1 2017-11-20 Jeff Law <law@redhat.com>
3 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
7 gimple-ssa-evrp-analyze.h.
9 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
11 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
13 2017-11-20 Bin Cheng <bin.cheng@arm.com>
15 * tree-predcom.c: Add general comment on Store-Store chains.
16 (split_data_refs_to_components): Postpone clearing eliminate_store_p
18 (get_chain_last_ref_at): Rename into...
19 (get_chain_last_write_at): ...this.
20 (get_chain_last_write_before_load): New function.
21 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
22 CT_STORE_STORE when write reference is added.
23 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
24 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
25 (initialize_root_vars_store_elim_1): Ditto.
26 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
27 definition is created.
28 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
29 chain by replacing it with dominant stored value.
31 2017-11-20 Bin Cheng <bin.cheng@arm.com>
33 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
35 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
37 * vr-values.c (extract_range_from_binary_expr): Use a full range
40 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
42 * config/arm/arm.md (R4_REGNUM): Define constant.
43 (nonsecure_call_internal): Remove r4 clobber.
44 (nonsecure_call_value_internal): Likewise.
45 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
46 clobber and resequence match_operands.
47 (nonsecure_call_value_reg_thumb1_v5): Likewise.
48 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
49 (nonsecure_call_value_reg_thumb2): Likewise.
51 2017-11-20 Jakub Jelinek <jakub@redhat.com>
53 PR tree-optimization/78821
54 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
55 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
56 to the address of the base rather than the base itself.
57 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
58 (find_bswap_or_nop_finalize): New function.
59 (find_bswap_or_nop): Use it.
60 (bswap_replace): Return a tree rather than bool, change first
61 argument from gimple * to gimple_stmt_iterator, allow inserting
62 into an empty sequence, allow ins_stmt to be NULL - then emit
63 all stmts into gsi. Fix up MEM_REF address gimplification.
64 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
65 (struct store_immediate_info): Add N and INS_STMT non-static
67 (store_immediate_info::store_immediate_info): Initialize them
68 from newly added ctor args.
69 (merged_store_group::apply_stores): Formatting fixes. Sort by
71 (stmts_may_clobber_ref_p): For stores call also
72 refs_anti_dependent_p.
73 (gather_bswap_load_refs): New function.
74 (imm_store_chain_info::try_coalesce_bswap): New method.
75 (imm_store_chain_info::coalesce_immediate_stores): Use it.
76 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
77 (imm_store_chain_info::output_merged_store): Fail if number of
78 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
79 and NOP_EXPR rhs_code.
80 (pass_store_merging::process_store): Compute n and ins_stmt, if
81 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
82 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
84 (pass_store_merging::execute): Calculate dominators.
86 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
87 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
88 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
89 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
90 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
91 class pass_optimize_bswap, bswap_replace,
92 pass_optimize_bswap::execute): Moved to ...
93 * gimple-ssa-store-merging.c: ... this file.
94 Include optabs-tree.h.
95 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
96 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
97 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
98 anonymous namespace, remove static keywords.
99 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
100 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
102 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
105 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
107 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
109 * vec.h (debug_helper): New function.
110 (DEFINE_DEBUG_VEC): New macro.
111 * hash-set.h (debug_helper): New function.
112 (DEFINE_DEBUG_HASH_SET): New macro.
113 * cfg.c (debug_slim (edge)): New function.
114 Call DEFINE_DEBUG_VEC for edges.
115 Call DEFINE_DEBUG_HASH_SET for edges.
116 * cfghooks.c (debug_slim (basic_block)): New function.
117 Call DEFINE_DEBUG_VEC for basic blocks.
118 Call DEFINE_DEBUG_HASH_SET for basic blocks.
119 * print-tree.c (debug_slim): New function to handle trees.
120 Call DEFINE_DEBUG_VEC for trees.
121 Call DEFINE_DEBUG_HASH_SET for trees.
122 (debug (vec<tree, va_gc>) &): Remove.
123 (debug (<vec<tree, va_gc>) *): Remove.
124 * print-rtl.c (debug_slim): New function to handle const_rtx.
125 Call DEFINE_DEBUG_VEC for rtx_def.
126 Call DEFINE_DEBUG_VEC for rtx_insn.
127 Call DEFINE_DEBUG_HASH_SET for rtx_def.
128 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
129 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
130 (debug (vec<rtx_insn *> *ptr): Remove.
131 (debug_insn_vector): Remove.
132 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
134 2017-11-20 Tom de Vries <tom@codesourcery.com>
136 PR rtl-optimization/82020
137 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
138 IF_THEN_ELSE condition.
140 2017-11-19 Jeff Law <law@redhat.com>
142 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
143 of degenerates resulting from ignoring an edge.
145 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
148 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
150 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
153 * profile-count.c (profile_count::to_sreal_scale): Fix return value
154 for uninitialied counts.
156 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
159 * tree-inline.c (estimate_num_insns): Set to 1 at least.
161 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
164 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
167 2017-11-19 Tom de Vries <tom@codesourcery.com>
169 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
172 2017-11-19 Tom de Vries <tom@codesourcery.com>
174 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
177 2017-11-19 Tom de Vries <tom@codesourcery.com>
179 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
182 2017-11-19 Tom de Vries <tom@codesourcery.com>
184 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
186 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
188 2017-11-19 Tom de Vries <tom@codesourcery.com>
190 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
192 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
193 semicolon after MCORE_EXPORT_NAME call.
195 2017-11-19 Tom de Vries <tom@codesourcery.com>
198 * vmsdbgout.c (vmsdbgout_early_finish): New function.
199 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
201 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
203 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
204 (cgraph_edge::clone): Cleanup updating of profile.
205 * ipa-cp.c (update_profiling_info): Likewise.
206 * ipa-inline-transform.c (inline_transform): Likewise.
207 * ipa-inline.c (inline_small_functions): Add missing space to dump.
208 * ipa-split.c (execute_split_functions): Do not split when function
210 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
211 * profile-count.c (profile_count::dump): Add global0.
212 (profile_count::to_cgraph_frequency): Do not ICE when entry is
214 (profile_count::to_sreal_scale): Likewise.
215 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
216 (profile_count::combine_with_ipa_count): New function.
217 * profile-count.h (profile_guessed_global0adjusted): New.
218 (profile_count::adjusted_zero): New.
219 (profile_count::global0adjusted): New.
220 (profile_count::combine_with_ipa_count): New.
221 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
222 correct profile of return block of split functions.
223 (copy_cfg_body): Remove unused profile_count.
224 (copy_body): Likewise.
225 (expand_call_inline): Update.
226 (tree_function_versioning): Update.
228 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
230 * hash-set.h (hash_set::empty): New.
231 * tree-ssa-threadbackward.h: Delete.
232 * tree-ssa-threadbackward.c (class thread_jumps): New.
233 Move max_threaded_paths into class.
234 (fsm_find_thread_path): Remove arguments that are now in class.
235 (profitable_jump_thread_path): Rename to...
236 (thread_jumps::profitable_jump_thread_path): ...this.
237 (convert_and_register_jump_thread_path): Rename to...
238 (thread_jumps::convert_and_register_current_path): ...this.
239 (check_subpath_and_update_thread_path): Rename to...
240 (thread_jumps::check_subpath_and_update_thread_path): ...this.
241 (register_jump_thread_path_if_profitable): Rename to...
242 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
243 (handle_phi): Rename to...
244 (thread_jumps::handle_phi): ...this.
245 (handle_assignment): Rename to...
246 (thread_jumps::handle_assignment): ...this.
247 (fsm_find_control_statement_thread_paths): Rename to...
248 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
249 (find_jump_threads_backwards): Rename to...
250 (thread_jumps::find_jump_threads_backwards): ...this.
251 Initialize path local data.
252 (pass_thread_jumps::execute): Call find_jump_threads_backwards
253 from within thread_jumps class.
254 (pass_early_thread_jumps::execute): Same.
256 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
258 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
260 2017-11-17 Steve Ellcey <sellcey@cavium.com>
262 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
263 to canonical location.
265 2017-11-17 Steve Ellcey <sellcey@cavium.com>
268 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
270 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
272 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
274 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
275 rcrt1.o%s/grcrt1.o%s for -static-pie.
277 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
279 * i386.c (ix86_multiplication_cost, ix86_division_cost,
280 ix86_shift_rotate_cost): Break out from ...
281 (ix86_rtx_costs): ... here.
282 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
285 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
287 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
288 when entry block was promoted unlikely.
289 (estimate_bb_frequencies): Increase frequency scale.
290 * profile-count.h (profile_count): Export precision info.
292 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
294 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
295 disturbin profile of entry block.
297 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
299 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
302 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
304 * ipa-cp.c (update_profiling_info): Handle conversion to local
306 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
308 2017-11-17 Jeff Law <law@redhat.com>
310 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
311 from evrp_dom_walker class. Various methods moved into new class.
312 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
313 (evrp_range_analyzer::enter): New method.
314 (evrp_range_analyzer::leave): New method.
315 (evrp_dom_walker): Remove delegators no longer needed by this class.
316 Replace vr_values data member with evrp_range_analyzer
318 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
319 method extracted from evrp_dom_walker::before_dom_children.
320 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
321 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
323 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
324 Add private copy constructor and move assignment operators.
325 Privatize methods and class data where trivially possible.
326 (evrp_dom_walker::cleanup): New function, extracted from
327 execute_early_vrp. Simplify access to class data.
329 * vr-values.h (get_output_for_vrp): Prototype.
330 * vr-values.c (get_output_for_vrp): New function extracted from
331 vrp_visit_assignment_or_call and extract_range_from_stmt.
332 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
334 2017-11-17 Luis Machado <luis.machado@linaro.org>
336 * config/aarch64/aarch64.c
337 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
338 (qdf24xx_tunings) <autoprefetcher_model>: Set to
339 tune_params::AUTOPREFETCHER_WEAK.
341 2017-11-17 Tamar Christina <tamar.christina@arm.com>
344 * config/arm/arm.c (arm_valid_target_attribute_rec):
345 Parse "arch=" and "+<ext>".
346 (arm_valid_target_attribute_tree): Re-init global options.
347 (arm_option_override): Make non-static.
348 (arm_options_perform_arch_sanity_checks): Make errors fatal.
349 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
350 (__ARM_FEATURE_CRC32): Support undef.
351 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
352 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
354 2017-11-17 David Malcolm <dmalcolm@redhat.com>
356 * gdbinit.in (break-on-diagnostic): New command.
358 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
360 * config/i386/i386.c (ix86_expand_epilogue): Change simple
361 return to indirect jump for EH return if control-flow
362 protection is enabled. Change explicit 'false' argument in
363 pro_epilogue_adjust_stack with a value of
365 * config/i386/i386.md (simple_return_indirect_internal):
366 Remove SImode restriction to support 64-bit.
368 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
370 * combine.c (added_notes_insn): New.
371 (try_combine): Handle added_notes_insn like added_links_insn.
372 Rewrite return value code.
373 (distribute_notes): Set added_notes_insn to the earliest insn we added
376 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
378 PR rtl-optimization/82621
379 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
380 dest of one of those SETs is unused.
382 2017-11-17 Richard Biener <rguenther@suse.de>
385 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
386 * tree-pretty-print.c (dump_generic_node): Handle
387 annot_expr_parallel_kind.
388 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
389 * gimplify.c (gimple_boolify): Likewise.
391 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
393 * config.gcc (extra_headers): Add cet.h for x86 targets.
394 * config/i386/cet.h: New file.
395 * doc/install.texi: Add --enable-cet/--disable-cet.
397 2017-11-17 Richard Biener <rguenther@suse.de>
399 PR tree-optimization/83017
400 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
401 (gen_parallel_loop): Properly count iterations.
402 (parallelize_loops): Handle loop->can_be_parallel independent
403 of flag_loop_parallelize_all. Make static profitability test match
405 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
406 * invoke.texi (parloops-min-per-thread): Document.
408 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
410 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
411 upstreaming review comments.
413 2017-11-17 Tamar Christina <tamar.christina@arm.com>
415 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
416 with fast unaligned access.
417 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
419 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
421 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
422 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
424 2017-11-17 Richard Biener <rguenther@suse.de>
426 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
427 folding of references.
429 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
432 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
433 of replacing call to strncmp with corresponding call to strcmp when
436 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
438 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
439 option prefer-avx256 for skylake-avx512 configuration.
440 * config/i386/i386.c (ix86_option_override_internal): Ditto.
441 (get_builtin_code_for_version): Ditto.
443 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
444 Monk Chiang <sh.chiang04@gmail.com>
446 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
447 (FIXED_REGISTERS): Reserve more register numbers.
448 (CALL_USED_REGISTERS): Likewise.
449 (REG_ALLOC_ORDER): Likewise.
450 (REG_CLASS_CONTENTS): Likewise.
451 (REGISTER_NAMES): Likewise.
453 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
454 Kito Cheng <kito.cheng@gmail.com>
456 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
458 * config/nds32/iterators.md: Add vector mode iterators and attributes.
460 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
462 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
463 parameter list for vec_splats.
465 2017-11-16 Joseph Myers <joseph@codesourcery.com>
467 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
469 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
472 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
475 * doc/install.texi: Document incompatibility of Perl >=5.6.26
476 with the required version of automake 1.11.6.
478 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
480 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
482 (DU_C2_3_power9): Correct reservation combinations.
483 (FP_DIV_power9, VEC_DIV_power9): New.
484 (power9-alu): Split out rotate/shift...
485 (power9-rot): ...to here, correct dispatch resource.
486 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
488 (power9-fp): Correct latency.
489 (power9-sdiv): Add div/sqrt resource.
490 (power9-ddiv): Correct latency, add div/sqrt resource.
491 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
492 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
494 (power9-qpdiv, power9-qpmul): Adjust resource usage.
496 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
498 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
499 switch statement mapping KF built-ins to TF built-ins if we don't
500 have the proper ISA 3.0 assembler support.
502 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
504 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
505 (lower_emutls_function_body): Do not compute it.
507 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
509 * ipa-split.c (split_bb_info): Turn time to sreal.
510 (split_point): Likewise.
511 (dump_split_point): Likewise.
512 (fine_split_points): Likewise.
513 (execute_split_functions): Only zero split_bbs; turn time to sreals.
515 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
517 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
520 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
522 * predict.c (combine_predictions_for_bb): Preserve zero predicted
524 (expensive_function_p): Remove useless assert.
525 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
527 2017-11-16 Martin Sebor <msebor@redhat.com>
529 PR tree-optimization/82588
530 PR tree-optimization/82583
531 * tree-vrp.c (check_array_ref): Handle flexible array members,
532 string literals, and inner indices.
533 (search_for_addr_array): Add detail to diagnostics.
535 2017-11-16 Nathan Sidwell <nathan@acm.org>
539 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
540 (SET_DECL_ASSEMBLER_NAME): Forward to
541 overwrite_decl_assembler_name.
542 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
543 (overwrite_decl_assembler_name): Declare.
544 * tree.c (overwrite_decl_assembler_name): New.
545 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
546 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
547 (LANG_HOOKS_INITIALIZER): Add it.
548 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
549 * langhooks.c (lhd_set_decl_assembler_name): Use
550 SET_DECL_ASSEMBLER_NAME.
551 (lhd_overwrite_decl_assembler_name): Default implementation.
553 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
554 Jackson Woodruff <jackson.woodruff@arm.com>
556 PR tree-optimization/71026
557 * match.pd: Canonicalize constant multiplies in division.
559 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
561 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
562 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
563 Remove OPT_fomit_frame_pointer entry.
564 * common/config/arc/arc-common.c: Likewise.
565 * common/config/arm/arm-common.c: Likewise.
566 * common/config/avr/avr-common.c: Likewise.
567 * common/config/c6x/c6x-common.c: Likewise.
568 * common/config/cr16/cr16-common.c: Likewise.
569 * common/config/cris/cris-common.c: Likewise.
570 * common/config/epiphany/epiphany-common.c: Likewise.
571 * common/config/fr30/fr30-common.c: Likewise.
572 * common/config/frv/frv-common.c: Likewise.
573 * common/config/ia64/ia64-common.c: Likewise.
574 * common/config/iq2000/iq2000-common.c: Likewise.
575 * common/config/lm32/lm32-common.c: Likewise.
576 * common/config/m32r/m32r-common.c: Likewise.
577 * common/config/mcore/mcore-common.c: Likewise.
578 * common/config/microblaze/microblaze-common.c: Likewise.
579 * common/config/mips/mips-common.c: Likewise.
580 * common/config/mmix/mmix-common.c: Likewise.
581 * common/config/mn10300/mn10300-common.c: Likewise.
582 * common/config/nios2/nios2-common.c: Likewise.
583 * common/config/pa/pa-common.c: Likewise.
584 * common/config/pdp11/pdp11-common.c: Likewise.
585 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
586 * common/config/riscv/riscv-common.c: Likewise.
587 * common/config/rs6000/rs6000-common.c: Likewise.
588 * common/config/rx/rx-common.c: Likewise.
589 * common/config/s390/s390-common.c: Likewise.
590 * common/config/sh/sh-common.c: Likewise.
591 * common/config/sparc/sparc-common.c: Likewise.
592 * common/config/tilegx/tilegx-common.c: Likewise.
593 * common/config/tilepro/tilepro-common.c: Likewise.
594 * common/config/v850/v850-common.c: Likewise.
595 * common/config/visium/visium-common.c: Likewise.
596 * common/config/xstormy16/xstormy16-common.c: Likewise.
597 * common/config/xtensa/xtensa-common.c: Likewise.
598 * invoke.texi (-fomit-frame-pointer): Update documentation.
600 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
602 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
604 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
606 * cfg.c (scale_bbs_frequencies_int,
607 cale_bbs_frequencies_gcov_type): Remove.
608 * cfg.h (scale_bbs_frequencies_int,
609 cale_bbs_frequencies_gcov_type): Remove.
611 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
613 * tree-ssa-loop-manip.c
614 (scale_dominated_blocks_in_loop): Update to profile counts.
615 (tree_transform_and_unroll_loop): Likewise.
617 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
619 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
620 scale_bbs_frequencies_int.
622 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
624 * final.c (compute_alignments): Use counts rather than frequencies.
626 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
628 * cfgloopanal.c: Include sreal.h
629 (average_num_loop_insns): Use counts and sreal for accounting.
631 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
633 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
636 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
637 Kito Cheng <kito.cheng@gmail.com>
639 * config/nds32/constraints.md: Provide more constraints.
640 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
641 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
642 support constraints usage.
644 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
646 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
648 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
649 Kito Cheng <kito.cheng@gmail.com>
651 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
652 * config/nds32/nds32.opt: Refine the layout.
653 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
654 TARGET_EXT_STRING): Support new options.
655 * config/nds32/nds32.h: Likewise.
656 * config/nds32/nds32.md: Likewise.
657 * config/nds32/nds32-predicates.c: Likewise.
658 * config/nds32/constraints.md: Likewise.
659 * common/config/nds32/nds32-common.c: Likewise.
661 2017-11-16 Julia Koval <julia.koval@intel.com>
664 * config/i386/gfniintrin.h: Add sse check.
665 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
667 2017-11-16 Julia Koval <julia.koval@intel.com>
669 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
670 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
671 (ix86_handle_option): Handle -mavx512vbmi2.
672 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
673 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
674 * config/i386/i386-c.c (__AVX512VBMI2__): New.
675 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
676 (ix86_valid_target_attribute_inner_p): Ditto.
677 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
678 * config/i386/i386.opt (mavx512vbmi2): New option.
679 * doc/invoke.texi: Add new option.
681 [2017-11-16 Julia Koval <julia.koval@intel.com>
683 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
684 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
685 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
686 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
687 _mm512_gf2p8mul_epi8): New intrinsics.
688 * config/i386/i386-builtin-types.def
689 (V64QI_FTYPE_V64QI_V64QI): New type.
690 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
691 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
692 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
693 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
694 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
695 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
697 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
699 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
700 explicitly as a stream of bytes.
702 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
704 * config/rs6000/altivec.h (vec_xst_be): New #define.
705 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
706 and externalize from *altivec_vperm_<mode>_internal.
707 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
709 (XL_BE_V8HI): Likewise.
710 (XL_BE_V4SI): Likewise.
711 (XL_BE_V4SI): Likewise.
712 (XL_BE_V2DI): Likewise.
713 (XL_BE_V4SF): Likewise.
714 (XL_BE_V2DF): Likewise.
715 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
716 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
717 all array entries with these keys: VSX_BUILTIN_VEC_XL,
718 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
719 VSX_BUILTIN_VEC_XST_BE.
720 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
721 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
723 (altivec_init_builtins): Replace conditional calls to def_builtin
724 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
725 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
726 with unconditional calls. Remove calls to def_builtin for
727 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
728 __builtin_vec_xst_be.
729 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
730 to define_expand, and add alternate RTL generation for P8.
731 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
733 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
734 add alternate RTL generation for P8.
735 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
736 vsx_ld_elemrev_v16qi.
737 (vsx_st_elemrev_v8hi): Convert define_insn
738 to define_expand, and add alternate RTL generation for P8.
739 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
741 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
742 add alternate RTL generation for P8.
743 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
744 vsx_st_elemrev_v16qi.
746 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
749 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
750 TARGET_AVX512ER check.
751 (ix86_option_override_internal): Set MASK_VZEROUPPER if
752 neither -mzeroupper nor -mno-zeroupper is used and
753 TARGET_EMIT_VZEROUPPER is set.
754 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
755 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
757 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
759 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
760 folding of vector compares.
761 (fold_build_vec_cmp): New helper function.
762 (fold_compare_helper): New helper function.
763 (builtin_function_type): Add compare builtins to the list of functions
764 having unsigned arguments. Cosmetic updates to comment indentation.
765 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
766 the not+eq combination.
768 2017-11-15 Bin Cheng <bin.cheng@arm.com>
770 PR tree-optimization/82726
771 PR tree-optimization/70754
772 * tree-predcom.c (order_drefs_by_pos): New function.
773 (combine_chains): Move code setting has_max_use_after to...
774 (try_combine_chains): ...here. New parameter. Sort combined chains
775 according to position information.
776 (tree_predictive_commoning_loop): Update call to above function.
777 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
779 2017-11-15 Bin Cheng <bin.cheng@arm.com>
781 PR tree-optimization/82726
783 2017-01-23 Bin Cheng <bin.cheng@arm.com>
785 PR tree-optimization/70754
786 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
787 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
788 combined stmt before it if not NULL.
789 (combine_chains): Process refs reversely and compute dominance point
793 2017-02-23 Bin Cheng <bin.cheng@arm.com>
795 PR tree-optimization/79663
796 * tree-predcom.c (combine_chains): Process refs in reverse order
797 only for ZERO length chains, and add explaining comment.
799 2017-11-15 Tamar Christina <tamar.christina@arm.com>
801 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
802 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
803 * doc/invoke.texi (ARM Options): Add armv8.3-a.
805 2017-11-15 Tamar Christina <tamar.christina@arm.com>
807 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
809 2017-11-15 Martin Liska <mliska@suse.cz>
811 * tree-cfg.c (pass_warn_function_return::execute):
812 Compare warn_return_type for greater than zero.
814 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
818 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
819 to return true on Xeon and not on Xeon Phi.
820 (ix86_check_avx256_register): Changed to ...
821 (ix86_check_avx_upper_register): ... this. Add extra check for
822 VALID_AVX512F_REG_OR_XI_MODE.
823 (ix86_avx_u128_mode_needed): Changed
824 ix86_check_avx256_register to ix86_check_avx_upper_register.
825 (ix86_check_avx256_stores): Changed to ...
826 (ix86_check_avx_upper_stores): ... this. Changed
827 ix86_check_avx256_register to ix86_check_avx_upper_register.
828 (ix86_avx_u128_mode_after): Changed
829 avx_reg256_found to avx_upper_reg_found. Changed
830 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
831 (ix86_avx_u128_mode_entry): Changed
832 ix86_check_avx256_register to ix86_check_avx_upper_register.
833 (ix86_avx_u128_mode_exit): Ditto.
834 * config/i386/i386.h: (host_detect_local_cpu): New define.
836 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
838 * config/arm/xgene1.md (xgene1): Split into automatons
839 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
840 (xgene1_f_load): Adjust reservations and/or types.
841 (xgene1_f_store): Likewise.
842 (xgene1_load_pair): Likewise.
843 (xgene1_store_pair): Likewise.
844 (xgene1_fp_load1): Likewise.
845 (xgene1_load1): Likewise.
846 (xgene1_store1): Likewise.
847 (xgene1_move): Likewise.
848 (xgene1_alu): Likewise.
849 (xgene1_simd): Likewise.
850 (xgene1_bfm): Likewise.
851 (xgene1_neon_load1): Likewise.
852 (xgene1_neon_store1): Likewise.
853 (xgene1_neon_logic): Likewise.
854 (xgene1_neon_st1): Likewise.
855 (xgene1_neon_ld1r): Likewise.
856 (xgene1_alu_cond): Added.
857 (xgene1_shift_reg): Likwise.
858 (xgene1_bfx): Likewise.
859 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
861 2017-11-15 Jakub Jelinek <jakub@redhat.com>
864 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
866 (can_widen_mult_without_libcall): New function.
867 (expand_mul_overflow): If only checking unsigned mul overflow,
868 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
869 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
870 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
871 (expand_DIVMOD): Formatting fix.
872 * expmed.h (expand_mult): Add NO_LIBCALL argument.
873 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
874 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
876 PR tree-optimization/82977
877 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
878 constructed temporary to strlen_to_stridx.put.
880 2017-11-15 Martin Liska <mliska@suse.cz>
882 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
883 * configure: Regenerate.
885 2017-11-15 Martin Liska <mliska@suse.cz>
888 * config/sh/sh-mem.cc: Use proper probability for
891 2017-11-14 Jeff Law <law@redhat.com>
893 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
894 the red zone for stack_clash_protection_final_dynamic_probe targets
895 when the total dynamic stack size is zero bytes.
897 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
898 blocks is post order.
900 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
902 * dumpfile.h (TDF_COMPARE_DEBUG): New.
903 * final.c (rest_of_clean_state): Set it for the
904 -fcompare-debug dump.
905 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
906 class when TDF_COMPARE_DEBUG is set.
908 * dwarf2out.c (gen_producer_string): Discard
911 2017-11-15 Joseph Myers <joseph@codesourcery.com>
914 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
915 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
916 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
917 (__TGMATH_REAL_2_3): Remove macros.
918 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
919 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
921 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
923 (remquo): Define using __TGMATH_REAL_3.
925 2017-11-14 Jeff Law <law@redhat.com>
927 * vr-values.c: New file with contents extracted from tree-vrp.c.
928 * Makefile.in (OBJS): Add vr-values.o
929 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
930 (set_value_range, set_and_canonicalize_value_range): Likewise.
931 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
932 (value_range_constant_singleton, symbolic_range_p): Likewise.
933 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
934 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
935 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
936 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
937 (find_case_label_range, find_case_label_index): Likewise.
938 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
939 (range_int_cst_singleton_p, value_inside_range): Likewise.
940 (get_single_symbol): Likewise.
941 (switch_update): Move structure definition here.
942 (to_remove_edges, to_update_switch_stmts): Provide externs.
943 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
944 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
945 (vrp_val_is_min, set_value_range): Likewise.
946 (set_and_canonicalize_value_range, copy_value_range): Likewise.
947 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
948 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
949 (range_is_nonnull, range_int_cst_p): Likewwise.
950 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
951 (get_single_symbol, operand_less_p): Likewise
952 (compare_values_warnv, compare_values): Likewise.
953 (value_inside_range, value_range_constant_singleton): Likewise.
954 (zero_nonzero_bitgs_from_vr): Likewise.
955 (extract_range_from_binary_expr_1): Likewise.
956 (overflow_comparison_p): Likewise.
957 (to_remove_edges, to_update_switch_stmts): Likewise.
958 (find_case_label-index, find_case_label_range): Likewise.
959 (switch_update, set_value_range_to_nonnegative): Remove.
960 (set_value_range_to_truthvalue): Likewise.
961 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
962 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
963 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
964 (find_case_label_ranges, test_for_singularity): Likewise.
965 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
966 (x_vr_values): Move to its remaining use site.
968 2017-11-10 Jeff Law <law@redhat.com>
970 * vr-values.h (VR_INITIALIZER): Move #define here.
971 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
972 * Makefile.in (OBJS): Add tree-evrp.o
973 * tree-vrp.h (assert_info): Move structure definition here.
974 (set_value_range_to_varying): Prototype.
975 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
976 (infer_value_range, register_edge_assert_for): Likewise.
977 (stmt_interesting_for_vrp): Likewise.
978 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
979 (set_value_range_to_varying): No longer static.
980 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
981 (infer_value_range, register_edge_assert_for): Likewise.
983 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
985 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
986 generate the XXBRD instruction.
988 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
989 (rs6000_builtin_type_compatible): Treat _Float128 and long double
990 as being compatible if -mabi=ieeelongdouble.
991 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
992 to setup float128 built-ins with hardware support.
993 (BU_FLOAT128_HW_2): Likewise.
994 (BU_FLOAT128_HW_3): Likewise.
995 (BU_FLOAT128_HW_VSX_1): Likewise.
996 (BU_FLOAT128_HW_VSX_2): Likewise.
997 (scalar_extract_expq): Change float128 built-in functions to
998 accommodate having both KFmode and TFmode functions. Use the
999 KFmode variant as the default.
1000 (scalar_extract_sigq): Likewise.
1001 (scalar_test_neg_qp): Likewise.
1002 (scalar_insert_exp_q): Likewise.
1003 (scalar_insert_exp_qp): Likewise.
1004 (scalar_test_data_class_qp): Likewise.
1005 (sqrtf128_round_to_odd): Delete processing the round to odd
1006 built-in functions as special built-in functions, and define them
1007 as float128 built-ins. Use the KFmode variant as the default.
1008 (truncf128_round_to_odd): Likewise.
1009 (addf128_round_to_odd): Likewise.
1010 (subf128_round_to_odd): Likewise.
1011 (mulf128_round_to_odd): Likewise.
1012 (divf128_round_to_odd): Likewise.
1013 (fmaf128_round_to_odd): Likewise.
1014 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1015 support for KFmode and TFmode xststdcqp calls.
1016 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
1017 point, switch the built-in handlers for the get/set float128
1018 exponent, get float128 mantissa, float128 test built-ins, and the
1019 float128 round to odd built-in functions. Eliminate creating the
1020 float128 round to odd built-in functions as special built-ins.
1021 (rs6000_init_builtins): Eliminate special creation of the float128
1022 round to odd built-in functions.
1023 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
1024 function insns to support both TFmode and KFmode variants.
1025 (xsxsigqp_<mode>): Likewise.
1026 (xsiexpqpf_<mode>): Likewise.
1027 (xsiexpqp_<mode>): Likewise.
1028 (xststdcqp_<mode>): Likewise.
1029 (xststdcnegqp_<mode>): Likewise.
1030 (xststdcqp_<mode>): Likewise.
1032 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1034 * tree-ssa-threadupdate.c (compute_path_counts): Remove
1035 unused path_in_freq_ptr parameter.
1036 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
1038 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
1040 * ipa-inline.c (edge_badness): Dump sreal frequency.
1041 (compute_inlined_call_time): Match natural implementaiton ...
1042 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
1043 forgotten division by CGRAPH_FREQ_BASE.
1045 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1047 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
1048 Solaris 11. Update comment.
1049 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
1051 * config/sol2.h (STARTFILE_SPEC): Likewise.
1052 * configure: Regenerate.
1054 2017-11-14 Carl Love <cel@us.ibm.com>
1056 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
1057 le_ and be_ prefixes to swap* variables. Remove
1058 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
1061 2017-11-14 Jason Merrill <jason@redhat.com>
1063 Support GTY((cache)) on hash_map.
1064 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
1065 (ggc_cache_remove): Override it instead of ggc_mx.
1066 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
1067 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
1068 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
1069 (simple_cache_map_traits): Override maybe_mx.
1070 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
1071 (hash_map): Friend gt_cleare_cache.
1072 (gt_cleare_cache): New.
1073 * tree.h (tree_cache_traits): New hash_map traits class.
1074 (tree_cache_map): New typedef.
1076 2017-11-14 Richard Biener <rguenther@suse.de>
1078 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
1079 paramter and handling.
1080 (cleanup_control_flow_bb): Likewise.
1081 (cleanup_control_flow_pre): New helper performing a DFS walk
1082 to call cleanup_control_flow_bb in PRE order.
1083 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
1084 via cleanup_control_flow_pre.
1086 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
1088 * config/aarch64/aarch64-simd.md
1089 (aarch64_simd_bsl<mode>_internal): Remove DImode.
1090 (*aarch64_simd_bsl<mode>_alt): Likewise.
1091 (aarch64_simd_bsldi_internal): New.
1092 (aarch64_simd_bsldi_alt): Likewise.
1094 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1096 * tracer.c (better_p): Do not compare frequencies.
1097 * reg-stack.c (better_edge): Likewise.
1098 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1101 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1103 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1104 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1105 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1106 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1107 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1108 * predict.c (maybe_hot_frequency_p): Inline to ...
1109 (maybe_hot_count_p): ... here; rewrite to counts.
1110 (counts_to_freqs): Rename to ...
1111 (update_max_bb_count): ... this one.
1112 (expensive_function_p): Use counts.
1113 (estimate_bb_frequencies): Update.
1114 (rebuild_frequencies): Update.
1115 * predict.h (counts_to_freqs): Rename to ...
1116 (update_max_bb_count): ... this one.
1117 * profile.c (compute_branch_probabilities): Add debug info
1118 * tree-inline.c (expand_call_inline): Update debug info.
1119 (optimize_inline_calls): Use update_max_bb_count..
1120 (tree_function_versioning): Use update_max_bb_count..
1121 * value-prof.c (gimple_value_profile_transformations):
1122 Do not use update_max_bb_count.
1124 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1126 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1127 always use frequencies.
1129 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1131 * bb-reorder.c: Remove frequencies from comments.
1132 (better_edge_p): Use profile counts.
1133 (find_traces): Dump profile counts.
1134 (rotate_loop): Use profile counts.
1135 (find_traces_1_round): Likewise.
1136 (connect_better_edge_p): Use counts instead of probabilities for
1138 (copy_bb_p): Drop early check for non-0 frequency.
1139 (sanitize_hot_paths): Update comments.
1141 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1143 * ipa-split.c (struct split_point): Add count.
1144 (consider_split): Do not compute incoming frequency; compute incoming
1145 count and store it to split_point.
1146 (split_function): Set count of the call to split part correctly.
1148 2017-11-13 Carl Love <cel@us.ibm.com>
1150 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1152 2017-11-13 Tom Tromey <tom@tromey.com>
1154 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1156 2017-11-13 Carl Love <cel@us.ibm.com>
1158 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1159 Add support for builtins:
1160 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1161 vector {un,}signed {char,int,short},
1162 vector {un,}signed {char,int,short}) arguments.
1163 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1164 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1165 Add BU_P9V_AV_2 expansions for the builtins.
1166 * config/rs6000/altivec.h (vec_first_match_index,
1167 vec_first_mismatch_index, vec_first_match_or_eos_index,
1168 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1169 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1170 new extern declaration.
1171 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1172 * config/rs6000/vsx.md (first_match_index_<mode>,
1173 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1174 first_mismatch_or_eos_index_<mode>): Add define expand.
1175 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1176 * doc/extend.texi: Update the built-in documenation file for the new
1179 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1181 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1182 into the min/max operations for _Float<N> and _Float<N>X types.
1184 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1187 * dwarf2out.c (do_eh_frame): New static variable.
1188 (dwarf2out_begin_prologue): Set it.
1189 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1191 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1193 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1195 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1197 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1200 * debug.h (dwarf2out_do_eh_frame): Declare.
1201 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1202 (dwarf2out_do_frame): Use it.
1203 (dwarf2out_do_cfi_asm): Likewise.
1204 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1205 (dwarf2out_assembly_start): Likewise.
1206 (dwarf2out_begin_prologue): Fix comment.
1207 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1208 if the target needs either debug or unwind DWARF2 info.
1209 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1210 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1211 -ftrapping-math, -ftrapv and -fwrapv.
1213 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1215 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1216 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1217 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1218 (estimate_edge_size_and_time): Likewise.
1219 (ipa_merge_fn_summary_after_inlining): Likewise.
1220 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1221 (compute_uninlined_call_time): Use sreal_frequency.
1222 (compute_inlined_call_time): Likewise.
1223 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1224 * profile-count.c: Include sreal.h.
1225 (profile_count::to_sreal_scale): New.
1226 * profile-count.h: Forward declare sreal.
1227 (profile_count::to_sreal_scale): Declare.
1229 2017-11-13 Nathan Sidwell <nathan@acm.org>
1231 * diagnostic.c (maybe_line_and_column): New.
1232 (diagnostic_get_location_text): Use it.
1233 (diagnostic_report_current_module): Likewise.
1234 (test_diagnostic_get_location_text): Add tests.
1236 2017-11-13 Luis Machado <luis.machado@linaro.org>
1238 * doc/md.texi (Specifying processor pipeline description): Fix
1239 incorrect latency for the div instruction example.
1241 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1243 PR tree-optimization/78821
1244 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1245 that bit_not_p is the same.
1246 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1247 (split_group): Count precisely bit_not_p bits in each statement.
1248 (invert_op): New function.
1249 (imm_store_chain_info::output_merged_store): Use invert_op to
1250 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1251 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1253 2017-11-13 Martin Liska <mliska@suse.cz>
1255 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1256 (struct source_info): Likewise.
1257 (add_branch_counts): Likewise.
1258 (add_line_counts): Likewise.
1259 (function_summary): Likewise.
1260 (output_intermediate_line): Likewise.
1261 (generate_results): Likewise.
1263 2017-11-13 Martin Liska <mliska@suse.cz>
1265 * gcov.c (struct block_info): Remove typedef for block_t.
1266 (struct line_info): Likewise.
1267 (line_info::has_block): Likewise.
1268 (EXIT_BLOCK): Likewise.
1269 (unblock): Likewise.
1270 (circuit): Likewise.
1271 (get_cycles_count): Likewise.
1272 (process_file): Likewise.
1273 (read_graph_file): Likewise.
1274 (solve_flow_graph): Likewise.
1275 (find_exception_blocks): Likewise.
1276 (add_line_counts): Likewise.
1277 (accumulate_line_info): Likewise.
1278 (output_line_details): Likewise.
1280 2017-11-13 Martin Liska <mliska@suse.cz>
1282 * gcov.c (struct arc_info): Remove typedef for arc_t.
1283 (struct line_info): Likewise.
1284 (add_branch_counts): Likewise.
1285 (output_branch_count): Likewise.
1286 (function_info::~function_info): Likewise.
1287 (circuit): Likewise.
1288 (output_intermediate_line): Likewise.
1289 (read_graph_file): Likewise.
1290 (solve_flow_graph): Likewise.
1291 (find_exception_blocks): Likewise.
1292 (add_line_counts): Likewise.
1293 (accumulate_line_info): Likewise.
1294 (output_line_details): Likewise.
1295 (output_function_details): Likewise.
1297 2017-11-13 Martin Liska <mliska@suse.cz>
1299 * gcov.c (struct function_info): Remove typedef for function_t.
1300 (struct source_info): Likewise.
1301 (source_info::get_functions_at_location): Likewise.
1302 (solve_flow_graph): Likewise.
1303 (find_exception_blocks): Likewise.
1304 (add_line_counts): Likewise.
1305 (output_intermediate_file): Likewise.
1306 (process_file): Likewise.
1307 (generate_results): Likewise.
1308 (release_structures): Likewise.
1309 (read_graph_file): Likewise.
1310 (read_count_file): Likewise.
1311 (accumulate_line_counts): Likewise.
1312 (output_lines): Likewise.
1314 2017-11-13 Martin Liska <mliska@suse.cz>
1316 * gcov.c (function_info::function_info): Remove num_counts
1317 and add vector<gcov_type>.
1318 (function_info::~function_info): Use the vector.
1319 (process_file): Likewise.
1320 (read_graph_file): Likewise.
1321 (read_count_file): Likewise.
1322 (solve_flow_graph): Likewise.
1324 2017-11-13 Martin Liska <mliska@suse.cz>
1326 * gcov.c (function_info::is_artificial): New function.
1327 (process_file): Erase all artificial early.
1328 (generate_results): Skip as all artificial are already
1331 2017-11-13 Martin Liska <mliska@suse.cz>
1333 * gcov.c (read_graph_file): Store to global vector of functions.
1334 (read_count_file): Iterate the vector.
1335 (process_file): Likewise.
1336 (generate_results): Likewise.
1337 (release_structures): Likewise.
1339 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1341 PR tree-optimization/82954
1342 * gimple-ssa-store-merging.c
1343 (imm_store_chain_info::coalesce_immediate_stores): If
1344 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1346 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1348 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1349 Upddate call to ENDIAN_LANE_N.
1350 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1351 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1352 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1353 and use aarch64_endian_lane_rtx.
1354 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1356 2017-11-12 Tom de Vries <tom@codesourcery.com>
1358 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1360 2017-11-12 Tom de Vries <tom@codesourcery.com>
1362 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1364 2017-11-12 Tom de Vries <tom@codesourcery.com>
1366 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1368 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1369 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1370 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1372 2017-11-11 Martin Sebor <msebor@redhat.com>
1375 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1378 * prefic.c (translate_name): Replace strncpy with memcpy to
1379 avoid -Wstringop-truncation.
1381 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1383 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1385 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1387 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1389 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1391 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1394 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1396 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1397 enable generating XXBRH if the value is in a vector register.
1398 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1399 value is in a vector register.
1400 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1401 register to register bswap64's instead of doing the GPR sequence
1402 used on previous machines.
1403 (bswapdi2_xxbrd): New insn.
1404 (bswapdi2_reg): Disallow on ISA 3.0.
1405 (register to register bswap64 splitter): Do not split the insn on
1406 ISA 3.0 systems that use XXBRD.
1408 2017-11-10 Martin Sebor <msebor@redhat.com>
1411 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1413 (find_subframework_file): Same.
1415 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1417 * auto-profile.c (afdo_indirect_call): Drop frequency.
1418 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1419 (cgraph_node::create_edge): Drop frequency argument.
1420 (cgraph_node::create_indirect_edge): Drop frequency argument.
1421 (cgraph_edge::make_speculative): Drop frequency arguments.
1422 (cgraph_edge::resolve_speculation): Do not update frequencies
1423 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1424 (cgraph_node::dump): Check consistency in IPA mode.
1425 (cgraph_edge::maybe_hot_p): Use IPA counter.
1426 (cgraph_edge::verify_count_and_frequency): Rename to ...
1427 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1428 (cgraph_node::verify_node): Update.
1429 * cgraph.h (struct cgraph_edge): Drop frequency.
1430 (cgraph_edge::frequency): New function.
1431 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1433 (cgraph_edge::rebuild_edges): Likewise.
1434 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1435 (duplicate_thunk_for_node): Do not pass frequency.
1436 (cgraph_node::create_clone): Scale only counts.
1437 (cgraph_node::create_virtual_clone): Do not pass frequency.
1438 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1439 (cgraph_node::create_version_clone): Do not pass frequency.
1440 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1441 (cgraph_node::expand_thunk): Do not pass frequency.
1442 (cgraph_node::create_wrapper): Do not pass frequency.
1443 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1445 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1446 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1447 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1448 (gather_caller_stats): Use frequency function.
1449 (ipcp_cloning_candidate_p): Use frequency function.
1450 (ipcp_propagate_stage): Use frequency function.
1451 (get_info_about_necessary_edges): Use frequency function.
1452 (update_profiling_info): Update only IPA profile.
1453 (update_specialized_profile): Use frequency functoin.
1454 (perhaps_add_new_callers): Update only IPA profile.
1455 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1456 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1457 (dump_ipa_call_summary): Use frequency function.
1458 (estimate_edge_size_and_time): Use frequency function.
1459 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1460 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1461 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1462 (update_noncloned_counts): ... ths one; scale counts only.
1463 (clone_inlined_nodes): Do not scale frequency.
1464 (inline_call): Do not pass frequency.
1465 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1466 (compute_inlined_call_time): Use IPA profile.
1467 (want_inline_small_function_p): Use IPA profile.
1468 (want_inline_self_recursive_call_p): Use IPA profile.
1469 (edge_badness): Use IPA profile.
1470 (lookup_recursive_calls): Use IPA profile.
1471 (recursive_inlining): Do not pass frequency.
1472 (resolve_noninline_speculation): Do not update frequency.
1473 (inline_small_functions): Collect max of IPA profile.
1474 (dump_overall_stats): Dump IPA porfile.
1475 (dump_inline_stats): Dump IPA porfile.
1476 (ipa_inline): Collect IPA stats.
1477 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1478 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1479 (ipa_propagate_frequency): Use frequency function.
1480 (ipa_profile): Cleanup.
1481 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1482 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1483 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1484 (input_node): Do not stream frequency.
1485 (input_edge): Do not stream frequency.
1486 (merge_profile_summaries): Scale only IPA profiles.
1487 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1488 * predict.c (drop_profile): Do not recompute frequency.
1489 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1490 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1491 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1492 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1493 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1494 * tree-inline.c (copy_bb): Do not scale frequency.
1495 (expand_call_inline): Do not scale frequency.
1496 (tree_function_versioning): Do not scale frequency.
1497 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1499 2017-11-10 Julia Koval <julia.koval@intel.com>
1501 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1502 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1503 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1504 (_mm256_maskz_gf2p8affine_epi64_epi8)
1505 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1506 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1507 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1508 (__builtin_ia32_vgf2p8affineqb_v32qi)
1509 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1510 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1512 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1516 (arm_option_override): Refactor.
1517 (arm_option_reconfigure_globals): New.
1518 (arm_options_perform_arch_sanity_checks): New.
1519 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1521 (arm_options_perform_arch_sanity_checks): Likewise
1523 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1525 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1526 (power9-qpmul): New.
1527 * rs6000/rs6000.md ("type" attr): Add qmul.
1528 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1529 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1530 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1532 2017-11-10 Martin Sebor <msebor@redhat.com>
1535 * builtins.c (compute_objsize): Handle arrays that
1536 compute_builtin_object_size likes to fail for. Make extern.
1537 * builtins.h (compute_objsize): Declare.
1538 (check_strncpy_sizes): New function.
1539 (expand_builtin_strncpy): Call check_strncpy_sizes.
1540 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1541 -Wstringop-truncation.
1542 (gimple_fold_builtin_strncat): Same.
1543 * gimple.c (gimple_build_call_from_tree): Set call location.
1544 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1545 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1546 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1547 (handle_builtin_strlen): Use strlen_to_stridx.
1548 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1550 Use strlen_to_stridx.
1551 (pass_strlen::execute): Release strlen_to_stridx.
1552 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1553 (-Wstringop-truncation): Document new option.
1555 2017-11-10 Martin Liska <mliska@suse.cz>
1557 PR gcov-profile/82702
1558 * gcov.c (main): Handle intermediate files in a different
1560 (get_gcov_intermediate_filename): New function.
1561 (output_gcov_file): Remove support of intermediate files.
1562 (generate_results): Allocate intermediate file.
1563 (release_structures): Clean-up properly fn_end.
1564 (output_intermediate_file): Start iterating with line 1.
1566 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1568 PR tree-optimization/82929
1569 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1570 ops_swapped_p non-static data member.
1571 (store_immediate_info::store_immediate_info): Clear it.
1572 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1573 ops set ops_swapped_p.
1574 (count_multiple_uses): Handle ops_swapped_p.
1576 2017-11-10 Martin Liska <mliska@suse.cz>
1578 * coverage.c (coverage_init): Stream information about
1579 support of has_unexecuted_blocks.
1580 * doc/gcov.texi: Document that.
1581 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1582 * gcov.c (read_graph_file): Likewise.
1583 (output_line_beginning): Fix a small issue with
1586 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1588 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1589 reference of trivial component.
1591 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1594 * gimple-ssa-store-merging.c
1595 (pass_store_merging::terminate_all_aliasing_chains): For
1596 gimple_store_p stmts also call refs_output_dependent_p.
1598 PR rtl-optimization/82913
1599 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1602 2017-11-09 Jeff Law <law@redhat.com>
1604 * vr-values.h: New file with vr_values class.
1605 * tree-vrp.c: Include vr-values.h
1606 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1607 data objects into the vr_values class.
1608 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1609 (get_value_range): Make it a member function within vr_values class.
1610 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1611 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1612 (op_with_constant_singleton_value_range): Likewise.
1613 (extract_range_for_var_from_comparison_expr): Likewise.
1614 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1615 (extract_range_from_binary_expr): Likewise.
1616 (extract_range_from_unary_expr): Likewise.
1617 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1618 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1619 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1620 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1621 (compare_name_with_value, compare_names): Likewise.
1622 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1623 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1624 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1625 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1626 (extract_range_from_phi_node): Likewise.
1627 (simplify_truth_ops_using_ranges): Likewise.
1628 (simplify_div_or_mod_using_ranges): Likewise.
1629 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1630 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1631 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1632 (simplify_float_conversion_using_ranges): Likewise.
1633 (simplify_internal_call_using_ranges): Likewise.
1634 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1635 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1636 poitner via x_vr_values for calls into gimple folder.
1637 (vrp_initialize_lattice): Make this the vr_values ctor.
1638 (vrp_free_lattice): Make this the vr_values dtor.
1639 (set_vr_value): New function.
1640 (class vrp_prop): Add vr_values data member. Add various member
1641 functions as well as member functions that delegate to vr_values.
1642 (check_array_ref): Make a member function within vrp_prop class.
1643 (search_for_addr_array, vrp_initialize): Likewise.
1644 (vrp_finalize): Likewise. Revamp to avoid direct access to
1645 vr_value, values_propagated, etc.
1646 (check_array_bounds): Extract vrp_prop class instance pointer from
1647 walk info structure. Use it to call member functions.
1648 (check_all_array_refs): Make a member function within vrp_prop class.
1649 Smuggle class instance pointer via walk info structure.
1650 (x_vr_values): New local static.
1651 (vrp_valueize): Use x_vr_values to get class instance.
1652 (vr_valueize_1): Likewise.
1653 (class vrp_folder): Add vr_values data member. Add various member
1654 functions as well as member functions that delegate to vr_values.
1655 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1656 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1657 class instance from vr_values. Use it to call member functions.
1658 (vrp_dom_walker): Add vr_values data member.
1659 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1660 instance via x_vr_values.
1661 (identify_jump_threads): Accept vr_values as argument. Store
1662 it into the walker structure.
1663 (evrp_dom_walker): Add vr_values class data member. Add various
1665 (evrp_dom_walker::try_find_new_range): Use vr_values data
1666 member to access the memory allocator.
1667 (evrp_dom_walker::before_dom_children): Store vr_values class
1668 instance into the vrp_folder class.
1669 (evrp_dom_walker::push_value_range): Rework to avoid direct
1670 access to num_vr_values and vr_value.
1671 (evrp_dom_walker::pop_value_range): Likewise.
1672 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1673 Use vr_values to get to dump_all_value_ranges member function.
1674 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1675 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1676 Pass vr_values class instance down to identify_jump_threads.
1677 Remove call to vrp_free_lattice.
1678 (debug_all_value_ranges): Remove.
1680 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1681 (vrp_folder): Likewise.
1683 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1684 Get it from the existing bitmap instead.
1685 (vrp_intersect_ranges_1): Likewise.
1687 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1689 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1691 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1692 set bit_not_p to it.
1693 (imm_store_chain_info::coalesce_immediate_stores): Break group
1694 if bit_not_p is different.
1695 (count_multiple_uses, split_group,
1696 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1697 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1698 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1699 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1700 to store_immediate_info ctor.
1702 2017-11-09 Jim Wilson <jimw@sifive.com>
1704 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1705 (scan_prog_file): Likewise.
1707 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1709 * bb-reorder.c (max_entry_frequency): Remove.
1710 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1711 connect_traces, push_to_next_round_p): Remove prototypes.
1712 (find_traces_1_round): Use counts only.
1713 (push_to_next_round_p): Likewise.
1714 (find_traces): Likewise.
1715 (rotate_loop): Likewise.
1716 (find_traces_1_round): Likewise.
1717 (connect_traces): Likewise.
1718 (edge_order): Likewise.
1720 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1722 * config/arm/arm.c (output_return_instruction): Add comments to
1723 indicate requirement for cmse_nonsecure_entry return to account
1724 for the size of clearing instruction output here.
1725 (thumb_exit): Likewise.
1726 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1727 return in hardfloat mode.
1729 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1731 * config/rs6000/rs6000.c (machine_function): Add a bool,
1732 "toc_is_wrapped_separately".
1733 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1734 if it wasn't explicitly set or unset, we are optimizing for speed, and
1735 doing separate shrink-wrapping.
1736 (rs6000_get_separate_components): Enable the TOC component if
1737 saving the TOC register in the prologue.
1738 (rs6000_components_for_bb): Handle the TOC component.
1739 (rs6000_emit_prologue_components): Store the TOC register where needed.
1740 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1741 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1743 2017-11-09 Martin Jambor <mjambor@suse.cz>
1745 * ipa-param-manipulation.c: New file.
1746 * ipa-param-manipulation.h: Likewise.
1747 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1748 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1749 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1750 (ipa_parm_adjustment): Likewise.
1751 (ipa_parm_adjustment_vec): Likewise.
1752 (ipa_get_vector_of_formal_parms): Moved declaration to
1753 ipa-param-manipulation.h.
1754 (ipa_get_vector_of_formal_parm_types): Likewise.
1755 (ipa_modify_formal_parameters): Likewise.
1756 (ipa_modify_call_arguments): Likewise.
1757 (ipa_combine_adjustments): Likewise.
