask_i32gather_epi32): Ditto.
34240 (_mm256_mmask_i32gather_epi64): Ditto.
34241 (_mm_mmask_i32gather_epi64): Ditto.
34242 (_mm256_mmask_i64gather_epi32): Ditto.
34243 (_mm_mmask_i64gather_epi32): Ditto.
34244 (_mm256_mmask_i64gather_epi64): Ditto.
34245 (_mm_mmask_i64gather_epi64): Ditto.
34246 (_mm256_i32scatter_ps): Change __addr type to void*.
34247 (_mm256_mask_i32scatter_ps): Ditto.
34248 (_mm_i32scatter_ps): Ditto.
34249 (_mm_mask_i32scatter_ps): Ditto.
34250 (_mm256_i32scatter_pd): Ditto.
34251 (_mm256_mask_i32scatter_pd): Ditto.
34252 (_mm_i32scatter_pd): Ditto.
34253 (_mm_mask_i32scatter_pd): Ditto.
34254 (_mm256_i64scatter_ps): Ditto.
34255 (_mm256_mask_i64scatter_ps): Ditto.
34256 (_mm_i64scatter_ps): Ditto.
34257 (_mm_mask_i64scatter_ps): Ditto.
34258 (_mm256_i64scatter_pd): Ditto.
34259 (_mm256_mask_i64scatter_pd): Ditto.
34260 (_mm_i64scatter_pd): Ditto.
34261 (_mm_mask_i64scatter_pd): Ditto.
34262 (_mm256_i32scatter_epi32): Ditto.
34263 (_mm256_mask_i32scatter_epi32): Ditto.
34264 (_mm_i32scatter_epi32): Ditto.
34265 (_mm_mask_i32scatter_epi32): Ditto.
34266 (_mm256_i32scatter_epi64): Ditto.
34267 (_mm256_mask_i32scatter_epi64): Ditto.
34268 (_mm_i32scatter_epi64): Ditto.
34269 (_mm_mask_i32scatter_epi64): Ditto.
34270 (_mm256_i64scatter_epi32): Ditto.
34271 (_mm256_mask_i64scatter_epi32): Ditto.
34272 (_mm_i64scatter_epi32): Ditto.
34273 (_mm_mask_i64scatter_epi32): Ditto.
34274 (_mm256_i64scatter_epi64): Ditto.
34275 (_mm256_mask_i64scatter_epi64): Ditto.
34276 (_mm_i64scatter_epi64): Ditto.
34277 (_mm_mask_i64scatter_epi64): Ditto.
34278 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34279 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34280 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34281 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34282 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34283 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34284 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34285 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34286 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34287 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34288 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34289 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34290 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34291 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34292 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34293 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34294 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34295 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34296 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34297 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34298 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34299 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34300 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34301 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34302 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34303 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34304 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34305 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34306 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34307 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34308 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34309 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34310 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34311 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34312 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34313 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34314 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34315 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34316 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34317 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34318 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34319 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34320 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34321 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34322 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34323 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34324 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34325 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34326 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34327 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34328 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34329 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34330 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34331 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34332 definitions accordingly.
34334 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34335 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34338 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34341 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34344 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34345 hard regs for inner regno.
34347 2017-01-17 Martin Liska <mliska@suse.cz>
34350 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34351 assumption and add comment.
34353 2017-01-17 Nathan Sidwell <nathan@acm.org>
34355 * ipa-visibility.c (localize_node): New function, broken out of ...
34356 (function_and_variable_visibility): ... here. Call it.
34358 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34360 PR middle-end/77445
34361 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34362 correctly set frequency of oudgoing edge.
34363 (duplicate_thread_path): Fix profile updating.
34365 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34368 * configure.ac: Add GCC_BASE_VER.
34369 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34370 version from BASE-VER file.
34371 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34372 (gcc.o): Depend on $(BASEVER).
34373 * common.opt (dumpfullversion): New option.
34374 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34375 * doc/invoke.texi: Document -dumpfullversion.
34376 * doc/install.texi: Document --with-gcc-major-version-only.
34377 * configure: Regenerated.
34379 2017-01-17 Richard Biener <rguenther@suse.de>
34381 PR tree-optimization/71433
34382 * tree-vrp.c (register_new_assert_for): Merge same asserts
34383 on all incoming edges.
34384 (process_assert_insertions_for): Handle insertions at the
34387 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34389 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34390 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34392 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34395 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34398 2017-01-17 Alan Modra <amodra@gmail.com>
34401 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34402 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34403 symbolic stack limit when pic.
34405 2017-01-16 Martin Sebor <msebor@redhat.com>
34407 PR tree-optimization/78608
34408 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34410 2017-01-16 Jeff Law <law@redhat.com>
34413 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34414 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34415 for several include directories that may be relative to sysroot.
34416 * config/i386/x-mingw32 (gplus_includedir): Define.
34417 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34418 (native_system_includedir): Likewise.
34419 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34420 override if TARGET_SYSTEM_ROOT is defined.
34421 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34423 PR tree-optimization/79090
34424 PR tree-optimization/33562
34425 PR tree-optimization/61912
34426 PR tree-optimization/77485
34427 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34428 and computed trims into the dump file.
34430 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34432 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34434 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34437 * gimplify.c (gimplify_init_constructor): If want_value and
34438 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34442 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34443 sequence. Formatting fixes.
34444 (doloop_optimize): Formatting fixes.
34447 * gcc.c (debug_level_greater_than_spec_func): New function.
34448 (static_spec_functions): Add debug-level-gt spec function.
34449 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34451 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34452 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34453 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34454 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34455 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34456 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34458 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34460 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34461 QImode fixups to general and mask registers only.
34463 2017-01-16 Carl Love <cel@us.ibm.com>
34465 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34466 for built-in functions
34467 vector signed char vec_nabs (vector signed char)
34468 vector signed short vec_nabs (vector signed short)
34469 vector signed int vec_nabs (vector signed int)
34470 vector signed long long vec_nabs (vector signed long long)
34471 vector float vec_nabs (vector float)
34472 vector double vec_nabs (vector double)
34473 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34475 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34476 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34477 * doc/extend.texi: Update the documentation file for the new built-in
34480 2017-01-16 Martin Sebor <msebor@redhat.com>
34482 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34485 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34487 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34488 UNSPEC_VSX__XXSPLTD to require special splat handling.
34490 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34493 * system.h: Poison strndup.
34495 2017-01-16 Alan Modra <amodra@gmail.com>
34498 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34501 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34503 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34505 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34507 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34508 call recog here. Assert that INSN_CODE (insn) is non-negative.
34510 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34513 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34515 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34516 in the currently scheduled RTL fragment.
34518 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34520 PR rtl-optimization/78751
34521 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34524 2017-01-14 Jeff Law <law@redhat.com>
34526 PR tree-optimization/79090
34527 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34528 variable length stores.
34529 (compute_trims): Delete dead assignment to *trim_tail.
34530 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34533 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34535 PR rtl-optimization/78626
34536 PR rtl-optimization/78727
34537 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34538 of a block, and split such blocks after everything else is finished.
34540 2017-01-14 Alan Modra <amodra@gmail.com>
34543 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34544 target legitimate_combined_insn.
34545 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34546 (rs6000_legitimate_combined_insn): New function.
34547 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34549 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34550 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34551 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34553 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34555 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34557 2017-01-13 Jeff Law <law@redhat.com>
34559 PR tree-optimization/33562
34560 PR tree-optimization/61912
34561 PR tree-optimization/77485
34562 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34564 (delete_dead_assignment): Likewise.
34565 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34566 statement to delete_dead_call and delete_dead_assignment.
34568 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34571 * substring-locations.c (format_warning_va): Strengthen case 1 so
34572 that both endpoints of the substring must be within the format
34573 range for just the substring to be printed.
34575 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34577 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34578 * config/i386/i386.c (ix86_target_string): Add missing options
34579 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34580 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34581 flags_other and ix86_target_other to flags2_other. Display unknown
34583 (ix86_valid_target_attribute_inner_p): Add missing options and
34584 reorder options by implied ISAs, as in ix86_target_string.
34586 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34588 * hash-table.h (hash_table::too_empty_p): New function.
34589 (hash_table::expand): Use it.
34590 (hash_table::traverse): Likewise.
34591 (hash_table::empty_slot): Use sizeof (value_type) instead of
34592 sizeof (PTR) to convert bytes to elements. Shrink the table
34593 if the current size is excessive for the current number of
34596 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34598 * ira-costs.c (record_reg_classes): Break from the inner loop
34599 early once alt_fail is known to be true. Update outer loop
34600 handling accordingly.
34602 2017-01-13 Jeff Law <law@redhat.com>
34604 * tree-ssa-dse.c (decrement_count): New function.
34605 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34606 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34607 when we know the partially dead statement is a mem* function.
34609 PR tree-optimization/61912
34610 PR tree-optimization/77485
34611 * tree-ssa-dse.c: Include expr.h.
34612 (maybe_trim_constructor_store): New function.
34613 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34615 PR tree-optimization/33562
34616 PR tree-optimization/61912
34617 PR tree-optimization/77485
34618 * doc/invoke.texi: Document new dse-max-object-size param.
34619 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34620 * tree-ssa-dse.c: Include params.h.
34621 (dse_store_status): New enum.
34622 (initialize_ao_ref_for_dse): New, partially extracted from
34624 (valid_ao_ref_for_dse, normalize_ref): New.
34625 (setup_live_bytes_from_ref, compute_trims): Likewise.
34626 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34627 (maybe_trim_partially_dead_store): Likewise.
34628 (maybe_trim_complex_store): Likewise.
34629 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34630 Track what bytes live from the original store. Return tri-state
34631 for dead, partially dead or live.
34632 (dse_dom_walker): Add constructor, destructor and new private members.
34633 (delete_dead_call, delete_dead_assignment): New extracted from
34635 (dse_optimize_stmt): Make a member of dse_dom_walker.
34636 Use initialize_ao_ref_for_dse.
34638 PR tree-optimization/33562
34639 PR tree-optimization/61912
34640 PR tree-optimization/77485
34641 * sbitmap.h (bitmap_count_bits): Prototype.
34642 (bitmap_clear_range, bitmap_set_range): Likewise.
34643 * sbitmap.c (bitmap_clear_range): New function.
34644 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34646 2017-01-13 Martin Liska <mliska@suse.cz>
34649 * function.c (set_cfun): Add new argument force.
34650 * function.h (set_cfun): Likewise.
34651 * ipa-inline-transform.c (inline_call): Use the function when
34652 strict alising from is dropped for function we inline to.
34654 2017-01-13 Richard Biener <rguenther@suse.de>
34656 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34657 for dumping GIMPLE INTEGER_CSTs.
34659 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34661 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34662 to 201112L since C++17.
34664 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34667 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34668 if -fsanitize=kernel-address is present.
34670 2017-01-13 Richard Biener <rguenther@suse.de>
34672 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34673 as _Literal ( type ) number in case usual suffixes do not
34674 preserve all information.
34676 2017-01-13 Richard Biener <rguenther@suse.de>
34678 PR tree-optimization/77283
34679 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34680 and ssa-iterators.h.
34681 (is_feasible_trace): Implement a cost model based on joiner
34684 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34687 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34688 char or short to __float128/_Float128 directly.
34690 2017-01-12 Martin Sebor <msebor@redhat.com>
34692 to -Wformat-overflow.
34693 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34694 (min_bytes_remaining): Same.
34695 (get_string_length): Same.
34696 (format_string): Same.