1758 (ipa_dump_param_adjustments): Likewise.
1759 (ipa_modify_expr): Likewise.
1760 (ipa_get_adjustment_candidate): Likewise.
1761 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1762 ipa-param-manipulation.c.
1763 (ipa_get_vector_of_formal_parm_types): Likewise.
1764 (ipa_modify_formal_parameters): Likewise.
1765 (ipa_modify_call_arguments): Likewise.
1766 (ipa_modify_expr): Likewise.
1767 (get_ssa_base_param): Likewise.
1768 (ipa_get_adjustment_candidate): Likewise.
1769 (index_in_adjustments_multiple_times_p): Likewise.
1770 (ipa_combine_adjustments): Likewise.
1771 (ipa_dump_param_adjustments): Likewise.
1772 * tree-sra.c: Also include ipa-param-manipulation.h
1773 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
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_masked_store): 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_align_stack_vars): 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_variable_length): 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_unaligned_possible): Document.
1800 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1801 Alan Hayward <alan.hayward@arm.com>
1802 David Sherwood <david.sherwood@arm.com>
1804 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1806 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1807 Alan Hayward <alan.hayward@arm.com>
1808 David Sherwood <david.sherwood@arm.com>
1810 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1811 previously undocumented selectors.
1812 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1814 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1816 * doc/rtl.texi (const_vector): Say that elements can be
1817 const_wide_ints too.
1818 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1819 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1820 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1821 * optabs.c (expand_vector_broadcast): Likewise.
1823 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1824 Alan Hayward <alan.hayward@arm.com>
1825 David Sherwood <david.sherwood@arm.com>
1827 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1828 scaled index even if the unscaled address was invalid.
1829 Don't increase the complexity of using a scale in that case.
1831 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1832 Alan Hayward <alan.hayward@arm.com>
1833 David Sherwood <david.sherwood@arm.com>
1835 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1836 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1837 UNITS_PER_WORD bytes.
1838 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1839 based on REGMODE_NATURAL_SIZE of the inner mode.
1840 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1841 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1842 * expmed.c (lowpart_bit_field_p): Divide the value up into
1843 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1844 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1845 whether something is likely to occupy more than one register.
1847 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1850 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1851 profile_count::adjust_for_ipa_scaling.
1852 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1853 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1855 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1857 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1859 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1860 (split_group): Add total_orig and total_new arguments, estimate the
1861 number of statements related to the store group without store merging
1862 and with store merging.
1863 (imm_store_chain_info::output_merged_store): Adjust split_group
1864 callers, punt if estimated number of statements with store merging
1865 is not smaller than estimated number of statements without it.
1867 (handled_load): Remove has_single_use checks.
1868 (pass_store_merging::process_store): Likewise.
1870 2017-11-09 Richard Biener <rguenther@suse.de>
1872 PR tree-optimization/82902
1873 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1875 2017-11-09 Martin Liska <mliska@suse.cz>
1878 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1881 2017-11-09 Martin Liska <mliska@suse.cz>
1883 PR tree-optimization/82669
1884 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1886 2017-11-09 Martin Liska <mliska@suse.cz>
1888 PR gcov-profile/48463
1889 * coverage.c (coverage_begin_function): Output also end locus
1890 of a function and information whether the function is
1892 * gcov-dump.c (tag_function): Parse and print the information.
1893 * gcov.c (INCLUDE_MAP): Add include.
1894 (INCLUDE_SET): Likewise.
1895 (struct line_info): Move earlier in the source file because
1896 of vector<line_info> in function_info structure.
1897 (line_info::line_info): Likewise.
1898 (line_info::has_block): Likewise.
1899 (struct source_info): Add new member index.
1900 (source_info::get_functions_at_location): New function.
1901 (function_info::group_line_p): New function.
1902 (output_intermediate_line): New function.
1903 (output_intermediate_file): Use the mentioned function.
1904 (struct function_start): New.
1905 (struct function_start_pair_hash): Likewise.
1906 (process_file): Add code that identifies group functions.
1907 Assign lines either to global or function scope.
1908 (generate_results): Skip artificial functions.
1909 (find_source): Assign index for each source file.
1910 (read_graph_file): Read new flag artificial and end_line.
1911 (add_line_counts): Assign it either to global of function scope.
1912 (accumulate_line_counts): Isolate core of the function to
1913 accumulate_line_info and call it for both function and global
1915 (accumulate_line_info): New function.
1916 (output_line_beginning): Fix GNU coding style.
1917 (print_source_line): New function.
1918 (output_line_details): Likewise.
1919 (output_function_details): Likewise.
1920 (output_lines): Iterate both source (global) scope and function
1922 (struct function_line_start_cmp): New class.
1923 * doc/gcov.texi: Reflect changes in documentation.
1925 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1928 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1929 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1930 and similarly for not instead of neg.
1932 2017-11-08 Andi Kleen <ak@linux.intel.com>
1934 * config/i386/i386.opt: Add -mforce-indirect-call.
1935 * config/i386/predicates.md: Check for flag_force_indirect_call.
1936 * doc/invoke.texi: Document -mforce-indirect-call
1938 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1940 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1942 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1943 config/riscv/riscv.c (predict.h): New include.
1944 (riscv_slow_unaligned_access_p): No longer static.
1945 (riscv_block_move_straight): Add require.
1946 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1948 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1951 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1952 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1953 nonimmediate_operand predicate for operand 1 instead of
1956 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1958 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1960 * config/aarch64/constraints.md (Uml): New constraint.
1961 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1964 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1966 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1967 of two vec_duplicates into a vec_concat.
1969 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1971 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1972 Simplify vec_merge of vec_duplicate and vec_concat.
1973 * config/aarch64/constraints.md (Utq): New constraint.
1974 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1977 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1979 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1980 Simplify vec_merge of vec_duplicate and const_vector.
1981 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1983 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1984 mode iterator. Update predicate on operand 1 to
1985 handle non-const_vec constants. Delete constraints.
1986 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1988 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1990 PR tree-optimization/78821
1991 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1993 (store_operand_info::store_operand_info): Initialize it to false.
1994 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1995 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1996 store in the group, and if chain_info is non-NULL, to ignore altogether
1998 (compatible_load_p): Fail if bit_not_p does not match.
1999 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
2000 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
2001 (pass_store_merging::process_store): Adjust
2002 terminate_all_aliasing_chains calls to pass NULL in all current spots,
2003 call terminate_all_aliasing_chains newly when adding a store into
2004 a chain with non-NULL chain_info.
2006 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2008 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
2010 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
2012 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2014 (aarch64_layout_frame): Initialise emit_frame_chain.
2015 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
2016 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
2018 2017-11-08 Martin Liska <mliska@suse.cz>
2020 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
2021 of gimple_call_internal_p.
2023 2017-11-07 Tom de Vries <tom@codesourcery.com>
2025 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
2027 2017-11-07 Tom de Vries <tom@codesourcery.com>
2029 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2032 2017-11-08 Martin Liska <mliska@suse.cz>
2035 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
2037 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2039 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
2041 2017-11-07 Jakub Jelinek <jakub@redhat.com>
2044 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
2045 (*cmp<mode>_ccz_1): New insn with $k alternative.
2048 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
2049 RTX_COMM_COMPARE as commutative as well.
2050 (ix86_binary_operator_ok): Formatting fix.
2051 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
2052 *<code><mode>3<mask_name><round_saeonly_name>,
2053 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
2054 *<s>mul<mode>3_highpart<mask_name>,
2055 *vec_widen_umult_even_v16si<mask_name>,
2056 *vec_widen_umult_even_v8si<mask_name>,
2057 *vec_widen_umult_even_v4si<mask_name>,
2058 *vec_widen_smult_even_v16si<mask_name>,
2059 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
2060 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
2061 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
2062 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
2063 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
2064 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
2065 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
2066 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
2067 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
2068 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
2069 ix86_binary_operator_ok. Formatting fixes.
2070 (*<plusminus_insn><mode>3<mask_name><round_name>,
2071 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
2074 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2076 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
2077 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
2078 eq and ne if TARGET_ISEL.
2079 (cmp): New code_iterator.
2080 (UNS, UNSU_, UNSIK): New code_attrs.
2081 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
2082 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
2084 ("eq<mode>3"): ... this.
2085 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
2087 ("ne<mode>3"): ... this.
2089 2017-11-07 Julia Koval <julia.koval@intel.com>
2092 * common/config/i386/i386-common.c
2093 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
2094 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
2095 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
2096 * config/i386/i386.opt: Ditto.
2097 * config/i386/i386.c (ix86_target_string): Ditto.
2098 (ix86_option_override_internal): Ditto.
2099 (ix86_init_mpx_builtins): Move MPX to args2.
2100 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2101 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2102 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2103 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2104 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2105 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2106 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2108 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2111 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2112 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2113 (zero-extendsidi peephole2): Remove peephole.
2115 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2118 * stor-layout.c: Include attribs.h.
2119 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2120 explicit lookup of "aligned" attribute.
2122 2017-11-07 Andrew Waterman <andrew@sifive.com>
2124 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2125 (riscv_expand_block_move): Likewise.
2126 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2128 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2129 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2130 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2131 (riscv_adjust_block_mem): Likewise.
2132 (riscv_block_move_loop): Likewise.
2133 (riscv_expand_block_move): Likewise.
2134 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2136 2017-11-07 Michael Clark <michaeljclark@mac.com>
2138 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2139 (MUSL_DYNAMIC_LINKER): Likewise.
2141 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2143 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2144 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2147 2017-11-07 Richard Biener <rguenther@suse.de>
2149 * match.pd: Fix build.
2151 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2152 Jackson Woodruff <jackson.woodruff@arm.com>
2154 PR tree-optimization/71026
2155 * match.pd: Canonicalize negate in division.
2157 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2160 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2162 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2164 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2165 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2167 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2169 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2170 non-scalar integral types.
2171 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2172 (-(A-B), -(~A)): New transformations.
2174 2017-11-07 Tom de Vries <tom@codesourcery.com>
2176 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2177 semicolon after "do {} while (0)".
2178 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2179 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2180 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2181 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2182 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2184 2017-11-07 Tom de Vries <tom@codesourcery.com>
2186 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2187 after "do {} while (0)".
2188 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2189 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2190 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2191 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2192 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2194 2017-11-07 Tom de Vries <tom@codesourcery.com>
2197 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2199 (arm_rtx_costs_internal): Add missing semicolon after
2200 HANDLE_NARROW_SHIFT_ARITH call.
2202 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2204 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2205 disable isel if it was not set explicitly.
2207 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2209 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2211 (add_type_attribute) likewise.
2213 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2215 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2216 of struct ix86_frame.
2217 (ix86_initial_elimination_offset): Likewise.
2218 (ix86_expand_split_stack_prologue): Likewise.
2220 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2222 * tree-vrp.h (enum value_range_type): Update stale comment.
2224 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2225 Alan Hayward <alan.hayward@arm.com>
2226 David Sherwood <david.sherwood@arm.com>
2228 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2229 (aarch64_expand_vec_perm_const): Take the number of units too.
2230 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2231 (aarch64_expand_vec_perm_const): Likewise.
2232 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2233 (vec_perm<mode>): Update accordingly.
2235 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2236 Alan Hayward <alan.hayward@arm.com>
2237 David Sherwood <david.sherwood@arm.com>
2239 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2240 Take the number of units too.
2241 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2242 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2243 but check for a vector mode before rather than after the call.
2244 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2245 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2246 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2247 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2248 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2249 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2250 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2251 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2252 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2253 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2254 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2255 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2256 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2257 (aarch64_sqdmull2_n<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_reverse_mask): Take
2264 the number of units too.
2265 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2266 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2267 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2268 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2269 (vec_store_lanesxi<mode>): Update accordingly.
2271 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2276 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2277 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2278 of units rather than the mode.
2279 * config/aarch64/iterators.md (nunits): New mode attribute.
2280 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2281 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2282 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2283 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2284 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2285 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2286 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2287 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2288 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2289 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2290 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2291 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2292 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2293 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2294 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2295 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2296 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2297 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2298 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2299 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2300 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2301 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2302 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2303 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2304 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2305 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2306 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2307 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2308 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2309 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2310 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2311 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2312 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2313 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2314 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2315 (aarch64_simd_vec_setv2di): Likewise.
2317 2017-11-06 Carl Love <cel@us.ibm.com>
2319 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2321 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2322 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2324 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2326 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2327 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2328 Add power 8 macro expansions.
2329 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2330 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2331 power 8 instructions. (VSX_XXBR): Add iterator.
2333 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2335 * config/arm/arm.md (predicable_short_it): Change default to "no",
2336 improve documentation, remove uses that are identical to the default.
2337 (enabled_for_depr_it): Rename to enabled_for_short_it.
2338 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2339 Remove default uses.
2340 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2341 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2342 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2343 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2345 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2348 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2349 float128 helper macros, which are no longer used after deleting
2350 the old 'q' built-in functions, and moving the round to odd
2351 built-in functions to being special built-in functions.
2352 (BU_FLOAT128_2): Likewise.
2353 (BU_FLOAT128_1_HW): Likewise.
2354 (BU_FLOAT128_2_HW): Likewise.
2355 (BU_FLOAT128_3_HW): Likewise.
2356 (FABSQ): Delete old 'q' built-in functions.
2357 (COPYSIGNQ): Likewise.
2358 (SQRTF128_ODD): Move round to odd built-in functions to be
2359 special built-in functions, so that we can handle
2360 -mabi=ieeelongdouble.
2361 (TRUNCF128_ODD): Likewise.
2362 (ADDF128_ODD): Likewise.
2363 (SUBF128_ODD): Likewise.
2364 (MULF128_ODD): Likewise.
2365 (DIVF128_ODD): Likewise.
2366 (FMAF128_ODD): Likewise.
2367 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2368 built-in names to 'f128'.
2369 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2370 old 'q' built-in functions, as the machine independent code for
2371 'f128' built-in functions handles this.
2372 (rs6000_expand_builtin): Add expansion for float128 round to odd
2373 functions, keying off on -mabi=ieeelongdouble of whether to use
2374 the KFmode or TFmode variant.
2375 (rs6000_init_builtins): Initialize the _Float128 round to odd
2377 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2378 _Float128 'q' built-in functions are now mapped into the new
2379 'f128' built-in functions.
2381 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2383 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2384 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2386 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2389 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2390 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2391 * toplev.c: Include "ipa-fnsummary.h".
2392 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2394 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2396 PR tree-optimization/82838
2397 * gimple-ssa-store-merging.c
2398 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2399 on a separate gimple_seq which is then appended to seq.
2401 2017-11-06 Jeff Law <law@redhat.com>
2404 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2405 (get_probe_interval): New functions.
2406 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2407 (ix86_adjust_stack_and_probe): Likewise.
2408 (output_adjust_stack_and_probe): Likewise.
2409 (ix86_emit_probe_stack_range): Likewise.
2410 (ix86_expand_prologue): Likewise.
2412 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2414 PR tree-optimization/82816
2415 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2416 if the modes of the two types are the same.
2417 (convert_plusminus_to_widen): Likewise.
2419 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2421 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2423 (usadv16qi): New define_expand.
2424 (usadv8hi): New define_expand.
2426 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2429 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2431 (inline_transform): Likewise.
2432 * predict.c (counts_to_freqs): Remove useless conditional.
2433 * profile-count.h (profile_count::apply_scale): Move sanity check.
2434 * tree-inline.c (copy_bb): Always scale.
2435 (copy_cfg_body): Likewise.
2437 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2440 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2442 (*cmp_ite0): Add enabled_for_depr_it attribute.
2443 (*cmp_ite1): Likewise.
2445 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2447 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2450 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2452 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2453 return the value by pointer.
2454 (extract_range_from_multiplicative_op_1): Update accordingly.
2455 Return as soon as an operation fails.
2457 2017-11-05 Tom de Vries <tom@codesourcery.com>
2460 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2461 (DEF_SANITIZER_BUILTIN): ... here.
2462 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2463 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2465 2017-11-05 Tom de Vries <tom@codesourcery.com>
2468 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2470 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2471 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2472 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2474 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2475 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2476 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2478 2017-11-05 Tom de Vries <tom@codesourcery.com>
2481 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2484 2017-11-04 Michael Clark <michaeljclark@mac.com>
2486 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2487 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2489 (*addsi3_extended): Likewise.
2490 (*addsi3_extended2): Likewise.
2491 (<optab>si3): Likewise.
2492 (<optab>di3): Likewise.
2493 (<optab><mode>3): Likewise.
2494 (<*optabe>si3_internal): Likewise.
2495 (zero_extendqi<SUPERQI:mode>2): Likewise.
2496 (*add<mode>hi3): Likewise.
2497 (*xor<mode>hi3): Likewise.
2498 (<optab>di3): Likewise.
2499 (*<optab>si3_extend): Likewise.
2500 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2501 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2502 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2504 2017-11-04 Andrew Waterman <andrew@sifive.com>
2506 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2507 TARGET_STRICT_ALIGN based upon -mtune argument.
2509 2017-11-04 Andrew Waterman <andrew@sifive.com>
2511 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2513 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2515 config/i386/i386.c (choose_basereg): Use optional scratch
2516 register and add assertion.
2517 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2518 needed, and don't allocate stack.
2519 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2520 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2521 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2523 2017-11-03 Jeff Law <law@redhat.com>
2525 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2526 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2527 to probe at the start of a noreturn function.
2529 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2531 PR tree-optimization/78821
2532 * gimple-ssa-store-merging.c: Update the file comment.
2533 (MAX_STORE_ALIAS_CHECKS): Define.
2534 (struct store_operand_info): New type.
2535 (store_operand_info::store_operand_info): New constructor.
2536 (struct store_immediate_info): Add rhs_code and ops data members.
2537 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2538 and op1r arguments to the ctor, initialize corresponding data members.
2539 (struct merged_store_group): Add load_align_base and load_align
2541 (merged_store_group::merged_store_group): Initialize them.
2542 (merged_store_group::do_merge): Update them.
2543 (merged_store_group::apply_stores): Pick the constant for
2544 encode_tree_to_bitpos from one of the two operands, or skip
2545 encode_tree_to_bitpos if neither operand is a constant.
2546 (class pass_store_merging): Add process_store method decl. Remove
2547 bool argument from terminate_all_aliasing_chains method decl.
2548 (pass_store_merging::terminate_all_aliasing_chains): Remove
2549 var_offset_p argument and corresponding handling.
2550 (stmts_may_clobber_ref_p): New function.
2551 (compatible_load_p): New function.
2552 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2553 if there is overlap and rhs_code is not INTEGER_CST. For
2554 non-overlapping stores terminate group if rhs is not mergeable.
2555 (get_alias_type_for_stmts): Change first argument from
2556 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2557 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2558 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2560 (get_location_for_stmts): Change first argument from
2561 auto_vec<gimple *> & to vec<gimple *> &.
2562 (struct split_store): Remove orig_stmts data member, add orig_stores.
2563 (split_store::split_store): Create orig_stores rather than orig_stmts.
2564 (find_constituent_stmts): Renamed to ...
2565 (find_constituent_stores): ... this. Change second argument from
2566 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2567 to info structures rather than the statements.
2568 (split_group): Rename ALLOW_UNALIGNED argument to
2569 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2570 it. Adjust find_constituent_stores caller.
2571 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2572 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2573 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2574 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2575 (mem_valid_for_store_merging): New function.
2576 (handled_load): New function.
2577 (pass_store_merging::process_store): New method.
2578 (pass_store_merging::execute): Use process_store method. Adjust
2579 terminate_all_aliasing_chains caller.
2581 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2583 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2584 Return true for more constants, symbols and label references.
2585 (aarch64_valid_floating_const): Remove unused function.
2587 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2590 * config/aarch64/aarch64.c (aarch64_layout_frame):
2591 Undo forcing of LR at bottom of frame.
2593 2017-11-03 Jeff Law <law@redhat.com>
2596 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2597 for int_registers_saved.
2599 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2600 extracted from tree-ssa-dom.c.
2601 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2602 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2603 (record_equivalences_from_incoming_edge): Add additional argument
2604 to single_pred_edge_ignoring_loop_edges call.
2605 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2606 (uncprop_dom_walker::before_dom_children): Add additional argument
2607 to single_pred_edge_ignoring_loop_edges call.
2608 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2609 single_pred_edge_ignoring_loop_edges rather than open coding.
2610 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2612 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2614 * match.pd (-(-A)): Rewrite.
2616 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2618 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2619 (rs6000_emit_int_cmove): New declaration.
2620 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2621 (rs6000_emit_sISEL): Delete.
2622 (rs6000_emit_int_cmove): Make non-static.
2623 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2624 instead of rs6000_emit_sISEL.
2626 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2628 * asan.c (create_cond_insert_point): Maintain profile.
2629 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2631 * basic-block.h (struct basic_block_def): Remove frequency.
2632 (EDGE_FREQUENCY): Use to_frequency
2633 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2635 (find_traces): Update to use to_frequency.
2636 (find_traces_1_round): Likewise; use only IPA counts.
2637 (bb_to_key): Likewise.
2638 (connect_traces): Use IPA counts only.
2639 (copy_bb_p): Update to use to_frequency.
2640 (fix_up_crossing_landing_pad): Likewise.
2641 (sanitize_hot_paths): Likewise.
2642 * bt-load.c (basic_block_freq): Likewise.
2643 * cfg.c (init_flow): Set count_max to uninitialized.
2644 (check_bb_profile): Remove frequencies; check counts.
2645 (dump_bb_info): Do not dump frequencies.
2646 (update_bb_profile_for_threading): Update counts only.
2647 (scale_bbs_frequencies_int): Likewise.
2648 (MAX_SAFE_MULTIPLIER): Remove.
2649 (scale_bbs_frequencies_gcov_type): Update counts only.
2650 (scale_bbs_frequencies_profile_count): Update counts only.
2651 (scale_bbs_frequencies): Update counts only.
2652 * cfg.h (struct control_flow_graph): Add count-max.
2653 (update_bb_profile_for_threading): Update prototype.
2654 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2655 (find_many_sub_basic_blocks): Likewise.
2656 * cfgcleanup.c (try_forward_edges): Likewise.
2657 (try_crossjump_to_edge): Likewise.
2658 * cfgexpand.c (expand_gimple_cond): Likewise.
2659 (expand_gimple_tailcall): Likewise.
2660 (construct_init_block): Likewise.
2661 (construct_exit_block): Likewise.
2662 * cfghooks.c (verify_flow_info): Check consistency of counts.
2663 (dump_bb_for_graph): Do not dump frequencies.
2664 (split_block_1): Do not update frequencies.
2665 (split_edge): Do not update frequencies.
2666 (make_forwarder_block): Do not update frequencies.
2667 (duplicate_block): Do not update frequencies.
2668 (account_profile_record): Do not update frequencies.
2669 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2670 for global heuristics.
2671 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2672 (expected_loop_iterations_unbounded): Use counts only.
2673 * cfgloopmanip.c (scale_loop_profile): Simplify.
2674 (create_empty_loop_on_edge): Simplify
2676 (duplicate_loop_to_header_edge): Simplify
2677 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2678 (update_br_prob_note): Take care of removing note when profile
2680 (relink_block_chain): Do not dump frequency.
2681 (rtl_account_profile_record): Use to_frequency.
2682 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2683 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2684 (cgraph_update_edges_for_call_stmt_node): Likewise.
2685 (cgraph_edge::verify_count_and_frequency): Update.
2686 (cgraph_node::verify_node): Temporarily disable frequency verification.
2687 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2688 to_cgraph_frequency.
2689 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2690 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2692 (cgraph_node::expand_thunk): Update profile.
2693 * except.c (dw2_build_landing_pads): Do not update frequency.
2694 * final.c (compute_alignments): Use to_frequency.
2695 (dump_basic_block_info): Do not dump frequency.
2696 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2697 (dump_gimple_bb_header): Do not dump frequency.
2698 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2700 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2701 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2702 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2703 (init_before_recovery): Do not update frequency.
2704 (sched_create_recovery_edges): Do not update frequency.
2705 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2706 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2707 (ipa_cp_c_finalize): Set max_count to uninitialized.
2708 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2709 (param_change_prob): Use counts.
2710 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2712 * ipa-split.c (consider_split): Use to_frequency.
2713 (split_function): Use to_frequency.
2714 * ira-build.c (loop_compare_func): Likewise.
2715 (mark_loops_for_removal): Likewise.
2716 (mark_all_loops_for_removal): Likewise.
2717 * loop-doloop.c (doloop_modify): Do not update frequency.
2718 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2720 * lto-streamer-in.c (input_function): Update count_max.
2721 * omp-expand.c (expand_omp_taskreg): Update count_max.
2722 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2723 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2724 (maybe_hot_count_p): Use ipa counts only.
2725 (maybe_hot_bb_p): Simplify.
2726 (maybe_hot_edge_p): Simplify.
2727 (probably_never_executed): Do not take frequency argument.
2728 (probably_never_executed_bb_p): Do not pass frequency.
2729 (probably_never_executed_edge_p): Likewise.
2730 (combine_predictions_for_bb): Check that profile is nonzero.
2731 (propagate_freq): Do not set frequency.
2732 (drop_profile): Simplify.
2733 (counts_to_freqs): Simplify.
2734 (expensive_function_p): Use to_frequency.
2735 (propagate_unlikely_bbs_forward): Simplify.
2736 (determine_unlikely_bbs): Simplify.
2737 (estimate_bb_frequencies): Add hack to silence graphite issues.
2738 (compute_function_frequency): Use ipa counts.
2739 (pass_profile::execute): Update.
2740 (rebuild_frequencies): Use counts only.
2741 (force_edge_cold): Use counts only.
2742 * profile-count.c (profile_count::dump): Dump new count types.
2743 (profile_count::differs_from_p): Check compatiblity.
2744 (profile_count::to_frequency): New function.
2745 (profile_count::to_cgraph_frequency): New function.
2746 * profile-count.h (struct function): Declare.
2747 (enum profile_quality): Add profile_guessed_local and
2748 profile_guessed_global0.
2749 (class profile_proability): Decrease number of bits to 29;
2750 update from_reg_br_prob_note and to_reg_br_prob_note.
2751 (class profile_count: Update comment; decrease number of bits
2752 to 61. Check compatibility.
2753 (profile_count::compatible_p): New private member function.
2754 (profile_count::ipa_p): New member function.
2755 (profile_count::operator<): Handle global zero correctly.
2756 (profile_count::operator>): Handle global zero correctly.
2757 (profile_count::operator<=): Handle global zero correctly.
2758 (profile_count::operator>=): Handle global zero correctly.
2759 (profile_count::nonzero_p): New member function.
2760 (profile_count::force_nonzero): New member function.
2761 (profile_count::max): New member function.
2762 (profile_count::apply_scale): Handle IPA scalling.
2763 (profile_count::guessed_local): New member function.
2764 (profile_count::global0): New member function.
2765 (profile_count::ipa): New member function.
2766 (profile_count::to_frequency): Declare.
2767 (profile_count::to_cgraph_frequency): Declare.
2768 * profile.c (OVERLAP_BASE): Delete.
2769 (compute_frequency_overlap): Delete.
2770 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2771 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2772 * sched-ebb.c (rank): Use counts only.
2773 * shrink-wrap.c (handle_simple_exit): Use counts only.
2774 (try_shrink_wrapping): Use counts only.
2775 (place_prologue_for_one_component): Use counts only.
2776 * tracer.c (find_best_predecessor): Use to_frequency.
2777 (find_trace): Use to_frequency.
2778 (tail_duplicate): Use to_frequency.
2779 * trans-mem.c (expand_transaction): Do not update frequency.
2780 * tree-call-cdce.c: Do not update frequency.
2781 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2782 (gimple_merge_blocks): Likewise.
2783 (gimple_split_edge): Likewise.
2784 (gimple_duplicate_sese_region): Likewise.
2785 (gimple_duplicate_sese_tail): Likewise.
2786 (move_sese_region_to_fn): Likewise.
2787 (gimple_account_profile_record): Likewise.
2788 (insert_cond_bb): Likewise.
2789 * tree-complex.c (expand_complex_div_wide): Likewise.
2790 * tree-eh.c (lower_resx): Update profile.
2791 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2793 (initialize_cfun): Do not initialize frequencies
2794 (freqs_to_counts): Delete.
2795 (copy_cfg_body): Ignore count parameter.
2796 (copy_body): Update.
2797 (expand_call_inline): Update count_max.
2798 (optimize_inline_calls): Update count_max.
2799 (tree_function_versioning): Update count_max.
2800 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2801 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2803 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2804 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2805 (try_peel_loop): Likewise.
2806 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2808 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2809 (tree_transform_and_unroll_loop): Do not use frequencies
2810 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2811 Use reliable prediction only.
2812 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2813 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2814 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2815 probability scaling.
2816 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2817 not update frequency
2818 (any_remaining_duplicated_blocks): Likewise.
2819 (update_profile): Likewise.
2820 (estimated_freqs_path): Delete.
2821 (freqs_to_counts_path): Delete.
2822 (clear_counts_path): Delete.
2823 (ssa_fix_duplicate_block_edges): Likewise.
2824 (duplicate_thread_path): Likewise.
2825 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2826 * tree-tailcall.c (decrease_profile): Do not update frequency.
2827 (eliminate_tail_call): Likewise.
2828 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2829 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2830 (optimize_mask_stores): Likewise.
2831 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2832 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2833 (ubsan_expand_ptr_ifn): Update profile.
2834 * value-prof.c (gimple_ic): Simplify.
2835 * value-prof.h (gimple_ic): Update prototype.
2836 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2837 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2839 (want_inline_self_recursive_call_p): Likewise.
2840 (resolve_noninline_speculation): Only cummulate defined counts.
2841 (inline_small_functions): Use nonzero_p.
2842 (ipa_inline): Do not access freed node.
2844 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2846 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2847 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2849 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2851 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2852 non-legitimate address.
2854 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2856 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2857 (*lt0_<mode>di, *lt0_<mode>si): New.
2859 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2861 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2862 TARGET_PAIRED_FLOAT.
2864 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2865 Jim Wilson <jim.wilson@linaro.org>
2867 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2868 * config/aarch64/aarch64-tune.md: Regenerated.
2869 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2870 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2872 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2874 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2875 (arc_expand_prologue): Restore blink for millicode.
2876 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2878 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2881 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2882 gen_vec_duplicate after forcing the scalar into a register.
2884 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2886 * combine (try_combine): Print the insns input to try_combine to the
2889 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2892 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2893 Remove second argument from aarch64_process_target_attr call.
2894 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2896 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2898 (aarch64_handle_attr_arch): Remove second argument.
2899 (aarch64_handle_attr_cpu): Ditto.
2900 (aarch64_handle_attr_tune): Ditto.
2901 (aarch64_handle_attr_isa_flags): Ditto.
2902 (aarch64_process_one_target_attr): Ditto.
2903 (aarch64_process_target_attr): Ditto.
2904 (aarch64_option_valid_attribute_p): Remove second argument.
2905 on aarch64_process_target_attr call.
2907 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2909 * diagnostic.c: Include "selftest-diagnostic.h".
2910 (selftest::assert_location_text): New function.
2911 (selftest::test_diagnostic_get_location_text): New function.
2912 (selftest::diagnostic_c_tests): Call it.
2914 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2916 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2917 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2918 (class selftest::test_diagnostic_context): Move to...
2919 * selftest-diagnostic.c: New file.
2920 * selftest-diagnostic.h: New file.
2922 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2924 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2925 offset range for FT32B.
2926 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2927 * config/ft32/ft32.md: Add TARGET_NOPM.
2928 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2929 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2931 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2933 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2935 2017-11-02 Jeff Law <law@redhat.com>
2937 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2938 virtual keyword on FINAL OVERRIDE members.
2940 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2941 virtuals together. Add virtual destructor.
2942 (substitute_and_fold_engine): Similarly.
2944 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2946 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2948 2017-11-02 Richard Biener <rguenther@suse.de>
2950 PR tree-optimization/82795
2951 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2953 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2955 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2957 (gcc_GAS_CHECK_FEATURE): Remove.
2958 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2959 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2960 * configure: Regenerate.
2962 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2964 * config/arc/arc.c (hwloop_optimize): Account for empty
2967 2017-11-02 Richard Biener <rguenther@suse.de>
2970 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2971 Truncate ARRAY_REF index and element size.
2973 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2975 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2977 2017-11-01 Jeff Law <law@redhat.com>
2979 * tree-ssa-ccp.c (ccp_folder): New class derived from
2980 substitute_and_fold_engine.
2981 (ccp_folder::get_value): New member function.
2982 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2983 (ccp_fold_stmt): Remove prototype.
2984 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2985 * tree-ssa-copy.c (copy_folder): New class derived from
2986 substitute_and_fold_engine.
2987 (copy_folder::get_value): Renamed from get_value.
2988 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2989 * tree-vrp.c (vrp_folder): New class derived from
2990 substitute_and_fold_engine.
2991 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2992 (vrp_folder::get_value): New member function.
2993 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2994 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2995 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2996 provide a class interface to folder/substitute routines.
2997 (ssa_prop_fold_stmt_fn): Remove typedef.
2998 (ssa_prop_get_value_fn): Likewise.
2999 (subsitute_and_fold): Remove prototype.
3000 (replace_uses_in): Likewise.
3001 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
3002 Renamed from replace_uses_in. Call the virtual member function
3003 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
3004 (substitute_and_fold_dom_walker): Remove initialization of
3005 data member entries for calbacks. Add substitute_and_fold_engine
3006 member and initialize it.
3007 (substitute_and_fold_dom_walker::before_dom_children0: Use the
3008 member functions for get_value, replace_phi_args_in c
3009 replace_uses_in, and fold_stmt calls.
3010 (substitute_and_fold_engine::substitute_and_fold): Renamed from
3011 substitute_and_fold. Remove assert. Update ctor call.
3013 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
3014 (ssa_prop_visit_phi_fn): Likewise.
3015 (class ssa_propagation_engine): New class to provide an interface
3017 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
3019 (ssa_prop_visit_phi): Likewise.
3020 (ssa_propagation_engine::simulate_stmt): Moved into class.
3021 Call visit_phi/visit_stmt from the class rather than via
3022 file scoped static variables.
3023 (ssa_propagation_engine::simulate_block): Moved into class.
3024 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
3025 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
3026 set file scoped statics for the visit_stmt/visit_phi callbacks.
3027 * tree-complex.c (complex_propagate): New class derived from
3028 ssa_propagation_engine.
3029 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
3030 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
3031 (tree_lower_complex): Call ssa_propagate via the complex_propagate
3033 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
3034 ssa_propagation_engine.
3035 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
3036 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
3037 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
3038 * tree-ssa-copy.c (copy_prop): New class derived from
3039 ssa_propagation_engine.
3040 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
3041 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
3042 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
3043 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
3044 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
3045 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
3046 (execute_vrp): Call ssa_propagate from the vrp_prop class.
3048 2017-11-01 Jakub Jelinek <jakub@redhat.com>
3050 PR rtl-optimization/82778
3051 PR rtl-optimization/82597
3052 * compare-elim.c (struct comparison): Add in_a_setter field.
3053 (find_comparison_dom_walker::before_dom_children): Remove killed
3054 bitmap and df_simulate_find_defs call, instead walk the defs.
3055 Compute last_setter and initialize in_a_setter. Merge definitions
3056 with first initialization for a few variables.
3057 (try_validate_parallel): Use insn_invalid_p instead of
3058 recog_memoized. Return insn rather than just the pattern.
3059 (try_merge_compare): Fix up comment. Don't uselessly test if
3060 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
3062 (execute_compare_elim_after_reload): Remove df_chain_add_problem
3065 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3066 Alan Hayward <alan.hayward@arm.com>
3067 David Sherwood <david.sherwood@arm.com>
3069 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
3070 aarch64_hard_regno_nregs to get the number of registers
3073 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3074 Alan Hayward <alan.hayward@arm.com>
3075 David Sherwood <david.sherwood@arm.com>
3077 * config/aarch64/constraints.md (Upl): Rename to...
3079 * config/aarch64/aarch64.md
3080 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
3083 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3084 Alan Hayward <alan.hayward@arm.com>
3085 David Sherwood <david.sherwood@arm.com>
3087 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
3088 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
3091 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3092 Alan Hayward <alan.hayward@arm.com>
3093 David Sherwood <david.sherwood@arm.com>
3095 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
3096 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
3097 (aarch64_evpc_dup): Generate rtl direcly, rather than using
3099 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
3100 of a single element.
3101 * config/aarch64/iterators.md: Add a comment above the permute
3102 unspecs to say that they are generated directly by
3103 aarch64_expand_vec_perm_const.
3104 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3106 2017-11-01 Nathan Sidwell <nathan@acm.org>
3108 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3110 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3112 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3113 and medany code models, and describe what they do.
3115 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3117 Revert accidental duplicate:
3119 * combine.c (can_change_dest_mode): Reject changes in
3120 REGMODE_NATURAL_SIZE.
3122 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3124 PR rtl-optimization/64682
3125 PR rtl-optimization/69567
3126 PR rtl-optimization/69737
3127 PR rtl-optimization/82683
3128 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3129 register mentioned in the note, drop the note, unless it came from I3,
3130 in which case it should go to I3 again.
3132 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3134 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3135 and return false if not.
3136 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3138 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3140 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3141 range is known to be empty.
3143 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3144 Alan Hayward <alan.hayward@arm.com>
3145 David Sherwood <david.sherwood@arm.com>
3147 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3148 and CONST_VECTOR_NUNITS instead of computing the number of units from
3149 the byte sizes of the vector and element.
3150 (simplify_binary_operation_1): Likewise.
3151 (simplify_const_binary_operation): Likewise.
3152 (simplify_ternary_operation): Likewise.
3154 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3155 Alan Hayward <alan.hayward@arm.com>
3156 David Sherwood <david.sherwood@arm.com>
3158 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3159 (int_mem_offset): ...this new function.
3160 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3161 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3162 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3165 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3166 Alan Hayward <alan.hayward@arm.com>
3167 David Sherwood <david.sherwood@arm.com>
3169 * lower-subreg.c (interesting_mode_p): New function.
3170 (compute_costs, find_decomposable_subregs, decompose_register)
3171 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3172 (resolve_clobber, dump_choices): Use it.
3174 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3175 Alan Hayward <alan.hayward@arm.com>
3176 David Sherwood <david.sherwood@arm.com>
3178 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3180 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3181 Alan Hayward <alan.hayward@arm.com>
3182 David Sherwood <david.sherwood@arm.com>
3184 * alias.c (find_base_value, find_base_term): Only process integer
3185 truncations. Check the precision rather than the size.
3187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3188 Alan Hayward <alan.hayward@arm.com>
3189 David Sherwood <david.sherwood@arm.com>
3191 * machmode.h (is_narrower_int_mode): New function
3192 * optabs.c (expand_float, expand_fix): Use it.
3193 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3195 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3196 Alan Hayward <alan.hayward@arm.com>
3197 David Sherwood <david.sherwood@arm.com>
3199 * rtl.h (narrower_subreg_mode): New function.
3200 * ira-color.c (update_costs_from_allocno): Use it.
3202 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3203 Alan Hayward <alan.hayward@arm.com>
3204 David Sherwood <david.sherwood@arm.com>
3206 * optabs-query.h (convert_optab_p): New function, split out from...
3207 (convert_optab_handler): ...here.
3208 (widening_optab_handler): Delete.
3209 (find_widening_optab_handler): Remove permit_non_widening parameter.
3210 (find_widening_optab_handler_and_mode): Likewise. Provide an
3211 override that operates on mode class wrappers.
3212 * optabs-query.c (widening_optab_handler): Delete.
3213 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3214 parameter. Assert that the two modes are the same class and that
3215 the "from" mode is narrower than the "to" mode. Use
3216 convert_optab_handler instead of widening_optab_handler.
3217 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3218 instead of widening_optab_handler.
3219 * expr.c (expand_expr_real_2): Update calls to
3220 find_widening_optab_handler.
3221 * optabs.c (expand_widen_pattern_expr): Likewise.
3222 (expand_binop_directly): Take the insn_code as a parameter.
3223 (expand_binop): Only call find_widening_optab_handler for
3224 conversion optabs; use optab_handler otherwise. Update calls
3225 to find_widening_optab_handler and expand_binop_directly.
3226 Use convert_optab_handler instead of widening_optab_handler.
3227 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3228 find_widening_optab_handler and use scalar_mode rather than
3230 (convert_plusminus_to_widen): Likewise.
3232 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3233 Alan Hayward <alan.hayward@arm.com>
3234 David Sherwood <david.sherwood@arm.com>
3236 * machmode.h (fixed_size_mode): New class.
3237 * rtl.h (get_pool_mode): Return fixed_size_mode.
3238 * gengtype.c (main): Add fixed_size_mode.
3239 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3240 (get_raw_arg_mode): Likewise.
3241 * doc/tm.texi: Regenerate.
3242 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3243 * targhooks.c (default_get_reg_raw_mode): Likewise.
3244 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3245 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3246 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3247 (msp430_get_raw_result_mode): Likewise.
3248 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3249 * dbxout.c (dbxout_parms): Require fixed-size modes.
3250 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3251 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3252 * omp-low.c (lower_oacc_reductions): Likewise.
3253 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3254 (simplify_subreg): Update accordingly.
3255 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3256 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3257 that aren't fixed-size.
3258 (get_pool_mode): Return a fixed_size_mode.
3259 (output_constant_pool_2): Take a fixed_size_mode.
3261 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3262 Alan Hayward <alan.hayward@arm.com>
3263 David Sherwood <david.sherwood@arm.com>
3265 * doc/rtl.texi (vec_series): Document.
3266 (const): Say that the operand can be a vec_series.
3267 * rtl.def (VEC_SERIES): New rtx code.
3268 * rtl.h (const_vec_series_p_1): Declare.
3269 (const_vec_series_p): New function.
3270 * emit-rtl.h (gen_const_vec_series): Declare.
3271 (gen_vec_series): Likewise.
3272 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3273 (gen_vec_series): Likewise.
3274 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3275 * simplify-rtx.c (simplify_unary_operation): Handle negations
3277 (simplify_binary_operation_series): New function.
3278 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3279 (test_vector_ops_series): New function.
3280 (test_vector_ops): Call it.
3281 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3282 gen_const_vec_series.
3283 (altivec_lvsr): Likewise.
3284 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3286 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3287 Alan Hayward <alan.hayward@arm.com>
3288 David Sherwood <david.sherwood@arm.com>
3290 * doc/rtl.texi (const): Update description of address constants.
3291 Say that vector constants are allowed too.
3292 * common.md (E, F): Use CONSTANT_P instead of checking for
3294 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3295 checking for CONST_VECTOR.
3296 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3298 * expr.c (expand_expr_real_2): Check for vector modes instead
3299 of checking for CONST_VECTOR.
3300 * rtl.h (const_vec_p): New function.
3301 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3302 (unwrap_const_vec_duplicate): Handle them here too.
3304 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3305 David Malcolm <dmalcolm@redhat.com>
3306 Alan Hayward <alan.hayward@arm.com>
3307 David Sherwood <david.sherwood@arm.com>
3309 * rtl.h (vec_duplicate_p): New function.
3310 * selftest-rtl.c (assert_rtx_eq_at): New function.
3311 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3312 (assert_rtx_eq_at): Declare.
3313 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3314 * selftest-run-tests.c (selftest::run_tests): Call it.
3315 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3316 (simplify_unary_operation_1): Recursively handle vector duplicates.
3317 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3319 (simplify_subreg): Handle subregs of vector duplicates.
3320 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3321 (selftest::simplify_rtx_c_tests): New functions.
3323 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3324 Alan Hayward <alan.hayward@arm.com>
3325 David Sherwood <david.sherwood@arm.com>
3327 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3328 (gen_vec_duplicate): Likewise.
3329 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3331 (gen_const_vector): ...here.
3332 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3333 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3334 whose elements are all equal.
3335 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3336 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3337 (simplify_relational_operation): Likewise.
3338 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3340 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3341 (aarch64_expand_vector_init): Likewise.
3342 * config/arm/arm.c (neon_vdup_constant): Likewise.
3343 (neon_expand_vector_init): Likewise.
3344 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3345 (arm_block_set_unaligned_vect): Likewise.
3346 (arm_block_set_aligned_vect): Likewise.
3347 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3348 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3349 (expand_vec_perm_even_odd_pack): Likewise.
3350 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3351 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3352 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3353 gen_const_vec_duplicate.
3354 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3355 * config/mips/mips.c (mips_gen_const_int_vector): Use
3356 gen_const_vec_duplicate.
3357 (mips_expand_vector_init): Use CONST0_RTX.
3358 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3359 (define_split): Use gen_const_vec_duplicate.
3360 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3361 (define_split): Use gen_const_vec_duplicate.
3362 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3363 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3364 * config/spu/spu.c (spu_const): Likewise.
3366 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3367 Alan Hayward <alan.hayward@arm.com>
3368 David Sherwood <david.sherwood@arm.com>
3370 * combine.c (can_change_dest_mode): Reject changes in
3371 REGMODE_NATURAL_SIZE.
3373 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3375 * configure.ac (--enable-libssp): New.
3376 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3377 trying to determine target-specific default. Adjust indentation.
3378 * configure: Regenerated.
3379 * doc/install.texi (Configuration): Expand --disable-libssp
3382 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3384 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3387 2017-10-31 Martin Jambor <mjambor@suse.cz>
3390 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3392 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3394 * auto-profile.c (autofdo_source_profile::read): Use
3395 UNKNOWN_LOCATION rather than 0.
3396 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3397 (warning_at): ...this overload.
3398 (warning_at_rich_loc_n): Rename to...
3399 (warning_n): ...this overload.
3400 (error_at_rich_loc): Rename to...
3401 (error_at): ...this overload.
3402 (pedwarn_at_rich_loc): Rename to...
3403 (pedwarn): ...this overload.
3404 (permerror_at_rich_loc): Rename to...
3405 (permerror): ...this overload.
3406 (inform_at_rich_loc): Rename to...
3407 (inform): ...this overload.
3408 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3409 (diagnostic_n_impl_richloc): Rename to...
3410 (diagnostic_n_impl): ...this rich_location *-based decl.
3411 (inform_at_rich_loc): Rename to...
3412 (inform): ...this, and add an assertion.
3413 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3414 (warning_at_rich_loc): Rename to...
3415 (warning_at): ...this, and add an assertion.
3416 (warning_at_rich_loc_n): Rename to...
3417 (warning_n): ...this, and add an assertion.
3418 (warning_n): Update location_t-based implementation for removal of
3419 location_t-based diagnostic_n_impl.
3420 (pedwarn_at_rich_loc): Rename to...
3421 (pedwarn): ...this, and add an assertion.
3422 (permerror_at_rich_loc): Rename to...
3423 (permerror): ...this, and add an assertion.
3424 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3425 (error_at_rich_loc): Rename to...
3426 (error_at): ...this, and add an assertion.
3427 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3428 (driver::do_spec_on_infiles): Likewise.
3429 * substring-locations.c (format_warning_va): Update for renaming
3430 of inform_at_rich_loc.
3432 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3434 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3435 _Float<N>X built-in functions so that the variant without the
3436 "__builtin_" prefix is only enabled for the GNU C and Objective C
3437 languages when they are in non-strict ANSI/ISO mode.
3438 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3439 * target.def (floatn_builtin_p): Add a target hook to control
3440 whether _Float<N> and _Float<N>X built-in functions without the
3441 "__builtin_" prefix are enabled, and return true for C and
3442 Objective C in the default hook. Include langhooks.h in
3444 * targhooks.h (default_floatn_builtin_p): Likewise.
3445 * targhooks.c (default_floatn_builtin_p): Likewise.
3446 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3447 floatn_builtin_p target hook.
3448 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3450 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3451 Eric Botcazou <ebotcazou@adacore.com>
3453 PR rtl-optimization/81803
3454 * lra-constraints.c (curr_insn_transform): Also reload the whole
3455 register for a strict subreg no wider than a word if this is for
3456 a WORD_REGISTER_OPERATIONS target.
3458 2017-10-31 Jason Merrill <jason@redhat.com>
3460 * gdbinit.in: Skip over inlines from timevar.h.
3462 2017-10-31 Martin Liska <mliska@suse.cz>
3464 * doc/gcov.texi: Document new option.
3465 * gcov.c (print_usage): Likewise print it.
3466 (process_args): Support the argument.
3467 (format_count): New function.
3468 (format_gcov): Use the function.
3470 2017-10-31 Martin Liska <mliska@suse.cz>
3472 * gcov.c (struct name_map): do not use typedef.
3473 Define operator== and operator<.
3474 (name_search): Remove.
3475 (name_sort): Remove.
3476 (main): Do not allocate names.
3477 (process_file): Add vertical space.
3478 (generate_results): Use std::find.
3479 (release_structures): Do not release memory.
3480 (find_source): Use std::find.
3482 2017-10-31 Martin Liska <mliska@suse.cz>
3484 * gcov.c (struct line_info): Remove it's typedef.
3485 (line_info::line_info): Add proper ctor.
3486 (line_info::has_block): Do not use a typedef.
3487 (struct source_info): Do not use typedef.
3488 (circuit): Likewise.
3489 (get_cycles_count): Likewise.
3490 (output_intermediate_file): Iterate via vector iterator.
3491 (add_line_counts): Use std::vector methods.
3492 (accumulate_line_counts): Likewise.
3493 (output_lines): Likewise.
3495 2017-10-31 Martin Liska <mliska@suse.cz>
3497 * gcov.c (struct source_info): Remove typedef.
3498 (source_info::source_info): Add proper ctor.
3499 (accumulate_line_counts): Use struct, not it's typedef.
3500 (output_gcov_file): Likewise.
3501 (output_lines): Likewise.
3502 (main): Do not allocate an array.
3503 (output_intermediate_file): Use size of vector container.
3504 (process_file): Resize the vector.
3505 (generate_results): Do not preallocate, use newly added vector
3507 (release_structures): Do not release sources.
3508 (find_source): Use vector methods.
3509 (add_line_counts): Do not use typedef.
3511 2017-10-31 Martin Liska <mliska@suse.cz>
3513 * doc/gcov.texi: Document that.
3514 * gcov.c (add_line_counts): Mark lines with a non-executed
3516 (output_line_beginning): Handle such lines.
3517 (output_lines): Pass new argument.
3518 (output_intermediate_file): Print it in intermediate format.
3520 2017-10-31 Martin Liska <mliska@suse.cz>
3522 * color-macros.h: New file.
3523 * diagnostic-color.c: Factor out color related to macros to
3525 * doc/gcov.texi: Document -k option.
3526 * gcov.c (INCLUDE_STRING): Include string.h.
3527 (print_usage): Add -k option.
3528 (process_args): Parse it.
3529 (pad_count_string): New function.
3530 (output_line_beginning): Likewise.
3531 (DEFAULT_LINE_START): New macro.
3532 (output_lines): Support color output.
3534 2017-10-31 Martin Liska <mliska@suse.cz>
3536 PR gcov-profile/82633
3537 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3538 their interaction with GCOV infrastructure.
3539 * configure.ac: Add -fkeep-{inline,static}-functions to
3541 * configure: Regenerate.
3543 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3546 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3548 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3551 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3552 into a register if it does not fit into an immediate offset field.
3554 2017-10-31 Olivier Hainque <hainque@adacore.com>
3556 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3558 2017-10-31 Julia Koval <julia.koval@intel.com>
3560 * config.gcc: Add gfniintrin.h.
3561 * config/i386/gfniintrin.h: New.
3562 * config/i386/i386-builtin-types.def
3563 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3564 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3565 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3566 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3567 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3568 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3569 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3570 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3571 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3572 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3573 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3574 * config/i386/immintrin.h: Include gfniintrin.h.
3575 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3577 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3579 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3581 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3583 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3584 (arm_ashldi3_1bit): Remove pattern.
3585 (ashrdi3): Remove shift by 1 expansion.
3586 (arm_ashrdi3_1bit): Remove pattern.
3587 (lshrdi3): Remove shift by 1 expansion.
3588 (arm_lshrdi3_1bit): Remove pattern.
3589 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3590 cost of ashldi3 by 1.
3591 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3592 (<shift>di3_neon): Likewise.
3594 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3596 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3597 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3598 and (*aarch64_simd_mov<VQ:mode>).
3599 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3600 pattern alternative.
3601 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3602 attributes to match pattern alternative.
3604 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3606 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3607 * config/rs6000/emmintrin.h: New file.
3608 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3610 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3612 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3613 * (movdi_vfp_cortexa8): Remove pattern.
3615 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3617 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3619 (Specific, alpha*-dec-osf5.1): Remove.
3620 (Specific, mips-sgi-irix5): Remove.
3621 (Specific, mips-sgi-irix6): Remove.
3623 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3626 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3627 arguments to clear_bit_region_be.
3629 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3631 * gimplify.c: Include memmodel.h.
3633 2017-10-30 Martin Jambor <mjambor@suse.cz>
3635 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3636 condition whether loop should be executed at all.
3638 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3640 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3641 gimple folding of vec_madd() intrinsics.
3642 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3643 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3644 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3646 2017-10-30 Richard Biener <rguenther@suse.de>
3648 PR tree-optimization/82762
3650 2017-10-23 Richard Biener <rguenther@suse.de>
3652 PR tree-optimization/82129
3654 2017-08-01 Richard Biener <rguenther@suse.de>
3656 PR tree-optimization/81181
3657 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3658 (compute_antic): ... end of iteration here.
3660 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3662 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3663 -std=iso9899:2017 and -std=gnu17.
3664 * doc/standards.texi (C Language): Document C17 support.
3665 * doc/cpp.texi (Overview): Mention -std=c17.