34697 (format_directive): Same.
34699 (pass_sprintf_length::handle_gimple_call): Same.
34701 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34703 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34704 info.nowrite calls with no lhs that can't throw. Return bool
34705 whether gsi_remove has been called or not.
34706 (pass_sprintf_length::handle_gimple_call): Return bool whether
34707 try_substitute_return_value called gsi_remove. Formatting fix.
34708 (pass_sprintf_length::execute): Don't use gsi_remove if
34709 handle_gimple_call returned true.
34712 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34713 be removed due to side-effects, don't remove following barrier nor
34714 turn the successor edge into fallthru edge.
34716 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34719 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34720 element-reversing loads and stores as not swappable.
34722 2017-01-12 Nathan Sidwell <nathan@acm.org>
34723 Nicolai Stange <nicstange@gmail.com>
34725 * combine.c (try_combine): Don't ignore result of overlap checking
34726 loop. Combine overlap & asm check into single loop.
34728 2017-01-12 Richard Biener <rguenther@suse.de>
34730 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34731 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34733 2017-01-12 Richard Biener <rguenther@suse.de>
34735 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34736 and TS_TARGET_OPTION directly derive from TS_BASE.
34737 * tree-core.h (tree_optimization_option): Derive from tree_base.
34738 (tree_target_option): Likewise.
34740 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34742 * config/i386/i386.c (memory_address_length): Increase len
34743 only when rip_relative_addr_p returns false.
34745 2017-01-11 Julia Koval <julia.koval@intel.com>
34747 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34748 (OPTION_MASK_ISA_SGX_SET): New.
34749 (ix86_handle_option): Handle OPT_msgx.
34750 * config.gcc: Added sgxintrin.h.
34751 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34752 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34753 * config/i386/i386.c (ix86_target_string): Add -msgx.
34755 (ix86_option_override_internal): Handle new options.
34756 (ix86_valid_target_attribute_inner_p): Add sgx.
34757 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34758 * config/i386/i386.opt: Add msgx.
34759 * config/i386/sgxintrin.h: New file.
34760 * config/i386/x86intrin.h: Add sgxintrin.h.
34762 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34765 * fold-const.c (maybe_nonzero_address): Return 1 for function
34767 (tree_single_nonzero_warnv_p): Don't handle function local objects
34771 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34774 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34777 * opts.c: Include "spellcheck.h"
34778 (struct string_fragment): New struct.
34779 (struct edit_distance_traits<const string_fragment &>): New
34781 (get_closest_sanitizer_option): New function.
34782 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34784 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34786 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34788 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34789 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34790 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34791 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34792 for initial die_offset if dwarf_split_debug_info.
34793 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34794 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34796 (output_skeleton_debug_sections): Formatting fix. Use
34797 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34798 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34800 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34802 * config/arm/cortex-a53.md: Add bypasses for
34803 cortex_a53_r2f_cvt.
34804 (cortex_a53_r2f): Only use for transfers.
34805 (cortex_a53_f2r): Likewise.
34806 (cortex_a53_r2f_cvt): Add reservation for conversions.
34807 (cortex_a53_f2r_cvt): Likewise.
34809 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34811 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34812 to all inlined functions, change static to extern.
34814 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34817 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34819 (arm_assemble_integer): Likewise.
34821 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34823 * config.gcc: Use new awk script to check CPU, FPU and architecture
34824 parameters for --with-... options.
34825 * config/arm/parsecpu.awk: New file
34826 * config/arm/arm-cpus.in: New file.
34827 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34829 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34831 * config/arm/t-arm: Update dependency rules.
34832 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34833 of processing .def files.
34834 * config/arm/genopt.sh: Deleted.
34835 * config/arm/gentune.sh: Deleted.
34836 * config/arm/arm-cores.def: Deleted.
34837 * config/arm/arm-arches.def: Deleted.
34838 * config/arm/arm-fpus.def: Deleted.
34839 * config/arm/arm-tune.md: Regenerated.
34840 * config/arm/arm-tables.opt: Regenerated.
34841 * config/arm/arm-cpu.h: New generated file.
34842 * config/arm/arm-cpu-data.h: New generated file.
34843 * config/arm/arm-cpu-cdata.h: New generated file.
34845 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34848 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34850 (input_varpool_node): Unpack dynamically_initialized bit.
34852 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34854 PR rtl-optimization/79032
34855 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34856 the alignment of the adjusted memory reference against that of MODE,
34857 instead of the alignment of the original memory reference.
34859 2017-01-11 Martin Jambor <mjambor@suse.cz>
34861 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34863 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34864 decorated functions.
34866 2017-01-11 Richard Biener <rguenther@suse.de>
34868 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34869 set range/nonnull info for PHI results. Do not set it on
34870 stmts marked for removal.
34872 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34874 * expr.c (store_field): In the bitfield case, fetch the return value
34875 from the registers before applying a single big-endian adjustment.
34876 Always do a final load for a BLKmode value not larger than a word.
34878 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34881 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34882 that we correctly handle column numbers greater than
34883 LINE_MAP_MAX_COLUMN_NUMBER.
34885 2017-01-10 Martin Sebor <msebor@redhat.com>
34887 PR middle-end/78245
34888 * gimple-ssa-sprintf.c (get_destination_size): Call
34889 {init,fini}object_sizes.
34890 * tree-object-size.c (addr_object_size): Adjust.
34891 (pass_through_call): Adjust.
34892 (pass_object_sizes::execute): Adjust.
34893 * tree-object-size.h (fini_object_sizes): Declare.
34895 2017-01-10 Martin Sebor <msebor@redhat.com>
34897 PR tree-optimization/78775
34898 * builtins.c (get_size_range): Move...
34899 * calls.c: ...to here.
34900 (alloc_max_size): Accept zero argument.
34901 (operand_signed_p): Remove.
34902 (maybe_warn_alloc_args_overflow): Call get_size_range.
34903 * calls.h (get_size_range): Declare.
34905 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34907 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34908 from TI's devices.csv file as of September 2016.
34909 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34911 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34913 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34914 * doc/invoke.texi: Likewise.
34915 * doc/md.texi: Likewise.
34916 * doc/objc.texi: Likewise.
34918 2017-01-10 Joshua Conner <joshconner@google.com>
34920 * config/arm/fuchsia-elf.h: New file.
34921 * config/fuchsia.h: New file.
34922 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34923 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34925 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34927 2016-01-10 Richard Biener <rguenther@suse.de>
34929 PR tree-optimization/79034
34930 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34931 Propagate out degenerate PHIs in the joiner.
34933 2017-01-10 Martin Liska <mliska@suse.cz>
34935 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34936 (sort_congruence_classes_by_decl_uid): Likewise.
34937 (sort_congruence_class_groups_by_decl_uid): Likewise.
34938 (sem_item_optimizer::merge_classes): Sort class, groups in these
34939 classes and members in the groups by DECL_UID of declarations.
34940 This would make merge operations stable.
34942 2017-01-10 Martin Liska <mliska@suse.cz>
34944 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34945 usage of m_classes_vec.
34946 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34947 (sem_item_optimizer::get_group_by_hash): Likewise.
34948 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34949 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34950 (sem_item_optimizer::verify_classes): Likewise.
34951 (sem_item_optimizer::process_cong_reduction): Likewise.
34952 (sem_item_optimizer::dump_cong_classes): Likewise.
34953 (sem_item_optimizer::merge_classes): Likewise.
34954 * ipa-icf.h (congruence_class_hash): Rename from
34955 congruence_class_group_hash. Remove declaration of m_classes_vec.
34957 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34959 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34960 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34961 * config.gcc: Add avx512vpopcntdqintrin.h.
34962 * config/i386/avx512vpopcntdqintrin.h: New.
34963 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34964 * config/i386/i386-builtin-types.def: Add new types.
34965 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34966 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34967 __builtin_ia32_vpopcountq_v8di_mask): New.
34968 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34969 __AVX512VPOPCNTDQ__.
34970 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34971 (PTA_AVX512VPOPCNTDQ): Define.
34972 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34973 TARGET_AVX512VPOPCNTDQ_P): Define.
34974 * config/i386/i386.opt: Add mavx512vpopcntdq.
34975 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34976 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34978 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34980 PR middle-end/77484
34981 * predict.def (PRED_CALL): Set to 67.
34983 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34985 * expr.c (store_field): In the bitfield case, if the value comes from
34986 a function call and is of an aggregate type returned in registers, do
34987 not modify the field mode; extract the value in all cases if the mode
34988 is BLKmode and the size is not larger than a word.
34990 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34993 * config/i386/cpuid.h: Fix undefined behavior.
34995 2017-01-04 Jeff Law <law@redhat.com>
34997 PR tree-optimization/79007
34998 PR tree-optimization/67955
34999 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35000 conservative for pt.null when flag_non_call_exceptions is on.
35002 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35004 PR translation/79019
35005 PR translation/79020
35006 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35007 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35008 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35010 * config/avr/avr.opt (maccumulate-args): Likewise.
35011 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35012 * common.opt (freport-bug): Likewise.
35013 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35014 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35015 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35016 translatable string.
35017 * config/i386/i386.c (function_value_32): Likewise.
35018 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35019 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35021 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35022 * common/config/msp430/msp430-common.c (msp430_handle_option):
35024 * symtab.c (symtab_node::verify_base): Likewise.
35025 * opts.c (set_debug_level): Likewise.
35026 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35027 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35028 missing whitespace to translatable strings.
35029 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35030 * config/sh/superh.h: Likewise.
35031 * config/i386/xopintrin.h: Likewise.
35032 * config/i386/znver1.md: Likewise.
35033 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35034 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35035 * double-int.h (struct double_int): Likewise.
35036 * double-int.c (div_and_round_double): Likewise.
35037 * wide-int.cc: Likewise.
35038 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35039 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35040 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35041 (crossjumps_occurred): ... this.
35042 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35045 PR tree-optimization/78899
35046 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35047 returning bool return struct loop *, NULL for failure and the new
35049 (versionable_outer_loop_p): Don't version outer loop if it has
35050 dont_vectorized bit set.
35051 (tree_if_conversion): When versioning outer loop, ensure
35052 tree_if_conversion is performed also on the inner loop of the
35053 non-vectorizable outer loop copy.
35054 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35055 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35056 prevent vectorization of it.
35057 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35058 the outer loop vectorization of the non-scalar version is attempted
35059 before vectorization of the inner loop in scalar version. If
35060 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35061 vectorization of its inner loop.
35062 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35063 has 2 inner loops, rename also on edges from bb whose single pred
35064 is outer_loop->header. Fix typo in function comment.
35066 2017-01-09 Martin Sebor <msebor@redhat.com>
35069 * asan.c (asan_emit_stack_protection): Increase local buffer size
35070 to avoid snprintf truncation warning.
35072 2017-01-09 Andrew Pinski <apinski@cavium.com>
35074 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35075 to reference thunderx2t99 for the tuning structure
35076 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35078 (thunderx2t99_extra_costs): This.
35079 * config/aarch64/aarch64-tune.md: Regenerate.
35080 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35081 (vulcan_addrcost_table): This.
35082 (vulcan_regmove_cost): Rename to ...
35083 (thunderx2t99_regmove_cost): This.
35084 (vulcan_vector_cost): Rename to ...
35085 (thunderx2t99_vector_cost): this.
35086 (vulcan_branch_cost): Rename to ...
35087 (thunderx2t99_branch_cost): This.
35088 (vulcan_tunings): Rename to ...