3666 (Standard Predefined Macros): Document C11 and C17 values of
3667 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3668 * doc/extend.texi (Inline): Do not list individual options for
3669 standards newer than C99.
3670 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3672 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3675 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3677 * asan.c (asan_finish_file): Align asan globals array by shadow
3680 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3683 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3684 (struct store_immediate_info): Add bitregion_start and bitregion_end
3686 (store_immediate_info::store_immediate_info): Add brs and bre
3687 arguments and initialize bitregion_{start,end} from those.
3688 (struct merged_store_group): Add bitregion_start, bitregion_end,
3689 align_base and mask fields. Drop unnecessary struct keyword from
3690 struct store_immediate_info. Add do_merge method.
3691 (clear_bit_region_be): Use memset instead of loop storing zeros.
3692 (merged_store_group::do_merge): New method.
3693 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3694 stores as long as the surrounding bitregions have no gaps.
3695 (merged_store_group::merge_overlapping): Use do_merge.
3696 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3697 is byte aligned, rather than requiring that start and width are
3698 byte aligned. Drop unnecessary struct keyword from
3699 struct store_immediate_info. Allocate and populate also mask array.
3700 Make start of the arrays relative to bitregion_start rather than
3701 start and size them according to bitregion_{end,start} difference.
3702 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3703 struct store_immediate_info.
3704 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3705 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3706 struct keyword from struct store_immediate_info.
3707 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3708 between stores as long as the surrounding bitregions have no gaps.
3710 (struct split_store): Add orig non-static data member.
3711 (split_store::split_store): Initialize orig to false.
3712 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3713 if there is exactly a single original stmt. Change stmts argument
3714 to pointer from reference, if NULL, don't push anything to it. Add
3715 first argument, use it to optimize skipping over orig stmts that
3716 are known to be before bitpos already. Simplify.
3717 (split_group): Return unsigned int count how many stores are or
3718 would be needed rather than a bool. Add allow_unaligned argument.
3719 Change split_stores argument from reference to pointer, if NULL,
3720 only do a dry run computing how many stores would be produced.
3721 Rewritten algorithm to use both alignment and misalign if
3722 !allow_unaligned and handle bitfield stores with gaps.
3723 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3724 from bitregion_start instead of start. Compute allow_unaligned
3725 here, if true, do 2 split_group dry runs to compute which one
3726 produces fewer stores and prefer aligned if equal. Punt if
3727 new count is bigger or equal than original before emitting any
3728 statements, rather than during that. Remove no longer needed
3729 new_ssa_names tracking. Replace num_stmts with
3730 split_stores.length (). Use 32-bit stack allocated entries
3731 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3732 if possible. Handle bitfields with gaps.
3733 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3734 Compute bitregion_{start,end} for the stores and construct
3735 store_immediate_info with that. Formatting fixes.
3737 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3740 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3741 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3743 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3745 * gimplify.c: Include tm_p.h.
3747 * common.opt (gcoff): Re-add as ignored option.
3748 (gcoff1, gcoff2, gcoff3): Likewise.
3750 * Makefile.in (OBJS): Delete sdbout.o.
3751 (GTFILES): Delete $(srcdir)/sdbout.c.
3752 * debug.h: Delete sdb_debug_hooks.
3753 * final.c: Delete sdbout.h include.
3754 (final_scan_insn): Delete SDB_DEBUG check.
3755 (rest_of_clean_state): Likewise.
3756 * output.h: Delete sdb_begin_function_line.
3759 * toplev.c: Delete sdbout.h include.
3760 (process_options): Delete SDB_DEBUG check.
3761 * tree-core.h (tree_type_common): Delete pointer field of
3763 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3764 TYPE_SYMTAB_POINTER.
3765 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3766 (TYPE_SYMTAB_IS_POINTER): Delete.
3767 (TYPE_SYMTAB_IS_DIE): Renumber.
3768 * xcoffout.c: Refer to former sdbout.c file.
3769 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3771 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3772 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3773 (Debugging Options): Delete -gcoff.
3774 (-gstabs): Delete SDB reference.
3776 (-gcoff@var{level}): Delete.
3777 * doc/passes.texi (Debugging information output): Delete SDB and
3778 sdbout.c references.
3779 * doc/tm.texi: Regenerate.
3780 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3781 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3783 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3784 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3785 (SDB_DEBUGGING_INFO): Delete.
3786 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3787 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3788 * target.def (output_source_filename): Delete COFF reference.
3790 * common.opt (gcoff): Delete.
3791 (gxcoff+): Update Negative chain.
3792 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3794 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3795 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3796 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3798 (expand_function_start): Change sdb reference to past tense.
3799 (expand_function_end): Change sdb reference to past tense.
3800 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3801 * opts.c (debug_type_names): Delete coff entry.
3802 (common_handle_option): Delete OPT_gcoff case.
3803 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3805 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3806 * config/cris/cris.h: Delete SDB reference in comment.
3807 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3808 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3809 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3810 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3812 (ix86_expand_prologue): Likewise.
3813 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3814 * config/ia64/ia64.h: Likewise.
3815 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3816 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3818 * config/mmix/mmix.h: Likewise.
3819 * config/nds32/nds32.c: Likewise.
3820 * config/stormy/storym16.h: Likewise.
3821 * config/visium/visium.h: Likewise.
3822 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3824 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3826 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3827 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3828 FRAME_POINTER_REGNUM point at high end of local var area.
3830 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3832 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3833 Move comment around. Do not reset best_edge for a copiable
3834 destination if the copy would cause a partition change.
3835 (better_edge_p): Remove redundant check.
3837 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3839 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3841 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3843 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3844 for math functions that have _Float<N> and _Float<N>X variants.
3845 (mathfn_built_in_2): Add support for math functions that have
3846 _Float<N> and _Float<N>X variants.
3847 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3848 (expand_builtin_mathfn_ternary): Add support for fma with
3849 _Float<N> and _Float<N>X variants.
3850 (expand_builtin): Likewise.
3851 (fold_builtin_3): Likewise.
3852 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3853 create math function _Float<N> and _Float<N>X variants as external
3855 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3856 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3857 the __builtin_ prefix and if not strict ansi, without the prefix.
3858 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3859 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3860 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3861 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3862 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3863 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3864 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3865 function signatures for fma _Float<N> and _Float<N>X variants.
3866 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3867 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3868 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3869 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3870 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3871 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3872 * gencfn-macros.c (print_case_cfn): Add support for math functions
3873 that have _Float<N> and _Float<N>X variants.
3874 (print_define_operator_list): Likewise.
3875 (fltfn_suffixes): Likewise.
3877 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3878 for math functions that have _Float<N> and _Float<N>X variants.
3879 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3880 and _Float<N>X variants.
3881 (COPYSIGN): Likewise.
3884 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3885 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3887 (integer_valued_read_call_p): Likewise.
3888 * fold-const-call.c (fold_const_call_ss): Likewise.
3889 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3890 _Float<N> and _Float<N>X variants.
3891 (fold_const_call_ssss): Add support for fma _Float<N> and
3892 _Float<N>X variants.
3893 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3894 support for copysign and fma _Float<N> and _Float<N>X variants.
3895 (backprop::process_builtin_call_use): Likewise.
3896 * tree-call-cdce.c (can_test_argument_range); Add support for
3897 sqrt _Float<N> and _Float<N>X variants.
3898 (edom_only_function): Likewise.
3899 (get_no_error_domain): Likewise.
3900 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3901 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3902 copysign _Float<N> and _Float<N>X variants.
3903 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3904 handled by machine independent code.
3905 (FMAF128): Likewise.
3906 * doc/cpp.texi (Common Predefined Macros): Document defining
3907 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3908 fma _Float<N> and _Float<N>X variants.
3910 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3913 * config/i386/i386-modes.def (CCFPU): Remove definition.
3914 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3915 (ix86_cc_modes_compatible): Ditto.
3916 (ix86_expand_carry_flag_compare): Ditto.
3917 (ix86_expand_int_movcc): Ditto.
3918 (ix86_expand_int_addcc): Ditto.
3919 (ix86_reverse_condition): Ditto.
3920 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3921 Return true/false for unordered/ordered fp comparisons.
3922 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3923 (ix86_prepare_fp_compare_args): Update for rename.
3924 (ix86_expand_fp_compare): Update for rename. Generate unordered
3925 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3926 (ix86_expand_sse_compare_and_jump): Ditto.
3927 * config/i386/predicates.md (fcmov_comparison_operator):
3928 Remove CCFPU mode handling.
3929 (ix86_comparison_operator): Ditto.
3930 (ix86_carry_flag_operator): Ditto.
3931 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3932 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3933 (*cmpu<mode>_cc_i387): Ditto.
3934 (FPCMP): Remove mode iterator.
3935 (unord): Remove mode attribute.
3936 (unord_subst): New define_subst transformation
3937 (unord): New define_subst attribute.
3939 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3940 (*cmpi<unord>xf_i387): Ditto.
3941 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3942 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3943 using unord_subst transformation.
3944 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3945 (round_saeonly): Also handle CCFP mode.
3946 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3947 Remove UNSPEC_SAHF unspec handling.
3949 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3951 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3953 2017-10-27 Jeff Law <law@redhat.com>
3955 * gimple-ssa-sprintf.c: Include domwalk.h.
3956 (class sprintf_dom_walker): New class, derived from dom_walker.
3957 (sprintf_dom_walker::before_dom_children): New function.
3958 (struct call_info): Moved into sprintf_dom_walker class
3959 (compute_formath_length, handle_gimple_call): Likewise.
3960 (sprintf_length::execute): Call the dominator walker rather
3961 than walking the statements.
3963 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3964 gimple statement locations.
3965 (check_array_bounds): Corresponding changes. Get the statement's
3966 location directly from wi->stmt.
3968 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3971 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3973 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3975 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3976 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3978 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3981 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3982 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3983 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3984 maybe_get_pool_constant.
3985 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3988 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3990 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3991 2.26 caveat. Update gas and gld versions.
3992 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3995 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3997 * cgraph.h (set_malloc_flag): Declare.
3998 * cgraph.c (set_malloc_flag_1): New function.
3999 (set_malloc_flag): Likewise.
4000 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
4001 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
4003 (read_ipa_call_summary): Add support for reading is_return_callee.
4004 (write_ipa_call_summary): Stream is_return_callee.
4005 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
4006 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
4007 ipa-prop.h, ipa-fnsummary.h.
4008 (pure_const_names): Change to static.
4009 (malloc_state_e): Define.
4010 (malloc_state_names): Define.
4011 (funct_state_d): Add field malloc_state.
4012 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
4013 (check_retval_uses): New function.
4014 (malloc_candidate_p): Likewise.
4015 (analyze_function): Add support for malloc attribute.
4016 (pure_const_write_summary): Stream malloc_state.
4017 (pure_const_read_summary): Add support for reading malloc_state.
4018 (dump_malloc_lattice): New function.
4019 (propagate_malloc): New function.
4020 (warn_function_malloc): New function.
4021 (ipa_pure_const::execute): Call propagate_malloc and
4022 ipa_free_fn_summary.
4023 (pass_local_pure_const::execute): Add support for malloc attribute.
4024 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
4025 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
4027 2017-10-27 Martin Liska <mliska@suse.cz>
4029 PR gcov-profile/82457
4030 * doc/invoke.texi: Document that one needs a non-strict ISO mode
4031 for fork-like functions to be properly instrumented.
4033 2017-10-27 Richard Biener <rguenther@suse.de>
4036 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
4037 info when we redirected EH.
4039 2017-10-26 Michael Collison <michael.collison@arm.com>
4041 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
4043 (<optab>_trunchf<GPI:mode>2: New pattern.
4044 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
4045 * config/aarch64/iterators.md (wv): New mode attribute.
4046 (vf, VF): New mode attributes.
4047 (vgp, VGP): New mode attributes.
4048 (s): Update attribute with SImode and DImode prefixes.
4050 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4052 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
4053 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
4054 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
4055 (nios2_option_overide): Initialize it. Don't allow R0-relative
4056 addressing with PIC.
4057 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
4058 (nios2_symbolic_constant_p): Likewise.
4059 (nios2_legitimate_address_p): Likewise.
4060 (nios2_r0rel_section_name_p): New.
4061 (nios2_symbol_ref_in_r0rel_data_p): New.
4062 (nios2_emit_move_sequence): Handle r0rel_constant_p.
4063 (r0rel_constant_p): New.
4064 (nios2_print_operand_address): Handle r0rel_constant_p.
4065 (nios2_cdx_narrow_form_p): Likewise.
4066 * config/nios2/nios2.opt (mr0rel-sec=): New option.
4067 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
4068 (Nios II Options): Document -mr0rel-sec.
4070 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
4072 * config/nios2/nios2.c: Include xregex.h.
4073 (nios2_gprel_sec_regex): New.
4074 (nios2_option_overide): Initialize it. Don't allow GP-relative
4075 addressing with PIC.
4076 (nios2_small_section_name_p): Check for regex match.
4077 * config/nios2/nios2.opt (mgprel-sec=): New option.
4078 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
4079 (Nios II Options): Document -mgprel-sec.
4081 2017-10-26 Jim Wilson <wilson@tuliptree.org>
4083 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
4085 2017-10-26 Tom de Vries <tom@codesourcery.com>
4087 PR tree-optimization/82707
4088 * gimple.c (gimple_copy): Fix unsharing of
4089 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
4091 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
4093 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
4094 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
4095 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
4096 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
4097 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
4098 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
4099 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
4100 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4101 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4102 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4103 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4104 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4105 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4106 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4107 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4108 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4109 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4111 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4113 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4115 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4116 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4117 warning to rs6000.c. Remove the Undocumented flag, since it has
4119 (-mabi=ibmlongdouble): Likewise.
4120 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4121 already set, set the default format for long double.
4122 (rs6000_debug_reg_global): Print whether long double is IBM or
4124 (rs6000_option_override_internal): Rework setting long double
4125 format. Only warn if the user is changing the long double default
4126 and they did not use -Wno-psabi.
4127 * doc/invoke.texi (PowerPC options): Update the documentation for
4128 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4130 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4131 Alan Hayward <alan.hayward@arm.com>
4132 David Sherwood <david.sherwood@arm.com>
4134 * rtl.h (wider_subreg_mode): New function.
4135 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4136 rather than an unsigned int *.
4137 * ira-color.c (regno_max_ref_width): Replace with...
4138 (regno_max_ref_mode): ...this new variable.
4139 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4140 Use wider_subreg_mode.
4141 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4142 rather than an unsigned int *.
4143 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4144 (process_alt_operands): Likewise.
4145 (invariant_p): Likewise.
4146 * lra-spills.c (assign_mem_slot): Likewise.
4147 (add_pseudo_to_slot): Likewise.
4148 * lra.c (collect_non_operand_hard_regs): Likewise.
4149 (add_regs_to_insn_regno_info): Likewise.
4150 * reload1.c (regno_max_ref_width): Replace with...
4151 (regno_max_ref_mode): ...this new variable.
4152 (reload): Update accordingly. Update call to
4153 ira_sort_regnos_for_alter_reg.
4154 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4155 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4156 (scan_paradoxical_subregs): Likewise.
4158 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4160 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4161 (aarch64_frame): Add emit_frame_chain boolean.
4162 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4163 Move eh_return case to aarch64_layout_frame.
4164 (aarch64_layout_frame): Initialize emit_frame_chain.
4165 (aarch64_expand_prologue): Use emit_frame_chain.
4167 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4169 * config/aarch64/aarch64.c (aarch64_layout_frame):
4170 Ensure LR is always stored at the bottom of the callee-saves.
4171 Remove rarely used frame layout which saves callee-saves at top of
4172 frame, so the store of LR can be used as a valid probe in all cases.
4174 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4176 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4177 Improve unaligned TImode/TFmode base/offset split.
4179 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4180 Alan Hayward <alan.hayward@arm.com>
4181 David Sherwood <david.sherwood@arm.com>
4183 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4184 * combine.c (find_single_use_1): Likewise.
4185 (expand_field_assignment): Likewise.
4186 (move_deaths): Likewise.
4187 * lra-constraints.c (simplify_operand_subreg): Likewise.
4188 (curr_insn_transform): Likewise.
4189 * lra.c (collect_non_operand_hard_regs): Likewise.
4190 (add_regs_to_insn_regno_info): Likewise.
4191 * rtlanal.c (reg_referenced_p): Likewise.
4192 (covers_regno_no_parallel_p): Likewise.
4194 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4196 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4197 Don't print any bits outside the precision of the value.
4198 * wide-int.cc (test_printing): Add some new tests.
4200 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4202 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4203 supports -xbrace_comment option.
4204 * configure: Regenerate.
4205 * config.in: Regenerate.
4206 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4207 (ASM_CPU_SPEC): Use it.
4209 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4211 * target.def (static_rtx_alignment): New hook.
4212 * targhooks.h (default_static_rtx_alignment): Declare.
4213 * targhooks.c (default_static_rtx_alignment): New function.
4214 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4215 * doc/tm.texi: Regenerate.
4216 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4217 instead of targetm.constant_alignment. Remove call to
4219 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4220 (cris_preferred_mininum_alignment): New function, split out from...
4221 (cris_constant_alignment): ...here.
4222 (cris_static_rtx_alignment): New function.
4223 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4225 (ix86_constant_alignment): ...here.
4226 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4227 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4228 (mmix_static_rtx_alignment): New function.
4229 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4230 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4232 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4235 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4236 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4238 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4240 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4241 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4242 mark as a sign-extending load.
4243 (local_pic_load_u): Define.
4245 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4248 * fold-const.c (operand_equal_for_comparison_p): Also return true
4249 if ARG0 is a simple variant of ARG1 with narrower precision.
4250 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4252 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4254 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4256 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4257 scatter_static, scatter_per_elt.
4258 * x86-tune-costs.h: Add new cost entries.
4260 2017-10-25 Richard Biener <rguenther@suse.de>
4262 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4263 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4264 class pass_fre): Move to ...
4265 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4266 class pass_fre): ... here and adjust for statistics.
4268 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4271 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4272 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4273 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4274 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4277 2017-10-25 Richard Biener <rguenther@suse.de>
4279 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4280 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4281 eliminate_push_avail, eliminate_insert): Move inside...
4282 (class eliminate_dom_walker): ... this class in preparation
4284 (fini_eliminate): Remove by merging with ...
4285 (eliminate): ... this function. Adjust for class changes.
4286 (pass_pre::execute): Remove fini_eliminate call.
4287 (pass_fre::execute): Likewise.
4289 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4292 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4293 (VPERMI2, VPERMI2I): New mode iterators.
4294 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4295 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4297 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4298 mode iterator. Remove 3 old define_insn patterns.
4299 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4300 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4301 VPERMI2 mode iterator, remove the other two expanders.
4302 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4303 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4304 instructions, remove the other two patterns.
4305 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4306 mode iterator, remove the other two patterns.
4307 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4308 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4309 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4310 and adjust argument order accordingly.
4311 (ix86_expand_vec_perm): Adjust caller.
4312 (expand_vec_perm_1): Likewise.
4313 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4314 (expand_vec_perm_vpermt2_vpshub2): ... this.
4315 (ix86_expand_vec_perm_const_1): Adjust caller.
4316 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4319 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4320 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4321 (vec_shl_<mode>): Remove unused expander.
4322 (avx512bw_<shift_insn><mode>3): New define_insn.
4323 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4324 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4326 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4329 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4332 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4334 PR rtl-optimization/82396
4335 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4336 (autopref_multipass_init): Simplify initialization.
4337 (autopref_rank_data): Simplify sort order.
4338 * gcc/sched-int.h (autopref_multipass_data_): Remove
4339 multi_mem_insn_p, min_offset and max_offset.
4341 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4344 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4345 Check special value of flag_omit_frame_pointer.
4346 (aarch64_can_eliminate): Likewise.
4347 (aarch64_override_options_after_change_1): Simplify handling of
4348 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4350 2017-10-24 Richard Biener <rguenther@suse.de>
4352 PR tree-optimization/82697
4353 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4354 zero for conditional load and unconditional store.
4356 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4358 * doc/install.texi: Document bootstrap-cet.
4360 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4363 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4364 ENDBR instruction at function entrance if function is only
4367 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4370 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4371 patterns to better describe from which operation the CF is computed.
4372 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4373 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4374 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4375 is 0, use _0 suffixed expanders instead of emitting a comparison
4378 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4380 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4381 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4383 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4386 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4387 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4388 * loop-iv.c (iv_get_reaching_def): Likewise.
4389 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4390 variable is promoted and the partition contains undefined values.
4392 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4394 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4396 (nios2_address_cost): Define.
4397 (nios2_legitimize_address): Recognize (exp + constant) directly.
4398 (TARGET_ADDRESS_COST): Define.
4400 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4402 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4403 (nios2_symbolic_memory_operand_p): Declare.
4404 (nios2_split_large_constant): Declare.
4405 (nios2_split_symbolic_memory_operand): Declare.
4406 * config/nios2/nios2.c: Adjust includes.
4407 (nios2_symbolic_constant_allowed): New.
4408 (nios2_symbolic_constant_p): New.
4409 (nios2_plus_symbolic_constant_p): New.
4410 (nios2_valid_addr_expr_p): Recognize addresses involving
4412 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4413 (nios2_symbolic_memory_operand_p): New.
4414 (nios2_large_constant_p): New.
4415 (nios2_split_large_constant): New.
4416 (nios2_split_plus_large_constant): New.
4417 (nios2_split_symbolic_memory_operand): New.
4418 (nios2_legitimize_address): Code refactoring. Handle addresses
4419 involving symbolic constants.
4420 (nios2_emit_move_sequence): Likewise.
4421 (nios2_print_operand): Improve error output.
4422 (nios2_print_operand_address): Handle LO_SUM.
4423 (nios2_cdx_narrow_form_p): Likewise.
4424 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4425 operands involving symbolic constants.
4426 (movhi_internal, movsi_internal): Likewise.
4427 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4428 (extendhisi2, extendqi<mode>2): Likewise.
4430 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4432 * tree-pass.h (PROP_rtl_split_insns): Define.
4433 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4435 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4437 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4439 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4442 * target.def (const_not_ok_for_debug_p): Default to
4443 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4444 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4445 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4446 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4447 which targetm.const_not_ok_for_debug_p returned true.
4448 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4450 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4452 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4453 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4454 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4456 (ix86_const_not_ok_for_debug_p): New function.
4457 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4458 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4460 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4463 * system.h: Conditionally include "unique-ptr.h" if
4464 INCLUDE_UNIQUE_PTR is defined.
4465 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4466 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4468 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4470 * config/rl78/rl78.md: New define_expand "subdi3".
4472 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4475 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4476 DF_REF_INSN if DF_REF_INSN_INFO is false.
4478 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4480 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4481 xmm_move instead of sse_move.
4482 (sse_store_index): New function.
4483 (ix86_register_move_cost): Be more sensible about mismatch stall;
4484 model AVX moves correctly; make difference between sse->integer and
4486 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4487 moves; make difference between SSE and AVX.
4488 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4489 and zmm_move. Increase size of sse load and store tables;
4490 add unaligned load and store tables; add ssemmx_to_integer.
4491 * x86-tune-costs.h: Update all entries according to real
4492 move latencies from Agner Fog's manual and chip documentation.
4494 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4497 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4498 * config/i386/constraints.md (Wf): New constraint.
4499 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4500 (cmp<dwi>_doubleword): Removed.
4501 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4502 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4503 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4504 expand with cmp<dwi>_doubleword. For LTU and GEU use
4505 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4507 * common.opt (gcolumn-info): Enable by default.
4508 * doc/invoke.texi (gcolumn-info): Document new default.
4510 2017-10-23 Richard Biener <rguenther@suse.de>
4512 PR tree-optimization/82672
4513 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4514 Fold the stmt if we propagated into it.
4516 2017-10-23 Richard Biener <rguenther@suse.de>
4518 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4519 (bitmap_remove_expr_from_set): ... this. All callers call this
4520 for non-constant values.
4521 (bitmap_set_subtract): Rename to...
4522 (bitmap_set_subtract_expressions): ... this. Adjust and
4524 (bitmap_set_contains_value): Remove superfluous check.
4525 (bitmap_set_replace_value): Inline into single caller ...
4526 (bitmap_value_replace_in_set): ... here and simplify.
4527 (dependent_clean): Merge into ...
4528 (clean): ... this using an overload. Adjust.
4529 (prune_clobbered_mems): Adjust.
4530 (compute_antic_aux): Likewise.
4531 (compute_partial_antic_aux): Likewise.
4533 2017-10-23 Richard Biener <rguenther@suse.de>
4535 PR tree-optimization/82129
4537 2017-08-01 Richard Biener <rguenther@suse.de>
4539 PR tree-optimization/81181
4540 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4541 (compute_antic): ... end of iteration here.
4543 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4545 * target.def (starting_frame_offset): New hook.
4546 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4547 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4548 * doc/tm.texi.in: Regenerate.
4549 * hooks.h (hook_hwi_void_0): Declare.
4550 * hooks.c (hook_hwi_void_0): New function.
4551 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4552 STARTING_FRAME_OFFSET.
4553 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4554 * reload1.c (reload): Likewise.
4555 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4556 instead of STARTING_FRAME_OFFSET.
4557 * function.c (try_fit_stack_local): Likewise.
4558 (assign_stack_local_1): Likewise
4559 (instantiate_virtual_regs): Likewise.
4560 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4561 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4562 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4563 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4564 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4565 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4566 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4567 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4568 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4569 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4570 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4571 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4572 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4573 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4574 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4575 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4576 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4577 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4578 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4579 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4580 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4581 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4582 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4583 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4584 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4585 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4586 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4587 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4588 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4589 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4590 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4591 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4592 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4593 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4594 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4595 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4596 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4597 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4598 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4599 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4600 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4601 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4602 return a HOST_WIDE_INT.
4603 (avr_builtin_setjmp_frame_value): Use it instead of
4604 STARTING_FRAME_OFFSET.
4605 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4606 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4607 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4609 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4610 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4611 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4612 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4613 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4614 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4615 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4616 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4617 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4618 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4619 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4620 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4622 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4623 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4624 * config/mips/mips.c (mips_compute_frame_info): Refer to
4625 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4626 (mips_starting_frame_offset): New function.
4627 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4628 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4629 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4630 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4631 and return a HOST_WIDE_INT.
4632 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4633 (mmix_initial_elimination_offset): Refer to
4634 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4635 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4636 * config/pa/pa.c (pa_starting_frame_offset): New function.
4637 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4638 (pa_expand_prologue): Likewise.
4639 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4640 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4641 !FRAME_GROWS_DOWNWARD handling to...
4642 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4643 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4644 !FRAME_GROWS_DOWNWARD handling to...
4645 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4646 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4647 !FRAME_GROWS_DOWNWARD handling to...
4648 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4649 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4651 (rs6000_starting_frame_offset): New function.
4652 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4653 !FRAME_GROWS_DOWNWARD handling to...
4654 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4655 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4656 !FRAME_GROWS_DOWNWARD handling to...
4657 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4658 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4659 !FRAME_GROWS_DOWNWARD handling to...
4660 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4661 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4662 (rs6000_starting_frame_offset): New function.
4663 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4664 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4665 * config/vax/vax.c (vax_starting_frame_offset): New function.
4666 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4667 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4668 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4669 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4670 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4671 * system.h (STARTING_FRAME_OFFSET): Poison.
4673 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4675 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4676 SCALAR_TYPE_MODE instead of TYPE_MODE.
4678 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4679 Alan Hayward <alan.hayward@arm.com>
4680 David Sherwood <david.sherwood@arm.com>
4682 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4684 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4685 Alan Hayward <alan.hayward@arm.com>
4686 David Sherwood <david.sherwood@arm.com>
4688 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4690 2017-10-23 Richard Biener <rguenther@suse.de>
4692 PR tree-optimization/82129
4693 * tree-ssa-pre.c (bitmap_set_and): Remove.
4694 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4695 canonicalizing expressions in the set to those with lowest
4696 ID rather than taking that from the first edge.
4698 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4700 * combine.c (rtx_equal_for_field_assignment_p): Use
4701 byte_lowpart_offset.
4703 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4704 Alan Hayward <alan.hayward@arm.com>
4705 David Sherwood <david.sherwood@arm.com>
4707 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4708 to a SUBREG_PROMOTED_VAR.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4715 (expand_debug_source_expr): Likewise.
4716 * combine.c (combine_simplify_rtx): Likewise.
4717 * cse.c (fold_rtx): Likewise.
4718 * optabs.c (expand_float): Likewise.
4719 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4720 (simplify_binary_operation_1): Likewise.
4722 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4723 Alan Hayward <alan.hayward@arm.com>
4724 David Sherwood <david.sherwood@arm.com>
4726 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4727 (record_promoted_value): Likewise.
4728 * expr.c (expand_expr_real_2): Likewise.
4729 * ree.c (update_reg_equal_equiv_notes): Likewise.
4730 (combine_set_extension): Likewise.
4731 * rtlanal.c (low_bitmask_len): Likewise.
4732 * simplify-rtx.c (neg_const_int): Likewise.
4733 (simplify_binary_operation_1): Likewise.
4735 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4736 Alan Hayward <alan.hayward@arm.com>
4737 David Sherwood <david.sherwood@arm.com>
4739 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4740 * regcprop.c (maybe_mode_change): Likewise.
4741 * reload1.c (alter_reg): Likewise.
4743 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4745 * inchash.h (inchash::hash::add_wide_int): New function.
4746 * lto-streamer-out.c (hash_tree): Use it.
4748 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4750 * inchash.h (inchash::hash::add_wide_int): Rename to...
4751 (inchash::hash::add_hwi): ...this.
4752 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4753 (polymorphic_call_target_hasher::hash): Likewise.
4754 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4755 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4756 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4757 * lto-streamer-out.c (hash_tree): Likewise.
4758 * optc-save-gen.awk: Likewise.
4759 * tree.c (add_expr): Likewise.
4761 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4764 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4765 for ordered inequality comparisons even with TARGET_IEEE_FP.
4767 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4770 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4771 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4772 Expand with cmp<dwi>_doubleword.
4774 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4776 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4777 List CET intrinsics.
4778 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4779 specific to -fcf-protection option.
4781 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4783 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4784 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4785 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4786 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4787 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4788 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4789 (extra_objs): Add cet.o for Linux/x86 targets.
4790 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4791 * config/i386/cet.c: New file.
4792 * config/i386/cetintrin.h: Likewise.
4793 * config/i386/t-cet: Likewise.
4794 * config/i386/cpuid.h (bit_SHSTK): New.
4795 (bit_IBT): Likewise.
4796 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4797 pass IBT and SHSTK bits.
4798 * config/i386/i386-builtin-types.def
4799 (VOID_FTYPE_UNSIGNED_PVOID): New.
4800 (VOID_FTYPE_UINT64_PVOID): Likewise.
4801 * config/i386/i386-builtin.def: Add CET intrinsics.
4802 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4803 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4804 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4805 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4807 * config/i386/i386.c (rest_of_insert_endbranch): New.
4808 (pass_data_insert_endbranch): Likewise.
4809 (pass_insert_endbranch): Likewise.
4810 (make_pass_insert_endbranch): Likewise.
4811 (ix86_notrack_prefixed_insn_p): Likewise.
4812 (ix86_target_string): Add -mibt, -mshstk flags.
4813 (ix86_option_override_internal): Add flag_cf_protection
4815 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4816 (ix86_print_operand): Add 'notrack' prefix output.
4817 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4818 (ix86_expand_builtin): Expand CET intrinsics.
4819 (x86_output_mi_thunk): Add 'endbranch' instruction.
4820 * config/i386/i386.h (TARGET_IBT): New.
4821 (TARGET_IBT_P): Likewise.
4822 (TARGET_SHSTK): Likewise.
4823 (TARGET_SHSTK_P): Likewise.
4824 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4825 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4826 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4827 (builtin_setjmp_setup): New pattern.
4828 (builtin_longjmp): Likewise.
4829 (rdssp<mode>): Likewise.
4830 (incssp<mode>): Likewise.
4831 (saveprevssp): Likewise.
4832 (rstorssp): Likewise.
4833 (wrss<mode>): Likewise.
4834 (wruss<mode>): Likewise.
4835 (setssbsy): Likewise.
4836 (clrssbsy): Likewise.
4837 (nop_endbr): Likewise.
4838 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4840 * config/i386/immintrin.h: Include <cetintrin.h>.
4841 * config/i386/linux-common.h
4842 (file_end_indicate_exec_stack_and_cet): New prototype.
4843 (TARGET_ASM_FILE_END): New.
4845 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4847 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4848 latencies instead of having separate table; make difference between
4849 integer and float costs.
4850 * i386.h (processor_costs): Remove scalar_stmt_cost,
4851 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4852 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4854 * x86-tune-costs.h: Remove entries which has been removed in
4855 procesor_costs from all tables; make cond_taken_branch_cost
4856 and cond_not_taken_branch_cost COST_N_INSNS based.
4858 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4860 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4862 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4864 * config/i386/i386.md (isa): Remove fma_avx512f.
4865 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4866 <avx512>_fmadd_<mode>_mask3<round_name>,
4867 <avx512>_fmsub_<mode>_mask<round_name>,
4868 <avx512>_fmsub_<mode>_mask3<round_name>,
4869 <avx512>_fnmadd_<mode>_mask<round_name>,
4870 <avx512>_fnmadd_<mode>_mask3<round_name>,
4871 <avx512>_fnmsub_<mode>_mask<round_name>,
4872 <avx512>_fnmsub_<mode>_mask3<round_name>,
4873 <avx512>_fmaddsub_<mode>_mask<round_name>,
4874 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4875 <avx512>_fmsubadd_<mode>_mask<round_name>,
4876 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4877 (*vec_widen_umult_even_v16si<mask_name>,
4878 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4879 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4881 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4883 * extend.texi: Add 'nocf_check' documentation.
4884 * gimple.texi: Add second parameter to
4885 gimple_build_call_from_tree.
4886 * invoke.texi: Add -fcf-protection documentation.
4887 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4889 2017-10-20 Richard Biener <rguenther@suse.de>
4891 PR tree-optimization/82473
4892 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4893 the largest input type.
4895 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4897 * c-attribs.c (handle_nocf_check_attribute): New function.
4898 (c_common_attribute_table): Add 'nocf_check' handling.
4899 * gimple-parser.c: Add second argument NULL to
4900 gimple_build_call_from_tree.
4901 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4902 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4904 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4905 * common.opt: Add fcf-protection flag.
4906 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4907 * flag-types.h: Add enum cf_protection_level.
4908 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4909 Add 'nocf_check' attribute propagation to gimple call.
4910 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4911 (gimple_build_call_from_tree): Update prototype.
4912 (gimple_call_nocf_check_p): New function.
4913 (gimple_call_set_nocf_check): Likewise.
4914 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4915 * ipa-icf.c: Add nocf_check attribute in statement hash.
4916 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4917 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4918 * toplev.c (process_options): Add flag_cf_protection handling.
4920 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4922 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4924 2017-10-20 Richard Biener <rguenther@suse.de>
4926 PR tree-optimization/82603
4927 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4928 remove false predicated stores.
4930 2017-10-20 Richard Biener <rguenther@suse.de>
4932 * graphite-isl-ast-to-gimple.c
4933 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4934 Remove return value and simplify, dump copied stmt after lhs
4936 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4937 Reduce dump verbosity.
4938 (gsi_insert_earliest): Likewise.
4939 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4940 * graphite.c (print_global_statistics): Adjust dumping.
4941 (print_graphite_scop_statistics): Likewise.
4942 (print_graphite_statistics): Do not dump loops here.
4943 (graphite_transform_loops): But here.
4945 2017-10-20 Nicolas Roche <roche@adacore.com>
4947 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4948 * configure: Regenerate.
4950 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4953 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4954 functions when optimizing replace GIMPLE_RETURN stmts with
4955 calls to __builtin_unreachable ().
4958 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4959 for -fsanitize=thread link of executables.
4960 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4961 link of executables.
4964 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4966 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4967 define_insns for logical vector shifts to use VI248_AVX512BW
4968 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4969 condition, useless isa and prefix attributes. Change the first
4970 2 of these define_insns to ...
4971 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4972 define_insn for avx512vl.
4973 (<shift_insn><mode>3): ... and this, new define_insn without
4974 masking for non-avx512vl.
4977 * config/i386/sse.md (*andnot<mode>3,
4978 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4979 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4980 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4981 not applied use empty suffix even for TARGET_AVX512VL.
4982 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4983 is applied, supply evex,evex or evex,evex,evex instead of just
4986 2017-10-20 Julia Koval <julia.koval@intel.com>
4988 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4989 (OPTION_MASK_ISA_GFNI_UNSET): New.
4990 (ix86_handle_option): Handle OPT_mgfni.
4991 * config/i386/cpuid.h (bit_GFNI): New.
4992 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4993 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4994 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4995 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4996 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4997 * config/i386/i386.opt: Add mgfni.
4999 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
5001 * config/msp430/msp430.c (msp430_option_override): Disable
5002 -fdelete-null-pointer-checks.
5003 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
5005 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5007 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
5008 of x87 and SSE instructions.
5010 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
5012 * asan.c (create_cond_insert_point): Do not update edge count.
5013 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
5014 (afdo_propagate_circuit): Likewise.
5015 (afdo_calculate_branch_prob): Likewise.
5016 (afdo_annotate_cfg): Likewise.
5017 * basic-block.h (struct edge_def): Remove count.
5018 (edge_def::count): New accessor.
5019 * bb-reorder.c (rotate_loop): Update.
5020 (find_traces_1_round): Update.
5021 (connect_traces): Update.
5022 (sanitize_hot_paths): Update.
5023 * cfg.c (unchecked_make_edge): Update.
5024 (make_single_succ_edge): Update.
5025 (check_bb_profile): Update.
5026 (dump_edge_info): Update.
5027 (update_bb_profile_for_threading): Update.
5028 (scale_bbs_frequencies_int): Update.
5029 (scale_bbs_frequencies_gcov_type): Update.
5030 (scale_bbs_frequencies_profile_count): Update.
5031 (scale_bbs_frequencies): Update.
5032 * cfganal.c (connect_infinite_loops_to_exit): Update.
5033 * cfgbuild.c (compute_outgoing_frequencies): Update.
5034 (find_many_sub_basic_blocks): Update.
5035 * cfgcleanup.c (try_forward_edges): Update.
5036 (try_crossjump_to_edge): Update
5037 * cfgexpand.c (expand_gimple_cond): Update
5038 (expand_gimple_tailcall): Update
5039 (construct_exit_block): Update
5040 * cfghooks.c (verify_flow_info): Update
5041 (redirect_edge_succ_nodup): Update
5042 (split_edge): Update
5043 (make_forwarder_block): Update
5044 (duplicate_block): Update
5045 (account_profile_record): Update
5046 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
5047 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
5048 * cfgloopmanip.c (scale_loop_profile): Update.
5050 (lv_adjust_loop_entry_edge): Update.
5051 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
5052 (force_nonfallthru_and_redirect): Update.
5053 (purge_dead_edges): Update.
5054 (rtl_flow_call_edges_add): Update.
5055 * cgraphunit.c (init_lowered_empty_function): Update.
5056 (cgraph_node::expand_thunk): Update.
5057 * gimple-pretty-print.c (dump_probability): Update.
5058 (dump_edge_probability): Update.
5059 * gimple-ssa-isolate-paths.c (isolate_path): Update.
5060 * haifa-sched.c (sched_create_recovery_edges): Update.
5061 * hsa-gen.c (convert_switch_statements): Update.
5062 * ifcvt.c (dead_or_predicable): Update.
5063 * ipa-inline-transform.c (inline_transform): Update.
5064 * ipa-split.c (split_function): Update.
5065 * ipa-utils.c (ipa_merge_profiles): Update.
5066 * loop-doloop.c (add_test): Update.
5067 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
5068 * lto-streamer-in.c (input_cfg): Update.
5069 (input_function): Update.
5070 * lto-streamer-out.c (output_cfg): Update.
5071 * modulo-sched.c (sms_schedule): Update.
5072 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
5073 * predict.c (maybe_hot_edge_p): Update.
5074 (unlikely_executed_edge_p): Update.
5075 (probably_never_executed_edge_p): Update.
5076 (dump_prediction): Update.
5077 (drop_profile): Update.
5078 (propagate_unlikely_bbs_forward): Update.
5079 (determine_unlikely_bbs): Update.
5080 (force_edge_cold): Update.
5081 * profile.c (compute_branch_probabilities): Update.
5082 * reg-stack.c (better_edge): Update.
5083 * shrink-wrap.c (handle_simple_exit): Update.
5084 * tracer.c (better_p): Update.
5085 * trans-mem.c (expand_transaction): Update.
5086 (split_bb_make_tm_edge): Update.
5087 * tree-call-cdce.c: Update.
5088 * tree-cfg.c (gimple_find_sub_bbs): Update.
5089 (gimple_split_edge): Update.
5090 (gimple_duplicate_sese_region): Update.
5091 (gimple_duplicate_sese_tail): Update.
5092 (gimple_flow_call_edges_add): Update.
5093 (insert_cond_bb): Update.
5094 (execute_fixup_cfg): Update.
5095 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
5096 * tree-complex.c (expand_complex_div_wide): Update.
5097 * tree-eh.c (lower_resx): Update.
5098 (unsplit_eh): Update.
5099 (cleanup_empty_eh_move_lp): Update.
5100 * tree-inline.c (copy_edges_for_bb): Update.
5101 (freqs_to_counts): Update.
5102 (copy_cfg_body): Update.
5103 * tree-ssa-dce.c (remove_dead_stmt): Update.
5104 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5105 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5106 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5107 (unloop_loops): Update.
5108 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5109 * tree-ssa-loop-split.c (connect_loops): Update.
5110 (split_loop): Update.
5111 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5112 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5113 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5114 * tree-ssa-reassoc.c (branch_fixup): Update.
5115 * tree-ssa-tail-merge.c (replace_block_by): Update.
5116 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5117 (compute_path_counts): Update.
5118 (update_profile): Update.
5119 (recompute_probabilities): Update.
5120 (update_joiner_offpath_counts): Update.
5121 (estimated_freqs_path): Update.
5122 (freqs_to_counts_path): Update.
5123 (clear_counts_path): Update.
5124 (ssa_fix_duplicate_block_edges): Update.
5125 (duplicate_thread_path): Update.
5126 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5127 (case_bit_test_cmp): Update.
5128 (collect_switch_conv_info): Update.
5129 (gen_inbound_check): Update.
5130 (do_jump_if_equal): Update.
5131 (emit_cmp_and_jump_insns): Update.
5132 * tree-tailcall.c (decrease_profile): Update.
5133 (eliminate_tail_call): Update.
5134 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5135 (vect_do_peeling): Update.
5136 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5137 * ubsan.c (ubsan_expand_null_ifn): Update.
5138 (ubsan_expand_ptr_ifn): Update.
5139 * value-prof.c (gimple_divmod_fixed_value): Update.
5140 (gimple_mod_pow2): Update.
5141 (gimple_mod_subtract): Update.
5142 (gimple_ic): Update.
5143 (gimple_stringop_fixed_value): Update.
5145 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5148 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5150 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5152 PR rtl-optimization/82395
5153 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5154 based on non_spilled_static_chain_regno_p.
5156 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5158 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5159 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5160 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5162 2017-10-19 Martin Sebor <msebor@redhat.com>
5164 PR tree-optimization/82596
5165 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5167 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5169 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5170 (get_mem_refs_of_builtin_call): Likewise.
5171 * builtins.c (expand_builtin_apply): Adjust call to
5172 allocate_dynamic_stack_space.
5173 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5174 the third argument to allocate_dynamic_stack_space, otherwise -1.
5175 (expand_builtin): Deal with all alloca variants.
5176 (is_inexpensive_builtin): Likewise.
5177 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5178 * calls.c (special_function_p): Deal with all alloca variants.
5179 (initialize_argument_information): Adjust call to
5180 allocate_dynamic_stack_space.
5181 (expand_call): Likewise.
5182 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5183 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5184 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5185 use it for the stack usage computation.
5186 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5187 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5188 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5189 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5190 (in_loop_p): Remove first argument and useless check.
5191 (pass_walloca::execute): Remove useless test and adjust call to above.
5192 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5193 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5194 (gimplify_call_expr): Deal with all alloca variants.
5195 * hsa-gen.c (gen_hsa_alloca): Likewise.
5196 (gen_hsa_insns_for_call): Likewise.
5197 * ipa-pure-const.c (special_builtin_state): Likewise.
5198 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5199 * tree-object-size.c (alloc_object_size): Likewise.
5200 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5201 (call_may_clobber_ref_p_1): Likewise.
5202 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5203 (ccp_fold_stmt): Likewise.
5204 (optimize_stack_restore): Likewise.
5205 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5206 (mark_all_reaching_defs_necessary_1): Likewise.
5207 (propagate_necessity): Likewise.
5208 (eliminate_unnecessary_stmts): Likewise.
5209 * tree.c (build_common_builtin_nodes): Build
5210 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5211 (build_alloca_call_expr): New function.
5212 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5213 (CASE_BUILT_IN_ALLOCA): Likewise.
5214 (build_alloca_call_expr): Declare.
5215 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5217 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5220 * dwarf2out.c (new_die_raw): New static inline function.
5221 (new_die): Use it to create the DIE.
5222 (add_AT_external_die_ref): Likewise.
5223 (clone_die): Likewise.
5224 (clone_as_declaration): Likewise.
5225 (dwarf2out_vms_debug_main_pointer): Likewise.
5226 (base_type_die): Likewise. Remove early return for corner cases.
5227 Do not call add_pubtype on the DIE here.
5228 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5229 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5230 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5231 native order exists for base types, attach the DIE manually and call
5232 add_pubtype on it. Do not equate a reverse order DIE to the type.
5234 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5236 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5237 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5239 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5241 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5243 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5244 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5245 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5246 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5247 * opts.c (sanitizer_opts): Add builtin.
5248 * ubsan.c (instrument_builtin): New function.
5249 (pass_ubsan::execute): Call it.
5250 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5251 * doc/invoke.texi: Document -fsanitize=builtin.
5253 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5254 builtins, store max (log2 (align), 0) into uchar field instead of
5255 align into uptr field.
5256 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5257 store uchar 0 field instead of uptr 0 field.
5258 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5259 instead of passing one address of struct with 2 locations pass
5260 two addresses of structs with 1 location each.
5261 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5262 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5263 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5264 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5265 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5266 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5267 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5268 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5270 2017-10-19 Martin Liska <mliska@suse.cz>
5273 * file-find.c (remove_prefix): Remove.
5274 * file-find.h (remove_prefix): Likewise.
5275 * gcc-ar.c: Remove smartness of lookup.
5277 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5279 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5280 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5281 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5283 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5286 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5287 (setcc + and to xor + setcc): New peephole2.
5289 2017-10-19 Tom de Vries <tom@codesourcery.com>
5291 * doc/sourcebuild.texi (Test Directives, Variants of
5292 dg-require-support): Add dg-require-stack-size.
5294 2017-10-19 Martin Liska <mliska@suse.cz>
5297 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5298 that have a large alignment.
5299 (gimplify_target_expr): Likewise.
5301 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5303 PR rtl-optimization/82602
5304 * ira.c (rtx_moveable_p): Return false for volatile asm.
5306 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5309 * config/i386/i386-modes.def (CCGZ): New CC mode.
5310 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5311 * config/i386/predicates.md (ix86_comparison_operator):
5313 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5314 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5315 with double-word subtraction.
5316 (put_condition_code): Handle CCGZmode.
5318 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5320 * wide-int.cc (debug (const wide_int &)): New.
5321 (debug (const wide_int *)): New.
5322 (debug (const widest_int &)): New.
5323 (debug (const widest_int *)): New.
5325 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5328 * lra-constraints.c (curr_insn_transform): Use non-input operand
5329 instead of output one for matched reload.
5331 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5333 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5334 (tree-ssa-loop-ivopts.h): New header file.
5335 (struct builtin_info): New fields.
5336 (classify_builtin_1): Compute and record base and offset parts for
5337 memset builtin partition by calling strip_offset.
5338 (offset_cmp, fuse_memset_builtins): New functions.
5339 (finalize_partitions): Fuse adjacent memset partitions by calling
5341 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5342 Expose the interface.
5343 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5345 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5347 PR tree-optimization/82574
5348 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5349 that data reference must be executed exactly once per iteration
5350 against the outermost loop in nest.
5351 (classify_partition): Update call to above function.
5353 2017-10-18 Richard Biener <rguenther@suse.de>
5355 PR tree-optimization/82591
5356 * graphite.c (graphite_transform_loops): Move code gen message
5358 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5359 Here. Handle scop_to_isl_ast failing.
5360 (scop_to_isl_ast): Limit the number of ISL operations.
5362 2017-10-18 Richard Biener <rguenther@suse.de>
5364 * graphite-isl-ast-to-gimple.c
5365 (translate_isl_ast_to_gimple::set_rename): Simplify.
5366 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5367 (graphite_copy_stmts_from_block): ... here.
5368 (copy_bb_and_scalar_dependences): Simplify.
5369 (add_parameters_to_ivs_params): Canonicalize.
5370 (generate_entry_out_of_ssa_copies): Simplify.
5371 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5372 by passing in ISL dimension.
5373 (parameter_index_in_region_1): Rename to ...
5374 (parameter_index_in_region): ... this.
5375 (extract_affine): Adjust assert, pass down parameter index.
5376 (add_param_constraints): Use range-info when available.
5377 (build_scop_context): Adjust.
5378 * sese.c (new_sese_info): Adjust.
5379 (free_sese_info): Likewise.
5380 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5381 Remove unused typedefs.
5382 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5384 2017-10-18 Martin Liska <mliska@suse.cz>
5386 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5388 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5390 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5391 (s390_sched_init): Do not reset s390_sched_state if we entered the
5392 current basic block via a fallthru edge and all others are unlikely.
5394 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5396 * config/s390/s390.c (NUM_SIDES): New variable.
5397 (LONGRUNNING_THRESHOLD): New variable.
5398 (LATENCY_FACTOR): New variable.
5399 (s390_sched_score): Decrease score for long-running instructions on
5401 (s390_sched_variable_issue): Perform bookkeeping for long-running
5404 2017-10-18 Richard Biener <rguenther@suse.de>
5406 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5407 Simplify with removal of the parameter rename map.
5408 (set_rename): Likewise.
5409 (should_copy_to_new_region): Likewise.
5410 (graphite_copy_stmts_from_block): Likewise.
5411 (copy_bb_and_scalar_dependences): Remove initialization of
5412 unused copied_bb_map.
5414 (copy_internal_parameters): Likewise.
5415 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5416 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5417 Use INTEGRAL_TYPE_P.
5418 (parameter_index_in_region_1): Rename to ...
5419 (assign_parameter_index_in_region): ... this. Assert we have
5420 a parameter we handle.
5421 (scan_tree_for_params): Adjust.
5422 * sese.h (parameter_rename_map_t): Remove.
5423 (struct sese_info_t): Remove unused parameter_rename_map and
5424 copied_bb_map members.
5425 * sese.c (new_sese_info): Adjust.
5426 (free_sese_info): Likewise.
5428 2017-10-18 Martin Liska <mliska@suse.cz>
5431 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5432 on an abnormal edge.
5434 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5436 * doc/invoke.texi (ffunction-sections and fdata-sections):
5439 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5441 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5442 the use statement can throw internally.
5444 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5446 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5447 any RTX present on the RHS of a SET.
5448 * compare-elim.c (try_eliminate_compare): Restore comment.
5450 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5452 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5453 may be also called on tcc_type nodes.
5454 * langhooks.c (lhd_tree_size): Likewise.
5456 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5458 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5459 format_warning_at_substring.
5460 (maybe_warn): Convert source_range * param to a location_t. Pass
5461 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5462 (format_directive): Remove code to extract source_ranges and
5463 source_range * in favor of just a location_t.
5464 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5466 * substring-locations.c (format_warning_va): Convert
5467 source_range * param to a location_t.
5468 (format_warning_at_substring): Likewise.
5469 * substring-locations.h (format_warning_va): Likewise.
5470 (format_warning_at_substring): Likewise.
5472 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5474 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5476 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5477 and scatter/gather ops.
5479 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5480 vec_gather_load and vec_scatter_store.
5481 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5482 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5483 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5484 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5485 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5486 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5488 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5490 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5491 Detect FTST instruction and handle its register pops. Only pop
5492 second operand if can_pop_second_op is true.
5493 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5494 set can_pop_second_op to false in the compare_for_stack_reg call.
5496 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5497 output_fp_compare for stack register operands.
5498 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5499 instructions here. Do not emit stack register pops here. Assert
5500 that FCOMPP pops next to top stack register. Rewrite function.
5502 2017-10-17 Nathan Sidwell <nathan@acm.org>
5505 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5506 use DECL_ASSEMBLER_NAME_RAW.
5509 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5512 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5513 Wilco Dijkstra <wilco.dijkstra@arm.com>
5515 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5516 converstion to Pmode from the buf_addr.
5518 2017-10-17 Richard Biener <rguenther@suse.de>
5520 * graphite-dependences.c (scop_get_reads_and_writes): Change
5521 output parameters to references.
5523 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5525 PR 71026/tree-optimization
5526 * fold-const.c (distribute_real_division): Removed.
5527 (fold_binary_loc): Remove calls to distribute_real_divison.
5529 2017-10-17 Richard Biener <rguenther@suse.de>
5531 * graphite-scop-detection.c
5532 (scop_detection::stmt_has_simple_data_refs_p): Always use
5533 the full nest as region.
5534 (try_generate_gimple_bb): Likewise.
5535 * sese.c (scalar_evolution_in_region): Simplify now that
5536 SCEV can handle instantiation in regions.
5537 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5538 in the non-loop part of a function if requested.