35089 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35090 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35092 2017-01-09 Martin Jambor <mjambor@suse.cz>
35096 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35097 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35098 (propagate_vr_accross_jump_function): Use the above function for all
35099 value range computations for pass-through jump functions and type
35100 converasion from explicit value range values.
35101 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35102 parameters from TYPE_ARG_TYPES.
35103 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35104 (ipa_write_node_info): Stream type of the actual argument.
35105 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35107 2017-01-09 Martin Liska <mliska@suse.cz>
35110 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35111 (lookup_compiler): Do not show error message with have_E.
35113 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35115 PR tree-optimization/78938
35116 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35117 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35118 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35119 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35122 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35124 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35127 2017-01-09 Richard Biener <rguenther@suse.de>
35129 PR tree-optimization/78997
35130 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35131 name condition properly.
35133 2017-01-09 Richard Biener <rguenther@suse.de>
35136 * dwarf2out.c (is_cxx): New overload with context.
35137 (is_naming_typedef_decl): Use it.
35139 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35141 * invoke.texi (Option Summary): Correct spacing in option lists
35142 and add line breaks to fix over-long lines.
35144 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35146 PR middle-end/17660
35148 * extend.texi (Common Variable Attributes): Add xref to GCC
35149 Internals manual to explain mode attribute keywords.
35151 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35154 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35155 and Preprocessor Options.
35156 (Options for Linking): Document -pthread here....
35157 (RS/6000 and PowerPC Options): ...not here.
35158 (Solaris 2 Options): ...or here.
35159 * doc/cppopts.texi: Document -pthread.
35161 2017-01-08 Martin Sebor <msebor@redhat.com>
35163 PR middle-end/77708
35164 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35165 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35166 New member functions.
35167 (format_directive): Used them.
35169 (pass_sprintf_length::handle_gimple_call): Same.
35170 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35171 to avoid truncation for any argument.
35172 (extract_affine_mul): Same.
35173 * tree.c (get_file_function_name): Same.
35175 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35177 PR middle-end/77484
35178 * predict.def (PRED_INDIR_CALL): Set to 86.
35180 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35182 PR preprocessor/54124
35183 * doc/cppopts.texi: Reformat -d subtable to list the full name
35184 of the options. Add cross-reference to the docs for the general
35185 compiler -d options.
35186 * doc/invoke.texi (Developer Options): Add cross-reference to the
35187 preprocessor-specific -d option documentation.
35189 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35191 PR preprocessor/13498
35192 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35193 redudant material, and reflect new command-line options.
35194 (System Headers): Likewise.
35196 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35198 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35199 -isystem, and -idirafter. Copy-edit.
35200 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35201 default for -ftrack-macro-expansion. Delete obsolete and
35202 badly-formatted implementation details about -fdebug-cpp output.
35203 * doc/cppwarnopts.texi: Copy-edit.
35205 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35208 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35209 that the transition from a max line width >= 1<<10 to narrower
35210 lines works correctly.
35212 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35214 * doc/options.texi (PerFunction): New.
35215 * opt-functions.awk (switch_flags): Map both Optimization and
35216 PerFunction to CL_OPTIMIZATION.
35217 * opth-gen.awk: Test for PerFunction flag along with
35219 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35220 it only when the latter is present. Skip those that don't in
35221 the hash function generator.
35222 * common.opt (fvar-tracking): Mark as PerFunction instead of
35224 (fvar-tracking-assignments): Likewise.
35225 (fvar-tracking-assignments-toggle): Likewise.
35226 (fvar-tracking-uninit): Likewise.
35228 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35230 PR translation/79018
35231 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35234 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35237 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35238 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35239 TARGET_LONG_JUMP_TABLE_OFFSETS.
35240 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35241 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35242 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35243 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35244 * config/m68k/m68k.md (tablejump expander): Likewise.
35245 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35246 TARGET_LONG_JUMP_TABLE_OFFSETS.
35247 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35248 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35250 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35251 David Holsgrove <david.holsgrove@xilinx.com>
35253 * common/config/microblaze/microblaze-common.c
35254 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35255 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35257 * config/microblaze/microblaze.c (microblaze_must_save_register)
35258 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35260 (microblaze_eh_return): New function.
35261 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35262 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35263 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35264 * config/microblaze/microblaze.md (eh_return): New pattern.
35266 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35268 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35269 GCC_DIAGNOSTIC_STRINGIFY): Define.
35271 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35273 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35275 * config/arm/arm.md (<mcrr>): New.
35277 * config/arm/arm.c (arm_arch5te): New.
35278 (arm_option_override): Set arm_arch5te.
35279 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35281 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35282 (arm_mcrr_qualifiers): ... this. New.
35283 (MRRC_QUALIFIERS): Define to...
35284 (arm_mrrc_qualifiers): ... this. New.
35285 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35287 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35288 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35289 (MRRCI, mrrc, MRRC): New.
35290 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35291 VUNSPEC_MRRC2): New.
35293 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35295 * config/arm/arm.md (<mcr>): New.
35297 * config/arm/arm.c (arm_coproc_builtin_available): Add
35298 support for mcr, mrc, mcr2 and mrc2.
35299 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35300 (arm_mcr_qualifiers): ... this. New.
35301 (MRC_QUALIFIERS): Define to ...
35302 (arm_mrc_qualifiers): ... this. New.
35303 (MCR_QUALIFIERS): Define to ...
35304 (arm_mcr_qualifiers): ... this. New.
35305 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35307 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35308 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35309 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35310 VUNSPEC_MRC2): New.
35312 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35314 * config/arm/arm.md (*ldc): New.
35318 * config/arm/arm.c (arm_coproc_builtin_available): Add
35319 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35320 (arm_coproc_ldc_stc_legitimate_address): New.
35321 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35322 'qualifier_const_pointer'.
35323 (LDC_QUALIFIERS): Define to...