5540 2017-10-17 Richard Biener <rguenther@suse.de>
5542 PR tree-optimization/82563
5543 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5545 (graphite_regenerate_ast_isl): Call it.
5546 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5548 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5550 PR tree-optimization/82549
5551 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5552 Formatting fixes. Instead of calling make_bit_field_ref with negative
5555 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5557 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5558 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5559 _mm_maskz_reduce_ss): New.
5560 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5561 __builtin_ia32_reducess_mask): Ditto..
5562 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5563 * config/i386/sse.md (reduces<mode>): Renamed to ...
5564 (reduces<mode><mask_scalar_name>): ... this.
5565 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5567 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5568 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5570 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5572 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5573 * selftest-run-tests.c (selftest::run_tests): Call
5574 selftest::unique_ptr_tests_cc_tests.
5575 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5576 * unique-ptr-tests.cc: New file.
5578 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5581 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5583 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5584 (make_hard_regno_born, make_hard_regno_dead): Update
5585 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5587 2017-10-16 Jeff Law <law@redhat.com>
5589 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5591 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5593 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5595 2017-10-16 Olivier Hainque <hainque@adacore.com>
5597 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5598 with_cpu if we were configured for an e500v2 target cpu name.
5600 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5602 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5603 * doc/invoke.texi: Document +nodsp as a valid extension for
5606 2017-10-16 Martin Liska <mliska@suse.cz>
5608 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5609 (test_set_range): Likewise.
5610 (test_range_functions): Rename to ...
5611 (test_bit_in_range): ... this.
5612 (sbitmap_c_tests): Add new test.
5614 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5616 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5618 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5619 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5621 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5623 * config/aarch64/aarch64-builtins.c
5624 (aarch64_types_quadopu_lane_qualifiers): New.
5625 (TYPES_QUADOPU_LANE): New.
5626 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5627 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5628 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5629 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5630 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5631 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5632 (Vdottype, DOTPROD): New.
5633 (sur): Add SDOT and UDOT.
5635 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5637 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5638 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5639 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5641 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5642 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5643 Enable TARGET_DOTPROD.
5644 (cortex-a75.cortex-a55): Likewise.
5645 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5647 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5649 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5650 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5652 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5654 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5655 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5656 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5657 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5658 * config/arm/types.md (neon_dot, neon_dot_q): New.
5659 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5661 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5663 * config/arm/arm.h (TARGET_DOTPROD): New.
5664 * config/arm/arm.c (arm_arch_dotprod): New.
5665 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5666 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5667 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5668 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5669 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5670 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5671 * doc/invoke.texi (armv8.2-a): Document dotprod
5673 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5675 * i386.c (ix86_vec_cost): New function.
5676 (ix86_rtx_costs): Handle vector operations better.
5677 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5678 * x86-tune-costs.h: Add new costs to all tables.
5680 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5682 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5684 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5685 divsd, sqrtss and sqrtsd
5686 * x86-tune-costs.h: Add new entries to all costs.
5687 (znver1_cost): Fix to match real instruction latencies.
5689 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5690 Michael Collison <michael.collison@arm.com>
5692 * compare-elim.c: Include emit-rtl.h.
5693 (can_merge_compare_into_arith): New function.
5694 (try_validate_parallel): Likewise.
5695 (try_merge_compare): Likewise.
5696 (try_eliminate_compare): Call the above when no previous clobber
5698 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5701 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5705 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5706 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5710 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5711 to be any operand_equal_p operands. For & (B - 1) require
5712 B to be power of 2. Recognize
5713 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5715 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5718 * optabs.c (expand_memory_blockage): Fix call of
5719 targetm.have_memory_blockage.
5721 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5724 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5725 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5726 objects to extra_objs instead of overwriting it.
5728 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5730 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5731 Use any_fp_register_operand as operand[3] predicate. Simplify
5732 equality test for operands[2] and operands[4] memory location.
5733 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5734 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5735 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5736 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5737 any_fp_register_operand as operand[1] predicate. Simplify
5738 equality test for operands[0] and operands[3] memory location.
5739 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5740 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5741 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5743 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5745 * target-insns.def: Add memory_blockage.
5746 * optabs.c (expand_memory_blockage): New function.
5747 (expand_asm_memory_barrier): Rename ...
5748 (expand_asm_memory_blockage): ... to this.
5749 (expand_mem_thread_fence): Call expand_memory_blockage
5750 instead of expand_asm_memory_barrier.
5751 (expand_mem_singnal_fence): Ditto.
5752 (expand_atomic_load): Ditto.
5753 (expand_atomic_store): Ditto.
5754 * doc/md.texi (Standard Pattern Names For Generation):
5755 Document memory_blockage instruction pattern.
5757 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5759 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5760 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5761 * config/rl78/rl78.md: New define_expand "adddi3".
5763 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5765 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5768 2017-10-13 Jeff Law <law@redhat.com>
5770 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5772 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5775 * internal-fn.c (expand_mul_overflow): If both operands have
5776 the same highpart of -1 or 0 and the topmost bit of lowpart
5777 is different, overflow is if res <= 0 rather than res < 0.
5779 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5781 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5782 TARGET_P9_VECTOR code for unaligned_load case.
5784 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5786 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5788 2017-10-13 Nathan Sidwell <nathan@acm.org>
5790 * tree-core.h (tree_contains_struct): Make bool.
5791 * tree.c (tree_contains_struct): Likewise.
5792 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5793 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5794 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5795 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5796 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5798 2017-10-13 Richard Biener <rguenther@suse.de>
5800 * graphite-isl-ast-to-gimple.c
5801 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5802 parameters and dominance check.
5803 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5804 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5805 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5806 Do not update SSA form here or do intermediate IL verification.
5807 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5808 (graphite_initialize): Remove check on the number of loops in
5809 the function and inline into graphite_transform_loops.
5810 (graphite_finalize): Inline into graphite_transform_loops.
5811 (graphite_transform_loops): Perform SSA update and IL verification
5813 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5815 2017-10-13 Richard Biener <rguenther@suse.de>
5817 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5818 graphite_expression_type_precision): Avoid global constructor
5820 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5821 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5822 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5823 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5824 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5825 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5827 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5830 * config/i386/i386.h (ix86_red_zone_size): New.
5831 * config/i386/i386.md (push peephole2s): Replace
5832 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5834 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5835 Alan Hayward <alan.hayward@arm.com>
5836 David Sherwood <david.sherwood@arm.com>
5838 * combine.c (can_change_dest_mode): Reject changes in
5839 REGMODE_NATURAL_SIZE.
5841 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5842 Alan Hayward <alan.hayward@arm.com>
5843 David Sherwood <david.sherwood@arm.com>
5845 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5846 (expand_debug_source_expr): Likewise.
5847 * combine.c (combine_simplify_rtx): Likewise.
5848 * cse.c (fold_rtx): Likewise.
5849 * fwprop.c (canonicalize_address): Likewise.
5850 * targhooks.c (default_shift_truncation_mask): 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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5857 (widened_mode): Likewise.
5858 (expand_unop): Likewise.
5859 * ree.c (transform_ifelse): Likewise.
5860 (merge_def_and_ext): Likewise.
5861 (combine_reaching_defs): Likewise.
5862 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5864 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5865 Alan Hayward <alan.hayward@arm.com>
5866 David Sherwood <david.sherwood@arm.com>
5868 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5869 * combine.c (gen_lowpart_for_combine): Likewise.
5870 * dwarf2out.c (rtl_for_decl_location): Likewise.
5871 * final.c (alter_subreg): Likewise.
5872 * rtlhooks.c (gen_lowpart_general): Likewise.
5873 (gen_lowpart_if_possible): Likewise.
5875 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5876 Alan Hayward <alan.hayward@arm.com>
5877 David Sherwood <david.sherwood@arm.com>
5879 * calls.c (expand_call): Use subreg_lowpart_offset.
5880 * cse.c (cse_insn): Likewise.
5881 * regcprop.c (copy_value): Likewise.
5882 (copyprop_hardreg_forward_1): Likewise.
5884 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5887 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5888 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5889 =Q constraints to +Q and into insn condition add check
5890 that operands[0] and operands[1] are equal.
5891 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5892 =Q constraints to +Q and into insn condition add check
5893 that operands[0] is equal to either operands[1] or operands[2].
5896 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5897 instead of handling MINUS_EXPR twice (once for each argument),
5898 canonicalize operand order and handle just once, use rtype where
5899 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5902 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5903 any values of __C while still being pattern recognizable as a simple
5906 2017-10-13 Richard Biener <rguenther@suse.de>
5908 PR tree-optimization/82451
5910 2017-10-02 Richard Biener <rguenther@suse.de>
5912 PR tree-optimization/82355
5913 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5914 a mapping for the enclosing loop but avoid generating one for
5916 (copy_bb_and_scalar_dependences): Remove premature codegen
5917 error on PHIs in blocks duplicated into multiple places.
5918 * graphite-scop-detection.c
5919 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5920 in the region use it as loop and nest to analyze the DR in.
5921 (try_generate_gimple_bb): Likewise.
5922 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5923 (add_loop_constraints): For blocks in a loop not in the region
5924 create a dimension with a single iteration.
5925 * sese.h (gbb_loop_at_index): Remove assert.
5927 * cfgloop.c (loop_preheader_edge): For the loop tree root
5928 return the single successor of the entry block.
5929 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5930 Reset the SCEV hashtable and niters.
5931 * graphite-scop-detection.c
5932 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5933 assert that we only have POLYNOMIAL_CHREC that vary in loops
5934 contained in the region.
5935 (scop_detection::graphite_can_represent_expr): Adjust.
5936 (scop_detection::stmt_has_simple_data_refs_p): For loops
5937 not in the region set loop to NULL. The nest is now the
5938 entry edge to the region.
5939 (try_generate_gimple_bb): Likewise.
5940 * sese.c (scalar_evolution_in_region): Adjust for
5941 instantiate_scev change.
5942 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5943 Make nest parameter the edge into the region.
5944 (create_data_ref): Likewise.
5945 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5946 entry edge into a region and adjust instantiate_scev calls.
5947 (create_data_ref): Likewise.
5948 (graphite_find_data_references_in_stmt): Likewise.
5949 (find_data_references_in_stmt): Pass the loop preheader edge
5950 from the nest argument.
5951 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5952 parameter the edge into the region.
5953 (instantiate_parameters): Use the loop preheader edge as entry.
5954 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5956 (get_instantiated_value_entry): Make instantiate_below parameter
5957 the edge into the region.
5958 (instantiate_scev_name): Likewise. Adjust dominance checks,
5959 when we cannot use loop-based instantiation instantiate by
5960 walking use-def chains.
5961 (instantiate_scev_poly): Adjust.
5962 (instantiate_scev_binary): Likewise.
5963 (instantiate_scev_convert): Likewise.
5964 (instantiate_scev_not): Likewise.
5965 (instantiate_array_ref): Remove.
5966 (instantiate_scev_3): Likewise.
5967 (instantiate_scev_2): Likewise.
5968 (instantiate_scev_1): Likewise.
5969 (instantiate_scev_r): Do not blindly handle N-operand trees.
5970 Do not instantiate array-refs. Handle all constants and invariants.
5971 (instantiate_scev): Make instantiate_below parameter
5972 the edge into the region.
5973 (resolve_mixers): Use the loop preheader edge for the region
5974 parameter to instantiate_scev_r.
5975 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5977 2017-10-13 Richard Biener <rguenther@suse.de>
5979 PR tree-optimization/82525
5980 * graphite-isl-ast-to-gimple.c
5981 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5983 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5984 Fail code generation when we cannot represent the isl integer.
5985 (binary_op_to_tree): Elide modulo operations that are no-ops
5986 in the type we code generate. Remove now superfluous code
5989 2017-10-13 Richard Biener <rguenther@suse.de>
5991 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5992 (scop_detection::harmful_loop_in_region): Remove premature
5993 IV type restriction.
5994 (scop_detection::graphite_can_represent_scev): We can handle
5995 pointer IVs just fine.
5997 2017-10-13 Alan Modra <amodra@gmail.com>
5999 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
6000 "Clobbers and Scratch Registers". Add paragraph on
6001 alternative to clobbers for scratch registers and OpenBLAS
6004 2017-10-13 Alan Modra <amodra@gmail.com>
6006 * doc/extend.texi (Clobbers): Correct vax example. Delete old
6007 example of a memory input for a string of known length. Move
6008 commentary out of table. Add a number of new examples
6009 covering array memory inputs.
6011 2017-10-12 Martin Liska <mliska@suse.cz>
6013 PR tree-optimization/82493
6014 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
6015 (test_range_functions): New function.
6016 (sbitmap_c_tests): Likewise.
6017 * selftest-run-tests.c (selftest::run_tests): Run new tests.
6018 * selftest.h (sbitmap_c_tests): New function.
6020 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
6022 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6024 * config/rs6000/amo.h: Fix spacing issue.
6026 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6029 * config/i386/i386.md (*ashl<mode>3_mask_1,
6030 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
6031 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
6034 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6036 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
6037 (profile_probability): Set max_probability
6038 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
6040 * profile-count.c (profile_probability::differs_from_p): Do not
6041 rely on max_probaiblity == 10000
6043 2017-10-12 Jeff Law <law@redhat.com>
6045 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
6048 2017-10-12 Martin Sebor <msebor@redhat.com>
6052 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
6053 (handle_alias_pairs): Call it.
6054 * common.opt (-Wattribute-alias): New option.
6055 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
6056 * doc/invoke.texi (-Wattribute-alias): Document.
6058 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
6061 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6063 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6065 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6066 (make_hard_regno_born, make_hard_regno_dead): Update
6067 bb_killed_pseudos and bb_gen_pseudos.
6069 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
6071 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
6073 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
6075 * config/alpha/alpha.c (alpha_split_conditional_move):
6076 Use std::swap instead of manually swapping.
6077 (alpha_stdarg_optimize_hook): Ditto.
6078 (alpha_canonicalize_comparison): Ditto.
6080 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6082 * tree-loop-distribution.c (struct builtin_info): New struct.
6083 (struct partition): Refactor fields into struct builtin_info.
6084 (partition_free): Free struct builtin_info.
6085 (build_size_arg_loc, build_addr_arg_loc): Delete.
6086 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
6087 information from struct builtin_info.
6088 (find_single_drs): New function refactored from classify_partition.
6089 Also moved builtin validity checks to this function.
6090 (compute_access_range, alloc_builtin): New functions.
6091 (classify_builtin_st, classify_builtin_ldst): New functions.
6092 (classify_partition): Refactor code into functions find_single_drs,
6093 classify_builtin_st and classify_builtin_ldst.
6094 (distribute_loop): Don't do runtime alias check when distributing
6096 (find_seed_stmts_for_distribution): New function.
6097 (pass_loop_distribution::execute): Refactor code finding seed
6098 stmts into above function. Support distribution for the innermost
6099 two-level loop nest. Adjust dump information.
6101 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6103 * tree-loop-distribution.c: Adjust the general comment.
6104 (NUM_PARTITION_THRESHOLD): New macro.
6105 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6106 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6107 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6108 in call to build_partition_graph.
6109 (finalize_partitions): New parameter. Make loop distribution more
6110 conservative by fusing more partitions.
6111 (distribute_loop): Don't do runtime alias check in case of loop nest
6113 (find_seed_stmts_for_distribution): New function.
6114 (prepare_perfect_loop_nest): New function.
6115 (pass_loop_distribution::execute): Refactor code finding seed stmts
6116 and loop nest into above functions. Support loop nest distribution.
6117 Adjust dump information accordingly.
6119 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6121 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6122 and set PTYPE_SEQUENTIAL for merged partition.
6124 2017-10-12 Richard Biener <rguenther@suse.de>
6126 PR tree-optimization/69728
6128 2017-09-19 Richard Biener <rguenther@suse.de>
6130 PR tree-optimization/69728
6131 * graphite-sese-to-poly.c (schedule_error): New global.
6132 (add_loop_schedule): Handle empty domain by failing the
6134 (build_original_schedule): Handle schedule_error.
6136 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6137 domain by returning an unchanged schedule.
6139 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6141 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6142 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6144 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6146 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6149 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6152 * expr.c (store_field): Don't optimize away bitsize == 0 store
6153 from CALL_EXPR with addressable return type.
6155 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6157 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6158 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6159 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6160 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6161 TARGET_ISEL instead of TARGET_ISEL<sel>.
6163 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6165 * config/rs6000/rs6000.c
6166 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6168 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6170 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6172 (reg_or_cint_operand): Fix comment.
6173 (reg_or_zero_operand): New predicate.
6174 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6175 * config/rs6000/rs6000.c (output_isel): Delete.
6176 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6177 instead of reg_or_cint_operand. Output instruction directly (not via
6179 (isel_unsigned_<mode>): Ditto.
6180 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6181 gpc_reg_operand. Add an instruction alternative for this. Output
6182 instruction directly.
6183 (*isel_reversed_unsigned_<mode>): Ditto.
6185 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6187 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6188 (TARGET_CANONICALIZE_COMPARISON): Define.
6190 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6193 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6194 Check whether the dest is REG before adding REG_EQUIV note.
6196 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6199 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6201 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6202 (make_hard_regno_born, make_hard_regno_dead): Update
6203 bb_killed_pseudos and bb_gen_pseudos.
6205 2017-10-11 Nathan Sidwell <nathan@acm.org>
6207 * incpath.h (enum incpath_kind): Name enum, prefix values.
6208 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6209 * incpath.c (heads, tails): Use INC_MAX.
6210 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6211 (merge_include_chains, split_quote_chain,
6212 register_include_chains): Update incpath_kind names.
6213 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6214 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6216 (add_framework_path, darwin_register_objc_includes): Likewise.
6217 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6219 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6221 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6222 Do not use float_operator operator predicate.
6223 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6224 * config/i386/predicates.md (float_operator): Remove predicate.
6226 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6228 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6229 (*jccxf_i387): Ditto.
6230 (*jcc<mode>_i387): Ditto.
6231 (*jccu<mode>_i387): Ditto.
6232 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6233 (*jcc_*_i387 splitters): Remove.
6234 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6235 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6236 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6239 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6241 * profile-count.h (slow_safe_scale_64bit): New function.
6242 (safe_scale_64bit): New inline.
6243 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6244 * profile-count.c: Include wide-int.h
6245 (slow_safe_scale_64bit): New.
6247 2017-10-11 Nathan Sidwell <nathan@acm.org>
6249 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6250 HAS_DECL_ASSEMBLER_NAME_P.
6251 * gimple-expr.c (gimple_decl_printable_name: Check
6252 HAS_DECL_ASSEMBLER_NAME_P too.
6253 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6254 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6255 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6256 * passes.c (rest_of_decl_compilation): Check
6257 HAS_DECL_ASSEMBLER_NAME_P too.
6258 * recog.c (verify_changes): Likewise.
6259 * tree-pretty-print.c (dump_decl_name): Likewise.
6260 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6262 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6263 (SET_DECL_ASSEMBLER_NAME): Use it.
6264 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6265 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6266 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6268 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6270 * config.gcc (i386, x86_64): Add extra objects.
6271 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6272 (ix86_min_insn_size): Declare.
6273 (ix86_issue_rate): Declare.
6274 (ix86_adjust_cost): Declare.
6275 (ia32_multipass_dfa_lookahead): Declare.
6276 (ix86_macro_fusion_p): Declare.
6277 (ix86_macro_fusion_pair_p): Declare.
6278 (ix86_bd_has_dispatch): Declare.
6279 (ix86_bd_do_dispatch): Declare.
6280 (ix86_core2i7_init_hooks): Declare.
6281 (ix86_atom_sched_reorder): Declare.
6282 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6283 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6284 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6285 (rip_relative_addr_p): Rename to ...
6286 (ix86_rip_relative_addr_p): ... this one; export.
6287 (memory_address_length): Update.
6288 (ix86_issue_rate): Move to x86-tune-sched.c.
6289 (ix86_flags_dependent): Move to x86-tune-sched.c.
6290 (ix86_agi_dependent): Move to x86-tune-sched.c.
6291 (exact_dependency_1): Move to x86-tune-sched.c.
6292 (exact_store_load_dependency): Move to x86-tune-sched.c.
6293 (ix86_adjust_cost): Move to x86-tune-sched.c.
6294 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6295 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6296 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6297 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6298 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6299 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6300 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6301 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6302 (min_insn_size): Rename to ...
6303 (ix86_min_insn_size): ... this one; export.
6304 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6305 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6306 (core2i7_first_cycle_multipass_backtrack): Move to
6307 x86-tune-sched-core.c.
6308 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6309 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6310 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6311 (ix86_avoid_jump_mispredicts): Update.
6312 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6313 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6314 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6315 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6316 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6317 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6318 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6319 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6320 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6321 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6322 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6323 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6324 (BIG): Move to ix86-tune-sched-bd.c.
6325 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6326 (enum insn_path): Move to ix86-tune-sched-bd.c.
6327 (get_mem_group): Move to ix86-tune-sched-bd.c.
6328 (is_cmp): Move to ix86-tune-sched-bd.c.
6329 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6330 (is_branch): Move to ix86-tune-sched-bd.c.
6331 (is_prefetch): Move to ix86-tune-sched-bd.c.
6332 (init_window): Move to ix86-tune-sched-bd.c.
6333 (allocate_window): Move to ix86-tune-sched-bd.c.
6334 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6335 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6336 (process_end_window): Move to ix86-tune-sched-bd.c.
6337 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6338 (find_constant): Move to ix86-tune-sched-bd.c.
6339 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6340 (has_immediate): Move to ix86-tune-sched-bd.c.
6341 (get_insn_path): Move to ix86-tune-sched-bd.c.
6342 (get_insn_group): Move to ix86-tune-sched-bd.c.
6343 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6344 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6345 (add_insn_window): Move to ix86-tune-sched-bd.c.
6346 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6347 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6348 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6349 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6350 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6351 (do_dispatch): Move to ix86-tune-sched-bd.c.
6352 (has_dispatch): Move to ix86-tune-sched-bd.c.
6353 * config/i386/t-i386: Add new object files.
6354 * config/i386/x86-tune-costs.h: New file.
6355 * config/i386/x86-tune-sched-atom.c: New file.
6356 * config/i386/x86-tune-sched-bd.c: New file.
6357 * config/i386/x86-tune-sched-core.c: New file.
6358 * config/i386/x86-tune-sched.c: New file.
6360 2017-10-11 Liu Hao <lh_mouse@126.com>
6362 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6363 the generic version below instead.
6364 (should_colorize): Recognize Windows consoles as terminals
6366 * pretty-print.c [__MINGW32__] (write_all): New function.
6367 [__MINGW32__] (find_esc_head): Likewise.
6368 [__MINGW32__] (find_esc_terminator): Likewise.
6369 [__MINGW32__] (eat_esc_sequence): Likewise.
6370 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6372 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6375 2017-10-11 Richard Biener <rguenther@suse.de>
6377 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6378 Properly call analyze_scalar_evolution with the loop of the stmt.
6380 2017-10-11 Richard Biener <rguenther@suse.de>
6382 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6383 * tree-core.h (tree_base): Add chrec_var union member.
6384 * tree.h (CHREC_VAR): Remove.
6385 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6386 * tree-chrec.h (build_polynomial_chrec): Adjust.
6387 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6388 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6390 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6392 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6393 * match.pd: ... here.
6394 ((T) X == (T) Y): Relax condition.
6396 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6398 PR tree-optimization/82472
6399 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6401 (break_alias_scc_partitions): Update postorder number.
6403 2017-10-11 Martin Liska <mliska@suse.cz>
6406 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6408 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6410 2017-10-11 Martin Liska <mliska@suse.cz>
6412 * print-rtl.c (print_insn): Move declaration of idbuf
6413 to same scope as name.
6415 2017-10-11 Martin Liska <mliska@suse.cz>
6420 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6423 2017-10-11 Martin Liska <mliska@suse.cz>
6426 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6429 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6431 * config.gcc (default_gnu_indirect_function): Default to yes for
6432 arm*-*-linux* with glibc.
6434 2017-10-11 Richard Biener <rguenther@suse.de>
6436 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6437 default-defs and types we do not want to analyze.
6438 (interpret_loop_phi): Replace unreachable code with an assert.
6439 (compute_scalar_evolution_in_loop): Remove and inline ...
6440 (analyze_scalar_evolution_1): ... here, replacing condition with
6441 what makes the intent clearer. Remove handling of cases
6442 get_scalar_evolution now handles.
6444 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6446 PR rtl-optimization/81434
6447 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6448 comment for main loop. In sched_group_found if, also add checks for
6449 pass and min_cost_group.
6451 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6453 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6454 (rs6000_insn_cost): New function.
6455 * config/rs6000/rs6000.md (cost): New attribute.
6457 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6458 H.J. Lu <hongjiu.lu@intel.com>
6462 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6463 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6464 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6465 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6466 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6467 OPTION_MASK_ISA_MMX and some other bit set require both
6468 mmx and the ISAs without the mmx bit.
6469 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6470 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6471 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6472 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6473 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6474 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6475 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6476 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6477 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6478 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6479 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6480 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6481 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6482 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6483 Add OPTION_MASK_ISA_MMX.
6485 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6487 * config.gcc (armv7*-*-freebsd*): New target.
6488 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6490 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6492 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6493 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6496 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6498 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6499 a template parameter.
6500 (WIDE_INT_REF_FOR): Update accordingly.
6501 * tree.h (wi::int_traits <const_tree>): Delete.
6502 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6503 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6504 (wi::tree_to_wide_ref): New typedef.
6505 (wi::to_wide): New function.
6506 * calls.c (get_size_range): Use wi::to_wide when operating on
6508 * cgraph.c (cgraph_node::create_thunk): Likewise.
6509 * config/i386/i386.c (ix86_data_alignment): Likewise.
6510 (ix86_local_alignment): Likewise.
6511 * dbxout.c (stabstr_O): Likewise.
6512 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6513 * expr.c (const_vector_from_tree): Likewise.
6514 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6515 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6516 (fold_negate_expr_1, int_const_binop_1, const_binop)
6517 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6518 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6519 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6520 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6521 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6522 (fold_not_const, round_up_loc): Likewise.
6523 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6524 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6525 (alloca_call_type): Likewise.
6526 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6527 * godump.c (go_output_typedef): Likewise.
6528 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6529 * internal-fn.c (get_min_precision): Likewise.
6530 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6531 * ipa-polymorphic-call.c
6532 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6533 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6534 (ipa_modify_call_arguments): Likewise.
6535 * match.pd: Likewise.
6536 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6537 * print-tree.c (print_node_brief, print_node): Likewise.
6538 * stmt.c (expand_case): Likewise.
6539 * stor-layout.c (layout_type): Likewise.
6540 * tree-affine.c (tree_to_aff_combination): Likewise.
6541 * tree-cfg.c (group_case_labels_stmt): Likewise.
6542 * tree-data-ref.c (dr_analyze_indices): Likewise.
6543 (prune_runtime_alias_test_list): Likewise.
6544 * tree-dump.c (dequeue_and_dump): Likewise.
6545 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6546 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6547 * tree-pretty-print.c (dump_generic_node): Likewise.
6548 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6549 (simple_iv_with_niters): Likewise.
6550 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6551 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6552 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6553 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6554 (refine_value_range_using_guard, number_of_iterations_ne_max)
6555 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6556 (get_cst_init_from_scev, record_nonwrapping_iv)
6557 (scev_var_range_cant_overflow): Likewise.
6558 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6559 * tree-ssa-pre.c (compute_avail): Likewise.
6560 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6561 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6562 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6563 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6564 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6565 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6566 (dump_case_nodes, try_switch_expansion): Likewise.
6567 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6568 (vect_do_peeling): Likewise.
6569 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6570 * tree-vect-stmts.c (vectorizable_load): Likewise.
6571 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6572 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6573 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6574 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6575 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6576 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6577 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6578 (evrp_dom_walker::before_dom_children): Likewise.
6579 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6580 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6581 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6582 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6583 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6584 (get_range_pos_neg): Likewise.
6585 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6586 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6587 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6588 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6589 * config/avr/avr.c (avr_fold_builtin): Likewise.
6590 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6591 * config/msp430/msp430.c (msp430_attr): Likewise.
6592 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6593 * config/powerpcspe/powerpcspe-c.c
6594 (altivec_resolve_overloaded_builtin): Likewise.
6595 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6596 (rs6000_expand_ternop_builtin): Likewise.
6597 * config/rs6000/rs6000-c.c
6598 (altivec_resolve_overloaded_builtin): Likewise.
6599 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6600 (rs6000_expand_ternop_builtin): Likewise.
6601 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6603 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6605 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6606 when copying loop nest with only one inner loop.
6608 2017-10-10 Richard Biener <rguenther@suse.de>
6610 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6611 blocks if SCEV is active.
6612 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6614 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6615 (scev_initialize): Assert we are not yet initialized.
6617 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6619 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6620 inner loop's exit stmt by making it always exit the loop, otherwise
6621 we would generate an infinite empty loop.
6623 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6625 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6626 renaming variables in new preheader if it's deleted.
6628 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6630 * tree-loop-distribution.c (struct partition): Remove unused field
6631 loops of the structure.
6632 (partition_alloc, partition_free): Ditto.
6633 (build_rdg_partition_for_vertex): Ditto.
6635 2017-10-09 Jeff Law <law@redhat.com>
6637 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6638 return type to match prototype and documentation.
6640 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6642 * config/rs6000/rs6000.c (processor_costs): Move to ...
6643 * config/rs6000/rs6000.h: ... here.
6644 (rs6000_cost): Declare.
6646 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6648 * except.c (setjmp_fn): New global variable.
6649 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6650 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6651 if DONT_USE_BUILTIN_SETJMP is defined.
6653 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6655 * target.def (insn_cost): New hook.
6656 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6657 * doc/tm.texi: Regenerate.
6658 * rtlanal.c (insn_cost): Use the new hook.
6660 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6662 * combine.c (combine_validate_cost): Compute the new insn_cost,
6663 not just pattern_cost.
6664 (try_combine): Adjust comment.
6666 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6668 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6670 * combine.c (uid_insn_cost): Adjust comment.
6671 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6673 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6674 * dse.c (find_shift_sequence): Ditto.
6675 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6676 (bb_valid_for_noce_process_p): Use pattern_cost.
6677 * rtl.h (insn_rtx_cost): Delete.
6678 (pattern_cost): New prototype.
6679 (insn_cost): New prototype.
6680 * rtlanal.c (insn_rtx_cost): Rename to...
6681 (pattern_cost): ... this.
6684 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6686 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6687 (*jcc<mode>_0_r_i387): Ditto.
6688 (*jccxf_r_i387): Ditto.
6689 (*jcc<mode>_r_i387): Ditto.
6690 (*jccu<mode>_r_i387): Ditto.
6691 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6692 (*jcc): Rename from *jcc_1.
6694 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6696 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6697 deferred rescans after the lvx/stvx recombination pre-pass.
6699 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6701 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6702 memory operation instruction support.
6703 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6704 (rs6000-ibm-aix[789]*): Likewise.
6705 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6706 Document new functions.
6708 2017-10-09 Richard Biener <rguenther@suse.de>
6710 PR tree-optimization/82397
6711 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6712 equality only for semantically equal trees.
6714 2017-10-09 Richard Biener <rguenther@suse.de>
6716 PR tree-optimization/82449
6717 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6718 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6719 allow constant addresses.
6720 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6723 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6725 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6728 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6731 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6734 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6737 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6739 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6742 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6743 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6745 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6747 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6748 (WI_BINARY_PREDICATE_RESULT): Likewise.
6749 (wi::binary_traits::operator_result): New type.
6750 (wi::binary_traits::predicate_result): Likewise.
6751 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6752 (generic_wide_int::operator==, generic_wide_int::operator!=)
6753 (generic_wide_int::operator&, generic_wide_int::and_not)
6754 (generic_wide_int::operator|, generic_wide_int::or_not)
6755 (generic_wide_int::operator^, generic_wide_int::operator+
6756 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6758 (operator~, unary operator-, operator==, operator!=, operator&)
6759 (operator|, operator^, operator+, binary operator-, operator*): New
6761 * expr.c (get_inner_reference): Use wi::bit_and_not.
6762 * fold-const.c (fold_binary_loc): Likewise.
6763 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6764 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6765 (bit_value_binop): Likewise.
6766 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6767 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6768 (extract_range_from_binary_expr_1): Likewise.
6769 (masked_increment): Likewise.
6770 (simplify_bit_ops_using_ranges): Likewise.
6772 2017-10-09 Martin Jambor <mjambor@suse.cz>
6775 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6776 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6777 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6779 (hsa_fixup_mov_insn_type): New function.
6780 (hsa_op_with_type::get_in_type): Use it.
6781 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6782 immediates in an assert.
6783 (hsa_op_with_type::extend_int_to_32bit): New method.
6784 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6785 types. Convert to dest type if necessary.
6786 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6787 (reg_for_gimple_ssa): Pass false as min32int to
6788 hsa_type_for_scalar_tree_type.
6789 (gen_hsa_addr): Fixup type when creating addresable temporary.
6790 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6791 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6792 necessary. Call hsa_fixup_mov_insn_type.
6793 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6794 extend operands and convert to dest type if necessary.
6795 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6796 to dest type if necessary.
6797 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6798 if conversion nt necessary and size matches.
6799 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6800 to dest type if necessary.
6801 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6802 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6804 (gen_hsa_clrsb): Likewise.
6805 (gen_hsa_ffs): Likewise.
6806 (gen_hsa_divmod): Extend operands and convert to dest type if
6808 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6810 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6812 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6813 Remove empty default arguments. Use a brace block as output
6815 (conditional return): Ditto.
6817 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6818 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6820 (group_ending_nop): Ditto.
6821 (doloop_end): Ditto.
6822 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6823 (splitters for those): Ditto.
6825 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6827 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6828 a conditional jump (and the compare for it) so that pc_rtx is the
6830 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6831 for the deleted and renamed ctr<mode>_internal[234] patterns.
6832 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6833 Delete second conditional return pattern.
6834 (ctr<mode>_internal2): Delete this second bdnz pattern.
6835 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6836 (ctr<mode>_internal4): Delete this second bdz pattern.
6838 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6840 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6841 (always_initialized_rtx_for_ssa_name_p): New predicate.
6842 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6843 (finish_out_of_ssa): Free new field of SA.
6844 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6845 * tree-ssa-coalesce.c: Include tree-ssa.h.
6846 (get_parm_default_def_partitions): Remove extern keyword.
6847 (get_undefined_value_partitions): New function.
6848 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6849 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6851 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6853 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6855 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6857 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6859 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6860 for targets that preffer 128bit.
6862 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6864 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6866 2017-10-08 Olivier Hainque <hainque@adacore.com>
6868 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6869 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6871 (thumb_set_return_address): Likewise.
6873 2017-10-08 Olivier Hainque <hainque@adacore.com>
6875 * common/config/arm/arm-common.c (arm_except_unwind_info):
6876 Handle DWARF2_UNWIND_INFO.
6878 2017-10-07 Michael Collison <michael.collison@arm.com>
6880 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6883 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6885 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6886 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6887 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6888 defined, force the creation of a new block for a dispatch label.
6890 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6892 * invoke.texi (Wsuggest-attribute=cold): Document.
6893 * common.opt (Wsuggest-attribute=cold): New
6894 * ipa-pure-const.c (warn_function_cold): New function.
6895 * predict.c (compute_function_frequency): Use it.
6896 * predict.h (warn_function_cold): Declare.
6898 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6900 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6903 2017-10-06 Martin Liska <mliska@suse.cz>
6905 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6906 keyword for member functions.
6907 (struct sanopt_tree_couple): New struct.
6908 (struct sanopt_tree_couple_hash): New function.
6909 (struct sanopt_ctx): Add new hash_map.
6910 (has_dominating_ubsan_ptr_check): New function.
6911 (record_ubsan_ptr_check_stmt): Likewise.
6912 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6913 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6914 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6916 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6919 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6920 aarch64_simd_valid_immediate on CONST_VECTORs.
6921 (aarch64_reg_or_bic_imm): Likewise.
6923 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6925 PR rtl-optimization/82396
6926 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6928 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6930 * graphite-dependences.c (scop_get_reads): Move code to...
6931 (scop_get_must_writes): Move code to...
6932 (scop_get_may_writes): Move code to...
6933 (scop_get_reads_and_writes): ... here.
6934 (scop_get_dependences): Call scop_get_reads_and_writes.
6936 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6938 PR tree-optimization/82434
6939 * fold-const.h (can_native_encode_type_p,
6940 can_native_encode_string_p): Remove.
6941 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6942 don't encode anything, just return what would be otherwise returned.
6943 (native_encode_fixed, native_encode_complex, native_encode_vector):
6945 (native_encode_string): Likewise. Inline by hand
6946 can_native_encode_string_p.
6947 (can_native_encode_type_p): Remove.
6948 (can_native_encode_string_p): Remove.
6949 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6950 STRING_CSTs using can_native_encode_string_p, test all
6951 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6952 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6953 argument from native_encode_expr.
6954 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6955 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6958 2017-10-06 Richard Biener <rguenther@suse.de>
6960 PR tree-optimization/82397
6961 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6962 operand_equal_p but rely on data_ref_compare_tree for detecting
6964 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6965 to match up with dr_group_sort_cmp.
6967 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6970 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6972 * config/s390/s390-builtin-types.def: Regenerate.
6974 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6977 * config/s390/s390-builtin-types.def: Regenerate.
6978 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6979 Change flag from B_VXE to B_VX.
6980 (s390_vec_min_dbl): Remove B_VXE flag.
6982 2017-10-06 Richard Biener <rguenther@suse.de>
6984 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6985 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6986 translate_isl_ast_to_gimple::is_valid_rename,
6987 translate_isl_ast_to_gimple::get_rename,
6988 translate_isl_ast_to_gimple::get_def_bb_for_const,
6989 translate_isl_ast_to_gimple::get_new_name,
6990 translate_isl_ast_to_gimple::collect_all_ssa_names,
6991 translate_isl_ast_to_gimple::copy_loop_phi_args,
6992 translate_isl_ast_to_gimple::collect_all_ssa_names,
6993 translate_isl_ast_to_gimple::copy_loop_phi_args,
6994 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6995 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6996 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6997 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6998 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6999 translate_isl_ast_to_gimple::copy_cond_phi_args,
7000 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
7001 translate_isl_ast_to_gimple::edge_for_new_close_phis,
7002 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
7003 translate_isl_ast_to_gimple::rename_uses,
7004 translate_isl_ast_to_gimple::rename_all_uses): Remove.
7005 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
7006 (set_rename_for_each_def): Likewise.
7007 (graphite_copy_stmts_from_block): Handle debug stmt resetting
7008 here. Handle rewriting SCEV analyzable uses here.
7009 (copy_bb_and_scalar_dependences): Generate code for PHI
7011 (graphite_regenerate_ast_isl): Adjust.
7012 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
7013 (add_write, add_read): New functions.
7014 (build_cross_bb_scalars_def): Use it and simplify.
7015 (build_cross_bb_scalars_use): Likewise.
7016 (graphite_find_cross_bb_scalar_vars): Inline into...
7017 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
7018 simulating out-of-SSA. Compute liveout and add dependencies.
7019 (build_scops): Force an empty entry block.
7020 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
7022 (sese_build_liveouts): Declare.
7023 (sese_trivially_empty_bb_p): Likewise.
7024 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
7025 compute liveout and debug_liveout.
7026 (sese_bad_liveouts_use): Remove.
7027 (sese_reset_debug_liveouts_bb): Likewise.
7028 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
7029 (sese_build_liveouts): Build liveout and debug_liveout and store
7031 (new_sese_info): Adjust.
7032 (free_sese_info): Likewise.
7033 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
7034 do not build liveout here.
7035 (move_sese_in_condition): Adjust region entry.
7036 (scev_analyzable_p): Match up with chrec_apply requirements.
7037 (sese_trivially_empty_bb_p): New.
7038 * tree-into-ssa.c (get_reaching_def): Properly support generating
7039 default-defs for incremental rewrite of anonymous names.
7041 2017-10-06 Richard Biener <rguenther@suse.de>
7043 * graphite-sese-to-poly.c (extract_affine): For casts increasing
7044 precision do not perform modulo reduction.
7046 2017-10-06 Richard Biener <rguenther@suse.de>
7048 PR tree-optimization/82436
7049 * tree-vect-slp.c (vect_supported_load_permutation_p): More
7050 conservatively choose the vectorization factor when checking
7051 whether we can perform the required load permutation.
7052 (vect_transform_slp_perm_load): Assert when we may not fail.
7054 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7056 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
7057 message for incompatible -msdata=* and -mcall-* options.
7059 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7061 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
7062 rate for post-reload scheduling.
7064 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7066 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
7068 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7070 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
7071 to improve monte carlo in scimark.
7073 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
7075 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
7076 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
7077 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
7078 pentium4_cost, nocona_cost): Set reassociation width to 1.
7079 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
7080 width to 2 for fp operations and 1 otherwise.
7081 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
7083 (atom_cost): Set reassociation width to 2.
7084 (slm_cost, generic_cost): Set fp reassociation width
7085 to 2 and 1 otherwise.
7086 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
7087 (core_cost): Set fp reassociation width to 4 and vector to 2.
7088 (ix86_reassociation_width): Rewrite using cost table; special case
7089 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
7090 and TARGET_AVX128_OPTIMAL.
7091 * config/i386/i386.h (processor_costs): Add
7092 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
7093 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
7094 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
7095 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
7096 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
7097 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
7099 2017-10-05 Nathan Sidwell <nathan@acm.org>
7101 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7103 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7105 * config/arm/arm.c (arm_test_fpu_data): New.
7106 (arm_run_selftests): Call arm_test_fpu_data.
7108 2017-10-04 Nathan Sidwell <nathan@acm.org>
7110 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7112 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7114 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7116 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7117 check type for aarch64_simd_valid_immediate.
7118 (aarch64_output_simd_mov_immediate): Update prototype.
7119 (aarch64_simd_valid_immediate): Update prototype.
7120 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7121 support for ORR-immediate.
7122 (and<mode>3): modified pattern to add support for BIC-immediate.
7123 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7124 now checks for valid immediate for BIC and ORR based on new enum
7126 (aarch64_output_simd_mov_immediate): Function now used to output
7127 BIC/ORR imm as well based on new enum argument.
7128 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7130 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7131 (aarch64_reg_or_bic_imm): Likewise.
7133 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7135 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7136 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7138 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7142 PR rtl-optimization/82396
7143 * haifa-sched.c (autopref_multipass_init): Simplify
7145 (autopref_rank_data): Simplify sort order.
7146 * sched-int.h (autopref_multipass_data_): Remove
7147 multi_mem_insn_p, min_offset and max_offset.
7149 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7151 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7153 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7155 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7158 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7160 * doc/sourcebuild.texi: Document vect_long_mult.
7162 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7164 PR tree-optimization/82413
7165 * fold-const.c (build_range_check): Use widest_int when comparing
7166 the maximum ETYPE value with HIGH.
7168 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7170 PR rtl-optimization/82396
7171 * haifa-sched.c (autopref_multipass_init): Simplify
7173 (autopref_rank_data): Simplify sort order.
7174 * sched-int.h (autopref_multipass_data_): Remove
7175 multi_mem_insn_p, min_offset and max_offset.
7177 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7179 PR tree-optimization/82381
7180 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7181 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7184 PR tree-optimization/82374
7185 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7186 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7187 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7188 current_function_decl to the new decl.
7190 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7192 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7193 helper macro for IEEE float128 hardware built-in functions.
7194 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7196 (TRUNCF128_ODD): Likewise.
7197 (ADDF128_ODD): Likewise.
7198 (SUBF128_ODD): Likewise.
7199 (MULF128_ODD): Likewise.
7200 (DIVF128_ODD): Likewise.
7201 (FMAF128_ODD): Likewise.
7202 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7203 UNSPEC_TRUNC_ROUND_TO_ODD.
7204 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7205 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7206 floating point round to odd instructions.
7207 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7208 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7209 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7210 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7211 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7212 (trunc<mode>sf2_hw): Change the truncate with round to odd
7213 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7214 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7215 to odd hardware instructions.
7216 (sub<mode>3_odd): Likewise.
7217 (mul<mode>3_odd): Likewise.
7218 (div<mode>3_odd): Likewise.
7219 (sqrt<mode>2_odd): Likewise.
7220 (fma<mode>4_odd): Likewise.
7221 (fms<mode>4_odd): Likewise.
7222 (nfma<mode>4_odd): Likewise.
7223 (nfms<mode>4_odd): Likewise.
7224 (trunc<mode>df2_odd): Change the truncate with round to odd
7225 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7227 * doc/extend.texi (PowerPC built-in functions): Update documentation
7228 for existing IEEE float128-bit built-in functions. Add built-in
7229 functions that generate the IEEE 128-bit floating point round to
7232 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7234 PR rtl-optimization/77729
7235 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7236 to (X&(C1&~C2))|C2 transformations.
7238 2017-10-03 Martin Jambor <mjambor@suse.cz>
7240 PR tree-optimization/82363
7241 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7242 mismatch, mark lacc written regardless of racc.
7244 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7246 PR tree-optimization/82381
7247 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7248 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7249 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7250 fallthrough into reassoc_stmt_dominates_stmt_p.
7253 * combine.c (combine_instructions): Don't combine in unreachable
7256 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7259 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7260 function to not use the have_cpu variable. Do not set cpu_index,
7261 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7263 (rs6000_valid_attribute_p): Remove duplicate initializations of
7264 old_optimize and func_optimize.
7265 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7266 (rs6000_activate_target_options): Make global.
7267 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7270 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7272 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7273 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7274 Set *poffset to 0 and *psize and *pmax_size to -1 if
7275 *poffset + *psize overflows in HOST_WIDE_INT.
7277 PR tree-optimization/82387
7278 PR tree-optimization/82388
7279 PR tree-optimization/82389
7280 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7281 instead of live_bytes non-NULL.
7283 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7286 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7289 2017-10-02 Richard Biener <rguenther@suse.de>
7291 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7292 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7293 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7295 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7297 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7298 requested precision matches the type's.
7299 * calls.c (alloc_max_size): Calculate the new candidate size as
7300 a widest_int and use wi::to_widest when comparing it with the
7301 current candidate size.
7302 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7303 zero rather than integer_zero_node.
7304 * match.pd: Check for a no-op conversion before using wi::add
7305 rather than after. Use tree_to_uhwi when summing small shift
7306 counts into an unsigned int.
7308 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7309 Alan Hayward <alan.hayward@arm.com>
7310 David Sherwood <david.sherwood@arm.com>
7313 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7314 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7315 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7316 POINTER_AND_FP_REGS.
7318 2017-10-02 Richard Biener <rguenther@suse.de>
7320 PR tree-optimization/82355
7321 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7322 a mapping for the enclosing loop but avoid generating one for
7324 (copy_bb_and_scalar_dependences): Remove premature codegen
7325 error on PHIs in blocks duplicated into multiple places.
7326 * graphite-scop-detection.c
7327 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7328 in the region use it as loop and nest to analyze the DR in.
7329 (try_generate_gimple_bb): Likewise.
7330 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7331 (add_loop_constraints): For blocks in a loop not in the region
7332 create a dimension with a single iteration.
7333 * sese.h (gbb_loop_at_index): Remove assert.
7335 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7337 * omp-expand.c (adjust_context_scope): New function.
7338 (expand_parallel_call): Call adjust_context_scope.
7340 2017-10-01 Jeff Law <law@redhat.com>
7342 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7343 dom_opt_dom_walker class with direct access to private members.
7344 Add comments. Call test_for_singularity.
7345 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7346 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7347 m_dummy_cond anymore.
7348 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7350 (pass_dominator:execute): Build the dummy_cond here and pass it
7351 to the dom_opt_dom_walker ctor.
7352 (test_for_singularity): New function.
7354 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7355 Maya Rashish <coypu@sdf.org>
7357 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7358 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7359 (alpha*-*-netbsd*) Use nbsd_tm_file.
7360 (arm*-*-netbsdelf*) Likewise.
7361 (i[34567]86-*-netbsdelf*) Likewise.
7362 (x86_64-*-netbsd*) Likewise.
7363 (mips*-*-netbsd*) Likewise.
7364 (powerpc-*-netbsd*) Likewise.
7365 (sh*-*-netbsd*) Likewise.
7366 (sparc-*-netbsdelf*) Likewise.
7367 (sparc64-*-netbsd*) Likewise.
7368 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7370 (vax-*-netbsdelf*) Likewise.
7371 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7372 (UINT_FAST8_TYPE) Likewise.
7373 (INT_FAST16_TYPE) Check CHAR_FAST16.
7374 (UINT_FAST16_TYPE) Likewise.
7376 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7379 * config/i386/i386.md
7380 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7381 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7382 *divmodsi4_zext_2): New define_insn_and_split.
7383 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7384 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7385 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7386 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7387 New define_insn_and_split.
7388 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7389 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7390 operands[1] having DImode when mode is SImode.
7392 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7393 always SImode for DIV and MOD in REG_EQUAL notes.
7395 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7398 * match.pd: Fix handling of NaNs in pattern.
7400 2017-09-29 Jeff Law <law@redhat.com>
7402 * sbitmap.c (bitmap_bit_in_range_p): New function.
7403 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7404 * tree-ssa-dse.c (live_bytes_read): New function.
7405 (dse_classify_store): Ignore reads of dead bytes.
7407 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7408 typos and whitespace errors.
7409 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7410 * config/s390/s390.c (s390_emit_prologue): Likewise.
7412 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7415 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7416 with a symbol for LRA.
7418 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7420 PR rtl-optimization/82338
7421 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7423 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7425 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7426 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7427 (qsort_chk): Declare.
7428 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7429 (qsort_chk): New function.
7431 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7433 PR tree-optimization/82337
7434 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7435 phi definition if the PHI result appears in an abnormal PHI.
7436 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7437 basis appears in an abnormal PHI.
7439 2017-09-29 Richard Biener <rguenther@suse.de>
7441 * graphite-isl-ast-to-gimple.c
7442 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7443 (binary_op_to_tree): Use it.
7444 (get_rename_from_scev): Likewise.
7445 (copy_loop_phi_nodes): Likewise.
7446 (copy_bb_and_scalar_dependences): Likewise.
7447 (translate_pending_phi_nodes): Likewise.
7449 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7452 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7453 for movabsq $(i32 << shift), r64.
7455 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7457 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7458 index when encoding %esp as %rsp to avoid 0x67 prefix.
7460 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7462 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7463 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7465 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7467 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7468 Extensions with more than 16 double VFP registers.
7469 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7470 to_clear_mask and all code related to it. Replace the remaining
7471 entry by a sbitmap and adapt code accordingly.
7473 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7475 * brig-builtins.def: Change pure attributes to const.
7477 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7479 * config.gcc (default_gnu_indirect_function): Default to yes for
7480 sparc*-*-linux* with glibc.
7482 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7484 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7485 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7486 when creating .init_array and .fini_array sections with priority
7489 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7492 * config/aarch64/aarch64.c
7493 (aarch64_builtin_support_vector_misalignment): Always return false
7494 when misalignment is unknown.
7496 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7498 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7499 this function to return false if the definition used by the swap
7500 instruction is artificial, or if the memory address from which the
7501 constant value is loaded is not represented by a base address held
7502 in a register or if the base address register is a frame or stack
7503 pointer. Additionally, return false if the base address of the
7504 loaded constant is a SYMBOL_REF but is not considered to be a
7506 (replace_swapped_load_constant): New function.
7507 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7508 loaded constant vector with a load of a swapped constant vector.
7510 2017-09-27 Carl Love <cel@us.ibm.com>
7512 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7513 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7514 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7517 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7519 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7520 first, always call autopref_rank_data otherwise.
7522 2017-09-27 Richard Biener <rguenther@suse.de>
7524 * graphite-scop-detection.c (find_scop_parameters): Move
7525 loop bound handling ...
7526 (gather_bbs::before_dom_children): ... here, avoiding the need
7527 to build scop_info->loop_nest.
7528 (record_loop_in_sese): Remove.
7529 * sese.h (sese_info_t::loop_nest): Remove.
7530 * sese.c (new_sese_info): Do not allocate loop_nest.
7531 (free_sese_info): Do not free loop_nest.
7533 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7536 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7537 lhs from calls if the lhs has addressable type.
7539 2017-09-27 Richard Biener <rguenther@suse.de>
7541 * graphite.h (scop::max_alias_set): New member.
7542 * graphite-scop-detection.c: Remove references to non-existing
7543 --param in comments.
7544 (build_alias_sets): Record the maximum alias set used for drs.
7545 (build_scops): Support zero as unlimited for
7546 --param graphite-max-arrays-per-scop.
7547 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7549 (build_poly_sr_1): ... here. Compute alias set based on the
7550 maximum alias set used for drs rather than
7551 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7553 2017-09-27 Richard Biener <rguenther@suse.de>
7555 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7556 --param loop-block-tile-size=0 to disable tiling.
7558 2017-09-27 Richard Biener <rguenther@suse.de>
7560 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7561 (graphite-max-nb-scop-params): Document special value zero.
7562 * domwalk.h (dom_walker::STOP): New symbolical constant.
7563 (dom_walker::dom_walker): Add optional parameter for bb to
7565 (dom_walker::~dom_walker): Declare.
7566 (dom_walker::before_dom_children): Document STOP return value.
7567 (dom_walker::m_user_bb_to_rpo): New member.
7568 (dom_walker::m_bb_to_rpo): Likewise.
7569 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7570 mapping here if not provided by the user.
7571 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7572 provided by the user.
7573 (dom_walker::STOP): Define.
7574 (dom_walker::walk): Do not compute bb to RPO mapping here.