35324 (arm_ldc_qualifiers): ... this. New.
35325 (STC_QUALIFIERS): Define to...
35326 (arm_stc_qualifiers): ... this. New.
35327 * config/arm/arm-protos.h
35328 (arm_coproc_ldc_stc_legitimate_address): New.
35329 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35330 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35331 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35332 stc2, stcl, stc2l): New.
35333 * config/arm/constraints.md (Uz): New.
35334 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35335 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35336 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35337 VUNSPEC_STC2L): New.
35339 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35341 * config/arm/arm.md (<cdp>): New.
35342 * config/arm/arm.c (neon_const_bounds): Rename this ...
35343 (arm_const_bounds): ... this.
35344 (arm_coproc_builtin_available): New.
35345 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35346 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35347 (CDP_QUALIFIERS): Define to...
35348 (arm_cdp_qualifiers): ... this. New.
35350 (arm_expand_builtin_args): Add case for 6 arguments.
35351 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35352 (arm_const_bounds): ... this.
35353 (arm_coproc_builtin_available): New.
35354 * config/arm/arm_acle.h (__arm_cdp): New.
35356 * config/arm/arm_acle_builtins.def (cdp): New.
35358 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35359 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35360 'arm_const_bounds'.
35361 * config/arm/types.md (coproc): New.
35362 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35363 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35364 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35365 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35367 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35369 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35370 (UBINOP_QUALIFIERS): New.
35372 (acle_builtin_data): New. Change comment.
35373 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35374 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35375 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35376 arm_acle_builtins.def.
35377 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35378 (arm_init_acle_builtins): New.
35379 (CRC32_BUILTIN): Remove.
35380 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35381 crc32cb, crc32ch and crc32cw.
35382 (arm_init_crc32_builtins): Remove.
35383 (arm_init_builtins): Use arm_init_acle_builtins rather
35384 than arm_init_crc32_builtins.
35385 (arm_expand_acle_builtin): New.
35386 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35387 * config/arm/arm_acle_builtins.def: New.
35389 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35391 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35392 (arm_builtin_datum): ... this.
35393 (arm_init_neon_builtin): Rename to ...
35394 (arm_init_builtin): ... this. Add a new parameters PREFIX
35395 and USE_SIG_IN_NAME.
35396 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35397 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35398 'arm_builtin_datum'.
35399 (arm_init_vfp_builtins): Likewise.
35400 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35401 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35402 (arm_expand_neon_args): Rename to ...
35403 (arm_expand_builtin_args): ... this. Rename builtin_arg
35404 enum values and differentiate between ARG_BUILTIN_MEMORY
35405 and ARG_BUILTIN_NEON_MEMORY.
35406 (arm_expand_neon_builtin_1): Rename to ...
35407 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35408 values, arm_expand_builtin_args and add bool parameter NEON.
35409 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35410 (arm_expand_vfp_builtin): Likewise.
35411 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35413 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35415 PR middle-end/77484
35416 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35417 * predict.c (tree_estimate_probability_bb): Reverse direction of
35418 polymorphic call predictor.
35420 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35422 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35423 (determine_pass_name_match): ...this new function and...
35424 (should_skip_pass_p): ...this new function.
35426 2017-01-06 Nathan Sidwell <nathan@acm.org>
35428 * ipa-visibility.c (function_and_variable_visibility): Reformat
35429 comments and long lines. Remove extrneous if.
35430 * symtab.c (symtab_node::make_decl_local): Fix code format.
35431 (symtab_node::set_section_for_node): Fix comment typo.
35433 2017-01-06 Martin Liska <mliska@suse.cz>
35436 * lra-constraints.c: Rename invariant to lra_invariant.
35437 * predict.c (set_even_probabilities): Initialize e to NULL.
35439 2017-01-05 Martin Sebor <msebor@redhat.com>
35441 PR tree-optimization/78910
35442 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35443 (format_integer): Correct off-by-one error in the handling
35444 of precision with negative numbers in signed conversions..
35446 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35448 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35450 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35452 PR tree-optimization/71016
35453 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35454 factor_out_conditional_conversion. Formatting fix.
35455 (factor_out_conditional_conversion): Add cond_stmt argument.
35456 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35457 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35460 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35462 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35463 read-rtl-function.o, and selftest-rtl.o.
35464 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35465 (selftest::aarch64_test_loading_full_dump): New function.
35466 (selftest::aarch64_run_selftests): New function.
35467 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35468 selftest::aarch64_run_selftests.
35469 * config/i386/i386.c
35470 (selftest::ix86_test_loading_dump_fragment_1): New function.
35471 (selftest::ix86_test_loading_call_insn): New function.
35472 (selftest::ix86_test_loading_full_dump): New function.
35473 (selftest::ix86_test_loading_unspec): New function.
35474 (selftest::ix86_run_selftests): Call the new functions.
35475 * emit-rtl.c (maybe_set_max_label_num): New function.
35476 * emit-rtl.h (maybe_set_max_label_num): New decl.
35477 * function.c (instantiate_decls): Guard call to
35478 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35479 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35481 * gensupport.c (gen_reader::gen_reader): Pass "false"
35482 for new "compact" param of rtx_reader.
35483 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35484 rather than an empty string for NULL strings.
35485 * read-md.c: Potentially include config.h rather than bconfig.h.
35486 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35487 (have_error): New global, copied from errors.c.
35488 (md_reader::read_name): Rename to...
35489 (md_reader::read_name_1): ...this, adding "out_loc" param,
35490 and converting "missing name or number" to returning false, rather
35492 (md_reader::read_name): Reimplement in terms of read_name_1.
35493 (md_reader::read_name_or_nil): New function.
35494 (md_reader::read_string): Handle "(nil)" by returning NULL.
35495 (md_reader::md_reader): Add new param "compact".