7575 Support STOP return value from before_dom_children to stop
7577 * graphite-optimize-isl.c (optimize_isl): If the schedule
7578 is the same still generate code if -fgraphite-identity
7579 or -floop-parallelize-all are given.
7580 * graphite-scop-detection.c: Include cfganal.h.
7581 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7583 (gather_bbs::before_dom_children): Return STOP for BBs
7585 (build_scops): Compute bb to RPO mapping and pass it to
7586 the domwalk. Treat --param graphite-max-nb-scop-params=0
7587 as not limiting the number of params.
7588 * graphite.c (graphite_initialize): Remove limit on the
7589 number of basic-blocks in a function.
7590 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7591 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7592 default value of 10.
7594 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7596 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7597 Adjust code to eliminate needing to do the shift right 32-bits
7598 operation after XSCVDPSPN.
7600 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7602 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7603 ((X / Y) != 0 -> X >= Y): Likewise.
7605 2017-09-26 Carl Love <cel@us.ibm.com>
7607 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7608 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7609 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7610 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7611 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7612 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7613 definitions and overloading.
7614 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7615 statement for P9V_BUILTIN_XST_LEN_R.
7616 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7617 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7618 define_expand and define_insn for the instructions and builtins.
7619 * doc/extend.texi: Update the built-in documentation file for the new
7621 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7622 define_insn for the instructions
7624 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7627 * gcc/config/netbsd-protos.h: New file.
7628 * gcc/config/netbsd.c: New file.
7629 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7630 * gcc/config/t-netbsd: New file.
7631 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7632 (tmake_file) Add t-netbsd.
7633 (extra_objs) Add netbsd.o.
7635 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7639 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7641 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7643 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7644 sign extension from a vector register to a GPR by doing a 32-bit
7645 direct move and then an EXTSW.
7646 (extendsi<mode>2 splitter): Likewise.
7647 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7648 right or vector extract after doing XSCVDPSPN. Use
7649 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7651 (movdi_from_sf_zero_ext): Likewise.
7652 (reload_gpr_from_vsxsf): Likewise.
7653 (p8_mfvsrd_4_disf): Delete, no longer used.
7654 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7655 then needing to move the SFmode to a GPR to use the XSCVDPSP
7656 instruction instead of FRSP and XSCVDPSPN.
7657 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7658 it is adjacent to the other XSCVSPDP insns.
7659 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7660 SFmode to be in traditional Altivec registers.
7661 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7662 (vsx_xscvspdpn): Likewise.
7663 (vsx_xscvspdpn_scalar): Likewise.
7665 2017-09-26 Martin Jambor <mjambor@suse.cz>
7667 * tree-sra.c (compare_access_positions): Put integral types first,
7668 stabilize sorting of integral types, remove conditions putting
7669 non-full-precision integers last.
7670 (sort_and_splice_var_accesses): Disable scalarization if a
7671 non-integert would be represented by a non-full-precision integer.
7673 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7675 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7676 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7677 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7678 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7679 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7680 conditionals inside the function instead of around it. Call
7681 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7682 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7684 2017-09-26 Richard Biener <rguenther@suse.de>
7686 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7688 (scop_detection::build_scop_breadth): ... this. Removed.
7689 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7690 (scop_detection::harmful_stmt_in_bb): Likewise.
7691 (scop_detection::graphite_can_represent_stmt): Likewise.
7692 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7693 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7694 (scop_detection::can_represent_loop_1): ... this. Removed.
7695 (scop_detection::harmful_loop_in_region): Simplify after inlining
7696 the above and remove more quadraticness.
7697 (build_scops): Adjust.
7698 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7701 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7704 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7705 REGNO (base) == SP_REG if base is a REG.
7708 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7709 if it is different SSA_NAME.
7710 (optimize_range_tests_cmp_bitwise): New function.
7711 (optimize_range_tests): Call it.
7713 2017-09-26 Richard Biener <rguenther@suse.de>
7715 PR tree-optimization/82321
7716 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7717 for the def being inside the loop.
7719 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7721 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7723 * config/s390/s390-builtins.def: Fix constraint on op4.
7725 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7727 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7728 independent expanders.
7729 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7730 ("vec_ordered", "vec_unordered"): New expanders.
7732 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7734 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7737 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7739 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7740 vec_unpacks_lo_v16qi.
7741 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7743 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7745 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7746 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7747 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7749 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7751 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7753 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7755 ("*vec_slb<mode>"): New insn pattern.
7756 ("vec_shr_<mode>"): New expander.
7757 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7758 and force the shift count operand to V16QImode.
7759 ("vec_srb<mode>"): Set shift count mode to V16QI.
7761 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7763 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7764 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7765 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7767 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7770 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7772 2017-09-26 Richard Biener <rguenther@suse.de>
7774 PR tree-optimization/82320
7775 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7778 2017-09-25 Jeff Law <law@redhat.com>
7780 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7781 prototype for new argument.
7782 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7783 mostly extracted from rs6000_emit_allocate_stack.
7784 (rs6000_emit_probe_stack_range_stack_clash): New function.
7785 (rs6000_emit_allocate_stack): Call
7786 rs6000_emit_probe_stack_range_stack_clash as needed.
7787 (rs6000_emit_probe_stack_range): Add additional argument
7788 to call to gen_probe_stack_range{si,di}.
7789 (output_probe_stack_range): New.
7790 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7791 (output_probe_stack_range_stack_clash): New.
7792 (rs6000_emit_prologue): Emit notes into dump file as requested.
7793 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7794 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7795 Add additional operand and pass it to output_probe_stack_range.
7797 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7799 PR tree-optimization/82163
7800 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7801 (checking_verify_loop_closed_ssa): New parameter.
7802 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7803 (check_loop_closed_ssa_stmt): Delete.
7804 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7805 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7806 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7809 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7811 * brig-builtins.def: Treat HSAIL barrier builtins as
7812 setjmp/longjump style functions.
7814 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7816 * target.def (constant_alignment): New hook.
7817 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7818 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7819 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7820 * doc/tm.texi: Regenerate.
7821 * targhooks.h (default_constant_alignment): Declare.
7822 (constant_alignment_word_strings): Likewise.
7823 * targhooks.c (default_constant_alignment): New function.
7824 (constant_alignment_word_strings): Likewise.
7825 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7826 instead of CONSTANT_ALIGNMENT.
7827 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7828 (force_const_mem): Likewise.
7829 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7830 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7831 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7832 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7833 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7835 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7836 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7837 constant_alignment_word_strings.
7838 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7839 (CONSTANT_ALIGNMENT): Likewise.
7840 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7841 (arm_constant_alignment): New function.
7842 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7843 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7844 constant_alignment_word_strings.
7845 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7846 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7847 constant_alignment_word_strings.
7848 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7849 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7850 (cris_constant_alignment): New function.
7851 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7852 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7853 (epiphany_constant_alignment): New function.
7854 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7855 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7856 constant_alignment_word_strings.
7857 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7858 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7859 constant_alignment_word_strings.
7860 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7861 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7862 constant_alignment_word_strings.
7863 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7864 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7865 * config/i386/i386.c (ix86_constant_alignment): Make static.
7866 Use the same interface as the target hook.
7867 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7868 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7869 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7870 constant_alignment_word_strings.
7871 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7872 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7873 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7874 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7875 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7876 constant_alignment_word_strings.
7877 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7878 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7879 constant_alignment_word_strings.
7880 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7881 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7882 constant_alignment_word_strings.
7883 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7884 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7886 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7887 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7888 * config/mips/mips.c (mips_constant_alignment): New function.
7889 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7890 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7891 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7892 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7893 (mmix_constant_alignment): Make static. Use the same interface
7895 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7896 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7897 constant_alignment_word_strings.
7898 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7899 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7900 constant_alignment_word_strings.
7901 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7902 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7903 constant_alignment_word_strings.
7904 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7905 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7906 (rs6000_constant_alignment): New function.
7907 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7908 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7909 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7910 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7911 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7912 (rs6000_constant_alignment): New function.
7913 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7914 * config/s390/s390.c (s390_constant_alignment): New function.
7915 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7916 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7917 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7918 constant_alignment_word_strings.
7919 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7920 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7921 (sparc_constant_alignment): New function.
7922 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7923 * config/spu/spu.c (spu_constant_alignment): New function.
7924 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7925 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7926 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7927 constant_alignment_word_strings.
7928 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7929 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7930 constant_alignment_word_strings.
7931 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7932 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7933 constant_alignment_word_strings.
7934 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7935 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7936 (visium_constant_alignment): New function.
7937 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7938 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7939 (xtensa_constant_alignment): New function.
7940 * system.h (CONSTANT_ALIGNMENT): Poison.
7942 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7945 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7946 (rs6000_builtin_valid_without_lhs): New helper function.
7947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7948 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7950 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7952 * target.h (vec_perm_indices): Use unsigned short rather than
7954 (auto_vec_perm_indices): Likewise.
7955 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7956 Use unsigned int rather than unsigned char.
7957 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7959 2017-09-25 Richard Biener <rguenther@suse.de>
7961 * cfgloop.h (sort_sibling_loops): Declare.
7962 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7963 (sort_sibling_loops): New function sorting the sibling loop list
7965 * graphite.c (graphite_transform_loops): Sort sibling loops.
7967 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7969 * target.def (vec_perm_const_ok): Change sel parameter to
7971 * optabs-query.c (can_vec_perm_p): Update accordingly.
7972 * doc/tm.texi: Regenerate.
7973 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7974 auto_vec_perm_indices and remove separate nelt field.
7975 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7976 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7977 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7978 (aarch64_expand_vec_perm_const): Update accordingly.
7979 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7980 to vec_perm_indices.
7981 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7982 auto_vec_perm_indices and remove separate nelt field.
7983 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7984 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7985 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7987 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7988 to vec_perm_indices.
7989 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7990 sel to vec_perm_indices.
7991 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7992 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7993 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7995 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7998 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
8001 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
8002 on the FUNCTION_DECL function context if it has a DIE that is a
8005 2017-09-25 Richard Biener <rguenther@suse.de>
8007 PR tree-optimization/82285
8008 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
8011 2017-09-25 Tom de Vries <tom@codesourcery.com>
8015 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
8018 2017-09-25 Richard Biener <rguenther@suse.de>
8020 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
8021 ISL errors other than isl_error_quota happen. Dump if the
8022 schedule is the same.
8023 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
8024 errors instead of aborting inside ISL.
8026 2017-09-25 Iain Sandoe <iain@codesourcery.com>
8029 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
8030 of libgcc_eh for m64.
8031 * config/i386/darwin64.h: Likewise.
8033 2017-09-25 Richard Biener <rguenther@suse.de>
8036 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
8037 attribute for incomplete types nor twice for complete ones.
8039 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
8042 * config/i386/i386.c (ix86_print_operand_address_as): Encode
8043 %esp as %rsp to avoid 0x67 prefix if there is no index or base
8046 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
8049 * config/i386/i386.opt (mprefer-avx256): Use
8050 ix86_target_flags variable.
8051 * config/i386/i386.c (ix86_target_string): Move
8052 -mprefer-avx256 to flag2_opts.
8054 2017-09-22 Jakub Jelinek <jakub@redhat.com>
8057 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
8058 and x != -1 | y != -1 into (x & y) != -1.
8060 2017-09-22 Steve Ellcey <sellcey@cavium.com>
8062 * config.gcc: Add new case statement to set
8063 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
8064 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
8065 s390x-*-linux* case statements. Added aarch64 to the list of
8066 supported architectures.
8068 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8070 PR tree-optimization/82289
8071 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
8072 STMT_VINFO_RELEVANT_P.
8074 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8075 Alan Hayward <alan.hayward@arm.com>
8076 David Sherwood <david.sherwood@arm.com>
8078 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
8079 for VR_RANGE only; don't allow VR_ANTI_RANGE.
8080 (extract_range_from_binary_expr_1): Don't call
8081 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
8083 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8084 Alan Hayward <alan.hayward@arm.com>
8085 David Sherwood <david.sherwood@arm.com>
8087 * target.def (preferred_vector_alignment): New hook.
8088 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
8090 * doc/tm.texi: Regenerate.
8091 * targhooks.h (default_preferred_vector_alignment): Declare.
8092 * targhooks.c (default_preferred_vector_alignment): New function.
8093 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
8095 (DR_TARGET_ALIGNMENT): New macro.
8096 (aligned_access_p): Update commentary.
8097 (vect_known_alignment_in_bytes): New function.
8098 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
8100 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8101 Calculate the misalignment based on the target alignment rather than
8103 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8104 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8105 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8106 the target alignment, rather than masking the element misalignment
8107 with the number of elements in a vector. Also use the target
8108 alignment when calculating the maximum number of peels.
8109 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8110 instead of TYPE_ALIGN_UNIT.
8111 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8112 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8113 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8114 (vect_create_data_ref_ptr): Likewise.
8115 (vect_setup_realignment): Realign by ANDing with
8116 -DR_TARGET_MISALIGNMENT.
8117 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8118 the number of peels based on DR_TARGET_ALIGNMENT.
8119 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8120 with the guaranteed alignment boundary when deciding whether
8122 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8123 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8124 (ensure_base_align): Remove stmt_info parameter. Get the
8125 target base alignment from DR_TARGET_ALIGNMENT.
8126 (vectorizable_store): Update call accordingly. Interpret
8127 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8129 (vectorizable_load): Likewise.
8131 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8132 Alan Hayward <alan.hayward@arm.com>
8133 David Sherwood <david.sherwood@arm.com>
8135 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8136 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8137 (vect_enhance_data_refs_alignment): Likewise.
8139 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8141 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8142 error. Only quit immediately if parsing is complete.
8143 (BEGIN): Initialize fatal_err and parse_done.
8144 (begin fpu, end fpu): Check number of arguments.
8145 (begin arch, end arch): Likewise.
8146 (begin cpu, end cpu): Likewise.
8147 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8148 (optalias): Likewise.
8150 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8152 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8153 * config/arm/arm-isa.h: Delete. Move definitions to ...
8154 * arm-cpus.in: ... here. Use new feature and fgroup values.
8155 * config/arm/arm.c (arm_option_override): Use lower case for feature
8157 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8158 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8159 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8161 (print_isa_bits_for): New function.
8162 (gen_isa): New function.
8163 (gen_comm_data): Use print_isa_bits_for.
8164 (define feature): New keyword.
8165 (define fgroup): New keyword.
8166 * config/arm/t-arm (TM_H): Remove.
8167 (GTM_H): Add arm-isa.h.
8168 (arm-isa.h): Add rule to generate file.
8169 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8170 case for feature bit names.
8172 2017-09-22 Richard Biener <rguenther@suse.de>
8174 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8176 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8177 print of no dependency loops ...
8178 * graphite.c (graphite_transform_loops): ... here.
8179 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8181 (same_close_phi_node, remove_duplicate_close_phi,
8182 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8183 (canonicalize_loop_closed_ssa): ... here and simplify.
8184 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8185 (optimize_isl): Use dump_printf_loc to tell when we stopped
8186 optimizing because of an ISL timeout.
8188 2017-09-22 Richard Biener <rguenther@suse.de>
8190 PR tree-optimization/82291
8191 * tree-if-conv.c (predicate_mem_writes): Make sure to
8192 remove writes in blocks predicated with false.
8194 2017-09-22 Richard Biener <rguenther@suse.de>
8196 * sese.c: Include cfganal.h.
8197 (if_region_set_false_region): Remove.
8198 (create_if_region_on_edge): Likewise.
8199 (move_sese_in_condition): Re-implement without destroying
8202 2017-09-22 Richard Biener <rguenther@suse.de>
8204 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8205 Verify both BBs contain loop PHI nodes before dispatching to
8207 (graphite_regenerate_ast_isl): Do not recompute dominators,
8208 do not verify three times. Restructure for clarity.
8209 * graphite-scop-detection.c (same_close_phi_node,
8210 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8211 defined_in_loop_p, canonicalize_loop_closed_ssa,
8212 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8213 checking and SSA rewrite, move to ...
8214 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8215 (graphite_initialize): Do not pass in ctx, do not reset the
8216 SCEV cache, compute only dominators.
8217 (graphite_transform_loops): Allocate ISL ctx after
8218 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8219 Maintain post-dominators only around build_scops.
8220 * sese.c (if_region_set_false_region): Make static. Free
8221 and recompute dominators.
8222 (move_sese_in_condition): Assert we don't get called with
8223 post-dominators computed.
8224 * sese.h (if_region_set_false_region): Remove.
8226 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8228 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8229 mode attribute for TARGET_AVX512VL.
8231 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8233 * config/i386/i386.opt (mprefer-avx256): New option.
8234 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8236 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8237 for TARGET_PREFER_AVX256.
8238 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8240 2017-09-21 Jeff Law <law@redhat.com>
8242 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8243 Fix dump output if the only stack space is for pushed registers.
8245 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8247 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8250 2017-09-21 Martin Sebor <msebor@redhat.com>
8253 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8254 code (in C++) or code that triggers warnings.
8256 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8258 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8260 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8262 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8263 * sched-rgn.c: Ditto.
8264 * sel-sched-ir.c: Ditto.
8266 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8268 * toplev.h (set_random_seed): Adjust return type.
8269 * toplev.c (init_local_tick): Move eager initialization of random_seed
8270 to get_random_seed. Adjust comment.
8271 (init_random_seed): Inline to get_random_seed, delete.
8272 (get_random_seed): Initialize random_seed lazily.
8273 (set_random_seed): Do not return previous value.
8274 (print_switch_value): Do not call get_random_seed.
8276 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8278 * cgraph.c (delete_function_version): New, broken out from...
8279 (cgraph_node::delete_function_version): ...here. Rename to
8280 cgraph_node::delete_function_version_by_decl. Update all uses.
8281 (cgraph_node::remove): Call delete_function_version.
8283 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8286 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8287 VAR_DECLs to which addressable non-volatile parameters are mapped
8288 and for id->retvar after the return value assignment. Clear
8289 id->retval and id->retbnd after inlining.
8291 2017-09-21 Richard Biener <rguenther@suse.de>
8293 PR tree-optimization/82276
8294 PR tree-optimization/82244
8295 * tree-vrp.c (build_assert_expr_for): Set
8296 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8298 (remove_range_assertions): Revert earlier change.
8300 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8303 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8305 2017-09-21 Richard Biener <rguenther@suse.de>
8307 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8308 Restore valid IL after code generation errors.
8309 * graphite.c (graphite_transform_loops): Diagnose code
8310 generation issues as MSG_MISSED_OPTIMIZATION and continue
8311 with processing SCOPs.
8313 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8314 Alan Hayward <alan.hayward@arm.com>
8315 David Sherwood <david.sherwood@arm.com>
8317 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8318 rather than choosing between plus_constant and gen_rtx_<CODE>.
8319 * expr.c (emit_push_insn): Likewise.
8320 (expand_expr_real_2): Likewise.
8322 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8323 Alan Hayward <alan.hayward@arm.com>
8324 David Sherwood <david.sherwood@arm.com>
8326 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8328 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8329 Alan Hayward <alan.hayward@arm.com>
8330 David Sherwood <david.sherwood@arm.com>
8332 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8333 calling tree_to_uhwi.
8335 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8336 Alan Hayward <alan.hayward@arm.com>
8337 David Sherwood <david.sherwood@arm.com>
8339 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8340 INTEGER_CST rather than a negative test for ADDR_EXPR.
8342 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8343 Alan Hayward <alan.hayward@arm.com>
8344 David Sherwood <david.sherwood@arm.com>
8346 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8347 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8349 2017-09-21 Richard Biener <rguenther@suse.de>
8351 PR tree-optimization/71351
8352 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8353 graphite_create_new_loop_guard): Remove, fold remaining parts
8355 (translate_isl_ast_node_for): ... here and simplify.
8357 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8360 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8361 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8362 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8363 alternative always use QI mode, for -Os imov (=R,R) alternative
8364 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8367 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8368 Jeff Law <law@redhat.com>
8370 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8371 (allocate_stack_space): New function, partially extracted from
8373 (s390_emit_prologue): Track offset to most recent stack probe.
8374 Code to allocate space moved into allocate_stack_space.
8375 Dump actions when no stack is allocated.
8376 (s390_prologue_plus_offset): New function.
8377 (s390_emit_stack_probe): Likewise.
8379 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8381 * common.opt (Wa, Wl, Wp, g, gz=): Add
8383 (gno-column-info): Remove.
8384 (gcolumn-info): Drop RejectNegative.
8386 (gno-record-gcc-switches): Remove.
8387 (grecord-gcc-switches): Drop RejectNegative.
8388 (gno-split-dwarf): Remove.
8389 (gsplit-dwarf): Drop RejectNegative.
8390 (gno-strict-dwarf): Remove.
8391 (gstrict-dwarf): Drop RejectNegative.
8392 * config/darwin.opt (gfull, gused): Add RejectNegative.
8393 * dwarf2out.c (gen_producer_string): Drop
8394 gno-record-gcc-switches handler.
8395 * optc-gen.awk: Add g to prefixes with negative forms.
8396 * opts-common.c (remapping_prefix_p): New.
8397 (find_opt): Check it.
8398 (generate_canonical_option): Test g prefix.
8399 (option_map): Add -gno- mapping.
8400 (add_misspelling_candidates): Check remapping_prefix_p.
8402 2017-09-20 Jeff Law <law@redhat.com>
8404 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8405 thinko in stack clash protection support.
8407 * explow.c (compute_stack_clash_protection_loop_data): Use
8408 CONST_INT_P instead of explicit test. Verify object is a
8409 CONST_INT_P before looking at INTVAL.
8410 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8411 instead of explicit test.
8413 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8416 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8417 address instead of to r1 and r11.
8419 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8421 * config.gcc: Support "knm".
8422 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8423 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8425 * config/i386/i386.c (m_KNM): Define.
8426 (processor_target_table): Add "knm".
8428 (ix86_option_override_internal): Add "knm".
8429 (ix86_issue_rate): Add PROCESSOR_KNM.
8430 (ix86_adjust_cost): Ditto.
8431 (ia32_multipass_dfa_lookahead): Ditto.
8432 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8433 (fold_builtin_cpu): Add M_INTEL_KNM.
8434 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8435 (processor_type): Add PROCESSOR_KNM.
8436 * config/i386/x86-tune.def: Add m_KNM.
8437 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8439 2017-09-20 Richard Biener <rguenther@suse.de>
8441 PR tree-optimization/80213
8442 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8443 are allowed in empty BBs as well.
8444 (canonicalize_loop_closed_ssa): Also look for other complex
8446 (scop_detection::get_sese): Include the loop-closed PHI block
8448 (scop_detection::merge_sese): Remove code adding extra blocks.
8449 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8450 (build_scops): Assert the final returned scop is invalid.
8452 2017-09-20 Richard Biener <rguenther@suse.de>
8454 PR tree-optimization/82264
8455 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8457 (vn_phi_lookup): Likewise.
8458 (vn_phi_insert): Likewise.
8460 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8462 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8463 that fits into uhwi or shwi, add DW_AT_const_value regardless
8464 of early_dwarf without going through RTL, using add_AT_unsigned
8467 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8468 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8469 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8470 (DEBUG_MACRO_SECTION): Likewise.
8471 (DEBUG_LTO_MACRO_SECTION): Likewise.
8472 (DEBUG_STR_DWO_SECTION): Likewise.
8473 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8474 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8475 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8476 (DEBUG_LTO_LINE_STR_SECTION): Define.
8477 (init_sections_and_labels): Initialize debug_line_str_section
8478 variable. Initialize debug_loc_section for -gdwarf-5 to
8479 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8481 2017-09-20 Richard Biener <rguenther@suse.de>
8483 * graphite-sese-to-poly.c (extract_affine): Properly handle
8484 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8486 2017-09-20 Richard Biener <rguenther@suse.de>
8488 PR tree-optimization/81373
8489 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8490 Force SESE live-out defs to be handled even if they are
8493 2017-09-19 Jeff Law <law@redhat.com>
8495 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8496 nothing for stack adjustments with REG_STACK_CHECK.
8497 * sched-deps.c (parse_add_or_inc): Reject insns with
8498 REG_STACK_CHECK from dependency breaking.
8499 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8500 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8501 * reg-notes.def (STACK_CHECK): New note.
8503 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8504 (ix86_expand_prologue): Dump stack clash info as needed.
8505 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8507 * function.c (dump_stack_clash_frame_info): New function.
8508 * function.h (dump_stack_clash_frame_info): Prototype.
8509 (enum stack_clash_probes): New enum.
8511 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8512 flag_stack_clash_protection.
8513 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8514 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8515 (arm_frame_pointer_required): Likewise.
8516 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8517 (ia64_expand_prologue): Likewise.
8518 * config/mips/mips.c (mips_expand_prologue): Likewise.
8519 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8520 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8521 (sparc_flat_expand_prologue): Likewise.
8522 * config/spu/spu.c (spu_expand_prologue): Likewise.
8524 * explow.c: Include "params.h".
8525 (anti_adjust_stack_and_probe_stack_clash): New function.
8526 (get_stack_check_protect): Likewise.
8527 (compute_stack_clash_protection_loop_data): Likewise.
8528 (emit_stack_clash_protection_loop_start): Likewise.
8529 (emit_stack_clash_protection_loop_end): Likewise.
8530 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8531 Use anti_adjust_stack_and_probe_stack_clash.
8532 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8533 (emit_stack_clash_protection_loop_start): Likewise.
8534 (emit_stack_clash_protection_loop_end): Likewise.
8535 * rtl.h (get_stack_check_protect): Prototype.
8536 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8537 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8538 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8540 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8542 * doc/tm.texi: Rebuilt.
8543 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8544 get_stack_check_protect.
8545 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8546 * config/arm/arm.c (arm_expand_prologue): Likewise.
8547 (arm_frame_pointer_required): Likewise.
8548 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8549 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8550 * config/mips/mips.c (mips_expand_prologue): Likewise.
8551 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8552 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8553 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8554 (sparc_flat_expand_prologue): Likewise.
8556 * common.opt (-fstack-clash-protection): New option.
8557 * flag-types.h (enum stack_check_type): Note difference between
8558 -fstack-check= and -fstack-clash-protection.
8559 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8560 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8561 * toplev.c (process_options): Issue warnings/errors for cases
8562 not handled with -fstack-clash-protection.
8563 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8564 (-fstack-check): Note additional problem with -fstack-check=generic.
8565 Note that -fstack-check is primarily for Ada and refer users
8566 to -fstack-clash-protection for stack-clash-protection.
8567 Document new params for stack clash protection.
8569 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8571 * config/i386/i386.c (ix86_split_long_move): Do not handle
8572 address used for LEA in a special way.
8574 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8576 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8578 2017-09-19 Martin Sebor <msebor@redhat.com>
8581 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8582 of incompatible types.
8584 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8586 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8587 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8588 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8589 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8591 2017-09-19 Richard Biener <rguenther@suse.de>
8593 PR tree-optimization/82244
8594 * tree-vrp.c (remove_range_assertions): Do not propagate
8595 a constant to abnormals but replace the assert with a copy.
8597 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8599 PR rtl-optimization/57878
8600 PR rtl-optimization/68988
8601 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8602 avoidance test involving non_reload_pseudos. Move frequency test
8603 below the general fragmentation avoidance test.
8605 2017-09-19 Richard Biener <rguenther@suse.de>
8607 PR tree-optimization/69728
8608 * graphite-sese-to-poly.c (schedule_error): New global.
8609 (add_loop_schedule): Handle empty domain by failing the
8611 (build_original_schedule): Handle schedule_error.
8613 2017-09-19 Richard Biener <rguenther@suse.de>
8615 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8616 Do not iterate to sibling loops but only to siblings of inner
8619 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8622 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8623 that the registers are different.
8625 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8627 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8628 to processor_model and "amdfam17h" to arch_names_table.
8629 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8631 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8634 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8636 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8637 Alan Hayward <alan.hayward@arm.com>
8638 David Sherwood <david.sherwood@arm.com>
8640 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8642 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8644 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8645 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8646 vect_schedule_slp_instance.
8647 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8648 Update call to vect_slp_analyze_node_operations. Simplify return
8650 (vect_slp_analyze_bb_1): Update call accordingly.
8651 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8652 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8653 (vect_schedule_slp): Update call accordingly.
8655 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8656 Alan Hayward <alan.hayward@arm.com>
8657 David Sherwood <david.sherwood@arm.com>
8659 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8660 with types that aren't in fact scalar.
8662 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8664 * tree-vect-slp.c (vect_record_max_nunits): New function,
8666 (vect_build_slp_tree_1): ...here.
8667 (vect_build_slp_tree_2): Call it for phis too.
8669 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8671 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8672 to vect_get_vec_def_for_operand when getting the mask operand.
8674 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8675 Alan Hayward <alan.hayward@arm.com>
8676 David Sherwood <david.sherwood@arm.com>
8678 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8681 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8682 Alan Hayward <alan.hayward@arm.com>
8683 David Sherwood <david.sherwood@arm.com>
8685 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8686 calculation for vector booleans.
8688 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8689 Alan Hayward <alan.hayward@arm.com>
8690 David Sherwood <david.sherwood@arm.com>
8692 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8694 (vect_transform_stmt): ...here.
8695 (vect_analyze_stmt): Use it instead of calling
8696 vectorizable_live_operation directly.
8698 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8700 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8701 non-SIMT targets in acc vector loops.
8703 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8705 * configure.ac: Add arc and check if assembler supports gdwarf2.
8706 * configure: Regenerate.
8708 2017-09-18 Richard Biener <rguenther@suse.de>
8710 PR tree-optimization/82220
8711 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8712 epilogue niters from the min_profitable_iters compute.
8714 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8717 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8718 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8719 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8721 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8724 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8725 switching to a new text section.
8727 2017-09-18 Richard Biener <rguenther@suse.de>
8729 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8731 (build_alias_set): Reject aliases with no access function.
8733 2017-09-18 Richard Biener <rguenther@suse.de>
8735 PR tree-optimization/79622
8736 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8738 (build_cross_bb_scalars_use): Likewise.
8740 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8742 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8744 2017-09-18 Alan Modra <amodra@gmail.com>
8747 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8748 stack_pointer_rtx for count 0. Update comments. Break up
8749 large rtl expression.
8751 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8753 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8754 Increase to 20 bytes.
8755 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8756 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8757 or avx version of the stub.
8759 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8762 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8763 compute the minimum stack alignment. Also update preferred stack
8764 boundary for leaf functions.
8766 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8768 PR tree-optimization/82228
8769 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8772 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8774 * common/config/nds32/nds32-common.c
8775 (nds32_option_optimization_table): Refine formatting.
8776 (nds32_option_optimization_table): Use -fsched-pressure and
8777 -fomit-frame-pointer for specific optimization level.
8779 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8781 * config/nds32/nds32.c: Refine formatting and comments.
8782 * config/nds32/nds32.h: Likewise.
8783 * config/nds32/nds32.md: Likewise.
8784 * config/nds32/nds32-cost.c: Likewise.
8785 * config/nds32/nds32-isr.c: Likewise.
8786 * config/nds32/nds32-md-auxiliary.c: Likewise.
8787 * config/nds32/nds32-multiple.md: Likewise.
8788 * config/nds32/nds32-predicates.c: Likewise.
8790 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8791 Jakub Jelinek <jakub@redhat.com>
8793 Add support for -std=c++2a.
8794 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8796 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8798 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8801 * doc/extend.texi (Common Function Attributes): Add
8802 references to ARM, AArch64, and S/390 specific attributes.
8803 (Function Specific Option Pragmas): Add AArch64 and S/390
8804 to list of back ends that support the target pragma.
8806 2017-09-15 Nathan Sidwell <nathan@acm.org>
8808 * doc/standards.texi: Fix C++17 description. Update URLs for
8811 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8813 * common.opt (Wcast-align=strict): New warning option.
8814 * doc/invoke.texi: Document -Wcast-align=strict.
8816 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8818 * cgraph.h (cgraph_thunk_info): Add comments.
8819 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8820 assert for VIRTUAL_* arguments stricter.
8822 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8824 PR tree-optimization/71026
8825 * match.pd: Move RDIV patterns from fold-const.c
8826 * fold-const.c (distribute_real_division): Removed.
8827 (fold_binary_loc): Remove calls to distribute_real_divison.
8829 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8831 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8832 c++1z and gnu++1z as deprecated. Change other references to
8833 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8834 Change -Wc++1z-compat to -Wc++17-compat.
8835 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8836 * dwarf2out.c (highest_c_language): Handle C++17.
8837 (gen_compile_unit_die): Likewise.
8839 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8841 PR rtl-optimization/82192
8842 * combine.c (make_extraction): Don't look through non-paradoxical
8843 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8846 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8847 Alan Hayward <alan.hayward@arm.com>
8848 David Sherwood <david.sherwood@arm.com>
8850 * target.def (function_arg_offset): New hook.
8851 * targhooks.h (default_function_arg_offset): Declare.
8852 * targhooks.c (default_function_arg_offset): New function.
8853 * function.c (locate_and_pad_parm): Use
8854 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8855 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8856 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8857 * doc/tm.texi: Regenerate.
8858 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8859 * config/spu/spu.c (spu_function_arg_offset): New function.
8860 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8861 * system.h (FUNCTION_ARG_OFFSET): Poison.
8863 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8864 Alan Hayard <alan.hayward@arm.com>
8865 David Sherwood <david.sherwood@arm.com>
8867 * target.def (truly_noop_truncation): New hook.
8868 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8869 than TRULY_NOOP_TRUNCATION.
8870 * hooks.h (hook_bool_uint_uint_true): Declare.
8871 * hooks.c (hook_bool_uint_uint_true): New function.
8872 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8873 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8874 * doc/tm.texi: Regenerate.
8875 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8876 rather than TRULY_NOOP_TRUNCATION in comments.
8877 (simplify_comparison): Likewise.
8878 (record_truncated_value): Likewise.
8879 * expmed.c (extract_bit_field_1): Likewise.
8880 (extract_split_bit_field): Likewise.
8881 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8882 instead of TRULY_NOOP_TRUNCATION.
8883 * function.c (assign_parm_setup_block): Likewise.
8884 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8885 * rtlhooks.c: Include target.h.
8886 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8887 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8888 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8889 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8890 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8891 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8892 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8893 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8894 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8895 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8896 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8897 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8898 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8899 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8900 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8901 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8902 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8903 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8904 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8905 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8906 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8907 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8908 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8909 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8910 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8911 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8912 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8913 rather than TRULY_NOOP_TRUNCATION in comments.
8914 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8915 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8916 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8917 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8918 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8919 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8920 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8921 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8922 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8923 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8924 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8925 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8926 rather than TRULY_NOOP_TRUNCATION in comments.
8927 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8928 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8929 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8930 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8931 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8932 TRULY_NOOP_TRUNCATION condition.
8933 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8934 (TRULY_NOOP_TRUNCATION): Delete.
8935 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8936 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8937 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8938 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8939 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8940 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8941 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8942 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8943 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8944 rather than TRULY_NOOP_TRUNCATION in comments.
8945 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8946 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8947 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8948 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8949 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8950 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8952 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8955 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8956 (*cmp_ior): Likewise.
8957 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8958 (*ior_scc_scc_cmp): Likewise.
8959 (*and_scc_scc): Likewise.
8960 (*and_scc_scc_cmp): Likewise.
8962 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8963 Alan Hayard <alan.hayward@arm.com>
8964 David Sherwood <david.sherwood@arm.com>
8966 * target.def (can_change_mode_class): New hook.
8967 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8968 (hard_regno_nregs): Likewise.
8969 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8970 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8971 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8972 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8973 (LOAD_EXTEND_OP): Update accordingly.
8974 * doc/tm.texi: Regenerate.
8975 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8976 CANNOT_CHANGE_MODE_CLASS.
8977 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8978 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8979 * combine.c (simplify_set): Update accordingly.
8980 * emit-rtl.c (validate_subreg): Likewise.
8981 * recog.c (general_operand): Likewise.
8982 * regcprop.c (mode_change_ok): Likewise.
8983 * reload1.c (choose_reload_regs): Likewise.
8984 (inherit_piecemeal_p): Likewise.
8985 * rtlanal.c (simplify_subreg_regno): Likewise.
8986 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8987 instead of CANNOT_CHANGE_MODE_CLASS.
8988 (reload_cse_simplify_operands): Likewise.
8989 * reload.c (push_reload): Use targetm.can_change_mode_class
8990 instead of CANNOT_CHANGE_MODE_CLASS.
8991 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8992 REG_CANNOT_CHANGE_MODE_P.
8993 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8994 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8995 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8996 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8997 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8998 (arm_can_change_mode_class): New function.
8999 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9000 than CANNOT_CHANGE_MODE_CLASS in comments.
9001 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9002 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
9003 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
9004 (ix86_can_change_mode_class): ...this new function, inverting the
9005 sense of the return value.
9006 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9007 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9008 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9009 (ia64_can_change_mode_class): New function.
9010 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9011 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
9012 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
9013 (m32c_can_change_mode_class): ...this new function, inverting the
9014 sense of the return value.
9015 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9016 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9017 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
9018 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
9019 (mips_can_change_mode_class): ...this new function, inverting the
9020 sense of the return value.
9021 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9022 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9023 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9024 (msp430_can_change_mode_class): New function.
9025 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9026 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
9027 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9028 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9029 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9030 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
9031 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9032 (pa_cannot_change_mode_class): Replace with...
9033 (pa_can_change_mode_class): ...this new function, inverting the
9034 sense of the return value.
9035 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
9036 than CANNOT_CHANGE_MODE_CLASS in comments.
9037 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9038 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
9039 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9040 (pdp11_cannot_change_mode_class): Replace with...
9041 (pdp11_can_change_mode_class): ...this new function, inverting the
9042 sense of the return value.
9043 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9044 * config/powerpcspe/powerpcspe-protos.h
9045 (rs6000_cannot_change_mode_class_ptr): Delete.
9046 * config/powerpcspe/powerpcspe.c
9047 (rs6000_cannot_change_mode_class_ptr): Delete.
9048 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9049 (rs6000_option_override_internal): Assign to
9050 targetm.can_change_mode_class instead of
9051 rs6000_cannot_change_mode_class_ptr.
9052 (rs6000_cannot_change_mode_class): Replace with...
9053 (rs6000_can_change_mode_class): ...this new function, inverting the
9054 sense of the return value.
9055 (rs6000_debug_cannot_change_mode_class): Replace with...
9056 (rs6000_debug_can_change_mode_class): ...this new function.
9057 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9058 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
9059 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9060 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9061 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
9063 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
9064 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9065 (rs6000_option_override_internal): Assign to
9066 targetm.can_change_mode_class instead of
9067 rs6000_cannot_change_mode_class_ptr.
9068 (rs6000_cannot_change_mode_class): Replace with...
9069 (rs6000_can_change_mode_class): ...this new function, inverting the
9070 sense of the return value.
9071 (rs6000_debug_cannot_change_mode_class): Replace with...
9072 (rs6000_debug_can_change_mode_class): ...this new function.
9073 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9074 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
9075 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
9076 (s390_can_change_mode_class): ...this new function, inverting the
9077 sense of the return value.
9078 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9079 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9080 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
9081 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9082 (sh_cannot_change_mode_class): Replace with...
9083 (sh_can_change_mode_class): ...this new function, inverting the
9084 sense of the return value.
9085 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9086 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9087 (sparc_can_change_mode_class): New function.
9088 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9089 * config/spu/spu.c (spu_can_change_mode_class): New function.
9090 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9091 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
9092 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
9093 (visium_can_change_mode_class): New function.
9094 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
9096 2017-09-15 Richard Biener <rguenther@suse.de>
9098 PR tree-optimization/82217
9099 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
9100 but not undefined case.
9102 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9105 * postreload.c (reload_cse_simplify_operands): Skip
9106 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9108 2017-09-15 Richard Biener <rguenther@suse.de>
9110 PR tree-optimization/68823
9111 * graphite-scop-detection.c (build_alias_set): If we have a
9112 possible dependence check whether we can handle them by just
9113 looking at the DRs DR_ACCESS_FNs.
9114 (build_scops): If build_alias_set fails, fail the SCOP.
9116 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9118 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9119 to support float128 built-in functions that require the ISA 3.0
9121 (BU_FLOAT128_3_HW): Likewise.
9122 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9124 (FMAF128): Likewise.
9126 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9127 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9128 floating point instructions.
9129 (rs6000_invalid_builtin): Likewise.
9130 (rs6000_builtin_mask_names): Likewise.
9131 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9132 (RS6000_BTM_FLOAT128_HW): Likewise.
9133 (RS6000_BTM_COMMON): Likewise.
9134 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9136 * doc/extend.texi (RS/6000 built-in functions): Document the
9137 IEEE 128-bit floating point square root and fused multiply-add
9140 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9142 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9143 reg (r2) isn't in the set of registers defined in the prologue.
9145 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9146 Alan Hayward <alan.hayward@arm.com>
9147 David Sherwood <david.sherwood@arm.com>
9149 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9150 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9151 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9152 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9153 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9155 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9156 max_vectorization_factor.
9157 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9159 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9160 Alan Hayward <alan.hayward@arm.com>
9161 David Sherwood <david.sherwood@arm.com>
9163 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9164 (vect_worthwhile_without_simd_p): Declare.
9165 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9166 (vectorizable_reduction): Use it.
9167 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9168 (vectorizable_operation): Likewise.
9170 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9171 Alan Hayward <alan.hayward@arm.com>
9172 David Sherwood <david.sherwood@arm.com>
9174 * tree-vectorizer.h (vect_get_num_copies): New function.
9175 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9176 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9177 (vectorizable_induction): Likewise.
9178 (vectorizable_live_operation): Likewise.
9179 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9180 (vectorizable_bswap): Likewise.
9181 (vectorizable_call): Likewise.
9182 (vectorizable_conversion): Likewise.
9183 (vectorizable_assignment): Likewise.
9184 (vectorizable_shift): Likewise.
9185 (vectorizable_operation): Likewise.
9186 (vectorizable_store): Likewise.
9187 (vectorizable_load): Likewise.
9188 (vectorizable_condition): Likewise.
9189 (vectorizable_comparison): Likewise.
9190 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9192 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9193 Alan Hayward <alan.hayward@arm.com>
9194 David Sherwood <david.sherwood@arm.com>
9196 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9197 of vect_init_vector.
9199 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9200 Alan Hayward <alan.hayward@arm.com>
9201 David Sherwood <david.sherwood@arm.com>
9203 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9204 an inline wrapper that provides a location.
9205 (gimple_build_vector): Likewise.
9206 * gimple-fold.c (gimple_build_vector_from_val): New function.
9207 (gimple_build_vector): Likewise.
9208 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9209 functions to build the initial value. Always return a gimple value.
9210 (get_initial_defs_for_reduction): Likewise. Only compute
9212 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9213 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9214 (vectorizable_induction): Use gimple_build_vector rather than
9217 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9218 Alan Hayward <alan.hayward@arm.com>
9219 David Sherwood <david.sherwood@arm.com>
9221 * target.h (vec_perm_indices): New typedef.
9222 (auto_vec_perm_indices): Likewise.
9223 * optabs-query.h: Include target.h
9224 (can_vec_perm_p): Take a vec_perm_indices *.
9225 * optabs-query.c (can_vec_perm_p): Likewise.
9226 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9227 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9228 * tree-vect-generic.c (lower_vec_perm): Likewise.
9229 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9230 (vect_grouped_load_supported): Likewise.
9231 (vect_shift_permute_load_chain): Likewise.
9232 (vect_permute_store_chain): Use auto_vec_perm_indices.
9233 (vect_permute_load_chain): Likewise.
9234 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9235 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9236 Update uses of can_vec_perm_p.
9237 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9238 mode with a number of elements. Take a vec_perm_indices *.
9239 (vect_create_epilog_for_reduction): Update accordingly.
9240 Use auto_vec_perm_indices.
9241 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9242 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9243 (vect_transform_slp_perm_load): Likewise.
9244 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9245 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9246 (vect_gen_perm_mask_checked): Likewise.
9247 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9248 (vect_gen_perm_mask_checked): Likewise.
9249 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9250 (vectorizable_store): Likewise.
9251 (vectorizable_load): Likewise.
9252 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9253 (vectorizable_bswap): Likewise.
9255 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9256 Alan Hayward <alan.hayward@arm.com>
9257 David Sherwood <david.sherwood@arm.com>
9259 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9260 * tree.c (build_vector): Likewise.
9261 (build_vector_from_ctor): Update accordingly.
9262 (build_vector_from_val): Likewise.
9263 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9264 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9265 * tree-vect-generic.c (add_rshift): Likewise.
9266 (expand_vector_divmod): Likewise.
9267 (optimize_vector_constructor): Likewise.
9268 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9269 (vect_transform_slp_perm_load): Likewise.
9270 (vect_schedule_slp_instance): Likewise.
9271 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9272 (vectorizable_call): Likewise.
9273 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9274 * expmed.c (make_tree): Likewise.
9275 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9276 a vector passed to build_vector.
9277 (fold_convert_const): Likewise.
9278 (exact_inverse): Likewise.
9279 (fold_ternary_loc): Likewise.
9280 (fold_relational_const): Likewise.
9281 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9282 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9283 (const_unop): Likewise. Store the reduction accumulator in a
9284 variable rather than an array.
9285 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9286 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9287 the new vector, rather than constructing it after the input arrays.
9288 (native_interpret_vector): Use auto_vec<tree> when building
9289 a vector passed to build_vector. Add elements in order.
9290 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9291 auto_vec<tree> when building a vector passed to build_vector.
9292 (vect_create_epilog_for_reduction): Likewise.
9293 (vectorizable_induction): Likewise.
9294 (get_initial_def_for_reduction): Likewise. Fix indentation of
9296 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9298 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9299 passed to build_vector.
9301 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9302 Alan Hayward <alan.hayward@arm.com>
9303 David Sherwood <david.sherwood@arm.com>
9305 * tree-core.h (tree_base::u): Add an "nelts" field.
9306 (tree_vector): Use VECTOR_CST_NELTS as the length.
9307 * tree.c (tree_size): Likewise.
9308 (make_vector): Initialize VECTOR_CST_NELTS.
9309 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9310 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9311 TYPE_VECTOR_SUBPARTS.
9312 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9313 as the number of units, setting it from VECTOR_CST_NELTS.
9314 (const_vector_from_tree): Likewise.
9315 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9316 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9317 (fold_negate_expr_1): Likewise.
9318 (fold_convert_const): Likewise.
9319 (const_binop): Likewise. Differentiate the number of output and
9321 (const_unop): Likewise.
9322 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9323 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9324 in cases that did the opposite.
9326 2017-09-14 Richard Biener <rguenther@suse.de>
9328 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9331 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9333 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9335 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9338 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9339 if and where to split a bb, except for splitting before debug insn
9340 sequences followed by non-label real insn. Delete debug insns
9341 in between basic blocks.
9343 * combine.c (make_compound_operation_int): Formatting fixes.
9345 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9346 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9347 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9348 * config/sol2.h (LINK_EH_SPEC): Likewise.
9349 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9350 * config/s390/linux.h (LINK_SPEC): Likewise.
9351 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9352 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9353 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9354 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9355 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9356 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9357 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9358 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9360 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9362 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9364 (ENDFILE_LINUX_SPEC): Likewise.
9365 (LINK_EH_SPEC): Likewise.
9366 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9367 (LINK_OS_LINUX_SPEC32): Likewise.
9368 (LINK_OS_LINUX_SPEC64): Likewise.
9369 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9370 (LINK_OS_LINUX_SPEC): Likewise.
9372 2017-09-13 Martin Liska <mliska@suse.cz>
9375 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9376 CASE_HIGH is NULL_TREE.
9378 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9379 Alan Hayward <alan.hayward@arm.com>
9380 David Sherwood <david.sherwood@arm.com>
9382 * target.def (secondary_memory_needed): New hook.
9383 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9384 instead of SECONDARY_MEMORY_NEEDED.
9385 (secondary_memory_needed_mode): Likewise.
9386 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9387 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9388 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9389 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9390 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9391 * doc/tm.texi: Regenerate.
9392 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9393 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9394 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9395 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9396 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9397 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9398 mode argument first and change the reg_class arguments to reg_class_t.
9399 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9400 Make static. Update the call to inline_secondary_memory_needed.
9401 (ix86_register_move_cost): Update the call to
9402 inline_secondary_memory_needed.
9403 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9404 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9406 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9407 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9409 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9410 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9411 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9412 and match hook interface. Add comment from mips.h.
9413 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9414 * config/mmix/mmix.md (truncdfsf2): Refer to
9415 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9417 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9418 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9419 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9420 (pa_secondary_memory_needed): New function.
9421 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9422 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9423 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9424 (pdp11_secondary_memory_needed): Make static and match hook interface.
9425 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9426 * config/powerpcspe/powerpcspe-protos.h
9427 (rs6000_secondary_memory_needed_ptr): Delete.
9428 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9430 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9431 (rs6000_option_override_internal): Assign to
9432 targetm.secondary_memory_needed rather than
9433 rs6000_secondary_memory_needed_ptr.
9434 (rs6000_secondary_memory_needed): Match hook interface.
9435 (rs6000_debug_secondary_memory_needed): Likewise.
9436 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9437 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9438 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9439 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9440 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9441 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9443 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9444 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9445 (rs6000_option_override_internal): Assign to
9446 targetm.secondary_memory_needed rather than
9447 rs6000_secondary_memory_needed_ptr.
9448 (rs6000_secondary_memory_needed): Match hook interface.
9449 (rs6000_debug_secondary_memory_needed): Likewise.
9450 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9451 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9452 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9453 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9454 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9455 (sparc_secondary_memory_needed): New function.
9456 * lra-constraints.c (check_and_process_move): Refer to
9457 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9459 (curr_insn_transform): Likewise.
9460 (process_alt_operands): Use targetm.secondary_memory_needed
9461 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9462 (check_secondary_memory_needed_p): Likewise.
9463 (choose_split_class): Likewise.
9464 * reload.c: Unconditionally include code that was previously
9465 conditional on SECONDARY_MEMORY_NEEDED.
9466 (push_secondary_reload): Use targetm.secondary_memory_needed
9467 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9468 (push_reload): Likewise.
9469 * reload1.c: Unconditionally include code that was previously
9470 conditional on SECONDARY_MEMORY_NEEDED.
9471 (choose_reload_regs): Use targetm.secondary_memory_needed
9472 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9473 (gen_reload): Likewise.
9474 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9476 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9477 Alan Hayward <alan.hayward@arm.com>
9478 David Sherwood <david.sherwood@arm.com>
9480 * target.def (secondary_memory_needed_mode): New hook:
9481 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9482 * targhooks.h (default_secondary_memory_needed_mode): New function.
9483 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9484 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9485 * doc/tm.texi: Regenerate.
9486 * lra-constraints.c (check_and_process_move): Use
9487 targetm.secondary_memory_needed_mode instead of
9488 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9489 (curr_insn_transform): Likewise.
9490 * reload.c (get_secondary_mem): Likewise.
9491 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9492 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9494 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9495 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9496 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9497 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9498 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9500 * config/powerpcspe/powerpcspe-protos.h
9501 (rs6000_secondary_memory_needed_mode): Delete.
9502 * config/powerpcspe/powerpcspe.c
9503 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9504 (rs6000_secondary_memory_needed_mode): Make static.
9505 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9506 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9508 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9510 (rs6000_secondary_memory_needed_mode): Make static.
9511 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9512 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9513 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9514 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9515 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9517 (sparc_secondary_memory_needed_mode): New function.
9518 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9520 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9522 * config/aarch64/constraints.md (Umq): New constraint.
9523 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9525 (mov<mode>): Update condition.
9527 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9529 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9530 when bitposition is the same.
9532 2017-09-13 Richard Biener <rguenther@suse.de>
9534 * dwarf2out.c (output_die_symbol): Remove.
9535 (output_die): Do not output a DIEs symbol.
9537 2017-09-13 Richard Biener <rguenther@suse.de>
9540 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9541 default-def to avoid breaking iterator update with the weird
9542 interaction with cgraph_update_edges_for_call_stmt_node.
9544 2017-09-13 Richard Biener <rguenther@suse.de>
9546 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9547 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9548 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9549 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9551 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9553 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9554 Disable pc relative literal load irrespective of
9555 TARGET_FIX_ERR_A53_84341 for default.
9557 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9559 * config/sparc/sparc.c (output_return): Output the source location of
9560 the insn in the delay slot, if any.
9561 (output_sibcall): Likewise.
9563 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9566 * common.opt (-static-pie): New alias.
9567 (shared): Negate static-pie.
9568 (-no-pie): Update help text.
9570 (static-pie): New option.
9571 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9572 -static-pie support.
9573 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9574 (LINK_EH_SPEC): Likewise.
9575 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9576 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9577 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9578 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9579 (init_gcc_specs): Likewise.
9580 (init_spec): Likewise.
9581 (display_help): Update help message for -pie.
9582 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9585 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9587 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9588 (movdi_aarch64): Likewise.
9589 (movti_aarch64): Likewise.
9591 2017-09-12 Simon Wright <simon@pushface.org>
9594 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9595 calculation of the minor version, always output as 0.
9597 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9600 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9601 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9602 on it early, rather than manual conversion late. For
9603 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9604 instead of performing manual conversion.
9606 2017-09-12 Carl Love <cel@us.ibm.com>
9608 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9609 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9611 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9612 VMULOSW): Add definitions.
9613 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9614 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9615 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9616 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9617 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9619 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9621 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9623 (movti_aarch64): Likewise.
9624 (movdf_aarch64): Likewise.
9625 (movtf_aarch64): Likewise.
9626 (load_pairdi): Likewise.
9627 (store_pairdi): Likewise.
9628 (load_pairdf): Likewise.
9629 (store_pairdf): Likewise.
9630 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9631 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9632 (ldr_got_small_<mode>): Likewise.
9633 (ldr_got_small_28k_<mode>): Likewise.
9634 (ldr_got_tiny): Likewise.
9635 * config/aarch64/iterators.md (ldst_sz): New.
9636 (ldpstp_sz): Likewise.
9637 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9639 (thunderx_load): Split load_8 to load_16.
9640 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9642 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9643 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9644 (xgene1_store_pair): Split store_8 to store_16.
9645 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9646 (falkor_st_0_st_sd): Split store_8 to store_16.
9648 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9650 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9651 and store1/2/3/4 to store_4/8/12/16.
9652 * config/aarch64/aarch64.md: Update for rename.
9653 * config/arm/arm.md: Likewise.: Likewise.
9654 * config/arm/arm.c: Likewise.
9655 * config/arm/thumb1.md: Likewise.
9656 * config/arm/thumb2.md: Likewise.
9657 * config/arm/vfp.md: Likewise.
9658 * config/arm/arm-generic.md: Likewise.
9659 * config/arm/arm1020e.md: Likewise.
9660 * config/arm/arm1026ejs.md: Likewise.
9661 * config/arm/arm1136jfs.md: Likewise.
9662 * config/arm/arm926ejs.md: Likewise.
9663 * config/arm/cortex-a15.md: Likewise.
9664 * config/arm/cortex-a17.md: Likewise.
9665 * config/arm/cortex-a5.md: Likewise.
9666 * config/arm/cortex-a53.md: Likewise.
9667 * config/arm/cortex-a57.md: Likewise.
9668 * config/arm/cortex-a7.md: Likewise.
9669 * config/arm/cortex-a8.md: Likewise.
9670 * config/arm/cortex-a9.md: Likewise.
9671 * config/arm/cortex-m4.md: Likewise.
9672 * config/arm/cortex-m7.md: Likewise.
9673 * config/arm/cortex-r4.md: Likewise.
9674 * config/arm/exynos-m1.md: Likewise.
9675 * config/arm/fa526.md: Likewise.
9676 * config/arm/fa606te.md: Likewise.
9677 * config/arm/fa626te.md: Likewise.
9678 * config/arm/fa726te.md: Likewise.
9679 * config/arm/fmp626.md: Likewise.
9680 * config/arm/iwmmxt.md: Likewise.
9681 * config/arm/ldmstm.md: Likewise.
9682 * config/arm/marvell-pj4.md: Likewise.
9683 * config/arm/xgene1.md: Likewise.
9684 * config/aarch64/thunderx.md: Likewise.
9685 * config/aarch64/thunderx2t99.md: Likewise.
9686 * config/aarch64/falkor.md: Likewise.
9688 2017-09-12 Martin Liska <mliska@suse.cz>
9690 * attribs.c (private_lookup_attribute): New function.
9691 * attribs.h (private_lookup_attribute): Declared here.
9692 (lookup_attribute): Called from this place.
9694 2017-09-12 Richard Biener <rguenther@suse.de>
9696 PR tree-optimization/82157
9697 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9698 stmts with side-effects.
9700 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9701 Alan Hayward <alan.hayward@arm.com>
9702 David Sherwood <david.sherwood@arm.com>
9704 * target.def (hard_regno_nregs): New hook.
9705 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9706 * targhooks.h (default_hard_regno_nregs): Declare.
9707 * targhooks.c (default_hard_regno_nregs): New function.
9708 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9709 (TARGET_HARD_REGNO_NREGS): ...this hook.
9710 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9711 (CLASS_MAX_NREGS): Likewise.
9712 * doc/tm.texi: Regenerate.
9713 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9714 instead of HARD_REGNO_NREGS.
9715 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9716 HARD_REGNO_NREGS in the comment.
9717 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9718 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9719 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9720 Return an unsigned int.
9721 (TARGET_HARD_REGNO_NREGS): Redefine.
9722 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9723 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9724 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9725 (arc_hard_regno_nregs): New function.
9726 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9727 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9728 (arm_hard_regno_nregs): New function.
9729 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9730 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9731 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9732 (TARGET_HARD_REGNO_NREGS): Redefine.
9733 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9734 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9735 (HARD_REGNO_NREGS): Delete.
9736 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9737 (cr16_hard_regno_nregs): New function.
9738 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9739 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9740 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9741 (cris_hard_regno_nregs): New function.
9742 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9743 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9744 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9745 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9746 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9747 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9748 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9749 (frv_hard_regno_nregs): Make static. Take and return an
9751 (frv_class_max_nregs): Remove outdated copy of documentation.
9752 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9753 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9754 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9755 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9756 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9757 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9758 (TARGET_HARD_REGNO_NREGS): Redefine.
9759 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9760 (CLASS_MAX_NREGS): Update comment.
9761 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9762 (ia64_hard_regno_nregs): New function.
9763 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9764 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9765 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9766 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9767 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9769 (m32c_hard_regno_nregs): Likewise. Make static.
9770 (TARGET_HARD_REGNO_NREGS): Redefine.
9771 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9772 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9773 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9774 (m68k_hard_regno_nregs): New function.
9775 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9776 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9777 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9778 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9779 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9780 Take and return an unsigned int.
9781 (TARGET_HARD_REGNO_NREGS): Redefine.
9782 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9783 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9784 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9785 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9786 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9787 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9788 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9789 (msp430_hard_regno_nregs): Make static. Take and return an
9791 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9792 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9793 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9794 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9795 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9796 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9797 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9798 (TARGET_HARD_REGNO_NREGS): Redefine.
9799 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9800 (PA_HARD_REGNO_NREGS): ...this.
9801 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9802 (PA_HARD_REGNO_NREGS): ...this.
9803 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9804 (pa_hard_regno_nregs): New function.
9805 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9806 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9807 (pdp11_hard_regno_nregs): New function.
9808 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9809 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9810 (rs6000_hard_regno_nregs_hook): New function.
9811 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9812 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9813 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9814 Take and return an unsigned int. Move earlier in file.
9815 (TARGET_HARD_REGNO_NREGS): Redefine.
9816 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9817 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9818 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9819 (rl78_hard_regno_nregs): Make static. Take and return an
9821 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9822 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9823 (rs6000_hard_regno_nregs_hook): New function.
9824 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9825 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9826 (TARGET_HARD_REGNO_NREGS): Redefine.
9827 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9828 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9829 instead of HARD_REGNO_NREGS.
9830 (s390_hard_regno_nregs): New function.
9831 (s390_hard_regno_mode_ok): Add comment from s390.h.
9832 (TARGET_HARD_REGNO_NREGS): Redefine.
9833 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9834 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9835 (sh_hard_regno_nregs): New function.
9836 (sh_pass_in_reg_p): Use it.
9837 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9838 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9839 (sparc_hard_regno_nregs): New function.
9840 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9841 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9842 (spu_function_arg_advance): Use it, supplying a valid register number.
9843 (TARGET_HARD_REGNO_NREGS): Redefine.
9844 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9845 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9846 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9847 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9848 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9849 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9850 (CLASS_MAX_NREGS): Remove copy of old documentation.
9851 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9852 (visium_hard_regno_nregs): New function.
9853 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9854 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9855 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9856 xtensa_hard_regno_nregs): New function.
9857 * system.h (HARD_REGNO_NREGS): Poison.
9859 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9861 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9862 hard_regno_nregs instead of HARD_REGNO_NREGS.
9863 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9864 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9865 (c6x_expand_epilogue): Likewise.
9866 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9867 (frv_read_iacc_argument): Likewise.
9868 * config/sh/sh.c: Include regs.h.
9869 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9870 (regs_used): Likewise.
9871 (output_stack_adjust): Likewise.
9872 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9873 * expmed.c: Include regs.h.
9874 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9875 * ree.c: Include regs.h.
9876 (combine_reaching_defs): Use hard_regno_nregs instead of
9878 (add_removable_extension): Likewise.
9880 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9882 * regs.h (hard_regno_nregs): Turn into a function.
9883 (end_hard_regno): Update accordingly.
9884 * caller-save.c (setup_save_areas): Likewise.
9885 (save_call_clobbered_regs): Likewise.
9886 (replace_reg_with_saved_mem): Likewise.
9887 (insert_restore): Likewise.
9888 (insert_save): Likewise.
9889 * combine.c (can_change_dest_mode): Likewise.
9890 (move_deaths): Likewise.
9891 (distribute_notes): Likewise.
9892 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9893 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9894 (rs6000_split_multireg_move): Likewise.
9895 (rs6000_register_move_cost): Likewise.
9896 (rs6000_memory_move_cost): Likewise.
9897 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9898 (rs6000_split_multireg_move): Likewise.
9899 (rs6000_register_move_cost): Likewise.
9900 (rs6000_memory_move_cost): Likewise.
9901 * cselib.c (cselib_reset_table): Likewise.
9902 (cselib_lookup_1): Likewise.
9903 * emit-rtl.c (set_mode_and_regno): Likewise.
9904 * function.c (aggregate_value_p): Likewise.
9905 * ira-color.c (setup_profitable_hard_regs): Likewise.
9906 (check_hard_reg_p): Likewise.
9907 (calculate_saved_nregs): Likewise.
9908 (assign_hard_reg): Likewise.
9909 (improve_allocation): Likewise.
9910 (calculate_spill_cost): Likewise.
9911 * ira-emit.c (modify_move_list): Likewise.
9912 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9913 (ira_hard_reg_in_set_p): Likewise.
9914 * ira.c (setup_reg_mode_hard_regset): Likewise.
9915 (clarify_prohibited_class_mode_regs): Likewise.
9916 (check_allocation): Likewise.
9917 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9918 (lra_setup_reg_renumber): Likewise.
9919 (setup_try_hard_regno_pseudos): Likewise.
9920 (spill_for): Likewise.
9921 (assign_hard_regno): Likewise.
9922 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9923 * lra-constraints.c (in_class_p): Likewise.
9924 (lra_constraint_offset): Likewise.
9925 (simplify_operand_subreg): Likewise.
9926 (lra_constraints): Likewise.
9927 (split_reg): Likewise.
9928 (split_if_necessary): Likewise.
9929 (invariant_p): Likewise.
9930 (inherit_in_ebb): Likewise.
9931 * lra-lives.c (process_bb_lives): Likewise.
9932 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9933 (get_hard_regs): Likewise.
9934 (do_remat): Likewise.
9935 * lra-spills.c (assign_spill_hard_regs): Likewise.
9936 * mode-switching.c (create_pre_exit): Likewise.
9937 * postreload.c (reload_combine_recognize_pattern): Likewise.
9938 * recog.c (peep2_find_free_register): Likewise.
9939 * regcprop.c (kill_value_regno): Likewise.
9940 (set_value_regno): Likewise.
9941 (copy_value): Likewise.
9942 (maybe_mode_change): Likewise.
9943 (find_oldest_value_reg): Likewise.
9944 (copyprop_hardreg_forward_1): Likewise.
9945 * regrename.c (check_new_reg_p): Likewise.
9946 (regrename_do_replace): Likewise.
9947 * reload.c (push_reload): Likewise.
9948 (combine_reloads): Likewise.
9949 (find_dummy_reload): Likewise.
9950 (operands_match_p): Likewise.
9951 (find_reloads): Likewise.
9952 (find_equiv_reg): Likewise.
9953 (reload_adjust_reg_for_mode): Likewise.
9954 * reload1.c (count_pseudo): Likewise.
9955 (count_spilled_pseudo): Likewise.
9956 (find_reg): Likewise.
9957 (clear_reload_reg_in_use): Likewise.
9958 (free_for_value_p): Likewise.
9959 (allocate_reload_reg): Likewise.
9960 (choose_reload_regs): Likewise.
9961 (reload_adjust_reg_for_temp): Likewise.
9962 (emit_reload_insns): Likewise.
9963 (delete_output_reload): Likewise.
9964 * rtlanal.c (subreg_get_info): Likewise.
9965 * sched-deps.c (sched_analyze_reg): Likewise.
9966 * sel-sched.c (init_regs_for_mode): Likewise.
9967 (mark_unavailable_hard_regs): Likewise.
9968 (choose_best_reg_1): Likewise.
9969 (verify_target_availability): Likewise.
9970 * valtrack.c (dead_debug_insert_temp): Likewise.
9971 * var-tracking.c (track_loc_p): Likewise.
9972 (emit_note_insn_var_location): Likewise.
9973 * varasm.c (make_decl_rtl): Likewise.
9974 * reginfo.c (choose_hard_reg_mode): Likewise.
9975 (init_reg_modes_target): Refer directly to
9976 this_target_regs->x_hard_regno_nregs.
9978 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9980 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9981 instead of hard_regno_nregs.
9983 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9985 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9986 end_hard_regno instead of hard_regno_nregs.
9987 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9988 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9989 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9990 * ira-color.c (improve_allocation): Likewise.
9991 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9992 * lra-lives.c (mark_regno_live): Likewise.
9993 (mark_regno_dead): Likewise.
9994 * lra-remat.c (operand_to_remat): Likewise.
9995 * lra.c (collect_non_operand_hard_regs): Likewise.
9996 * postreload.c (reload_combine_note_store): Likewise.
9997 (move2add_valid_value_p): Likewise.
9998 * reload.c (regno_clobbered_p): Likewise.
10000 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10002 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
10004 * config/v850/v850.c (v850_reorg): Likewise.
10005 * reload.c (refers_to_regno_for_reload_p): Likewise.
10006 (find_equiv_reg): Likewise.
10007 * reload1.c (reload_reg_reaches_end_p): Likewise.
10009 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10011 * caller-save.c (add_used_regs): Use REG_NREGS instead of
10013 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
10014 * config/arm/arm.c (output_move_neon): Likewise.
10015 (arm_attr_length_move_neon): Likewise.
10016 (neon_split_vcombine): Likewise.
10017 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10018 (c6x_mark_reg_written): Likewise.
10019 (c6x_dwarf_register_span): Likewise.
10020 * config/i386/i386.c (ix86_save_reg): Likewise.
10021 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
10022 (rws_access_reg): Likewise.
10023 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
10024 * mode-switching.c (create_pre_exit): Likewise.
10025 * ree.c (combine_reaching_defs): Likewise.
10026 (add_removable_extension): Likewise.
10027 * regcprop.c (find_oldest_value_reg): Likewise.
10028 (copyprop_hardreg_forward_1): Likewise.
10029 * reload.c (reload_inner_reg_of_subreg): Likewise.
10030 (push_reload): Likewise.
10031 (combine_reloads): Likewise.
10032 (find_dummy_reload): Likewise.
10033 (reload_adjust_reg_for_mode): Likewise.
10034 * reload1.c (find_reload_regs): Likewise.
10035 (forget_old_reloads_1): Likewise.
10036 (reload_reg_free_for_value_p): Likewise.
10037 (reload_adjust_reg_for_temp): Likewise.
10038 (emit_reload_insns): Likewise.
10039 (delete_output_reload): Likewise.
10040 * sel-sched.c (choose_best_reg_1): Likewise.
10041 (choose_best_pseudo_reg): Likewise.
10043 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10044 Alan Hayward <alan.hayward@arm.com>
10045 David Sherwood <david.sherwood@arm.com>
10047 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
10048 * target.def (slow_unaligned_access): New hook.
10049 * targhooks.h (default_slow_unaligned_access): Declare.
10050 * targhooks.c (default_slow_unaligned_access): New function.
10051 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
10052 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
10053 * doc/tm.texi: Regenerate.
10054 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
10055 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
10056 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
10058 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
10059 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
10061 (rs6000_slow_unaligned_access): New function.
10062 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10063 (expand_block_compare): Likewise.
10064 (expand_strn_compare): Likewise.
10065 (rs6000_rtx_costs): Likewise.
10066 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
10067 (riscv_slow_unaligned_access): Likewise.
10068 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
10069 (riscv_slow_unaligned_access_p): ...this and make static.
10070 (riscv_option_override): Update accordingly.
10071 (riscv_slow_unaligned_access): New function.
10072 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10073 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
10074 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
10075 (rs6000_slow_unaligned_access): New function.
10076 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
10077 (rs6000_rtx_costs): Likewise.
10078 * config/rs6000/rs6000-string.c (expand_block_compare)
10079 (expand_strn_compare): Use targetm.slow_unaligned_access instead
10080 of SLOW_UNALIGNED_ACCESS.
10081 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
10082 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
10083 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
10084 of SLOW_UNALIGNED_ACCESS.
10085 * expmed.c (simple_mem_bitfield_p): Likewise.
10086 * expr.c (alignment_for_piecewise_move): Likewise.
10087 (emit_group_load_1): Likewise.
10088 (emit_group_store): Likewise.
10089 (copy_blkmode_from_reg): Likewise.
10090 (emit_push_insn): Likewise.
10091 (expand_assignment): Likewise.
10092 (store_field): Likewise.
10093 (expand_expr_real_1): Likewise.
10094 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10095 * lra-constraints.c (simplify_operand_subreg): Likewise.
10096 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10097 * gimple-ssa-store-merging.c: Likewise in block comment at start
10099 * tree-ssa-strlen.c: Include target.h.
10100 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10101 of SLOW_UNALIGNED_ACCESS.
10102 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10104 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10106 PR rtl-optimization/82185
10107 * expmed.c (emit_store_flag_int): Only test tem if it has been
10110 2017-09-12 Richard Biener <rguenther@suse.de>
10112 PR middle-end/82149
10113 * match.pd ((FTYPE) N CMP CST): Fix typo.
10115 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10117 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10119 (mips_near_type_p): Add 'short_call' attribute as a synonym
10121 * doc/extend.texi (short_call): Document new function attribute.
10123 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10126 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10127 assertion check that in the condition.
10128 (get_atomic_generic_size): Likewise. Before testing if parameter
10129 has pointer type, if it has array type, call for C++
10130 default_conversion to perform array-to-pointer conversion.
10132 2017-09-12 Richard Biener <rguenther@suse.de>
10134 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10135 for operations we cannot scalarize.
10137 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10139 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10140 vectors heap vectors. Clean up comments.
10141 Make visited_bbs a reference.
10142 (profitable_jump_thread_path): Make GC
10143 vectors heap vectors. Clean up comments.
10145 (convert_and_register_jump_thread_path): Make GC vectors heap
10147 (check_subpath_and_update_thread_path): Same. Clean up comments.
10148 Make visited_bbs a reference.
10149 (handle_phi): Abstract common code to to
10150 register_jump_thread_path_if_profitable.
10151 Rename VAR_BB to DEF_BB.
10153 Make GC vectors heap vectors.
10154 Make visited_bbs a reference.
10155 (handle_assignment): Same.
10156 (register_jump_thread_path_if_profitable): New.
10157 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10159 Make GC vectors heap vectors. Clean up comments.
10160 Make visited_bbs a reference.
10161 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10162 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10165 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10168 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10169 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10171 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10173 Revert r251800 and r251799.
10175 2017-09-11 Martin Jambor <mjambor@suse.cz>
10178 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10179 arguments in advance.
10180 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10181 use it to find predecessor edges.
10182 (naive_outof_ssa): Collect vector of predecessors.
10184 2017-09-08 Jason Merrill <jason@redhat.com>
10186 PR c++/70029 - ICE with ref-qualifier and -flto
10187 * langhooks.h (struct lang_hooks_for_types): Add
10188 copy_lang_qualifiers.
10189 * attribs.c (build_type_attribute_qual_variant): Use it.
10190 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10192 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10193 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10195 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10198 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10199 (*mulsi3_sp64): New instruction.
10200 (mulsi3): New expander.
10202 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10204 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10206 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10208 * sancov.c: Include memmodel.h.
10210 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10213 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10216 2017-09-07 Carl Love <cel@us.ibm.com>
10218 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10219 the sldi instruction.
10221 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10223 * sancov.c: Include tm_p.h.
10225 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10228 * output.h (switch_to_other_text_partition): New declaration.
10229 * varasm.c (switch_to_other_text_partition): New function.
10230 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10231 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10232 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10233 to the other text partition before emitting LCL label and switch back
10234 after emitting the word after it.
10236 2017-09-07 Richard Biener <rguenther@suse.de>
10238 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10239 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10240 critical edges here, after loop init.
10241 (pass_data_pre): Remove PROP_no_crit_edges flags.
10242 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10243 for valueization of call args to avoid leaking VN_TOP.
10244 (visit_use): Assert we do not visit default defs.
10245 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10246 Use error_mark_node to more easily detect leaking VN_TOP.
10247 All default-defs are varying, not VN_TOP. Mark them visited.
10248 (run_scc_vn): Make code match comment.
10250 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10252 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10253 OPTION_MASK_FLOAT128_KEYWORD.
10254 (POWERPC_MASKS): Likewise.
10255 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10256 support for the -mfloat128-type option, and make -mfloat128
10257 default on PowerPC Linux systems. Define or undefine
10258 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10259 Define __float128 to be __ieee128 if IEEE 128-bit support is
10260 enabled, or undefine it.
10261 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10262 Delete defining __FLOAT128_TYPE__.
10263 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10264 -mfloat128-type option and make -mfloat128 default on PowerPC
10266 (TARGET_FLOAT128_TYPE): Likewise.
10267 (-mfloat128-type): Likewise.
10268 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10269 Delete the -mfloat128-type option and make -mfloat128 default on
10270 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10271 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10272 128-bit floating point is enabled. Change tests from using
10273 -mfloat128-type to -mfloat128.
10274 (rs6000_mangle_type): Use the correct mangling for the __float128
10275 type even if normal long double is restricted to 64-bits.
10276 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10278 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10279 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10280 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10282 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10283 documentation for -mfloat128.
10285 2017-09-06 Olivier Hainque <hainque@adacore.com>
10287 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10289 2017-09-06 Wish Wu <wishwu007@gmail.com>
10290 Jakub Jelinek <jakub@redhat.com>
10292 * asan.c (initialize_sanitizer_builtins): Add
10293 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10294 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10295 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10296 BT_FN_VOID_UINT64_PTR variables.
10297 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10298 (BT_FN_VOID_UINT16_UINT16): Likewise.
10299 (BT_FN_VOID_UINT32_UINT32): Likewise.
10300 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10301 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10302 (BT_FN_VOID_UINT64_PTR): Likewise.
10303 * common.opt (flag_sanitize_coverage): New variable.
10304 (fsanitize-coverage=trace-pc): Remove.
10305 (fsanitize-coverage=): Add.
10306 * flag-types.h (enum sanitize_coverage_code): New enum.
10307 * fold-const.c (fold_range_test): Disable non-short-circuit
10308 optimization if flag_sanitize_coverage.
10309 (fold_truth_andor): Likewise.
10310 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10311 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10312 (coverage_sanitizer_opts): New array.
10313 (get_closest_sanitizer_option): Add OPTS argument, handle also
10314 OPT_fsanitize_coverage_.
10315 (parse_sanitizer_options): Adjusted to also handle
10316 OPT_fsanitize_coverage_.
10317 (common_handle_option): Add OPT_fsanitize_coverage_.
10318 * sancov.c (instrument_comparison, instrument_switch): New function.
10319 (sancov_pass): Add trace-cmp support.
10320 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10321 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10322 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10323 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10324 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10325 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10326 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10327 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10328 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10329 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10331 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10333 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10334 error. Only quit immediately if parsing is complete.
10335 (BEGIN): Initialize fatal_err and parse_done.
10336 (begin fpu, end fpu): Check number of arguments.
10337 (begin arch, end arch): Likewise.
10338 (begin cpu, end cpu): Likewise.
10339 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10340 (optalias): Likewise.
10342 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10344 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10345 * config/arm/arm-isa.h: Delete. Move definitions to ...
10346 * arm-cpus.in: ... here. Use new feature and fgroup values.
10347 * config/arm/arm.c (arm_option_override): Use lower case for feature
10349 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10350 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10351 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10353 (print_isa_bits_for): New function.
10354 (gen_isa): New function.
10355 (gen_comm_data): Use print_isa_bits_for.
10356 (define feature): New keyword.
10357 (define fgroup): New keyword.
10358 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10359 (arm-isa.h): Add rule to generate file.
10360 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10361 case for feature bit names.
10363 2017-09-06 Richard Biener <rguenther@suse.de>
10365 * tree-ssa-pre.c (NECESSARY): Remove.
10366 (create_expression_by_pieces): Do not touch pass-local flags.
10367 (insert_into_preds_of_block): Likewise.
10368 (do_pre_regular_insertion): Likewise.
10369 (eliminate_insert): Likewise.
10370 (eliminate_dom_walker::before_dom_children): Likewise.
10371 (fini_eliminate): Do not look at inserted_exprs.
10372 (mark_operand_necessary): Remove.
10373 (remove_dead_inserted_code): Replace with simple work-list
10374 algorithm based on inserted_exprs and SSA uses.
10375 (pass_pre::execute): Re-order fini_eliminate and
10376 remove_dead_inserted_code.
10378 2017-09-06 Olivier Hainque <hainque@adacore.com>
10380 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10381 for VxWorks 7. Adjust surrounding comments.
10383 2017-09-06 Richard Biener <rguenther@suse.de>
10385 * gimple-ssa-strength-reduction.c
10386 (find_candidates_dom_walker::before_dom_children): Also allow
10389 2017-09-06 Richard Biener <rguenther@suse.de>
10391 PR tree-optimization/82108
10392 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10393 for gap in the non-permutation SLP case.
10395 2017-09-06 Martin Jambor <mjambor@suse.cz>
10397 PR tree-optimization/82078
10398 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10399 add_access_to_work_queue...
10400 (build_accesses_from_assign): ...here.
10401 (propagate_all_subaccesses): Make sure racc is the group
10402 representative, if there is one.
10404 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10406 PR middle-end/82095
10407 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10410 2017-09-06 Richard Biener <rguenther@suse.de>
10412 * gimple-ssa-strength-reduction.c
10413 (find_candidates_dom_walker::before_doom_children): Use a
10414 type and not a mode check.
10416 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10419 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10421 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10423 2017-09-05 Jeff Law <law@redhat.com>
10425 PR tree-optimization/64910
10426 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10427 cases where we have 3 or more operands.
10429 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10431 PR middle-end/81768
10432 * omp-low.c (lower_omp_for): Recompute tree invariant if
10433 gimple_omp_for_initial/final is ADDR_EXPR.
10435 PR middle-end/81768
10436 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10437 into gimple val before gimplification fo the COND_EXPR.
10439 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10441 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10442 REGION_COPY argument.
10443 (thread_through_all_blocks): Remove unused argument to
10444 duplicate_thread_path.
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_gen_adjusted_ldpstp):
10451 Take a scalar_mode rather than a machine_mode.
10452 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10453 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10454 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10455 (aarch64_gen_adjusted_ldpstp): Likewise.
10456 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10458 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10459 Alan Hayward <alan.hayward@arm.com>
10460 David Sherwood <david.sherwood@arm.com>
10462 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10463 Take a scalar_int_mode instead of a machine_mode.
10464 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10465 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10466 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10467 (aarch64_simd_attr_length_rglist): Delete.
10468 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10469 a scalar_int_mode instead of a machine_mode.
10470 (aarch64_add_offset): Likewise.
10471 (aarch64_internal_mov_immediate): Likewise
10472 (aarch64_add_constant_internal): Likewise.
10473 (aarch64_add_constant): Likewise.
10474 (aarch64_movw_imm): Likewise.
10475 (aarch64_rtx_arith_op_extract_p): Likewise.
10476 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10477 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10478 Remove assert that the mode isn't a vector.
10479 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10480 (aarch64_expand_mov_immediate): Update calls after above changes.
10481 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10482 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10483 (aarch64_move_imm): Likewise.
10484 (aarch64_can_const_movi_rtx_p): Likewise.
10485 (aarch64_strip_extend): Likewise.
10486 (aarch64_extr_rtx_p): Likewise.
10487 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10488 a CONST_INT when the mode parameter is VOIDmode.
10489 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10491 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10493 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10494 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10495 (bitwise_type_for_mode): Update accordingly.
10497 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10499 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10500 * stor-layout.c (mode_for_size_tree): Likewise.
10501 (mode_for_array): Update accordingly.
10502 (layout_decl): Likewise.
10503 (compute_record_mode): Likewise. Only set the mode once.
10505 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10507 * target.def (get_mask_mode): Change return type to opt_mode.
10509 * doc/tm.texi: Regenerate.
10510 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10511 * targhooks.c (default_get_mask_mode): Likewise.
10512 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10513 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10514 targetm.get_mask_mode.
10515 * tree.c (build_truth_vector_type): Likewise.
10517 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10519 * machmode.h (mode_for_vector): Return an opt_mode.
10520 * stor-layout.c (mode_for_vector): Likewise.
10521 (mode_for_int_vector): Update accordingly.
10522 (layout_type): Likewise.
10523 * config/i386/i386.c (emit_memmov): Likewise.
10524 (ix86_expand_set_or_movmem): Likewise.
10525 (ix86_expand_vector_init): Likewise.
10526 (ix86_get_mask_mode): Likewise.
10527 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10529 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10530 * expmed.c (extract_bit_field_1): Likewise.
10531 * expr.c (expand_expr_real_2): Likewise.
10532 * optabs-query.c (can_vec_perm_p): Likewise.
10533 (can_vec_mask_load_store_p): Likewise.
10534 * optabs.c (expand_vec_perm): Likewise.
10535 * targhooks.c (default_get_mask_mode): Likewise.
10536 * tree-vect-stmts.c (vectorizable_store): Likewise.
10537 (vectorizable_load): Likewise.
10538 (get_vectype_for_scalar_type_and_size): Likewise.
10540 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10542 * machmode.h (mode_for_int_vector): New function.
10543 * stor-layout.c (mode_for_int_vector): Likewise.
10544 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10545 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10546 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10547 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10548 (s390_expand_vcond): Likewise.
10550 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10552 * machmode.h (opt_machine_mode): New type.
10553 (opt_mode<T>): Allow construction from anything that can be
10555 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10556 (mode_for_size): Return an opt_machine_mode.
10557 * stor-layout.c (mode_for_size): Likewise.
10558 (mode_for_size_tree): Update call accordingly.
10559 (bitwise_mode_for_mode): Likewise.
10560 (make_fract_type): Likewise.
10561 (make_accum_type): Likewise.
10562 * caller-save.c (replace_reg_with_saved_mem): Update call
10564 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10565 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10566 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10567 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10568 * expmed.c (extract_bit_field_1): Likewise.
10569 * reload.c (get_secondary_mem): Likewise.
10570 * varasm.c (assemble_integer): Likewise.
10571 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10574 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10576 * machmode.h (decimal_float_mode_for_size): New function.
10577 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10578 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10579 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10580 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10581 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10582 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10584 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10586 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10587 (get_builtin_sync_mode): Likewise.
10588 (expand_ifn_atomic_compare_exchange): Likewise.
10589 (expand_builtin_atomic_clear): Likewise.
10590 (expand_builtin_atomic_test_and_set): Likewise.
10591 (fold_builtin_atomic_always_lock_free): Likewise.
10592 * calls.c (compute_argument_addresses): Likewise.
10593 (emit_library_call_value_1): Likewise.
10594 (store_one_arg): Likewise.
10595 * combine.c (combine_instructions): Likewise.
10596 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10597 * config/arm/arm.c (arm_function_value): Likewise.
10598 (aapcs_allocate_return_reg): Likewise.
10599 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10600 * config/i386/i386.c (construct_container): Likewise.
10601 (ix86_gimplify_va_arg): Likewise.
10602 (ix86_expand_sse_cmp): Likewise.
10603 (emit_memmov): Likewise.
10604 (emit_memset): Likewise.
10605 (expand_small_movmem_or_setmem): Likewise.
10606 (ix86_expand_pextr): Likewise.
10607 (ix86_expand_pinsr): Likewise.
10608 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10609 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10611 * config/mips/mips.c (mips_function_value_1) Likewise.
10612 (mips_block_move_straight): Likewise.
10613 (mips_expand_ins_as_unaligned_store): Likewise.
10614 * config/powerpcspe/powerpcspe.c
10615 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10616 (rs6000_darwin64_record_arg_flush): Likewise.
10617 * config/rs6000/rs6000.c
10618 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10619 (rs6000_darwin64_record_arg_flush): Likewise.
10620 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10621 (sparc_function_value_1): Likewise.
10622 * config/spu/spu.c (adjust_operand): Likewise.
10623 (spu_emit_branch_or_set): Likewise.
10624 (arith_immediate_p): Likewise.
10625 * emit-rtl.c (gen_lowpart_common): Likewise.
10626 * expr.c (expand_expr_real_1): Likewise.
10627 * function.c (assign_parm_setup_block): Likewise.
10628 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10629 * reload1.c (alter_reg): Likewise.
10630 * stor-layout.c (mode_for_vector): Likewise.
10631 (layout_type): Likewise.
10633 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10635 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10636 (spu_convert_move): Likewise.
10637 * lower-subreg.c (resolve_simple_move): Likewise.
10639 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10642 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10643 define_insn to a define_expand.
10644 (altivec_vsum2sws_direct): New define_insn.
10645 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10647 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10649 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10652 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10657 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10658 realign stack if stack alignment needed is less than incoming
10661 2017-09-05 Marek Polacek <polacek@redhat.com>
10664 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10667 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10669 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10671 2017-09-05 Richard Biener <rguenther@suse.de>
10673 PR tree-optimization/82084
10674 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10676 2017-09-05 Richard Biener <rguenther@suse.de>
10678 PR tree-optimization/82102
10679 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10681 2017-09-05 Martin Liska <mliska@suse.cz>
10683 PR tree-optimization/82032
10684 * tree-cfg.c (generate_range_test): New function.
10685 * tree-cfg.h (generate_range_test): Declared here.
10686 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10687 (cleanup_control_expr_graph): Use it.
10688 * tree-switch-conversion.c (try_switch_expansion): Remove
10690 (emit_case_nodes): Use generate_range_test.
10692 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10695 * config/i386/i386.md (*<btsc><mode>_mask): Add
10696 TARGET_USE_BT to insn constraint.
10697 (*btr<mode>_mask): Ditto.
10699 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10701 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10702 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10704 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10707 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10708 TARGET_NEON and TARGET_IWMMXT.
10709 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10710 TARGET_NEON and TARGET_IWMMXT.
10711 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10713 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10715 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10716 (ix86_rewrite_tls_address): Ditto.
10717 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10718 (ix86_rewrite_tls_address_1): Ditto.
10719 (ix86_rewrite_tls_address): Ditto.
10720 * config/i386/predicates.md (tls_address_pattern): New predicate.
10721 * config/i386/i386.md (TLS address splitter): New splitter.
10723 2017-09-04 Richard Biener <rguenther@suse.de>
10725 PR tree-optimization/82084
10726 * fold-const.h (can_native_encode_string_p): Declare.
10727 * fold-const.c (can_native_encode_string_p): Factor out from ...
10728 (native_encode_string): ... here.
10729 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10730 vectorizing stores from constants we later cannot handle.
10732 2017-09-04 Marek Polacek <polacek@redhat.com>
10735 * doc/invoke.texi: Update -Wtautological-compare documentation.
10737 2017-09-04 Jeff Law <law@redhat.com>
10739 PR tree-optimization/64910
10740 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10741 swap the first and last operand if the last is a constant.
10743 2017-09-04 Marek Polacek <polacek@redhat.com>
10746 * convert.c (do_narrow): When sanitizing signed integer overflows,
10747 bail out for signed types.
10748 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10750 2017-09-04 Richard Biener <rguenther@suse.de>
10752 PR tree-optimization/82060
10753 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10754 Move devirtualization after stmt folding and before EH/AB/noreturn
10755 cleanup to get the stmt refs canonicalized. Use a bool instead
10756 of gimple_modified_p since that doesn't work for NOPs. Schedule
10757 NOPs generated by folding for removal.
10759 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10760 Alan Hayward <alan.hayward@arm.com>
10761 David Sherwood <david.sherwood@arm.com>
10763 * coretypes.h (pad_direction): New enum.
10764 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10765 (FUNCTION_ARG_PADDING): Likewise.
10766 * target.def (function_arg_padding): New hook.
10767 * targhooks.h (default_function_arg_padding): Declare.
10768 * targhooks.c (default_function_arg_padding): New function.
10769 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10770 (TARGET_FUNCTION_ARG_PADDING): ...this.
10771 * doc/tm.texi: Regenerate.
10772 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10773 instead of direction.
10774 (compute_argument_addresses): Likewise.
10775 (load_register_parameters): Likewise.
10776 (emit_library_call_value_1): Likewise.
10777 (store_one_arg): Use targetm.calls.function_arg_padding instead
10778 of FUNCTION_ARG_PADDING.
10779 (must_pass_in_stack_var_size_or_pad): Likewise.
10780 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10781 (emit_group_store): Likewise.
10782 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10783 instead of FUNCTION_ARG_PADDING.
10784 (emit_push_insn): Likewise, and propagate enum change throughout
10786 * function.h (direction): Delete.
10787 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10789 * function.c (assign_parm_find_stack_rtl): Likewise.
10790 (assign_parm_setup_block_p): Likewise.
10791 (assign_parm_setup_block): Likewise.
10792 (gimplify_parameters): Likewise.
10793 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10794 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10796 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10797 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10798 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10799 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10800 (aarch64_function_arg_padding): ...this new function.
10801 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10802 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10803 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10804 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10805 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10806 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10807 (arm_pad_arg_upward): Replace with...
10808 (arm_function_arg_padding): ...this new function.
10809 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10811 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10812 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10813 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10814 (ia64_hpux_function_arg_padding): Replace with...
10815 (ia64_function_arg_padding): ...this new function. Use pad_direction
10816 instead of direction. Check for TARGET_HPUX.
10817 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10818 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10819 (iq2000_function_arg_padding): New function.
10820 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10821 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10822 (mips_function_arg_padding): ...this new function.
10823 (mips_pad_reg_upward): Update accordingly.
10824 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10825 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10826 targetm.calls.function_arg_padding.
10827 (FUNCTION_ARG_PADDING): Delete.
10828 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10829 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10830 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10831 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10832 (nios2_block_reg_padding): Return pad_direction instead of direction.
10833 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10834 instead of direction.
10835 (nios2_function_arg_padding): Likewise. Make static.
10836 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10837 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10838 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10839 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10840 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10841 (pa_function_arg_padding): Make static. Return pad_direction instead
10843 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10844 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10845 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10846 instead of direction. Use targetm.calls.function_arg_padding.
10847 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10848 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10849 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10850 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10851 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10853 (function_arg_padding): Rename to...
10854 (rs6000_function_arg_padding): ...this. Make static. Return
10855 pad_direction instead of direction.
10856 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10857 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10858 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10859 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10860 instead of direction. Use targetm.calls.function_arg_padding.
10861 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10862 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10863 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10864 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10865 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10866 (function_arg_padding): Rename to...
10867 (rs6000_function_arg_padding): ...this. Make static. Return
10868 pad_direction instead of direction.
10869 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10870 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10871 * config/s390/s390.c (s390_function_arg_padding): New function.
10872 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10873 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10874 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10875 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10876 (function_arg_padding): Rename to...
10877 (sparc_function_arg_padding): ...this. Make static. Return
10878 pad_direction instead of direction.
10879 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10880 * config/spu/spu.c (spu_function_arg_padding): New function.
10881 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10882 * system.h (FUNCTION_ARG_PADDING): Poison.
10884 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10885 Alan Hayward <alan.hayward@arm.com>
10886 David Sherwood <david.sherwood@arm.com>
10888 * target.def (modes_tieable_p): New hook.
10889 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10890 (TARGET_MODES_TIEABLE_P): ...this.
10891 * doc/tm.texi.in: Regenerate.
10892 * hooks.h (hook_bool_mode_mode_true): Declare.
10893 * hooks.c (hook_bool_mode_mode_true): New function.
10894 * combine.c (subst): Use targetm.modes_tieable_p instead of
10896 * dse.c (find_shift_sequence): Likewise.
10897 * expmed.c (extract_low_bits): Likewise.
10898 * lower-subreg.c: Include target.h.
10899 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10901 * rtlanal.c (rtx_cost): Likewise.
10902 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10903 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10904 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10905 (TARGET_MODES_TIEABLE_P): Redefine.
10906 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10907 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10908 (TARGET_MODES_TIEABLE_P): Redefine.
10909 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10910 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10911 (arc_modes_tieable_p): New function.
10912 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10913 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10914 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10915 (arm_modes_tieable_p): Make static.
10916 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10917 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10918 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10919 (TARGET_MODES_TIEABLE_P): Redefine.
10920 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10921 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10922 (TARGET_MODES_TIEABLE_P): Redefine.
10923 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10924 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10925 (cr16_modes_tieable_p): New function.
10926 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10927 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10928 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10929 (TRULY_NOOP_TRUNCATION): Update comment.
10930 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10931 (TRULY_NOOP_TRUNCATION): Update comment.
10932 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10933 (frv_modes_tieable_p): New function.
10934 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10935 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10936 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10937 (TARGET_MODES_TIEABLE_P): Redefine.
10938 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10939 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10940 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10941 (TARGET_MODES_TIEABLE_P): Redefine.
10942 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10943 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10944 (ia64_modes_tieable_p): New function.
10945 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10946 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10947 (iq2000_modes_tieable_p): New function.
10948 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10949 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10950 (lm32_modes_tieable_p): New function.
10951 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10952 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10953 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10954 (TARGET_MODES_TIEABLE_P): Redefine.
10955 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10956 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10957 (m32r_modes_tieable_p): New function.
10958 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10959 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10960 (m68k_modes_tieable_p): New function.
10961 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10962 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10963 (mcore_modes_tieable_p): New function.
10964 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10965 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10967 (TARGET_MODES_TIEABLE_P): Redefine.
10968 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10969 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10970 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10971 (TARGET_MODES_TIEABLE_P): Redefine.
10972 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10973 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10974 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10975 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10976 (mn10300_modes_tieable_p): ...this and make static.
10977 (TARGET_MODES_TIEABLE_P): Redefine.
10978 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10979 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10980 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10981 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10982 (msp430_modes_tieable_p): Make static.
10983 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10984 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10985 (TARGET_MODES_TIEABLE_P): Redefine.
10986 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10987 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10988 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10989 (TARGET_MODES_TIEABLE_P): Redefine.
10990 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10991 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10992 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10993 (TARGET_MODES_TIEABLE_P): Redefine.
10994 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10995 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10996 (pdp11_modes_tieable_p): New function.
10997 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10998 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10999 (rs6000_modes_tieable_p): New function.
11000 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11001 * config/powerpcspe/powerpcspe.md: Update comment.
11002 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
11003 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
11004 (TARGET_MODES_TIEABLE_P): Redefine.
11005 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
11006 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
11007 (rl78_modes_tieable_p): New function.
11008 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
11009 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
11010 (rs6000_modes_tieable_p): New function.
11011 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
11012 * config/rs6000/rs6000.md: Update comment.
11013 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
11014 * config/rx/rx.c (rx_modes_tieable_p): New function.
11015 (TARGET_MODES_TIEABLE_P): Redefine.
11016 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
11017 * config/s390/s390.c (s390_modes_tieable_p): New function.
11018 (TARGET_MODES_TIEABLE_P): Redefine.
11019 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
11020 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
11021 (sh_modes_tieable_p): New function.
11022 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
11023 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
11024 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
11025 (sparc_modes_tieable_p): Make static.
11026 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
11027 * config/spu/spu.c (spu_modes_tieable_p): New function.
11028 (TARGET_MODES_TIEABLE_P): Redefine.
11029 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
11030 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
11031 (TARGET_MODES_TIEABLE_P): Redefine.
11032 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
11033 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
11034 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
11035 * config/v850/v850.c (v850_modes_tieable_p): New function.
11036 (TARGET_MODES_TIEABLE_P): Redefine.
11037 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
11038 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
11039 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
11040 (visium_modes_tieable_p): New function.
11041 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
11042 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
11043 (xtensa_modes_tieable_p): New function.
11044 * system.h (MODES_TIEABLE_P): Poison.
11046 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11047 Alan Hayward <alan.hayward@arm.com>
11048 David Sherwood <david.sherwood@arm.com>
11050 * target.def (hard_regno_mode_ok): New hook.
11051 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
11052 (TARGET_HARD_REGNO_MODE_OK): ...this.
11053 * doc/tm.texi.in: Regenerate.
11054 * hooks.h (hook_bool_uint_mode_true): Declare.
11055 * hooks.c (hook_bool_uint_mode_true): New function.
11056 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
11057 HARD_REGNO_MODE_OK.
11058 * genpreds.c (write_insn_preds_c): Add an include of target.h.
11059 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
11060 instead of HARD_REGNO_MODE_OK.
11061 * caller-save.c: Include target.h.
11062 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
11063 HARD_REGNO_MODE_OK.
11064 * combine.c (can_combine_p): Likewise.
11065 (combinable_i3pat): Likewise.
11066 (can_change_dest_mode): Likewise.
11067 * expr.c (init_expr_target): Likewise.
11068 (convert_move): Likewise.
11069 (convert_modes): Likewise.
11070 * ira.c (setup_prohibited_class_mode_regs): Likewise.
11071 (setup_prohibited_mode_move_regs): Likewise.
11072 * ira.h (target_ira): Likewise.
11073 * lra-assigns.c (find_hard_regno_for_1): Likewise.
11074 * lra-constraints.c (process_alt_operands): Likewise.
11075 (split_reg): Likewise.
11076 * recog.c (peep2_find_free_register): Likewise.
11077 * ree.c (combine_reaching_defs): Likewise.
11078 * regcprop.c (maybe_mode_change): Likewise.
11079 * reginfo.c (init_reg_sets_1): Likewise.
11080 (choose_hard_reg_mode): Likewise.
11081 (simplifiable_subregs): Likewise.
11082 * regrename.c (check_new_reg_p): Likewise.
11083 * reload.c (find_valid_class): Likewise.
11084 (find_valid_class_1): Likewise.
11085 (reload_inner_reg_of_subreg): Likewise.
11086 (push_reload): Likewise.
11087 (combine_reloads): Likewise.
11088 (find_dummy_reload): Likewise.
11089 (find_reloads): Likewise.
11090 * reload1.c (find_reg): Likewise.
11091 (set_reload_reg): Likewise.
11092 (allocate_reload_reg): Likewise.
11093 (choose_reload_regs): Likewise.
11094 (reload_adjust_reg_for_temp): Likewise.
11095 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
11096 (simplify_subreg_regno): Likewise.
11097 * sel-sched.c (init_regs_for_mode): Likewise.
11098 * varasm.c (make_decl_rtl): Likewise.
11099 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
11100 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11101 HARD_REGNO_MODE_OK.
11102 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11103 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11104 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11105 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11106 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11108 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11109 (arc_mode_class): Delete.
11110 (HARD_REGNO_MODE_OK): Delete.
11111 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11112 (arc_hard_regno_mode_ok): Rename old array to...
11113 (arc_hard_regno_mode_ok_modes): ...this.
11114 (arc_conditional_register_usage): Update accordingly.
11115 (arc_mode_class): Make static.
11116 (arc_hard_regno_mode_ok): New function.
11117 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11118 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11119 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11120 (arm_hard_regno_mode_ok): Make static.
11121 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11122 HARD_REGNO_MODE_OK.
11123 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11124 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11125 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11127 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11128 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11129 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11130 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11131 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11132 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11133 * config/bfin/predicates.md (valid_reg_operand): Use
11134 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11135 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11136 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11137 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11138 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11139 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11140 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11141 (cr16_hard_regno_mode_ok): Make static and return a bool.
11142 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11143 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11144 (cris_hard_regno_mode_ok): New function.
11145 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11146 (epiphany_mode_class): Delete.
11147 (HARD_REGNO_MODE_OK): Delete.
11148 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11149 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11150 (hard_regno_mode_ok): Rename to...
11151 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11152 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11153 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11154 HARD_REGNO_MODE_OK.
11155 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11156 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11157 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11158 (frv_hard_regno_mode_ok): Make static and return a bool.
11159 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11160 HARD_REGNO_MODE_OK.
11161 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11162 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11163 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11164 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11166 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11167 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11168 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11169 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11171 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11172 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11173 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11174 (ia64_hard_regno_mode_ok): New function.
11175 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11176 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11177 (iq2000_hard_regno_mode_ok): New function.
11178 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11179 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11180 (lm32_hard_regno_mode_ok): New function.
11181 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11182 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11183 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11184 instead of HARD_REGNO_MODE_OK.
11185 (m32c_hard_regno_ok): Rename to...
11186 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11187 (m32c_cannot_change_mode_class): Update accordingly.
11188 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11189 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11190 (m32r_mode_class): Delete.
11191 (HARD_REGNO_MODE_OK): Delete.
11192 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11193 (m32r_hard_regno_mode_ok): Rename to...
11194 (m32r_hard_regno_modes): ...this.
11195 (m32r_mode_class): Make static.
11196 (m32r_hard_regno_mode_ok): New function.
11197 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11198 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11199 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11200 (m68k_hard_regno_mode_ok): Make static.
11201 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11202 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11203 (mcore_hard_regno_mode_ok): New function.
11204 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11205 (HARD_REGNO_MODE_OK): Delete.
11206 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11208 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11209 (microblaze_hard_regno_mode_ok): New function.
11210 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11211 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11212 (mips_hard_regno_mode_ok): Delete.
11213 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11214 (mips_hard_regno_mode_ok_p): ...this and make static.
11215 (mips_hard_regno_mode_ok_p): Rename to...
11216 (mips_hard_regno_mode_ok_uncached): ...this.
11217 (mips_hard_regno_mode_ok): New function.
11218 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11219 of HARD_REGNO_MODE_OK.
11220 (mips_option_override): Update after above name changes.
11221 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11222 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11223 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11224 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11225 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11226 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11227 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11228 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11229 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11230 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11231 (msp430_hard_regno_mode_ok): Make static and return a bool.
11232 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11233 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11234 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11236 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11237 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11238 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11239 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11240 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11241 (PA_HARD_REGNO_MODE_OK): ...this
11242 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11243 (PA_HARD_REGNO_MODE_OK): ...this.
11244 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11245 (pa_hard_regno_mode_ok): New function.
11246 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11247 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11248 (pdp11_hard_regno_mode_ok): New function.
11249 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11250 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11252 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11255 (rs6000_hard_regno_mode_ok): Rename to...
11256 (rs6000_hard_regno_mode_ok_uncached): ...this.
11257 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11258 (rs6000_hard_regno_mode_ok): New function.
11259 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11260 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11261 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11262 (riscv_hard_regno_mode_ok): ...this and make static.
11263 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11264 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11265 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11266 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11267 (rl78_hard_regno_mode_ok): Make static and return bool.
11268 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11269 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11271 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11273 (rs6000_hard_regno_mode_ok): Rename to...
11274 (rs6000_hard_regno_mode_ok_uncached): ...this.
11275 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11276 (rs6000_hard_regno_mode_ok): New function.
11277 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11278 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11279 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11280 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11281 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11282 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11284 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11285 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11286 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11287 (sh_hard_regno_mode_ok): Make static.
11288 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11289 instead of HARD_REGNO_MODE_OK.
11290 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11291 (sparc_mode_class): Delete.
11292 (HARD_REGNO_MODE_OK): Delete.
11293 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11294 (hard_regno_mode_classes): Make static.
11295 (sparc_mode_class): Likewise.
11296 (sparc_hard_regno_mode_ok): New function.
11297 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11298 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11299 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11301 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11302 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11303 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11304 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11305 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11306 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11307 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11308 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11309 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11310 (visium_hard_regno_mode_ok): New function.
11311 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11312 instead of HARD_REGNO_MODE_OK.
11313 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11314 (HARD_REGNO_MODE_OK): Delete.
11315 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11316 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11317 (xtensa_option_override): Update accordingly.
11318 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11319 (xtensa_hard_regno_mode_ok): New function.
11320 * system.h (HARD_REGNO_MODE_OK): Poison.
11322 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11323 Alan Hayward <alan.hayward@arm.com>
11324 David Sherwood <david.sherwood@arm.com>
11326 * target.def (hard_regno_call_part_clobbered): New hook.
11327 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11328 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11329 * doc/tm.texi: Regenerate.
11330 * hooks.h (hook_bool_uint_mode_false): Declare.
11331 * hooks.c (hook_bool_uint_mode_false): New function.
11332 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11333 * cselib.c (cselib_process_insn): Use
11334 targetm.hard_regno_call_part_clobbered instead of
11335 HARD_REGNO_CALL_PART_CLOBBERED.
11336 * ira-conflicts.c (ira_build_conflicts): Likewise.
11337 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11338 * lra-constraints.c (need_for_call_save_p): Likewise.
11339 * lra-lives.c: Include target.h.
11340 (check_pseudos_live_through_calls): Use
11341 targetm.hard_regno_call_part_clobbered instead of
11342 HARD_REGNO_CALL_PART_CLOBBERED.
11343 * regcprop.c: Include target.h.
11344 (copyprop_hardreg_forward_1): Use
11345 targetm.hard_regno_call_part_clobbered instead of
11346 HARD_REGNO_CALL_PART_CLOBBERED.
11347 * reginfo.c (choose_hard_reg_mode): Likewise.
11348 * regrename.c (check_new_reg_p): Likewise.
11349 * reload.c (find_equiv_reg): Likewise.
11350 * reload1.c (emit_reload_insns): Likewise.
11351 * sched-deps.c (deps_analyze_insn): Likewise.
11352 * sel-sched.c (init_regs_for_mode): Likewise.
11353 (mark_unavailable_hard_regs): Likewise.
11354 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11355 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11356 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11358 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11359 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11360 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11362 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11364 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11365 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11366 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11368 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11369 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11370 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11372 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11373 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11375 * config/powerpcspe/powerpcspe.c
11376 (rs6000_hard_regno_call_part_clobbered): New function.