35496 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35497 (md_reader::read_file): New method.
35498 * read-md.h (md_reader::md_reader): Add new param "compact".
35499 (md_reader::read_file): New method.
35500 (md_reader::is_compact): New accessor.
35501 (md_reader::read_name): Convert return type from void to file_location.
35502 (md_reader::read_name_or_nil): New decl.
35503 (md_reader::read_name_1): New decl.
35504 (md_reader::m_compact): New field.
35505 (noop_reader::noop_reader): Pass "false" for new "compact" param
35507 (rtx_reader::rtx_reader): Add new "compact" param.
35508 (rtx_reader::read_rtx_operand): Make virtual and convert return
35509 type from void to rtx.
35510 (rtx_reader::read_until): New decl.
35511 (rtx_reader::handle_any_trailing_information): New virtual function.
35512 (rtx_reader::postprocess): New virtual function.
35513 (rtx_reader::finalize_string): New virtual function.
35514 (rtx_reader::m_in_call_function_usage): New field.
35515 (rtx_reader::m_reuse_rtx_by_id): New field.
35516 * read-rtl-function.c: New file.
35517 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35518 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35519 (selftest::verify_three_block_rtl_cfg): New decl.
35520 * read-rtl-function.h: New file.
35521 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35522 For host, include function.h, memmodel.h, and emit-rtl.h.
35523 (one_time_initialization): New function.
35524 (struct compact_insn_name): New struct.
35525 (compact_insn_names): New array.
35526 (find_code): Handle insn codes in compact dumps.
35527 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35528 (bind_subst_iter_and_attr): Likewise.
35529 (add_condition_to_string): Likewise.
35530 (add_condition_to_rtx): Likewise.
35531 (apply_attribute_uses): Likewise.
35532 (add_current_iterators): Likewise.
35533 (apply_iterators): Likewise.
35534 (initialize_iterators): Guard usage of apply_subst_iterator with
35535 #ifdef GENERATOR_FILE.
35536 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35537 (md_reader::read_mapping): Likewise.
35538 (add_define_attr_for_define_subst): Likewise.
35539 (add_define_subst_attr): Likewise.
35540 (read_subst_mapping): Likewise.
35541 (check_code_iterator): Likewise.
35542 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35544 (one_time_initialization): New function.
35545 (rtx_reader::read_until): New method.
35546 (read_flags): New function.
35547 (parse_reg_note_name): New function.
35548 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35549 Handle reuse_rtx ids.
35550 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35551 Add parsing support for RTL dumps, mirroring the special-cases in
35552 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35553 values, and calling handle_any_trailing_information.
35554 (rtx_reader::read_rtx_operand): Convert return type from void
35555 to rtx, returning return_rtx. Handle case 'e'. Call
35556 finalize_string on XSTR and XTMPL fields.
35557 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35558 "(nil)" values were omitted. Call the postprocess vfunc on the
35560 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35561 class ctor. Initialize m_in_call_function_usage. Call
35562 one_time_initialization.
35563 * rtl-tests.c (selftest::test_uncond_jump): Call
35564 set_new_first_and_last_insn.
35565 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35566 * selftest-rtl.c: New file.
35567 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35568 (selftest::get_insn_by_uid): New decl.
35569 * selftest-run-tests.c (selftest::run_tests): Call
35570 read_rtl_function_c_tests.
35571 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35572 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35575 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35577 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35578 operands in a special way. Assert that pos+len <= mode precision.
35580 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35582 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35583 3 argument Alias with unlimited for the negative form.
35584 (fno-vect-cost-model): Removed.
35586 2017-01-05 Martin Liska <mliska@suse.cz>
35588 * hsa-gen.c (gen_hsa_divmod): New function.
35589 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35591 2017-01-05 Martin Liska <mliska@suse.cz>
35594 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35597 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35599 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35600 small constant length operands.
35602 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35604 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35605 between loop iterations.
35607 2017-01-05 Martin Liska <mliska@suse.cz>
35610 * gimplify.c (gimplify_decl_expr): Compare to
35611 asan_poisoned_variables instread of checking flags.
35612 (gimplify_target_expr): Likewise.
35613 (gimplify_expr): Likewise.
35614 (gimplify_function_tree): Conditionally initialize
35615 asan_poisoned_variables.
35617 2017-01-04 Jeff Law <law@redhat.com>
35619 PR tree-optimizatin/78812
35620 * rtl.h (contains_mem_rtx_p): Prototype.
35621 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35622 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35623 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35624 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35625 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35627 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35629 * input.c (assert_char_at_range): Default-initialize actual_range.
35631 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35633 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35634 to match the caller.
35636 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35638 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35639 insns after final jump in test to emit dummy move.
35641 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35643 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35644 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35646 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35648 * multiple_target.c (create_dispatcher_calls): Init e_next.
35649 * tree-ssa-loop-split.c (split_loop): Init border.
35650 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35653 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35658 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35659 (altivec_register_operand): Do not return true if the operand
35660 contains a SUBREG mixing SImode and SFmode.
35661 (vsx_register_operand): Likewise.
35662 (vsx_reg_sfsubreg_ok): New predicate.
35663 (vfloat_operand): Do not return true if the operand contains a
35664 SUBREG mixing SImode and SFmode.
35665 (vint_operand): Likewise.
35666 (vlogical_operand): Likewise.
35667 (gpc_reg_operand): Likewise.
35668 (int_reg_operand): Likewise.
35669 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35670 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35671 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35673 (rs6000_emit_move_si_sf_subreg): New helper function.
35674 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35675 fixup SUBREGs involving SImode and SFmode.
35676 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35677 numbers for the new peephole2 optimization.
35678 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35679 the GLIBC math library that do AND/IOR/XOR operations on single
35680 precision floating point.
35681 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35682 target macros to say whether we need to avoid SUBREGs mixing
35684 (TARGET_ALLOW_SF_SUBREG): Likewise.