11377 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11378 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11379 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11381 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11382 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11383 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11385 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11386 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11387 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11389 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11390 Alan Hayward <alan.hayward@arm.com>
11391 David Sherwood <david.sherwood@arm.com>
11393 * rtl.h (subreg_memory_offset): Declare.
11394 * emit-rtl.c (subreg_memory_offset): New function.
11395 * expmed.c (store_bit_field_1): Use it.
11396 * expr.c (undefined_operand_subword_p): Likewise.
11397 * simplify-rtx.c (simplify_subreg): Likewise.
11399 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11401 PR rtl-optimization/57448
11404 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11405 using atomic_load pattern.
11406 (expand_atomic_store): Likewise.
11408 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11411 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11412 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11413 handling. Use replace_call_with_value with NULL instead of
11414 gsi_replace, unlink_stmt_vdef and release_defs.
11416 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11419 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11421 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11424 * rtl.h (emit_library_call_value_1): Declare.
11425 (emit_library_call): Replace declaration with a series of overloads.
11426 Remove the parameter count argument.
11427 (emit_library_call_value): Likewise.
11428 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11429 with an "rtx_mode_t *".
11430 (emit_library_call_value): Delete.
11431 (emit_library_call): Likewise.
11432 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11433 (asan_emit_allocas_unpoison): Likewise.
11434 * builtins.c (expand_builtin_powi): Likewise.
11435 (expand_asan_emit_allocas_unpoison): Likewise.
11436 * cfgexpand.c (expand_main_function): Likewise.
11437 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11438 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11439 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11440 * config/arm/arm.c (arm_trampoline_init): Likewise.
11441 (arm_call_tls_get_addr): Likewise.
11442 (arm_expand_divmod_libfunc): Likewise.
11443 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11444 (smulsi3_highpart): Likewise.
11445 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11446 (c6x_expand_compare): Likewise.
11447 (c6x_expand_movmem): Likewise.
11448 * config/frv/frv.c (frv_trampoline_init): Likewise.
11449 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11450 (ix86_expand_divmod_libfunc): Likewise.
11451 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11452 (ia64_expand_compare): Likewise.
11453 (ia64_profile_hook): Likewise.
11454 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11455 (nonlocal_goto): Likewise.
11456 (restore_stack_nonlocal): Likewise.
11457 * config/m32r/m32r.c (block_move_call): Likewise.
11458 (m32r_trampoline_init): Likewise.
11459 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11460 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11461 (m68k_call_m68k_read_tp): Likewise.
11462 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11463 (microblaze_expand_divide): Likewise.
11464 * config/mips/mips.h (mips_args): Likewise.
11465 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11466 (MIPS_ICACHE_SYNC): Likewise.
11467 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11468 (nios2_trampoline_init): Likewise.
11469 * config/pa/pa.c (hppa_tls_call): Likewise.
11470 (pa_trampoline_init): Likewise.
11471 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11472 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11473 (expand_strn_compare): Likewise.
11474 (rs6000_generate_compare): Likewise.
11475 (rs6000_expand_float128_convert): Likewise.
11476 (output_profile_hook): Likewise.
11477 (rs6000_trampoline_init): Likewise.
11478 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11479 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11480 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11481 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11482 (rs6000_generate_compare): Likewise.
11483 (rs6000_expand_float128_convert): Likewise.
11484 (output_profile_hook): Likewise.
11485 (rs6000_trampoline_init): Likewise.
11486 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11487 * config/sh/sh.c (sh_trampoline_init): Likewise.
11488 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11489 (sparc_emit_float_lib_cmp): Likewise.
11490 (sparc32_initialize_trampoline): Likewise.
11491 (sparc64_initialize_trampoline): Likewise.
11492 (sparc_profile_hook): Likewise.
11493 * config/spu/spu.c (ea_load_store): Likewise.
11494 * config/spu/spu.md (floatunssidf2): Likewise.
11495 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11496 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11497 * config/visium/visium.c (expand_block_move_4): Likewise.
11498 (expand_block_move_2): Likewise.
11499 (expand_block_move_1): Likewise.
11500 (expand_block_set_4): Likewise.
11501 (expand_block_set_2): Likewise.
11502 (expand_block_set_1): Likewise.
11503 (visium_trampoline_init): Likewise.
11504 (visium_profile_hook): Likewise.
11505 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11506 (xtensa_setup_frame_addresses): Likewise.
11507 (xtensa_trampoline_init): Likewise.
11508 * except.c (sjlj_emit_function_enter): Likewise.
11509 (sjlj_emit_function_exit): Likewise.
11510 * explow.c (allocate_dynamic_stack_space): Likewise.
11511 (probe_stack_range): Likewise.
11512 * expr.c (convert_mode_scalar): Likewise.
11513 * optabs.c (expand_binop): Likewise.
11514 (expand_twoval_binop_libfunc): Likewise.
11515 (expand_unop): Likewise.
11516 (prepare_cmp_insn): Likewise.
11517 (prepare_float_lib_cmp): Likewise.
11518 (expand_float): Likewise.
11519 (expand_fix): Likewise.
11520 (expand_fixed_convert): Likewise.
11521 (maybe_emit_sync_lock_test_and_set): Likewise.
11522 (expand_atomic_compare_and_swap): Likewise.
11523 (expand_mem_thread_fence): Likewise.
11524 (expand_atomic_fetch_op): Likewise.
11526 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11528 * doc/generic.texi (OpenACC): Adjust URL.
11529 * doc/invoke.texi (C Dialect Options): Ditto.
11531 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11533 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11534 predicate for operand 1. Add (m,<S>) constraint.
11535 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11536 Prevent memory operand 1 with register operand 2.
11538 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11540 PR rtl-optimization/82024
11541 * combine.c (try_combine): If the combination result is a PARALLEL,
11542 and we only need to retain the SET in there that would be placed
11543 at I2, check that we can place that at I3 instead, before doing so.
11545 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11548 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11550 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11551 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11554 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11555 Jeff Law <law@redhat.com>
11557 * varasm.c (bss_initializer_p): Do not put constants into .bss
11558 (categorize_decl_for_section): Handle bss_initializer_p returning
11559 false when DECL_INITIAL is NULL.
11561 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11564 * config/s390/s390.c (s390_can_inline_p): New function.
11566 2017-09-01 Jeff Law <law@redhat.com>
11568 PR tree-optimization/82052
11569 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11570 Always initialize the returned slot after a hash table miss
11571 when INSERT is true.
11573 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11575 * config/s390/s390.md (mem_signal_fence): Remove.
11576 * doc/md.texi (mem_signal_fence): Remove.
11577 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11579 * target-insns.def (mem_signal_fence): Remove.
11581 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11584 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11587 * asan.c (create_odr_indicator): Strip name encoding from assembler
11588 name before appending it after __odr_asan_.
11590 2017-09-01 Martin Liska <mliska@suse.cz>
11592 PR tree-optimization/82059
11593 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11594 frequency only when an edge is redirected.
11596 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11598 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11599 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11600 (arc_conditional_register_usage): Remove ARC600 lp_count
11602 (arc_file_start): Emit Tag_ARC_CPU_variation.
11603 (arc_can_use_doloop_p): New conditions to use ZOLs.
11604 (hwloop_fail): New function.
11605 (hwloop_optimize): Likewise.
11606 (hwloop_pattern_reg): Likewise.
11607 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11608 (arc_reorg_loops): New function, calls reorg_loops.
11609 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11610 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11612 (arc_loop_hazard): Remove function, functionality moved into
11614 (arc_hazard): Remove arc_loop_hazard call.
11615 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11616 into hwloop_optimize.
11617 (arc_label_align): Remove ZOL handling.
11618 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11619 * config/arc/arc.md (doloop_begin): Remove pattern.
11620 (doloop_begin_i): Likewise.
11621 (doloop_end_i): Likewise.
11622 (doloop_fallback): Likewise.
11623 (doloop_fallback_m): Likewise.
11624 (doloop_end): Reimplement expand.
11625 (arc_lp): New pattern for LP instruction.
11626 (loop_end): New pattern.
11627 (loop_fail): Likewise.
11628 (decrement_and_branch_until_zero): Likewise.
11629 * config/arc/arc.opt (mlpc-width): New option.
11630 * doc/invoke.texi (mlpc-width): Document option.
11632 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11634 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11635 (arc_ccfsm_advance): Fix checking for delay slots.
11636 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11638 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11640 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11642 (movhi_insn): Update store instruction constraint which are saving
11643 6-bit short immediates.
11644 (movsi_insn): Consider also short scaled load operations.
11645 (zero_extendhisi2_i): Use Usd constraint instead of T.
11646 (extendhisi2_i): Add q constraint.
11647 (arc_clzsi2): Add type and length attributes.
11648 (arc_ctzsi2): Likewise.
11649 * config/arc/constraints.md (Usc): Update constraint, the
11650 assembler can parse two relocations for a single instruction.
11652 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11654 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11655 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11657 2017-08-31 Olivier Hainque <hainque@adacore.com>
11659 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11660 match as powerpc-wrs-vxworks*.
11662 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11664 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11665 register constraint for by-element operand.
11666 (aarch64_mls_elt_merge<mode>): Likewise.
11668 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11670 * config/arc/arc.c (arc_can_follow_jump): Check for short
11673 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11675 * config.gcc: Use g.opt for arc.
11676 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11677 functionality moved to ...
11678 (legitimate_scaled_address_p): New function, ...here.
11679 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11680 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11681 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11683 (arc_override_options): Handle G option.
11684 (arc_output_pic_addr_const): Correct function definition.
11685 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11686 (arc_decl_anon_ns_mem_p): Delete.
11687 (arc_in_small_data_p): Overhaul this function to take into
11688 consideration the value given via G option.
11689 (arc_rewrite_small_data_1): Renamed and corrected old
11690 arc_rewrite_small_data function.
11691 (arc_rewrite_small_data): New function.
11692 (small_data_pattern): Don't use pic_offset_table_rtx.
11693 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11694 * config/arc/simdext.md (movmisalignv2hi): Use
11695 prepare_move_operands function.
11697 (movmisalign*): Likewise.
11698 * doc/invoke.texi (ARC options): Document -G option.
11700 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11702 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11704 * config/arc/arc.c (arc_print_operand): Output scalled address for
11705 sdata whenever is possible.
11706 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11707 load/stores are available.
11708 (compact_sda_memory_operand): Check for the alignment required by
11709 code density instructions.
11710 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11712 * config/arc/constraints.md (Usd): Update constraint.
11713 (Us0): New constraint.
11714 (Usc): Update constraint.
11716 2017-08-31 Richard Biener <rguenther@suse.de>
11718 PR middle-end/82054
11719 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11720 function only once.
11722 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11724 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11725 Resize type_signature.
11727 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11728 Alan Hayward <alan.hayward@arm.com>
11729 David Sherwood <david.sherwood@arm.com>
11731 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11732 subregs whose inner modes can be stored in GPRs.
11733 (aarch64_classify_index): Likewise.
11735 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11736 Alan Hayward <alan.hayward@arm.com>
11737 David Sherwood <david.sherwood@arm.com>
11739 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11740 (V_INT_EQUIV): ...this.
11741 (v_cmp_result): Rename to...
11742 (v_int_equiv): ...this.
11743 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11744 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11745 (copysign<mode>3): Likewise.
11746 (aarch64_simd_bsl<mode>_internal): Likewise.
11747 (aarch64_simd_bsl<mode>): Likewise.
11748 (vec_cmp<mode><mode>): Likewise.
11749 (vcond<mode><mode>): Likewise.
11750 (vcond<v_cmp_mixed><mode>): Likewise.
11751 (vcondu<mode><v_cmp_mixed>): Likewise.
11752 (aarch64_cm<optab><mode>): Likewise.
11753 (aarch64_cmtst<mode>): Likewise.
11754 (aarch64_fac<optab><mode>): Likewise.
11755 (vec_perm_const<mode>): Likewise.
11756 (vcond_mask_<mode><v_cmp_result>): Rename to...
11757 (vcond_mask_<mode><v_int_equiv>): ...this.
11758 (vec_cmp<mode><v_cmp_result>): Rename to...
11759 (vec_cmp<mode><v_int_equiv>): ...this.
11761 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11762 Alan Hayward <alan.hayward@arm.com>
11763 David Sherwood <david.sherwood@arm.com>
11765 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11767 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11768 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11769 (UNSPEC_LD4_DREG): New unspecs.
11770 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11771 (aarch64_ld2<mode>_dreg_be): Replace with...
11772 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11774 (aarch64_ld3<mode>_dreg_le)
11775 (aarch64_ld3<mode>_dreg_be): Replace with...
11776 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11778 (aarch64_ld4<mode>_dreg_le)
11779 (aarch64_ld4<mode>_dreg_be): Replace with...
11780 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11783 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11785 PR tree-optimization/81987
11786 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11787 insert an initializer in a location not dominated by the stride
11790 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11792 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11793 on the entire header of the finally block in the fallthru case.
11795 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11797 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11799 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11801 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11802 rs6000_emit_move_from_cr and call renamed function.
11803 (rs6000_emit_prologue): Call renamed functions.
11804 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11805 movesi_from_cr, remove volatile CRs.
11807 2017-08-30 Jon Beniston <jon@beniston.com>
11808 Richard Biener <rguenther@suse.de>
11810 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11811 of VECTOR_MODE_P check.
11812 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11813 element vector types.
11815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11817 * df.h (df_read_modify_subreg_p): Remove in favor of...
11818 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11819 const_rtx instead of an rtx.
11820 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11821 * df-problems.c (df_word_lr_mark_ref): Likewise.
11822 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11823 (mark_pseudo_reg_dead): Likewise.
11824 (mark_ref_dead): Likewise.
11825 * reginfo.c (init_subregs_of_mode): Likewise.
11826 * sched-deps.c (sched_analyze_1): Likewise.
11827 * df-scan.c (df_def_record_1): Likewise.
11828 (df_uses_record): Likewise.
11829 (df_read_modify_subreg_p): Remove in favor of...
11830 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11831 const_rtx instead of an rtx.
11833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11834 Alan Hayward <alan.hayward@arm.com>
11835 David Sherwood <david.sherwood@arm.com>
11837 * rtl.h (partial_subreg_p): New function.
11838 * caller-save.c (save_call_clobbered_regs): Use it.
11839 * calls.c (expand_call): Likewise.
11840 * combine.c (combinable_i3pat): Likewise.
11841 (simplify_set): Likewise.
11842 (make_extraction): Likewise.
11843 (make_compound_operation_int): Likewise.
11844 (gen_lowpart_or_truncate): Likewise.
11845 (force_to_mode): Likewise.
11846 (make_field_assignment): Likewise.
11847 (reg_truncated_to_mode): Likewise.
11848 (record_truncated_value): Likewise.
11849 (move_deaths): Likewise.
11850 * cse.c (record_jump_cond): Likewise.
11851 (cse_insn): Likewise.
11852 * cselib.c (cselib_lookup_1): Likewise.
11853 * expmed.c (extract_bit_field_using_extv): Likewise.
11854 * function.c (assign_parm_setup_reg): Likewise.
11855 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11856 * ira-build.c (create_insn_allocnos): Likewise.
11857 * lra-coalesce.c (merge_pseudos): Likewise.
11858 * lra-constraints.c (match_reload): Likewise.
11859 (simplify_operand_subreg): Likewise.
11860 (curr_insn_transform): Likewise.
11861 * lra-lives.c (process_bb_lives): Likewise.
11862 * lra.c (new_insn_reg): Likewise.
11863 (lra_substitute_pseudo): Likewise.
11864 * regcprop.c (mode_change_ok): Likewise.
11865 (maybe_mode_change): Likewise.
11866 (copyprop_hardreg_forward_1): Likewise.
11867 * reload.c (push_reload): Likewise.
11868 (find_reloads): Likewise.
11869 (find_reloads_subreg_address): Likewise.
11870 * reload1.c (alter_reg): Likewise.
11871 (eliminate_regs_1): Likewise.
11872 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11874 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11876 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11877 back to if statements, including unpack.
11879 2017-08-30 Martin Liska <mliska@suse.cz>
11881 PR inline-asm/82001
11882 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11884 (func_checker::compare_asm_inputs_outputs): ... this function.
11885 (func_checker::compare_gimple_asm): Use the function to compare
11886 also ASM constrains.
11887 * ipa-icf-gimple.h: Rename the function.
11889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11890 Alan Hayward <alan.hayward@arm.com>
11891 David Sherwood <david.sherwood@arm.com>
11893 * coretypes.h (complex_mode): New type.
11894 * gdbhooks.py (build_pretty_printer): Handle it.
11895 * machmode.h (complex_mode): New class.
11896 (complex_mode::includes_p): New function.
11897 (is_complex_int_mode): Likewise.
11898 (is_complex_float_mode): Likewise.
11899 * genmodes.c (get_mode_class): Handle complex mode classes.
11900 * function.c (expand_function_end): Use is_complex_int_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 * coretypes.h (scalar_mode_pod): New typedef.
11907 * gdbhooks.py (build_pretty_printer): Handle it.
11908 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11909 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11910 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11911 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11912 as_a <scalar_mode>.
11914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11915 Alan Hayward <alan.hayward@arm.com>
11916 David Sherwood <david.sherwood@arm.com>
11918 * machmode.h (mode_for_vector): Take a scalar_mode instead
11920 * stor-layout.c (mode_for_vector): Likewise.
11921 * explow.c (promote_mode): Use as_a <scalar_mode>.
11922 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11925 Alan Hayward <alan.hayward@arm.com>
11926 David Sherwood <david.sherwood@arm.com>
11928 * target.def (preferred_simd_mode): Take a scalar_mode
11929 instead of a machine_mode.
11930 * targhooks.h (default_preferred_simd_mode): Likewise.
11931 * targhooks.c (default_preferred_simd_mode): Likewise.
11932 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11933 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11934 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11935 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11936 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11937 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11938 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11939 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11941 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11942 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11943 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11944 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11945 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11946 * doc/tm.texi: Regenerate.
11947 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11951 Alan Hayward <alan.hayward@arm.com>
11952 David Sherwood <david.sherwood@arm.com>
11954 * target.def (scalar_mode_supported_p): Take a scalar_mode
11955 instead of a machine_mode.
11956 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11957 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11958 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11959 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11960 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11961 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11962 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11963 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11964 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11965 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11966 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11967 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11968 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11969 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11971 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11972 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11973 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11974 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11975 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11977 * doc/tm.texi: Regenerate.
11979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11980 Alan Hayward <alan.hayward@arm.com>
11981 David Sherwood <david.sherwood@arm.com>
11983 * coretypes.h (opt_scalar_mode): New typedef.
11984 * gdbhooks.py (build_pretty_printers): Handle it.
11985 * machmode.h (mode_iterator::get_2xwider): Add overload for
11987 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11989 * expr.c (convert_mode_scalar): Likewise.
11990 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11991 * optabs.c (expand_float): Likewise.
11992 (expand_fix): Likewise.
11993 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11996 Alan Hayward <alan.hayward@arm.com>
11997 David Sherwood <david.sherwood@arm.com>
11999 * optabs.c (expand_float): Explicitly check for scalars before
12000 using a branching expansion.
12001 (expand_fix): Likewise.
12003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12004 Alan Hayward <alan.hayward@arm.com>
12005 David Sherwood <david.sherwood@arm.com>
12007 * expr.c (convert_mode): Split scalar handling out into...
12008 (convert_mode_scalar): ...this new function. Treat the modes
12011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12012 Alan Hayward <alan.hayward@arm.com>
12013 David Sherwood <david.sherwood@arm.com>
12015 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
12017 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
12019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12020 Alan Hayward <alan.hayward@arm.com>
12021 David Sherwood <david.sherwood@arm.com>
12023 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
12024 rather than a machine_mode.
12025 (fixed_from_string): Likewise.
12026 (fixed_convert): Likewise.
12027 (fixed_convert_from_int): Likewise.
12028 (fixed_convert_from_real): Likewise.
12029 (real_convert_from_fixed): Likewise.
12030 * fixed-value.c (fixed_from_double_int): Likewise.
12031 (fixed_from_string): Likewise.
12032 (fixed_convert): Likewise.
12033 (fixed_convert_from_int): Likewise.
12034 (fixed_convert_from_real): Likewise.
12035 (real_convert_from_fixed): Likewise.
12036 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
12038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12039 Alan Hayward <alan.hayward@arm.com>
12040 David Sherwood <david.sherwood@arm.com>
12042 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
12043 of separate mode class checks. Do not allow vector modes here.
12044 (immed_wide_int_const): Use as_a <scalar_mode>.
12045 * explow.c (trunc_int_for_mode): Likewise.
12046 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
12047 (wi::shwi): Likewise.
12048 (wi::min_value): Likewise.
12049 (wi::max_value): Likewise.
12050 * dwarf2out.c (loc_descriptor): Likewise.
12051 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
12052 for CONST_WIDE_INT.
12054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12055 Alan Hayward <alan.hayward@arm.com>
12056 David Sherwood <david.sherwood@arm.com>
12058 * tree.h (SCALAR_TYPE_MODE): New macro.
12059 * expr.c (expand_expr_addr_expr_1): Use it.
12060 (expand_expr_real_2): Likewise.
12061 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
12062 (fold_convert_const_fixed_from_int): Likewise.
12063 (fold_convert_const_fixed_from_real): Likewise.
12064 (native_encode_fixed): Likewise
12065 (native_encode_complex): Likewise
12066 (native_encode_vector): Likewise.
12067 (native_interpret_fixed): Likewise.
12068 (native_interpret_real): Likewise.
12069 (native_interpret_complex): Likewise.
12070 (native_interpret_vector): Likewise.
12071 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
12072 (simd_clone_adjust_argument_types): Likewise.
12073 (simd_clone_init_simd_arrays): Likewise.
12074 (simd_clone_adjust): Likewise.
12075 * stor-layout.c (layout_type): Likewise.
12076 * tree.c (build_minus_one_cst): Likewise.
12077 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12078 * tree-inline.c (estimate_move_cost): Likewise.
12079 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
12080 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
12081 (vectorizable_reduction): Likewise.
12082 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
12083 (vect_recog_mixed_size_cond_pattern): Likewise.
12084 (check_bool_pattern): Likewise.
12085 (adjust_bool_pattern): Likewise.
12086 (search_type_for_mask_1): Likewise.
12087 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12088 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12089 (vectorizable_load): Likewise.
12090 (vectorizable_store): Likewise.
12091 * ubsan.c (ubsan_encode_value): Likewise.
12092 * varasm.c (output_constant): Likewise.
12094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12095 Alan Hayward <alan.hayward@arm.com>
12096 David Sherwood <david.sherwood@arm.com>
12098 * coretypes.h (scalar_mode): New class.
12099 * machmode.h (scalar_mode): Likewise.
12100 (scalar_mode::includes_p): New function.
12101 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12102 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12103 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12104 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12105 * expmed.c (store_bit_field_1): Likewise.
12106 (extract_bit_field_1): Likewise.
12107 * expr.c (write_complex_part): Likewise.
12108 (read_complex_part): Likewise.
12109 (emit_move_complex_push): Likewise.
12110 (expand_expr_real_2): Likewise.
12111 * function.c (assign_parm_setup_reg): Likewise.
12112 (assign_parms_unsplit_complex): Likewise.
12113 * optabs.c (expand_binop): Likewise.
12114 * rtlanal.c (subreg_get_info): Likewise.
12115 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12116 * varasm.c (output_constant_pool_2): Likewise.
12118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12119 Alan Hayward <alan.hayward@arm.com>
12120 David Sherwood <david.sherwood@arm.com>
12122 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12124 (expmed_mult_highpart_optab): Likewise.
12125 (expmed_mult_highpart): Likewise.
12127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12128 Alan Hayward <alan.hayward@arm.com>
12129 David Sherwood <david.sherwood@arm.com>
12131 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12132 instead of a machine_mode.
12133 (builtin_memset_read_str): Likewise.
12134 * builtins.c (c_readstr): Likewise.
12135 (builtin_memcpy_read_str): Likewise.
12136 (builtin_strncpy_read_str): Likewise.
12137 (builtin_memset_read_str): Likewise.
12138 (builtin_memset_gen_str): Likewise.
12139 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12140 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12141 instead of a machine_mode.
12142 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12144 (make_extraction): Likewise.
12145 (try_widen_shift_mode): Take and return scalar_int_modes instead
12147 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12148 a scalar_int_mode instead of a machine_mode.
12149 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12150 (avr_addr_space_pointer_mode): Likewise.
12151 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12152 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12153 (msp430_unwind_word_mode): Likewise.
12154 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12155 (spu_addr_space_pointer_mode): Likewise.
12156 (spu_addr_space_address_mode): Likewise.
12157 (spu_libgcc_cmp_return_mode): Likewise.
12158 (spu_libgcc_shift_count_mode): Likewise.
12159 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12160 (rl78_addr_space_pointer_mode): Likewise.
12161 (fl78_unwind_word_mode): Likewise.
12162 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12164 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12165 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12166 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12167 (mips_valid_pointer_mode): Likewise.
12168 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12169 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12170 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12172 (ft32_addr_space_address_mode): Likewise.
12173 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12174 scalar_int_mode instead of a machine_mode.
12175 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12177 (m32c_addr_space_address_mode): Likewise.
12178 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12179 (rs6000_eh_return_filter_mode): Likewise.
12180 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12181 (rs6000_eh_return_filter_mode): Likewise.
12182 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12183 (s390_libgcc_shift_count_mode): Likewise.
12184 (s390_unwind_word_mode): Likewise.
12185 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12187 * target.def (mode_rep_extended): Likewise.
12188 (valid_pointer_mode): Likewise.
12189 (addr_space.valid_pointer_mode): Likewise.
12190 (eh_return_filter_mode): Return a scalar_int_mode rather than
12192 (libgcc_cmp_return_mode): Likewise.
12193 (libgcc_shift_count_mode): Likewise.
12194 (unwind_word_mode): Likewise.
12195 (addr_space.pointer_mode): Likewise.
12196 (addr_space.address_mode): Likewise.
12197 * doc/tm.texi: Regenerate.
12198 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12200 (do_jump): Use scalar_int_mode for local variables.
12201 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12202 rather than a machine_mode.
12203 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12204 (scompare_loc_descriptor_wide): Likewise.
12205 (scompare_loc_descriptor_narrow): Likewise.
12206 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12208 * except.c (sjlj_emit_dispatch_table): Likewise.
12209 (expand_builtin_eh_copy_values): Likewise.
12210 * explow.c (convert_memory_address_addr_space_1): Likewise.
12211 Take a scalar_int_mode rather than a machine_mode.
12212 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12213 than a machine_mode.
12214 (memory_address_addr_space): Use scalar_int_mode for local variables.
12215 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12216 rather than a machine_mode.
12217 * expmed.c (mask_rtx): Likewise.
12218 (init_expmed_one_conv): Likewise.
12219 (expand_mult_highpart_adjust): Likewise.
12220 (extract_high_half): Likewise.
12221 (expmed_mult_highpart_optab): Likewise.
12222 (expmed_mult_highpart): Likewise.
12223 (expand_smod_pow2): Likewise.
12224 (expand_sdiv_pow2): Likewise.
12225 (emit_store_flag_int): Likewise.
12226 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12228 (extract_low_bits): Likewise.
12229 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12231 * expr.c (pieces_addr::adjust): Likewise.
12232 (can_store_by_pieces): Likewise.
12233 (store_by_pieces): Likewise.
12234 (clear_by_pieces_1): Likewise.
12235 (expand_expr_addr_expr_1): Likewise.
12236 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12237 (expand_expr_real_1): Likewise.
12238 (try_casesi): Likewise.
12239 * final.c (shorten_branches): Likewise.
12240 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12241 type of "mode" to machine_mode.
12242 * internal-fn.c (expand_arith_overflow_result_store): Take a
12243 scalar_int_mode rather than a machine_mode.
12244 (expand_mul_overflow): Use scalar_int_mode for local variables.
12245 * loop-doloop.c (doloop_modify): Likewise.
12246 (doloop_optimize): Likewise.
12247 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12248 than a machine_mode.
12249 (expand_doubleword_shift_condmove): Likewise.
12250 (expand_doubleword_shift): Likewise.
12251 (expand_doubleword_clz): Likewise.
12252 (expand_doubleword_popcount): Likewise.
12253 (expand_doubleword_parity): Likewise.
12254 (expand_absneg_bit): Use scalar_int_mode for local variables.
12255 (prepare_float_lib_cmp): Likewise.
12256 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12257 rather than a machine_mode.
12258 (convert_memory_address_addr_space): Likewise.
12259 (get_mode_bounds): Likewise.
12260 (get_address_mode): Return a scalar_int_mode rather than a
12262 * rtlanal.c (get_address_mode): Likewise.
12263 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12264 than a machine_mode.
12265 * targhooks.c (default_mode_rep_extended): Likewise.
12266 (default_valid_pointer_mode): Likewise.
12267 (default_addr_space_valid_pointer_mode): Likewise.
12268 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12269 than a machine_mode.
12270 (default_libgcc_cmp_return_mode): Likewise.
12271 (default_libgcc_shift_count_mode): Likewise.
12272 (default_unwind_word_mode): Likewise.
12273 (default_addr_space_pointer_mode): Likewise.
12274 (default_addr_space_address_mode): Likewise.
12275 * targhooks.h (default_eh_return_filter_mode): Likewise.
12276 (default_libgcc_cmp_return_mode): Likewise.
12277 (default_libgcc_shift_count_mode): Likewise.
12278 (default_unwind_word_mode): Likewise.
12279 (default_addr_space_pointer_mode): Likewise.
12280 (default_addr_space_address_mode): Likewise.
12281 (default_mode_rep_extended): Take a scalar_int_mode rather than
12283 (default_valid_pointer_mode): Likewise.
12284 (default_addr_space_valid_pointer_mode): Likewise.
12285 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12287 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12288 rather than a machine_mode.
12289 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12290 for local variables.
12291 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12292 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12293 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12300 the mode argument to scalar_int_mode.
12301 (do_jump_by_parts_zero_rtx): Likewise.
12302 (do_jump_by_parts_equality_rtx): Likewise.
12303 (do_jump_by_parts_greater): Take a mode argument.
12304 (do_jump_by_parts_equality): Likewise.
12305 (do_jump_1): Update calls accordingly.
12307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12308 Alan Hayward <alan.hayward@arm.com>
12309 David Sherwood <david.sherwood@arm.com>
12311 * is-a.h (safe_dyn_cast): New function.
12312 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12313 (jump_table_for_label): Likewise.
12314 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12315 instead of an rtx_insn *.
12316 (shorten_branches): Use dyn_cast instead of LABEL_P and
12317 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12318 rtx_jump_table_data::get_data_mode.
12319 (final_scan_insn): Likewise.
12321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12322 Alan Hayward <alan.hayward@arm.com>
12323 David Sherwood <david.sherwood@arm.com>
12325 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12326 trying to combine a full-register integer set with a subreg
12329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12330 Alan Hayward <alan.hayward@arm.com>
12331 David Sherwood <david.sherwood@arm.com>
12333 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12334 that is always either address_mode or pointer_mode.
12336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12337 Alan Hayward <alan.hayward@arm.com>
12338 David Sherwood <david.sherwood@arm.com>
12340 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12341 when the two are known to be equal.
12343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12344 Alan Hayward <alan.hayward@arm.com>
12345 David Sherwood <david.sherwood@arm.com>
12347 * simplify-rtx.c (simplify_const_unary_operation): Use
12348 is_a <scalar_int_mode> instead of checking for a nonzero
12349 precision. Forcibly convert op_mode to a scalar_int_mode
12350 in that case. More clearly differentiate the operand and
12351 result modes and use the former when deciding what the value
12352 of a count-bits operation should be. Use is_int_mode instead
12353 of checking for a MODE_INT. Remove redundant check for whether
12354 this mode has a zero precision.
12356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12357 Alan Hayward <alan.hayward@arm.com>
12358 David Sherwood <david.sherwood@arm.com>
12360 * optabs.c (widen_leading): Change the type of the mode argument
12361 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12362 (widen_bswap): Likewise.
12363 (expand_parity): Likewise.
12364 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12365 (expand_ffs): Likewise.
12366 (epand_unop): Check for scalar integer modes before calling the
12369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12370 Alan Hayward <alan.hayward@arm.com>
12371 David Sherwood <david.sherwood@arm.com>
12373 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12375 (expand_expr_real_1): Update call accordingly.
12377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12378 Alan Hayward <alan.hayward@arm.com>
12379 David Sherwood <david.sherwood@arm.com>
12381 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12382 value_mode arguments. Use scalar_int_mode internally.
12383 (store_bit_field_1): Rename the new integer mode from imode
12384 to op0_mode and use it instead of GET_MODE (op0). Update calls
12385 to store_split_bit_field, store_bit_field_using_insv and
12386 store_fixed_bit_field.
12387 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12388 Use scalar_int_mode internally. Use a bit count rather than a mode
12389 when calculating the largest bit size for get_best_mode.
12390 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12391 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12392 Remove assertion that OP0 has a scalar integer mode.
12393 (store_split_bit_field): Add op0_mode and value_mode arguments.
12394 Update calls to extract_fixed_bit_field.
12395 (extract_bit_field_using_extv): Add an op0_mode argument.
12396 Use scalar_int_mode internally.
12397 (extract_bit_field_1): Rename the new integer mode from imode to
12398 op0_mode and use it instead of GET_MODE (op0). Update calls to
12399 extract_split_bit_field, extract_bit_field_using_extv and
12400 extract_fixed_bit_field.
12401 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12402 to extract_split_bit_field and extract_fixed_bit_field_1.
12403 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12404 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12405 on the target mode.
12406 (extract_split_bit_field): Add an op0_mode argument. Update call
12407 to extract_fixed_bit_field.
12409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12410 Alan Hayward <alan.hayward@arm.com>
12411 David Sherwood <david.sherwood@arm.com>
12413 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12414 * explow.c (hard_function_value): Likewise.
12415 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12416 convert_to_mode call outside the loop.
12417 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12418 for the mode iterator. Require the mode specified by max_pieces
12420 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12422 (copy_blkmode_to_reg): Likewise.
12423 (set_storage_via_setmem): Likewise.
12424 * optabs.c (prepare_cmp_insn): Likewise.
12425 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12426 * stor-layout.c (finish_bitfield_representative): Likewise.
12428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12429 Alan Hayward <alan.hayward@arm.com>
12430 David Sherwood <david.sherwood@arm.com>
12432 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12433 * expr.c (convert_move): Use them.
12434 (convert_modes): Likewise.
12435 (store_expr_with_bounds): Likewise.
12437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12438 Alan Hayward <alan.hayward@arm.com>
12439 David Sherwood <david.sherwood@arm.com>
12441 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12442 parameter for the mode of "x". Remove the "known_x", "known_mode"
12443 and "known_ret" arguments. Change the type of the mode argument
12444 to scalar_int_mode.
12445 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12446 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12447 (reg_num_sign_bit_copies_for_combine): Likewise.
12448 * rtlanal.c (nonzero_bits1): Likewise.
12449 (num_sign_bit_copies1): Likewise.
12450 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12451 (reg_num_sign_bit_copies_general): Likewise.
12452 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12453 (reg_nonzero_bits_general): Likewise.
12455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12456 Alan Hayward <alan.hayward@arm.com>
12457 David Sherwood <david.sherwood@arm.com>
12459 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12460 than in subroutines. Return 1 for non-integer modes.
12461 (cached_num_sign_bit_copies): Change the type of the mode parameter
12462 to scalar_int_mode.
12463 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12464 classes. Handle CONST_INT_P first and then check whether X also
12465 has a scalar integer mode. Check the same thing for inner registers
12466 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12473 in subroutines. Return the mode mask for non-integer modes.
12474 (cached_nonzero_bits): Change the type of the mode parameter
12475 to scalar_int_mode.
12476 (nonzero_bits1): Likewise. Remove early exit for other mode
12477 classes. Handle CONST_INT_P first and then check whether X
12478 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12485 Return a scalar_int_mode and assert that the size is greater than
12487 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12488 (op_by_pieces_d::op_by_pieces_d): Likewise.
12489 (op_by_pieces_d::run): Likewise.
12490 (can_store_by_pieces): Likewise.
12492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12493 Alan Hayward <alan.hayward@arm.com>
12494 David Sherwood <david.sherwood@arm.com>
12496 * combine.c (extract_left_shift): Add a mode argument and update
12498 (make_compound_operation_int): Change the type of the mode parameter
12499 to scalar_int_mode and update the call to extract_left_shift.
12501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12502 Alan Hayward <alan.hayward@arm.com>
12503 David Sherwood <david.sherwood@arm.com>
12505 * combine.c (simplify_and_const_int): Change the type of the mode
12506 parameter to scalar_int_mode.
12507 (simplify_and_const_int_1): Likewise. Update recursive call.
12509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12510 Alan Hayward <alan.hayward@arm.com>
12511 David Sherwood <david.sherwood@arm.com>
12513 * combine.c (simplify_compare_const): Check that the mode is a
12514 scalar_int_mode (rather than VOIDmode) before testing its
12516 (simplify_comparison): Move COMPARISON_P handling out of the
12517 loop and restrict the latter part of the loop to scalar_int_modes.
12518 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12519 and when considering SUBREG_REGs. Use is_int_mode instead of
12520 checking GET_MODE_CLASS against MODE_INT.
12522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12523 Alan Hayward <alan.hayward@arm.com>
12524 David Sherwood <david.sherwood@arm.com>
12526 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12527 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12528 shift_unit_mode and for modes involved in scalar shifts.
12530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12531 Alan Hayward <alan.hayward@arm.com>
12532 David Sherwood <david.sherwood@arm.com>
12534 * combine.c (force_int_to_mode): New function, split out from...
12535 (force_to_mode): ...here. Keep xmode up-to-date and use it
12536 instead of GET_MODE (x).
12538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12539 Alan Hayward <alan.hayward@arm.com>
12540 David Sherwood <david.sherwood@arm.com>
12542 * optabs-query.h (extraction_insn::struct_mode): Change type to
12543 opt_scalar_int_mode and update comment.
12544 (extraction_insn::field_mode): Change type to scalar_int_mode.
12545 (extraction_insn::pos_mode): Likewise.
12546 * combine.c (make_extraction): Update accordingly.
12547 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12548 (get_optab_extraction_insn): Likewise.
12549 * recog.c (simplify_while_replacing): Likewise.
12550 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12551 parameter to opt_scalar_int_mode.
12553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12554 Alan Hayward <alan.hayward@arm.com>
12555 David Sherwood <david.sherwood@arm.com>
12557 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12558 to a scalar_int_mode instead of a machine_mode.
12559 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12560 (get_best_mode): Return a boolean and use a pointer argument to store
12561 the selected mode. Replace the limit mode parameter with a bit limit.
12562 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12563 for the values returned by bit_field_mode_iterator::next_mode.
12564 (store_bit_field): Update call to get_best_mode.
12565 (store_fixed_bit_field): Likewise.
12566 (extract_fixed_bit_field): Likewise.
12567 * expr.c (optimize_bitfield_assignment_op): Likewise.
12568 * fold-const.c (optimize_bit_field_compare): Likewise.
12569 (fold_truth_andor_1): Likewise.
12570 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12571 Update for new type of m_mode.
12572 (get_best_mode): As above.
12574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12575 Alan Hayward <alan.hayward@arm.com>
12576 David Sherwood <david.sherwood@arm.com>
12578 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12579 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12580 (store_bit_field): Check is_a <scalar_int_mode> before calling
12581 strict_volatile_bitfield_p.
12582 (extract_bit_field): Likewise.
12584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12585 Alan Hayward <alan.hayward@arm.com>
12586 David Sherwood <david.sherwood@arm.com>
12588 * target.def (cstore_mode): Return a scalar_int_mode.
12589 * doc/tm.texi: Regenerate.
12590 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12591 * targhooks.h (default_cstore_mode): Likewise.
12592 * targhooks.c (default_cstore_mode): Likewise, using a forced
12594 * expmed.c (emit_cstore): Expect the target of the cstore to be
12597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12598 Alan Hayward <alan.hayward@arm.com>
12599 David Sherwood <david.sherwood@arm.com>
12601 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12603 (niter_desc): Likewise mode.
12604 (iv_analyze): Add a mode parameter.
12606 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12607 and change its type to scalar_int_mode.
12608 * loop-iv.c: Update commentary at head of file.
12609 (iv_constant): Pass the mode paraeter before the rtx it describes
12610 and change its type to scalar_int_mode. Remove VOIDmode handling.
12611 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12612 (iv_extend): Likewise.
12613 (shorten_into_mode): Likewise.
12614 (iv_add): Use scalar_int_mode.
12615 (iv_mult): Likewise.
12616 (iv_shift): Likewise.
12617 (canonicalize_iv_subregs): Likewise.
12618 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12619 it describes and change its mode to scalar_int_mode. Also change
12620 the type of the returned inner_mode to scalar_int_mode.
12621 (get_biv_step): Likewise, turning outer_mode from a pointer
12622 into a direct parameter. Update call to get_biv_step_1.
12623 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12624 iv_constant and get_biv_step.
12625 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12626 and change its type to scalar_int_mode. Don't initialise iv->mode
12627 to VOIDmode and remove later checks for its still being VOIDmode.
12628 Update calls to iv_analyze_op and iv_analyze_expr. Check
12629 is_a <scalar_int_mode> when changing the mode under consideration.
12630 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12631 Update call to iv_analyze_expr.
12632 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12633 inner register is not also a scalar_int_mode. Update call to
12635 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12636 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12637 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12638 separate mode class checks. Update calls to iv_analyze. Remove
12639 fix-up of VOIDmodes after iv_analyze_biv.
12640 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12641 don't have a scalar_int_mode. Update call to biv_p.
12643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12644 Alan Hayward <alan.hayward@arm.com>
12645 David Sherwood <david.sherwood@arm.com>
12647 * cfgexpand.c (convert_debug_memory_address): Use
12648 as_a <scalar_int_mode>.
12649 * combine.c (expand_compound_operation): Likewise.
12650 (make_extraction): Likewise.
12651 (change_zero_ext): Likewise.
12652 (simplify_comparison): Likewise.
12653 * cse.c (cse_insn): Likewise.
12654 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12655 (mem_loc_descriptor): Likewise.
12656 (loc_descriptor): Likewise.
12657 * expmed.c (init_expmed_one_mode): Likewise.
12658 (synth_mult): Likewise.
12659 (emit_store_flag_1): Likewise.
12660 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12661 of a comparison with size.
12662 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12663 (reduce_to_bit_field_precision): Likewise.
12664 * function.c (expand_function_end): Likewise.
12665 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12666 * loop-doloop.c (doloop_modify): Likewise.
12667 * optabs.c (expand_binop): Likewise.
12668 (expand_unop): Likewise.
12669 (expand_copysign_absneg): Likewise.
12670 (prepare_cmp_insn): Likewise.
12671 (maybe_legitimize_operand): Likewise.
12672 * recog.c (const_scalar_int_operand): Likewise.
12673 * rtlanal.c (get_address_mode): Likewise.
12674 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12675 (simplify_cond_clz_ctz): Likewise.
12676 * tree-nested.c (get_nl_goto_field): Likewise.
12677 * tree.c (build_vector_type_for_mode): Likewise.
12678 * var-tracking.c (use_narrower_mode): Likewise.
12680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12681 Alan Hayward <alan.hayward@arm.com>
12682 David Sherwood <david.sherwood@arm.com>
12684 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12685 * builtins.c (expand_builtin_signbit): Use it.
12686 * cfgexpand.c (expand_debug_expr): Likewise.
12687 * dojump.c (do_jump): Likewise.
12688 (do_compare_and_jump): Likewise.
12689 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12690 * expmed.c (make_tree): Likewise.
12691 * expr.c (expand_expr_real_2): Likewise.
12692 (expand_expr_real_1): Likewise.
12693 (try_casesi): Likewise.
12694 * fold-const-call.c (fold_const_call_ss): Likewise.
12695 * fold-const.c (unextend): Likewise.
12696 (extract_muldiv_1): Likewise.
12697 (fold_single_bit_test): Likewise.
12698 (native_encode_int): Likewise.
12699 (native_encode_string): Likewise.
12700 (native_interpret_int): Likewise.
12701 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12702 * internal-fn.c (expand_addsub_overflow): Likewise.
12703 (expand_neg_overflow): Likewise.
12704 (expand_mul_overflow): Likewise.
12705 (expand_arith_overflow): Likewise.
12706 * match.pd: Likewise.
12707 * stor-layout.c (layout_type): Likewise.
12708 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12709 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12710 * tree-ssanames.c (get_range_info): Likewise.
12711 * tree-switch-conversion.c (array_value_type) Likewise.
12712 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12713 (vect_recog_divmod_pattern): Likewise.
12714 (vect_recog_mixed_size_cond_pattern): Likewise.
12715 * tree-vrp.c (extract_range_basic): Likewise.
12716 (simplify_float_conversion_using_ranges): Likewise.
12717 * tree.c (int_fits_type_p): Likewise.
12718 * ubsan.c (instrument_bool_enum_load): Likewise.
12719 * varasm.c (mergeable_string_section): Likewise.
12720 (narrowing_initializer_constant_valid_p): Likewise.
12721 (output_constant): 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 * machmode.h (NARROWEST_INT_MODE): New macro.
12728 * expr.c (alignment_for_piecewise_move): Use it instead of
12729 GET_CLASS_NARROWEST_MODE (MODE_INT).
12730 (push_block): Likewise.
12731 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12733 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12736 Alan Hayward <alan.hayward@arm.com>
12737 David Sherwood <david.sherwood@arm.com>
12739 * postreload.c (move2add_valid_value_p): Change the type of the
12740 mode parameter to scalar_int_mode.
12741 (move2add_use_add2_insn): Add a mode parameter and use it instead
12743 (move2add_use_add3_insn): Likewise.
12744 (reload_cse_move2add): Update accordingly.
12746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12747 Alan Hayward <alan.hayward@arm.com>
12748 David Sherwood <david.sherwood@arm.com>
12750 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12752 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12753 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12760 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12761 (popcount_loc_descriptor): Likewise.
12762 (bswap_loc_descriptor): Likewise.
12763 (rotate_loc_descriptor): Likewise.
12764 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12765 calling the functions above.
12767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12768 Alan Hayward <alan.hayward@arm.com>
12769 David Sherwood <david.sherwood@arm.com>
12771 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12773 (try_combine): Likewise.
12774 (simplify_if_then_else): Likewise.
12775 * cse.c (cse_insn): Likewise.
12776 * dwarf2out.c (mem_loc_descriptor): Likewise.
12777 * emit-rtl.c (gen_lowpart_common): Likewise.
12778 * simplify-rtx.c (simplify_truncation): Likewise.
12779 (simplify_binary_operation_1): Likewise.
12780 (simplify_const_relational_operation): Likewise.
12781 (simplify_ternary_operation): Likewise.
12782 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12785 Alan Hayward <alan.hayward@arm.com>
12786 David Sherwood <david.sherwood@arm.com>
12788 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12789 * reload.c (push_reload): Likewise.
12790 (find_reloads): Likewise.
12792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12793 Alan Hayward <alan.hayward@arm.com>
12794 David Sherwood <david.sherwood@arm.com>
12796 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12797 (make_compound_operation_int): Likewise.
12798 (change_zero_ext): Likewise.
12799 * expr.c (convert_move): Likewise.
12800 (convert_modes): Likewise.
12801 * fwprop.c (forward_propagate_subreg): Likewise.
12802 * loop-iv.c (get_biv_step_1): Likewise.
12803 * optabs.c (widen_operand): Likewise.
12804 * postreload.c (move2add_valid_value_p): Likewise.
12805 * recog.c (simplify_while_replacing): Likewise.
12806 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12807 (simplify_binary_operation_1): Likewise. Remove redundant
12808 mode equality check.
12810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12811 Alan Hayward <alan.hayward@arm.com>
12812 David Sherwood <david.sherwood@arm.com>
12814 * combine.c (combine_simplify_rtx): Add checks for
12815 is_a <scalar_int_mode>.
12816 (simplify_if_then_else): Likewise.
12817 (make_field_assignment): Likewise.
12818 (simplify_comparison): Likewise.
12819 * ifcvt.c (noce_try_bitop): Likewise.
12820 * loop-invariant.c (canonicalize_address_mult): Likewise.
12821 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12824 Alan Hayward <alan.hayward@arm.com>
12825 David Sherwood <david.sherwood@arm.com>
12827 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12828 is_a <scalar_int_mode> instead of != BLKmode.
12830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12831 Alan Hayward <alan.hayward@arm.com>
12832 David Sherwood <david.sherwood@arm.com>
12834 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12835 instead of != VOIDmode.
12836 * combine.c (if_then_else_cond): Likewise.
12837 (change_zero_ext): Likewise.
12838 * dwarf2out.c (mem_loc_descriptor): Likewise.
12839 (loc_descriptor): Likewise.
12840 * rtlanal.c (canonicalize_condition): Likewise.
12841 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12844 Alan Hayward <alan.hayward@arm.com>
12845 David Sherwood <david.sherwood@arm.com>
12847 * simplify-rtx.c (simplify_binary_operation_1): Use
12848 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12851 Alan Hayward <alan.hayward@arm.com>
12852 David Sherwood <david.sherwood@arm.com>
12854 * wide-int.h (int_traits<unsigned char>) New class.
12855 (int_traits<unsigned short>) Likewise.
12856 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12857 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12859 * combine.c (set_nonzero_bits_and_sign_copies): Use
12860 is_a <scalar_int_mode>.
12861 (find_split_point): Likewise.
12862 (combine_simplify_rtx): Likewise.
12863 (simplify_logical): Likewise.
12864 (expand_compound_operation): Likewise.
12865 (expand_field_assignment): Likewise.
12866 (make_compound_operation): Likewise.
12867 (extended_count): Likewise.
12868 (change_zero_ext): Likewise.
12869 (simplify_comparison): Likewise.
12870 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12871 (ucompare_loc_descriptor): Likewise.
12872 (minmax_loc_descriptor): Likewise.
12873 (mem_loc_descriptor): Likewise.
12874 (loc_descriptor): Likewise.
12875 * expmed.c (init_expmed_one_mode): Likewise.
12876 * lra-constraints.c (lra_constraint_offset): Likewise.
12877 * optabs.c (prepare_libcall_arg): Likewise.
12878 * postreload.c (move2add_note_store): Likewise.
12879 * reload.c (operands_match_p): Likewise.
12880 * rtl.h (load_extend_op): Likewise.
12881 * rtlhooks.c (gen_lowpart_general): Likewise.
12882 * simplify-rtx.c (simplify_truncation): Likewise.
12883 (simplify_unary_operation_1): Likewise.
12884 (simplify_binary_operation_1): Likewise.
12885 (simplify_const_binary_operation): Likewise.
12886 (simplify_const_relational_operation): Likewise.
12887 (simplify_subreg): Likewise.
12888 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12889 * var-tracking.c (adjust_mems): Likewise.
12890 (prepare_call_arguments): Likewise.
12892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12893 Alan Hayward <alan.hayward@arm.com>
12894 David Sherwood <david.sherwood@arm.com>
12896 * machmode.h (is_int_mode): New fuction.
12897 * combine.c (find_split_point): Use it.
12898 (combine_simplify_rtx): Likewise.
12899 (simplify_if_then_else): Likewise.
12900 (simplify_set): Likewise.
12901 (simplify_shift_const_1): Likewise.
12902 (simplify_comparison): Likewise.
12903 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12904 * cse.c (notreg_cost): Likewise.
12905 (cse_insn): Likewise.
12906 * cselib.c (cselib_lookup_1): Likewise.
12907 * dojump.c (do_jump_1): Likewise.
12908 (do_compare_rtx_and_jump): Likewise.
12909 * dse.c (get_call_args): Likewise.
12910 * dwarf2out.c (rtl_for_decl_init): Likewise.
12911 (native_encode_initializer): Likewise.
12912 * expmed.c (emit_store_flag_1): Likewise.
12913 (emit_store_flag): Likewise.
12914 * expr.c (convert_modes): Likewise.
12915 (store_field): Likewise.
12916 (expand_expr_real_1): Likewise.
12917 * fold-const.c (fold_read_from_constant_string): Likewise.
12918 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12919 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12920 * optabs.c (expand_binop): Likewise.
12921 (expand_unop): Likewise.
12922 (expand_abs_nojump): Likewise.
12923 (expand_one_cmpl_abs_nojump): Likewise.
12924 * simplify-rtx.c (mode_signbit_p): Likewise.
12925 (val_signbit_p): Likewise.
12926 (val_signbit_known_set_p): Likewise.
12927 (val_signbit_known_clear_p): Likewise.
12928 (simplify_relational_operation_1): Likewise.
12929 * tree.c (vector_type_mode): Likewise.
12931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12932 Alan Hayward <alan.hayward@arm.com>
12933 David Sherwood <david.sherwood@arm.com>
12935 * machmode.h (smallest_mode_for_size): Fix formatting.
12936 (smallest_int_mode_for_size): New function.
12937 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12938 instead of smallest_mode_for_size.
12939 * combine.c (make_extraction): Likewise.
12940 * config/arc/arc.c (arc_expand_movmem): Likewise.
12941 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12942 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12943 * config/s390/s390.c (s390_expand_insv): Likewise.
12944 * config/sparc/sparc.c (assign_int_registers): Likewise.
12945 * config/spu/spu.c (spu_function_value): Likewise.
12946 (spu_function_arg): Likewise.
12947 * coverage.c (get_gcov_type): Likewise.
12948 (get_gcov_unsigned_t): Likewise.
12949 * dse.c (find_shift_sequence): Likewise.
12950 * expmed.c (store_bit_field_1): Likewise.
12951 * expr.c (convert_move): Likewise.
12952 (store_field): Likewise.