35685 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35686 (UNSPEC_SI_FROM_SF): Likewise.
35687 (iorxor): Change spacing.
35688 (and_ior_xor): New iterator for AND, IOR, and XOR.
35689 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35690 (movdi_from_sf_zero_ext): Likewise.
35691 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35692 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35693 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35694 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35695 (fms<mode>4): Likewise.
35696 (fnma<mode>4): Likewise.
35697 (fnms<mode>4): Likewise.
35698 (nfma<mode>4): Likewise.
35699 (nfms<mode>4): Likewise.
35701 2017-01-04 Marek Polacek <polacek@redhat.com>
35704 * doc/invoke.texi: Document -Wpointer-compare.
35706 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35708 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35711 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35712 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35715 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35718 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35719 documentation of the powerpc_popcntb_ok attribute.
35720 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35721 code to issue warning messages if a requested CPU configuration is
35722 not supported by the binary (assembler and loader) toolchain.
35723 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35724 made to define a built-in function that has been disabled.
35725 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35726 made to define a built-in function that has been disabled.
35727 (altivec_init_builtins): Add comment explaining why definition
35728 of the DST built-in functions is not preceded by an assertion
35729 check. Add assertions to prevent ICE if attempts are made to
35730 define an altivec predicate or an abs* built-in function that has
35732 (htm_init_builtins): Add comment explaining why definition of the
35733 htm built-in functions is not preceded by an assertion check.
35735 2017-01-04 Jeff Law <law@redhat.com>
35737 PR tree-optimizatin/67955
35738 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35739 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35740 the points-to solution does not include pt_null. Use DECL_PT_UID
35743 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35745 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35746 Use gen_int_mode instead of gen_lopwart for const_int operands.
35748 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35750 PR tree-optimization/71563
35751 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35752 out of range value - has low bits known to be zero.
35754 2017-01-04 Alan Modra <amodra@gmail.com>
35756 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35757 * configure: Regenerate.
35758 * config.in: Regenerate.
35760 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35763 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35764 a substring of the message, but strcmp with the whole message. Ifdef
35765 ENABLE_NLS, translate the message first using dgettext.
35767 2017-01-03 Jeff Law <law@redhat.com>
35769 PR tree-optimizatin/78856
35770 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35771 (mark_threaded_blocks): Remove code to truncate thread paths that
35772 cross multiple loop headers. Instead invalidate the cached loop
35773 iteration information and handle case of a thread path walking
35774 into an irreducible region.
35776 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35779 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35780 assertions. Add support for doing the signbit if the IEEE 128-bit
35781 floating point value is in a GPR.
35782 * config/rs6000/rs6000.md (Fsignbit): Delete.
35783 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35784 Update the length attribute if the value is in a GPR.
35785 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35786 the sign or zero extension instruction, since the value is always 0/1.
35787 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35790 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35791 extracting SImode to a GPR register so that we can generate a
35792 store, limit the vector to be in a traditional Altivec register
35793 for the vextuwrx instruction.
35795 2017-01-03 Ian Lance Taylor <iant@google.com>
35797 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35799 2017-01-03 Martin Sebor <msebor@redhat.com>
35801 PR tree-optimization/78696
35802 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35803 precision. Use MPFR for %f for greater fidelity. Correct handling
35805 (pass_sprintf_length::compute_format_length): Set width and precision
35806 specified by asrerisk to void_node for vararg functions.
35807 (try_substitute_return_value): Adjust dump output.
35809 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35811 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35813 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35815 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35816 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35817 -mlra/-mno-lra was passed to the compiler.
35819 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35821 PR rtl-optimization/65618
35822 * emit-rtl.c (try_split): Move initialization of "before" and
35823 "after" to just before the call to emit_insn_after_setloc.
35825 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35827 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35829 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35831 * dwarf2out.c (gen_enumeration_type_die): When
35832 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35834 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35836 PR tree-optimization/78965
35837 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35838 Change first argument from const call_info & to call_info &. For %n
35839 set info.nowrite to false.
35841 PR middle-end/78901
35842 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35843 possibly throwing calls.
35845 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35846 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35847 and fns handling, rather than in a separate case SSA_NAME.
35849 2017-01-02 Jeff Law <law@redhat.com>
35851 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35852 fixes for first_period and second_period variables.
35854 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35857 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35858 (*insvqi_1): New insn pattern.
35859 (*insvqi_1_mem_rex64): Ditto.
35860 (*insvqi_2): Ditto.
35861 (*insvqi_3): Rename from *insvqi.
35863 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35865 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35867 * doc/cfg.texi (Edges): Remove reference to Java.
35868 (Maintaining the CFG): Ditto.
35870 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35872 PR middle-end/77674
35873 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35874 transparent aliases.
35876 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35878 PR middle-end/77484
35879 * predict.def (PRED_CALL): Update hitrate.
35880 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35881 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35882 into direct/indirect/polymorphic variants.
35884 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35886 Update copyright years.
35888 * gcc.c (process_command): Update copyright notice dates.
35889 * gcov-dump.c (print_version): Ditto.
35890 * gcov.c (print_version): Ditto.
35891 * gcov-tool.c (print_version): Ditto.
35892 * gengtype.c (create_file): Ditto.
35893 * doc/cpp.texi: Bump @copying's copyright year.
35894 * doc/cppinternals.texi: Ditto.
35895 * doc/gcc.texi: Ditto.
35896 * doc/gccint.texi: Ditto.
35897 * doc/gcov.texi: Ditto.
35898 * doc/install.texi: Ditto.
35899 * doc/invoke.texi: Ditto.
35901 Copyright (C) 2017 Free Software Foundation, Inc.
35903 Copying and distribution of this file, with or without modification,
35904 are permitted in any medium without royalty provided the copyright
35905 notice and this notice are preserved.