12953 * internal-fn.c (expand_arith_overflow): Likewise.
12954 * optabs-query.c (get_best_extraction_insn): Likewise.
12955 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12956 * stor-layout.c (layout_type): Likewise.
12957 (initialize_sizetypes): Likewise.
12958 * targhooks.c (default_get_mask_mode): Likewise.
12959 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12962 Alan Hayward <alan.hayward@arm.com>
12963 David Sherwood <david.sherwood@arm.com>
12965 * machmode.h (opt_mode::else_blk): New function.
12966 (int_mode_for_mode): Declare.
12967 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12968 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12970 * cfgexpand.c (expand_debug_expr): Likewise.
12971 * combine.c (gen_lowpart_or_truncate): Likewise.
12972 (gen_lowpart_for_combine): Likewise.
12973 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12974 * config/avr/avr.c (avr_to_int_mode): Likewise.
12975 (avr_out_plus_1): Likewise.
12976 (avr_out_plus): Likewise.
12977 (avr_out_round): Likewise.
12978 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12979 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12980 (s390_expand_vcond): Likewise.
12981 * config/spu/spu.c (spu_split_immediate): Likewise.
12982 (spu_expand_mov): Likewise.
12983 * dse.c (get_stored_val): Likewise.
12984 * expmed.c (store_bit_field_1): Likewise.
12985 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12987 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12988 (extract_low_bits): Likewise.
12989 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12990 handling rather than repeating the check.
12991 (emit_group_store): Likewise.
12992 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12993 * optabs.c (expand_absneg_bit): Likewise.
12994 (expand_copysign_absneg): Likewise.
12995 (expand_copysign_bit): Likewise.
12996 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12997 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12998 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12999 * var-tracking.c (prepare_call_arguments): Likewise.
13000 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
13001 int_mode_for_mode instead of mode_for_size.
13002 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13005 Alan Hayward <alan.hayward@arm.com>
13006 David Sherwood <david.sherwood@arm.com>
13008 * machmode.h (int_mode_for_size): New function.
13009 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
13010 instead of mode_for_size.
13011 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
13013 * combine.c (expand_field_assignment): Use int_mode_for_size
13014 instead of mode_for_size.
13015 (make_extraction): Likewise.
13016 (simplify_shift_const_1): Likewise.
13017 (simplify_comparison): Likewise.
13018 * dojump.c (do_jump): Likewise.
13019 * dwarf2out.c (mem_loc_descriptor): Likewise.
13020 * emit-rtl.c (init_derived_machine_modes): Likewise.
13021 * expmed.c (flip_storage_order): Likewise.
13022 (convert_extracted_bit_field): Likewise.
13023 * expr.c (copy_blkmode_from_reg): Likewise.
13024 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
13025 * internal-fn.c (expand_mul_overflow): Likewise.
13026 * lower-subreg.c (simple_move): Likewise.
13027 * optabs-libfuncs.c (init_optabs): Likewise.
13028 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13029 * tree.c (vector_type_mode): Likewise.
13030 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
13031 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
13032 * tree-vect-generic.c (expand_vector_parallel): Likewise.
13033 * tree-vect-stmts.c (vectorizable_load): Likewise.
13034 (vectorizable_store): Likewise.
13036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13037 Alan Hayward <alan.hayward@arm.com>
13038 David Sherwood <david.sherwood@arm.com>
13040 * coretypes.h (pod_mode): New type.
13041 (scalar_int_mode_pod): New typedef.
13042 * machmode.h (pod_mode): New class.
13043 (int_n_data_t::m): Change type to scalar_int_mode_pod.
13044 * genmodes.c (emit_mode_int_n): Update accordingly.
13045 * lower-subreg.h (target_lower_subreg): Change type to
13046 scalar_int_mode_pod.
13047 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
13048 scalar_int_mode_pod.
13050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13051 Alan Hayward <alan.hayward@arm.com>
13052 David Sherwood <david.sherwood@arm.com>
13054 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
13055 machine_mode to scalar_int_mode.
13056 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
13057 (rs6000_option_override_internal): Remove cast to int.
13058 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
13059 machine_mode to scalar_int_mode.
13060 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
13061 (rs6000_option_override_internal): Remove cast to int.
13062 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
13063 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
13065 * config/s390/s390.c (s390_expand_builtin): Likewise.
13066 * coretypes.h (scalar_int_mode): New type.
13067 (opt_scalar_int_mode): New typedef.
13068 * machmode.h (scalar_int_mode): New class.
13069 (scalar_int_mode::includes_p): New function.
13070 (byte_mode): Change type to scalar_int_mode.
13071 (word_mode): Likewise.
13072 (ptr_mode): Likewise.
13073 * emit-rtl.c (byte_mode): Likewise.
13074 (word_mode): Likewise.
13075 (ptr_mode): Likewise.
13076 (init_derived_machine_modes): Update accordingly.
13077 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
13078 and MODE_PARTIAL_INT.
13079 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
13080 opt_scalar_int_mode.
13082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13083 Alan Hayward <alan.hayward@arm.com>
13084 David Sherwood <david.sherwood@arm.com>
13086 * target.def (libgcc_floating_mode_supported_p): Take a
13088 * doc/tm.texi: Regenerate.
13089 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
13091 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13092 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
13095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13096 Alan Hayward <alan.hayward@arm.com>
13097 David Sherwood <david.sherwood@arm.com>
13099 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
13100 * doc/tm.texi: Regenerate.
13101 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13102 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13103 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13104 * targhooks.h (default_floatn_mode): Likewise.
13105 * targhooks.c (default_floatn_mode): Likewise.
13106 * tree.c (build_common_tree_nodes): Update accordingly.
13108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13109 Alan Hayward <alan.hayward@arm.com>
13110 David Sherwood <david.sherwood@arm.com>
13112 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13113 (mode_iterator::iterate_p): Likewise.
13114 (mode_iterator::get_wider): Likewise.
13115 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13118 Alan Hayward <alan.hayward@arm.com>
13119 David Sherwood <david.sherwood@arm.com>
13121 * coretypes.h (opt_scalar_float_mode): New typedef.
13122 * machmode.h (float_mode_for_size): New function.
13123 * emit-rtl.c (double_mode): Delete.
13124 (init_emit_once): Use float_mode_for_size.
13125 * stor-layout.c (layout_type): Likewise.
13126 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13129 Alan Hayward <alan.hayward@arm.com>
13130 David Sherwood <david.sherwood@arm.com>
13132 * output.h (assemble_real): Take a scalar_float_mode.
13133 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13134 * config/arm/arm.md (consttable_4): Likewise.
13135 (consttable_8): Likewise.
13136 (consttable_16): Likewise.
13137 * config/mips/mips.md (consttable_float): Likewise.
13138 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13139 * varasm.c (assemble_real): Take a scalar_float_mode.
13140 (output_constant_pool_2): Update accordingly.
13141 (output_constant): 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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13148 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13149 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13150 (native_encode_real): Likewise.
13151 (native_interpret_real): Likewise.
13152 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13153 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13156 Alan Hayward <alan.hayward@arm.com>
13157 David Sherwood <david.sherwood@arm.com>
13159 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13160 <scalar_float_mode>. Simplify.
13161 (gen_extend_conv_libfunc): Likewise.
13163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13164 Alan Hayward <alan.hayward@arm.com>
13165 David Sherwood <david.sherwood@arm.com>
13167 * coretypes.h (scalar_float_mode): New type.
13168 * machmode.h (mode_traits::from_int): Use machine_mode if
13169 USE_ENUM_MODES is defined.
13170 (is_a): New function.
13172 (dyn_cast): Likewise.
13173 (scalar_float_mode): New class.
13174 (scalar_float_mode::includes_p): New function.
13175 (is_float_mode): Likewise.
13176 * gdbhooks.py (MachineModePrinter): New class.
13177 (build_pretty_printer): Use it for scalar_float_mode.
13178 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13179 (format_helper::format_helper): Turn into a template.
13180 * genmodes.c (get_mode_class): New function.
13181 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13182 or machine_mode if none.
13183 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13184 as_a <scalar_float_mode>.
13185 * dwarf2out.c (mem_loc_descriptor): Likewise.
13186 (insert_float): Likewise.
13187 (add_const_value_attribute): Likewise.
13188 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13189 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13190 (expand_unop): Update accordingly.
13191 (expand_abs_nojump): Likewise.
13192 (expand_copysign_absneg): Take a scalar_float_mode.
13193 (expand_copysign_bit): Likewise.
13194 (expand_copysign): Update accordingly.
13196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13197 Alan Hayward <alan.hayward@arm.com>
13198 David Sherwood <david.sherwood@arm.com>
13200 * coretypes.h (opt_mode): New class.
13201 * machmode.h (opt_mode): Likewise.
13202 (opt_mode::else_void): New function.
13203 (opt_mode::require): Likewise.
13204 (opt_mode::exists): Likewise.
13205 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13206 (GET_MODE_2XWIDER_MODE): Likewise.
13207 (mode_iterator::get_wider): Update accordingly.
13208 (mode_iterator::get_2xwider): Likewise.
13209 (mode_iterator::get_known_wider): Likewise, turning into a template.
13210 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13211 forcing a wider mode to exist.
13212 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13213 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13214 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13215 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13216 * lower-subreg.c (init_lower_subreg): Likewise.
13217 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13218 on the final iteration.
13219 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13220 a wider mode exists before asking for a move pattern.
13221 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13222 forcing a wider mode to exist.
13223 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13224 returning false if no such mode exists.
13225 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13226 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13227 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13228 Avoid checking for a MODE_INT if we already know the mode is not a
13230 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13231 forcing a wider mode to exist.
13232 (expmed_mult_highpart_optab): Likewise.
13233 (expmed_mult_highpart): Likewise.
13234 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13236 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13237 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13238 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13239 * internal-fn.c (expand_mul_overflow): Update use of
13240 GET_MODE_2XWIDER_MODE.
13241 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13242 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13243 GET_MODE_WIDER_MODE.
13244 (convert_plusminus_to_widen): Likewise.
13245 * tree-switch-conversion.c (array_value_type): Likewise.
13246 * var-tracking.c (emit_note_insn_var_location): Likewise.
13247 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13248 Return false inside rather than outside the loop if no wider mode
13250 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13251 and GET_MODE_2XWIDER_MODE
13252 (can_compare_p): Use else_void.
13253 * gdbhooks.py (OptMachineModePrinter): New class.
13254 (build_pretty_printer): Use it for opt_mode.
13256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13257 Alan Hayward <alan.hayward@arm.com>
13258 David Sherwood <david.sherwood@arm.com>
13260 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13261 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13264 Alan Hayward <alan.hayward@arm.com>
13265 David Sherwood <david.sherwood@arm.com>
13267 * machmode.h (mode_traits): New structure.
13268 (get_narrowest_mode): New function.
13269 (mode_iterator::start): Likewise.
13270 (mode_iterator::iterate_p): Likewise.
13271 (mode_iterator::get_wider): Likewise.
13272 (mode_iterator::get_known_wider): Likewise.
13273 (mode_iterator::get_2xwider): Likewise.
13274 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13275 (FOR_EACH_MODE): Likewise.
13276 (FOR_EACH_MODE_FROM): Likewise.
13277 (FOR_EACH_MODE_UNTIL): Likewise.
13278 (FOR_EACH_WIDER_MODE): Likewise.
13279 (FOR_EACH_2XWIDER_MODE): Likewise.
13280 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13281 * combine.c (simplify_comparison): Likewise
13282 * config/i386/i386.c (type_natural_mode): Likewise.
13283 * cse.c (cse_insn): Likewise.
13284 * dse.c (find_shift_sequence): Likewise.
13285 * emit-rtl.c (init_derived_machine_modes): Likewise.
13286 (init_emit_once): Likewise.
13287 * explow.c (hard_function_value): Likewise.
13288 * expmed.c (extract_fixed_bit_field_1): Likewise.
13289 (extract_bit_field_1): Likewise.
13290 (expand_divmod): Likewise.
13291 (emit_store_flag_1): Likewise.
13292 * expr.c (init_expr_target): Likewise.
13293 (convert_move): Likewise.
13294 (alignment_for_piecewise_move): Likewise.
13295 (widest_int_mode_for_size): Likewise.
13296 (emit_block_move_via_movmem): Likewise.
13297 (copy_blkmode_to_reg): Likewise.
13298 (set_storage_via_setmem): Likewise.
13299 (compress_float_constant): Likewise.
13300 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13301 * optabs-query.c (get_best_extraction_insn): Likewise.
13302 * optabs.c (expand_binop): Likewise.
13303 (expand_twoval_unop): Likewise.
13304 (expand_twoval_binop): Likewise.
13305 (widen_leading): Likewise.
13306 (widen_bswap): Likewise.
13307 (expand_parity): Likewise.
13308 (expand_unop): Likewise.
13309 (prepare_cmp_insn): Likewise.
13310 (prepare_float_lib_cmp): Likewise.
13311 (expand_float): Likewise.
13312 (expand_fix): Likewise.
13313 (expand_sfix_optab): Likewise.
13314 * postreload.c (move2add_use_add2_insn): Likewise.
13315 * reg-stack.c (reg_to_stack): Likewise.
13316 * reginfo.c (choose_hard_reg_mode): Likewise.
13317 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13318 * stor-layout.c (mode_for_size): Likewise.
13319 (smallest_mode_for_size): Likewise.
13320 (mode_for_vector): Likewise.
13321 (finish_bitfield_representative): Likewise.
13322 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13323 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13324 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13325 * var-tracking.c (prepare_call_arguments): Likewise.
13327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13328 Alan Hayward <alan.hayward@arm.com>
13329 David Sherwood <david.sherwood@arm.com>
13331 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13332 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13333 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13334 * machmode.h (mode_size): Move earlier in file.
13335 (mode_precision): Likewise.
13336 (mode_inner): Likewise.
13337 (mode_nunits): Likewise.
13338 (mode_unit_size): Likewise.
13339 (unit_unit_precision): Likewise.
13340 (mode_wider): Likewise.
13341 (mode_2xwider): Likewise.
13342 (machine_mode): New class.
13343 (mode_to_bytes): New function.
13344 (mode_to_bits): Likewise.
13345 (mode_to_precision): Likewise.
13346 (mode_to_inner): Likewise.
13347 (mode_to_unit_size): Likewise.
13348 (mode_to_unit_precision): Likewise.
13349 (mode_to_nunits): Likewise.
13350 (GET_MODE_SIZE): Use mode_to_bytes.
13351 (GET_MODE_BITSIZE): Use mode_to_bits.
13352 (GET_MODE_PRECISION): Use mode_to_precision.
13353 (GET_MODE_INNER): Use mode_to_inner.
13354 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13355 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13356 (GET_MODE_NUNITS): Use mode_to_nunits.
13357 * system.h (ALWAYS_INLINE): New macro.
13358 * config/powerpcspe/powerpcspe-c.c
13359 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13360 int for arg1_mode and arg2_mode.
13362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13363 Alan Hayward <alan.hayward@arm.com>
13364 David Sherwood <david.sherwood@arm.com>
13366 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13367 Prefix mode names with E_ in case statements.
13368 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13369 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13370 (aarch64_split_simd_move): Likewise.
13371 (aarch64_gen_storewb_pair): Likewise.
13372 (aarch64_gen_loadwb_pair): Likewise.
13373 (aarch64_gen_store_pair): Likewise.
13374 (aarch64_gen_load_pair): Likewise.
13375 (aarch64_get_condition_code_1): Likewise.
13376 (aarch64_constant_pool_reload_icode): Likewise.
13377 (get_rsqrte_type): Likewise.
13378 (get_rsqrts_type): Likewise.
13379 (get_recpe_type): Likewise.
13380 (get_recps_type): Likewise.
13381 (aarch64_gimplify_va_arg_expr): Likewise.
13382 (aarch64_simd_container_mode): Likewise.
13383 (aarch64_emit_load_exclusive): Likewise.
13384 (aarch64_emit_store_exclusive): Likewise.
13385 (aarch64_expand_compare_and_swap): Likewise.
13386 (aarch64_gen_atomic_cas): Likewise.
13387 (aarch64_emit_bic): Likewise.
13388 (aarch64_emit_atomic_swap): Likewise.
13389 (aarch64_emit_atomic_load_op): Likewise.
13390 (aarch64_evpc_trn): Likewise.
13391 (aarch64_evpc_uzp): Likewise.
13392 (aarch64_evpc_zip): Likewise.
13393 (aarch64_evpc_ext): Likewise.
13394 (aarch64_evpc_rev): Likewise.
13395 (aarch64_evpc_dup): Likewise.
13396 (aarch64_gen_ccmp_first): Likewise.
13397 (aarch64_gen_ccmp_next): Likewise.
13398 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13399 (alpha_emit_xfloating_libcall): Likewise.
13400 (emit_insxl): Likewise.
13401 (alpha_arg_type): Likewise.
13402 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13403 (arc_preferred_simd_mode): Likewise.
13404 (arc_secondary_reload): Likewise.
13405 (get_arc_condition_code): Likewise.
13406 (arc_print_operand): Likewise.
13407 (arc_legitimate_constant_p): Likewise.
13408 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13409 * config/arc/arc.md (casesi_load): Likewise.
13410 (casesi_compact_jump): Likewise.
13411 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13412 (cc_use_register): Likewise.
13413 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13414 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13415 (arm_init_iwmmxt_builtins): Likewise.
13416 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13417 (neon_expand_vector_init): Likewise.
13418 (arm_attr_length_move_neon): Likewise.
13419 (maybe_get_arm_condition_code): Likewise.
13420 (arm_emit_vector_const): Likewise.
13421 (arm_preferred_simd_mode): Likewise.
13422 (arm_output_iwmmxt_tinsr): Likewise.
13423 (thumb1_output_casesi): Likewise.
13424 (thumb2_output_casesi): Likewise.
13425 (arm_emit_load_exclusive): Likewise.
13426 (arm_emit_store_exclusive): Likewise.
13427 (arm_expand_compare_and_swap): Likewise.
13428 (arm_evpc_neon_vuzp): Likewise.
13429 (arm_evpc_neon_vzip): Likewise.
13430 (arm_evpc_neon_vrev): Likewise.
13431 (arm_evpc_neon_vtrn): Likewise.
13432 (arm_evpc_neon_vext): Likewise.
13433 (arm_validize_comparison): Likewise.
13434 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13435 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13436 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13437 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13438 (c6x_preferred_simd_mode): Likewise.
13439 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13440 (epiphany_rtx_costs): Likewise.
13441 * config/epiphany/predicates.md (proper_comparison_operator):
13443 * config/frv/frv.c (condexec_memory_operand): Likewise.
13444 (frv_emit_move): Likewise.
13445 (output_move_single): Likewise.
13446 (output_condmove_single): Likewise.
13447 (frv_hard_regno_mode_ok): Likewise.
13448 (frv_matching_accg_mode): Likewise.
13449 * config/h8300/h8300.c (split_adds_subs): Likewise.
13450 (h8300_rtx_costs): Likewise.
13451 (h8300_print_operand): Likewise.
13452 (compute_mov_length): Likewise.
13453 (output_logical_op): Likewise.
13454 (compute_logical_op_length): Likewise.
13455 (compute_logical_op_cc): Likewise.
13456 (h8300_shift_needs_scratch_p): Likewise.
13457 (output_a_shift): Likewise.
13458 (compute_a_shift_length): Likewise.
13459 (compute_a_shift_cc): Likewise.
13460 (expand_a_rotate): Likewise.
13461 (output_a_rotate): Likewise.
13462 * config/i386/i386.c (classify_argument): Likewise.
13463 (function_arg_advance_32): Likewise.
13464 (function_arg_32): Likewise.
13465 (function_arg_64): Likewise.
13466 (function_value_64): Likewise.
13467 (ix86_gimplify_va_arg): Likewise.
13468 (ix86_legitimate_constant_p): Likewise.
13469 (put_condition_code): Likewise.
13470 (split_double_mode): Likewise.
13471 (ix86_avx256_split_vector_move_misalign): Likewise.
13472 (ix86_expand_vector_logical_operator): Likewise.
13473 (ix86_split_idivmod): Likewise.
13474 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13475 (ix86_build_const_vector): Likewise.
13476 (ix86_build_signbit_mask): Likewise.
13477 (ix86_match_ccmode): Likewise.
13478 (ix86_cc_modes_compatible): Likewise.
13479 (ix86_expand_branch): Likewise.
13480 (ix86_expand_sse_cmp): Likewise.
13481 (ix86_expand_sse_movcc): Likewise.
13482 (ix86_expand_int_sse_cmp): Likewise.
13483 (ix86_expand_vec_perm_vpermi2): Likewise.
13484 (ix86_expand_vec_perm): Likewise.
13485 (ix86_expand_sse_unpack): Likewise.
13486 (ix86_expand_int_addcc): Likewise.
13487 (ix86_split_to_parts): Likewise.
13488 (ix86_vectorize_builtin_gather): Likewise.
13489 (ix86_vectorize_builtin_scatter): Likewise.
13490 (avx_vpermilp_parallel): Likewise.
13491 (inline_memory_move_cost): Likewise.
13492 (ix86_tieable_integer_mode_p): Likewise.
13493 (x86_maybe_negate_const_int): Likewise.
13494 (ix86_expand_vector_init_duplicate): Likewise.
13495 (ix86_expand_vector_init_one_nonzero): Likewise.
13496 (ix86_expand_vector_init_one_var): Likewise.
13497 (ix86_expand_vector_init_concat): Likewise.
13498 (ix86_expand_vector_init_interleave): Likewise.
13499 (ix86_expand_vector_init_general): Likewise.
13500 (ix86_expand_vector_set): Likewise.
13501 (ix86_expand_vector_extract): Likewise.
13502 (emit_reduc_half): Likewise.
13503 (ix86_emit_i387_round): Likewise.
13504 (ix86_mangle_type): Likewise.
13505 (ix86_expand_round_sse4): Likewise.
13506 (expand_vec_perm_blend): Likewise.
13507 (canonicalize_vector_int_perm): Likewise.
13508 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13509 (expand_vec_perm_1): Likewise.
13510 (expand_vec_perm_interleave3): Likewise.
13511 (expand_vec_perm_even_odd_pack): Likewise.
13512 (expand_vec_perm_even_odd_1): Likewise.
13513 (expand_vec_perm_broadcast_1): Likewise.
13514 (ix86_vectorize_vec_perm_const_ok): Likewise.
13515 (ix86_expand_vecop_qihi): Likewise.
13516 (ix86_expand_mul_widen_hilo): Likewise.
13517 (ix86_expand_sse2_abs): Likewise.
13518 (ix86_expand_pextr): Likewise.
13519 (ix86_expand_pinsr): Likewise.
13520 (ix86_preferred_simd_mode): Likewise.
13521 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13522 * config/i386/sse.md (*andnot<mode>3): Likewise.
13523 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13524 (*<code><mode>3): Likewise.
13525 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13526 (ia64_expand_atomic_op): Likewise.
13527 (ia64_arg_type): Likewise.
13528 (ia64_mode_to_int): Likewise.
13529 (ia64_scalar_mode_supported_p): Likewise.
13530 (ia64_vector_mode_supported_p): Likewise.
13531 (expand_vec_perm_broadcast): Likewise.
13532 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13533 (iq2000_function_arg_advance): Likewise.
13534 (iq2000_function_arg): Likewise.
13535 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13536 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13537 (m68k_libcall_value): Likewise.
13538 (m68k_function_value): Likewise.
13539 (sched_attr_op_type): Likewise.
13540 * config/mcore/mcore.c (mcore_output_move): Likewise.
13541 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13543 (microblaze_function_arg): Likewise.
13544 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13545 (mips_print_operand): Likewise.
13546 (mips_mode_ok_for_mov_fmt_p): Likewise.
13547 (mips_vector_mode_supported_p): Likewise.
13548 (mips_preferred_simd_mode): Likewise.
13549 (mips_expand_vpc_loongson_even_odd): Likewise.
13550 (mips_expand_vec_unpack): Likewise.
13551 (mips_expand_vi_broadcast): Likewise.
13552 (mips_expand_vector_init): Likewise.
13553 (mips_expand_vec_reduc): Likewise.
13554 (mips_expand_msa_cmp): Likewise.
13555 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13556 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13557 (cc_flags_for_mode): Likewise.
13558 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13559 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13560 (nds32_output_casesi_pc_relative): Likewise.
13561 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13562 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13563 (nvptx_gen_unpack): Likewise.
13564 (nvptx_gen_pack): Likewise.
13565 (nvptx_gen_shuffle): Likewise.
13566 (nvptx_gen_wcast): Likewise.
13567 (nvptx_preferred_simd_mode): Likewise.
13568 * config/pa/pa.c (pa_secondary_reload): Likewise.
13569 * config/pa/predicates.md (base14_operand): Likewise.
13570 * config/powerpcspe/powerpcspe-c.c
13571 (altivec_resolve_overloaded_builtin): Likewise.
13572 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13574 (rs6000_preferred_simd_mode): Likewise.
13575 (output_vec_const_move): Likewise.
13576 (rs6000_expand_vector_extract): Likewise.
13577 (rs6000_split_vec_extract_var): Likewise.
13578 (reg_offset_addressing_ok_p): Likewise.
13579 (rs6000_legitimate_offset_address_p): Likewise.
13580 (rs6000_legitimize_address): Likewise.
13581 (rs6000_emit_set_const): Likewise.
13582 (rs6000_const_vec): Likewise.
13583 (rs6000_emit_move): Likewise.
13584 (spe_build_register_parallel): Likewise.
13585 (rs6000_darwin64_record_arg_recurse): Likewise.
13586 (swap_selector_for_mode): Likewise.
13587 (spe_init_builtins): Likewise.
13588 (paired_init_builtins): Likewise.
13589 (altivec_init_builtins): Likewise.
13590 (do_load_for_compare): Likewise.
13591 (rs6000_generate_compare): Likewise.
13592 (rs6000_expand_float128_convert): Likewise.
13593 (emit_load_locked): Likewise.
13594 (emit_store_conditional): Likewise.
13595 (rs6000_output_function_epilogue): Likewise.
13596 (rs6000_handle_altivec_attribute): Likewise.
13597 (rs6000_function_value): Likewise.
13598 (emit_fusion_gpr_load): Likewise.
13599 (emit_fusion_p9_load): Likewise.
13600 (emit_fusion_p9_store): Likewise.
13601 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13602 (fusion_gpr_mem_load): Likewise.
13603 (fusion_addis_mem_combo_load): Likewise.
13604 (fusion_addis_mem_combo_store): Likewise.
13605 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13606 (fusion_gpr_mem_load): Likewise.
13607 (fusion_addis_mem_combo_load): Likewise.
13608 (fusion_addis_mem_combo_store): Likewise.
13609 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13611 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13612 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13613 (rs6000_preferred_simd_mode): Likewise.
13614 (output_vec_const_move): Likewise.
13615 (rs6000_expand_vector_extract): Likewise.
13616 (rs6000_split_vec_extract_var): Likewise.
13617 (reg_offset_addressing_ok_p): Likewise.
13618 (rs6000_legitimate_offset_address_p): Likewise.
13619 (rs6000_legitimize_address): Likewise.
13620 (rs6000_emit_set_const): Likewise.
13621 (rs6000_const_vec): Likewise.
13622 (rs6000_emit_move): Likewise.
13623 (rs6000_darwin64_record_arg_recurse): Likewise.
13624 (swap_selector_for_mode): Likewise.
13625 (paired_init_builtins): Likewise.
13626 (altivec_init_builtins): Likewise.
13627 (rs6000_expand_float128_convert): Likewise.
13628 (emit_load_locked): Likewise.
13629 (emit_store_conditional): Likewise.
13630 (rs6000_output_function_epilogue): Likewise.
13631 (rs6000_handle_altivec_attribute): Likewise.
13632 (rs6000_function_value): Likewise.
13633 (emit_fusion_gpr_load): Likewise.
13634 (emit_fusion_p9_load): Likewise.
13635 (emit_fusion_p9_store): Likewise.
13636 * config/rx/rx.c (rx_gen_move_template): Likewise.
13637 (flags_from_mode): Likewise.
13638 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13639 (s390_slb_comparison): Likewise.
13640 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13641 (s390_vector_mode_supported_p): Likewise.
13642 (s390_cc_modes_compatible): Likewise.
13643 (s390_match_ccmode_set): Likewise.
13644 (s390_canonicalize_comparison): Likewise.
13645 (s390_emit_compare_and_swap): Likewise.
13646 (s390_branch_condition_mask): Likewise.
13647 (s390_rtx_costs): Likewise.
13648 (s390_secondary_reload): Likewise.
13649 (__SECONDARY_RELOAD_CASE): Likewise.
13650 (s390_expand_cs): Likewise.
13651 (s390_preferred_simd_mode): Likewise.
13652 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13653 * config/sh/sh.c (sh_print_operand): Likewise.
13654 (dump_table): Likewise.
13655 (sh_secondary_reload): Likewise.
13656 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13657 * config/sh/sh.md (casesi_worker_1): Likewise.
13658 (casesi_worker_2): Likewise.
13659 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13660 (fcc_comparison_operator): Likewise.
13661 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13662 (emit_soft_tfmode_cvt): Likewise.
13663 (sparc_preferred_simd_mode): Likewise.
13664 (output_cbranch): Likewise.
13665 (sparc_print_operand): Likewise.
13666 (sparc_expand_vec_perm_bmask): Likewise.
13667 (vector_init_bshuffle): Likewise.
13668 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13669 (spu_vector_mode_supported_p): Likewise.
13670 (spu_expand_insv): Likewise.
13671 (spu_emit_branch_or_set): Likewise.
13672 (spu_handle_vector_attribute): Likewise.
13673 (spu_builtin_splats): Likewise.
13674 (spu_builtin_extract): Likewise.
13675 (spu_builtin_promote): Likewise.
13676 (spu_expand_sign_extend): Likewise.
13677 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13678 (tilegx_simd_int): Likewise.
13679 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13680 (tilepro_simd_int): Likewise.
13681 * config/v850/v850.c (const_double_split): Likewise.
13682 (v850_print_operand): Likewise.
13683 (ep_memory_offset): Likewise.
13684 * config/vax/vax.c (vax_rtx_costs): Likewise.
13685 (vax_output_int_move): Likewise.
13686 (vax_output_int_add): Likewise.
13687 (vax_output_int_subtract): Likewise.
13688 * config/visium/predicates.md (visium_branch_operator): Likewise.
13689 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13690 (visium_print_operand_address): Likewise.
13691 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13692 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13693 (xtensa_expand_conditional_branch): Likewise.
13694 (xtensa_copy_incoming_a7): Likewise.
13695 (xtensa_output_literal): Likewise.
13696 * dfp.c (decimal_real_maxval): Likewise.
13697 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13700 Alan Hayward <alan.hayward@arm.com>
13701 David Sherwood <david.sherwood@arm.com>
13703 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13704 (mode_nunits_inline): Likewise.
13705 (mode_inner_inline): Likewise.
13706 (mode_unit_size_inline): Likewise.
13707 (mode_unit_precision_inline): Likewise.
13708 (emit_insn_modes_h): Likewise. Also emit a #define of the
13710 (emit_mode_wider): Add an E_ prefix to mode names.
13711 (emit_mode_complex): Likewise.
13712 (emit_mode_inner): Likewise.
13713 (emit_mode_adjustments): Likewise.
13714 (emit_mode_int_n): Likewise.
13715 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13716 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13717 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13718 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13719 (CRC32_BUILTIN, ENTRY): Likewise.
13720 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13721 (aarch64_pop_regs): Likewise.
13722 (aarch64_process_components): Likewise.
13723 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13724 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13725 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13726 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13727 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13728 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13730 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13731 (pcmpistr): Likewise.
13732 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13733 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13734 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13736 * config/rl78/rl78.c (mduc_regs): Likewise.
13737 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13738 (htm_expand_builtin): Likewise.
13739 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13740 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13741 * config/xtensa/xtensa.c (print_operand): Likewise.
13742 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13743 (NUM_MODE_VECTOR_INT): Likewise.
13744 * genoutput.c (null_operand): Likewise.
13745 (output_operand_data): Likewise.
13746 * genrecog.c (print_parameter_value): Likewise.
13747 * lra.c (debug_operand_data): Likewise.
13749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13751 * dwarf2out.c (scompare_loc_descriptor_wide)
13752 (scompare_loc_descriptor_narrow): New functions, split out from...
13753 (scompare_loc_descriptor): ...here.
13754 * expmed.c (emit_store_flag_int): New function, split out from...
13755 (emit_store_flag): ...here.
13757 2017-08-30 Richard Biener <rguenther@suse.de>
13759 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13760 (dwarf2out_early_finish): Move setting of AT_pubnames from
13761 early debug output to early finish.
13763 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13765 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13766 and -mdata-region to the assembler.
13768 2017-08-30 Richard Biener <rguenther@suse.de>
13770 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13772 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13773 (dwarf2out_early_global_decl): Only generate a DIE for the
13774 abstract origin if it doesn't already exist or is a declaration DIE.
13775 (resolve_addr): Do not add the linkage name twice when
13776 generating a stub DIE for the DW_TAG_GNU_call_site target.
13778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13780 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13781 Use machine_mode rather than int for arg1_mode.
13783 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13786 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13787 that the second argument of the built-in functions to unpack
13788 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13789 switch statement instead a lot of if statements.
13790 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13791 Allow 64-bit values to be in Altivec registers as well as
13792 traditional floating point registers.
13793 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13795 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13797 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13798 MAX_REGS_PER_ADDRESS == 1.
13800 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13802 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13803 * config/i386/i386.c (ix86_option_override_internal): Simplify
13804 setting of opts->x_flag_entry.
13806 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13807 Jakub Jelinek <jakub@redhat.com>
13808 Richard Biener <rguenther@suse.de>
13810 PR tree-optimization/81503
13811 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13812 folded constant fits in the target type; reorder tests for clarity.
13814 2017-08-29 Martin Liska <mliska@suse.cz>
13816 * passes.def: Include pass_lower_switch.
13817 * stmt.c (dump_case_nodes): Remove and move to
13818 tree-switch-conversion.
13819 (case_values_threshold): Likewise.
13820 (expand_switch_as_decision_tree_p): Likewise.
13821 (emit_case_decision_tree): Likewise.
13822 (expand_case): Likewise.
13823 (balance_case_nodes): Likewise.
13824 (node_has_low_bound): Likewise.
13825 (node_has_high_bound): Likewise.
13826 (node_is_bounded): Likewise.
13827 (emit_case_nodes): Likewise.
13828 (struct simple_case_node): New struct.
13829 (add_case_node): Remove.
13830 (emit_case_dispatch_table): Use vector instead of case_list.
13831 (reset_out_edges_aux): Remove.
13832 (compute_cases_per_edge): Likewise.
13833 (expand_case): Build list of simple_case_node.
13834 (expand_sjlj_dispatch_table): Use it.
13835 * tree-switch-conversion.c (struct case_node): Moved from
13836 stmt.c and adjusted.
13837 (emit_case_nodes): Likewise.
13838 (node_has_low_bound): Likewise.
13839 (node_has_high_bound): Likewise.
13840 (node_is_bounded): Likewise.
13841 (case_values_threshold): Likewise.
13842 (reset_out_edges_aux): Likewise.
13843 (compute_cases_per_edge): Likewise.
13844 (add_case_node): Likewise.
13845 (dump_case_nodes): Likewise.
13846 (balance_case_nodes): Likewise.
13847 (expand_switch_as_decision_tree_p): Likewise.
13848 (emit_jump): Likewise.
13849 (emit_case_decision_tree): Likewise.
13850 (try_switch_expansion): Likewise.
13851 (do_jump_if_equal): Likewise.
13852 (emit_cmp_and_jump_insns): Likewise.
13853 (fix_phi_operands_for_edge): New function.
13854 (record_phi_operand_mapping): Likewise.
13855 (class pass_lower_switch): New pass.
13856 (pass_lower_switch::execute): New function.
13857 (make_pass_lower_switch): Likewise.
13858 (conditional_probability):
13859 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13860 * tree-pass.h: Add make_pass_lower_switch.
13862 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13865 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13868 2017-08-29 Richard Biener <rguenther@suse.de>
13870 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13871 we do not add a DW_AT_inline attribute twice.
13872 (gen_subprogram_die): Remove code setting DW_AT_inline on
13873 DECL_ABSTRACT_P nodes.
13875 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13877 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13878 calls to internal functions.
13879 (gimplify_modify_expr): Likewise.
13880 * tree-call-cdce.c (use_internal_fn): Likewise.
13881 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13882 (convert_to_divmod): Set the nothrow flag.
13883 * tree-if-conv.c (predicate_mem_writes): Likewise.
13884 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13885 (vectorizable_call): Likewise.
13886 (vectorizable_store): Likewise.
13887 (vectorizable_load): Likewise.
13888 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13889 (vect_recog_mask_conversion_pattern): Likewise.
13891 2017-08-29 Martin Liska <mliska@suse.cz>
13894 * gcc.c (driver_handle_option): Add new argument.
13895 * opts-common.c (handle_option): Pass
13896 target_option_override_hook.
13897 * opts-global.c (lang_handle_option): Add new option.
13898 (set_default_handlers): Add new argument.
13899 (decode_options): Likewise.
13900 * opts.c (target_handle_option): Likewise.
13901 (common_handle_option): Call target_option_override_hook.
13902 * opts.h (struct cl_option_handler_func): Add hook for
13903 target option override.
13904 (struct cl_option_handlers): Likewise.
13905 (set_default_handlers): Add new argument.
13906 (decode_options): Likewise.
13907 (common_handle_option): Likewise.
13908 (target_handle_option): Likewise.
13909 * toplev.c (toplev::main): Pass targetm.target_option.override
13912 2017-08-29 Richard Biener <rguenther@suse.de>
13913 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13915 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13916 life to the active subtree.
13918 2017-08-28 Jeff Law <law@redhat.com>
13920 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13921 derive_equivalences.
13922 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13923 Code moved into....
13924 (edge_info::derive_equivalences): New private member function
13926 * tree-ssa-dom.c (class edge_info): Changed from a struct
13927 to a class. Add ctor/dtor, methods and data members.
13928 (edge_info::edge_info): Renamed from allocate_edge_info.
13929 Initialize additional members.
13930 (edge_info::~edge_info): New.
13931 (free_dom_edge_info): Delete the edge info.
13932 (record_edge_info): Use new class & associated member functions.
13933 Tighten forms for testing for edge equivalences.
13934 (record_temporary_equivalences): Iterate over the simple
13935 equivalences rather than assuming there's only one per edge.
13936 (cprop_into_successor_phis): Iterate over the simple
13937 equivalences rather than assuming there's only one per edge.
13938 (optimize_stmt): Use operand_equal_p rather than pointer
13939 equality for mini-DSE code.
13941 2017-08-28 Nathan Sidwell <nathan@acm.org>
13943 * gcc.c (execute): Fold SIGPIPE handling into switch
13944 statement. Adjust internal error message.
13946 2017-08-28 Richard Biener <rguenther@suse.de>
13949 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13950 Do nothing for removed DIEs.
13952 2017-08-28 Richard Biener <rguenther@suse.de>
13954 PR tree-optimization/81977
13955 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13958 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13961 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13962 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13963 using targetm.gen_mem_thread_fence.
13965 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13968 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13969 predicate to register_operand. Reorder operands.
13970 (*btr<mode>): Ditto.
13971 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13972 (*btr<mode>_mask): Ditto.
13974 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13976 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13977 * config/rs6000/xmmintrin.h: New file.
13978 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13980 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13983 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13984 parameter and_insn and return it.
13985 (recombine_lvx_pattern): Insert a copy to ensure availability of
13986 the base register of the copied masking operation at the point of
13987 the instruction replacement.
13988 (recombine_stvx_pattern): Likewise.
13990 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13992 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13993 undocumented switches.
13994 (-mpower9-dform-vector): Likewise.
13995 (-mpower9-dform): Likewise.
13996 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13997 comments to delete references to -mpower9-dform* switches.
13998 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13999 Delete reference to -mpower9-dform* switches, test for
14000 -mpower9-vector instead.
14001 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
14002 (OTHER_P9_VECTOR_MASKS): Likewise.
14003 (POWERPC_MASKS): Likewise.
14004 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
14005 tests against -mpower9-dform* to -mpower9-vector. Delete code
14006 that checked for -mpower9-dform* consistancy with other options.
14007 Add test for -mpower9-misc to enable other power9 switches.
14008 (rs6000_init_hard_regno_mode_ok): Likewise.
14009 (rs6000_option_override_internal): Likewise.
14010 (rs6000_emit_prologue): Likewise.
14011 (rs6000_emit_epilogue): Likewise.
14012 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
14013 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
14014 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
14016 (emit_fusion_p9_store): Likewise.
14017 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
14018 resetting these macros if the assembler does not support ISA 3.0
14020 (TARGET_P9_DFORM_VECTOR): Likewise.
14021 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
14022 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
14024 2017-08-25 Alan Modra <amodra@gmail.com>
14026 PR rtl-optimization/81747
14027 * cse.c (cse_extended_basic_block): Don't attempt to record
14028 equivalences for degenerate conditional branches that branch
14029 to their fall-through.
14031 2017-08-24 Martin Sebor <msebor@redhat.com>
14033 PR middle-end/81908
14034 * gimple-fold.c (size_must_be_zero_p): New function.
14035 (gimple_fold_builtin_memory_op): Call it.
14037 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
14039 * config/rs6000/mm_malloc.h: New file.
14041 2017-08-24 Bin Cheng <bin.cheng@arm.com>
14043 PR tree-optimization/81913
14044 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
14045 analysis when either IVs in condition can wrap.
14047 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
14049 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
14050 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
14052 2017-08-24 Richard Biener <rguenther@suse.de>
14055 * targhooks.c (default_target_can_inline_p): Properly
14056 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
14057 is present and always compare.
14058 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
14059 infer -mfpmath=sse from TARGET_SSE_P.
14060 (ix86_can_inline_p): Properly use target_option_default_node when
14061 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
14063 2017-08-24 Richard Biener <rguenther@suse.de>
14066 * dwarf2out.c (output_die): Handle flag_generate_offload like
14068 (output_comp_unit): Likewise.
14069 (gen_array_type_die): Likewise.
14070 (dwarf2out_early_finish): Likewise.
14071 (note_variable_value_in_expr): Likewise.
14072 (dwarf2out_finish): Likewise. Adjust assert.
14073 * cgraphunit.c (symbol_table::compile): Move setting of
14074 flag_generate_offload earlier ...
14075 (symbol_table::finalize_compilation_unit): ... here, before
14076 early debug finalization.
14078 2017-08-24 Richard Biener <rguenther@suse.de>
14080 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
14081 and ipa-fnsummary.h.
14082 (ix86_can_inline_p): When ix86_fpmath flags do not match
14083 check whether the callee uses FP math at all.
14085 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
14087 PR middle-end/81931
14088 * tree-ssanames.c (get_nonzero_bits): Use element_precision
14089 instead of TYPE_PRECISION.
14091 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
14092 Alan Hayward <alan.hayward@arm.com>
14093 David Sherwood <david.sherwood@arm.com>
14095 * combine.c (make_extraction): Use subreg_offset_from_lsb.
14097 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
14099 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14101 (ix86_frame::stack_realign_allocate): New field.
14102 (struct machine_frame_state): Modify comments.
14103 (machine_frame_state::sp_realigned_fp_end): New field.
14104 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14105 layout calculation.
14106 (sp_valid_at): Add assertion to assure no attempt to access invalid
14107 offset of a realigned stack.
14108 (fp_valid_at): Likewise.
14109 (choose_baseaddr): Modify comments.
14110 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14111 ix86_expand_prologue.
14112 (ix86_expand_prologue): Modify stack realignment and allocation.
14113 (ix86_expand_epilogue): Modify comments.
14114 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14115 avx2_runtime, avx512f, and avx512f_runtime.
14117 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14119 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14120 (mstackrealign): Do not init to -1.
14121 * config/i386/i386.c (ix86_option_override_internal):
14122 Check opts_set, not opts when setting default value of
14123 opts->x_ix86_force_align_arg_pointer.
14125 2017-08-23 Richard Biener <rguenther@suse.de>
14127 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14128 lang_hooks.decl_printable_name.
14129 * print-rtl-function.c (print_rtx_function): Likewise.
14130 * tree-pretty-print.c (dump_function_header): Likewise.
14132 2017-08-23 Richard Biener <rguenther@suse.de>
14135 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14136 -g0 at compile-time.
14138 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14140 PR middle-end/19706
14141 * doc/sourcebuild.texi (Other hardware attributes):
14144 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14146 PR middle-end/19706
14147 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14148 Fix single-use check.
14150 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14152 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14154 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14156 * doc/install.texi: Modify to add more details on running selected
14159 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14161 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14162 is combined with -mabi=ms.
14163 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14166 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14168 PR tree-optimization/81488
14169 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14170 and cached_basis fields.
14171 (MAX_SPREAD): New constant.
14172 (alloc_cand_and_find_basis): Initialize new fields.
14173 (clear_visited): New function.
14174 (create_phi_basis_1): Rename from create_phi_basis, set visited
14175 and cached_basis fields.
14176 (create_phi_basis): New wrapper function.
14177 (phi_add_costs_1): Rename from phi_add_costs, add spread
14178 parameter, set visited field, short-circuit when limits reached.
14179 (phi_add_costs): New wrapper function.
14180 (record_phi_increments_1): Rename from record_phi_increments, set
14182 (record_phi_increments): New wrapper function.
14183 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14184 (phi_incr_cost): New wrapper function.
14185 (all_phi_incrs_profitable_1): Rename from
14186 all_phi_incrs_profitable, set visited field.
14187 (all_phi_incrs_profitable): New wrapper function.
14189 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14190 Alan Hayward <alan.hayward@arm.com>
14191 David Sherwood <david.sherwood@arm.com>
14193 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14194 that takes the outer and inner modes.
14195 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14196 comparison as the canonical test for a paradoxical subreg.
14197 * combine.c (simplify_set): Use paradoxical_subreg_p.
14198 (make_extraction): Likewise.
14199 (force_to_mode): Likewise.
14200 (rtx_equal_for_field_assignment_p): Likewise.
14201 (gen_lowpart_for_combine): Likewise.
14202 (simplify_comparison): Likewise.
14203 * cse.c (equiv_constant): Likewise.
14204 * expmed.c (store_bit_field_1): Likewise.
14205 * final.c (alter_subreg): Likewise.
14206 * fwprop.c (propagate_rtx): Likewise.
14207 (forward_propagate_subreg): Likewise.
14208 * ira-conflicts.c (ira_build_conflicts): Likewise.
14209 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14210 * lra-constraints.c (curr_insn_transform): Likewise.
14211 (split_reg): Likewise.
14212 * lra-eliminations.c (move_plus_up): Likewise.
14213 (lra_eliminate_regs_1): Likewise.
14214 * recog.c (general_operand): Likewise.
14215 * ree.c (combine_reaching_defs): Likewise.
14216 * reload.c (push_reload): Likewise.
14217 (find_reloads): Likewise.
14218 * reload1.c (elimination_effects): Likewise.
14219 (compute_reload_subreg_offset): Likewise.
14220 (choose_reload_regs): Likewise.
14221 * rtlanal.c (subreg_lsb_1): Likewise.
14222 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14223 (simplify_subreg): Likewise.
14224 * var-tracking.c (track_loc_p): Likewise.
14225 * emit-rtl.c (byte_lowpart_offset): Likewise.
14226 (paradoxical_subreg_p): Delete out-of-line definition.
14228 2017-08-22 Jeff Law <law@redhat.com>
14230 PR tree-optimization/81741
14231 PR tree-optimization/71947
14232 * tree-ssa-dom.c: Include tree-inline.h.
14233 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14234 equivalences if one is more expensive to compute than the other.
14235 * tree-ssa-scopedtables.h (class const_or_copies): Make
14236 record_const_or_copy_raw method private.
14237 (class avail_exprs_stack): New method simplify_binary_operation.
14238 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14239 avail_exprs_stack::simplify_binary_operation as needed.
14240 (avail_exprs_stack::simplify_binary_operation): New function.
14242 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14244 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14245 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14246 (DOT_SYMBOLS): Likewise.
14247 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14248 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14249 (RS6000_ABI_NAME): Likewise.
14250 (TARGET_CMODEL): Likewise.
14251 (TOC_SECTION_ASM_OP): Likewise.
14252 (SET_CMODEL): New macro.
14253 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14255 2017-08-22 Richard Biener <rguenther@suse.de>
14257 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14258 to allow for free-lang-data replacements similar to verify_type_variant.
14260 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14261 Alan Hayward <alan.hayward@arm.com>
14262 David Sherwood <david.sherwood@arm.com>
14264 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14265 VOIDmode for the LABEL_REF.
14267 2017-08-22 Richard Biener <rguenther@suse.de>
14269 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14271 2017-08-22 Richard Biener <rguenther@suse.de>
14273 * common.opt (feliminate-dwarf2-dups): Ignore.
14274 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14275 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14276 same_die_p_wrap, compute_section_prefix,
14277 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14278 (comdat_symbol_id, comdat_symbol_number): Likewise.
14279 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14281 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14282 (output_die): Mark unreachable path unreachable.
14283 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14284 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14285 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14286 (dwarf2out_early_finish): Likewise.
14288 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14290 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14292 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14295 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14296 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14297 (avr_attribute_table) <io, io_low, address>: Initialize
14298 .decl_required with true.
14300 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14302 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14303 undocumented debugging options.
14304 (-mvsx-scalar-double): Likewise.
14305 (-mallow-df-permute): Likewise.
14306 (-mvectorize-builtins): Likewise.
14307 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14308 (rs6000_builtin_vectorized_function): Likewise.
14309 (rs6000_builtin_md_vectorized_function): Likewise.
14310 (rs6000_opt_vars): Likewise.
14312 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14315 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14316 (*btrq_imm): Rename from *btrq.
14317 (*btcq_imm): Rename from *btcq.
14318 (btsc): New code attribute.
14319 (*<btsc><mode>): New insn pattern.
14320 (*btr<mode>): Ditto.
14321 (*<btsc><mode>_mask): New insn_and_split pattern.
14322 (*btr<mode>_mask): Ditto.
14324 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14325 Alan Hayward <alan.hayward@arm.com>
14326 David Sherwood <david.sherwood@arm.com>
14328 * function.c (pad_below): Simplify padding calculation.
14330 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14331 Alan Hayward <alan.hayward@arm.com>
14332 David Sherwood <david.sherwood@arm.com>
14334 * target.def (function_prologue): Remove frame size argument.
14335 (function_epilogue): Likewise.
14336 * doc/tm.texi: Regenerate.
14337 * final.c (final_start_function): Update call to function_prologue.
14338 (final_end_function): Update call to function_epilogue.
14339 (default_function_pro_epilogue): Remove frame size argument.
14340 * output.h (default_function_pro_epilogue): Likewise.
14341 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14342 (arm_output_function_prologue): Likewise.
14343 * config/frv/frv.c (frv_function_prologue): Likewise.
14344 (frv_function_epilogue): Likewise.
14345 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14346 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14347 (ia64_output_function_epilogue): Likewise.
14348 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14349 (m32r_output_function_epilogue): Likewise.
14350 * config/microblaze/microblaze.c (microblaze_function_prologue)
14351 (microblaze_function_epilogue): Likewise.
14352 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14353 (mips_output_function_epilogue): Likewise.
14354 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14355 (mmix_target_asm_function_epilogue): Likewise.
14356 * config/msp430/msp430.c (msp430_start_function): Likewise.
14357 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14358 (nds32_asm_function_epilogue): Likewise.
14359 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14360 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14361 (pa_output_function_epilogue): Likewise.
14362 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14363 (rs6000_output_function_epilogue): Likewise.
14364 * config/rl78/rl78.c (rl78_start_function): Likewise.
14365 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14366 (rs6000_output_function_epilogue): Likewise.
14367 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14368 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14369 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14370 (sparc_asm_function_epilogue): Likewise.
14372 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14374 * tree.h (type_has_mode_precision_p): New function.
14375 * convert.c (convert_to_integer_1): Use it.
14376 * expr.c (expand_expr_real_2): Likewise.
14377 (expand_expr_real_1): Likewise.
14378 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14379 * match.pd: Likewise.
14380 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14381 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14382 * tree-tailcall.c (process_assignment): Likewise.
14383 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14384 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14385 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14386 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14387 (vectorizable_assignment): Likewise.
14388 (vectorizable_shift): Likewise.
14389 (vectorizable_operation): Likewise.
14390 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14392 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14394 * match.pd: Add pow (C, x) simplification.
14396 2017-08-21 Richard Biener <rguenther@suse.de>
14398 PR tree-optimization/81900
14399 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14400 for blocks with abnormal predecessors.
14401 (compute_antic): Do not set visited flag prematurely.
14403 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14406 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14408 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14410 * stor-layout.h (vector_type_mode): Move to...
14411 * tree.h (vector_type_mode): ...here.
14412 * stor-layout.c (vector_type_mode): Move to...
14413 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14415 2017-08-21 Richard Biener <rguenther@suse.de>
14417 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14418 register_external_die hooks.
14419 (debug_false_tree_charstarstar_uhwistar): Declare.
14420 (debug_nothing_tree_charstar_uhwi): Likewise.
14421 * debug.c (do_nothing_debug_hooks): Adjust.
14422 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14423 (debug_nothing_tree_charstar_uhwi): Likewise.
14424 * dbxout.c (dbx_debug_hooks): Adjust.
14425 (xcoff_debug_hooks): Likewise.
14426 * sdbout.c (sdb_debug_hooks): Likewise.
14427 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14428 * dwarf2out.c (macinfo_label_base): New global.
14429 (dwarf2out_register_external_die): New function for the
14430 register_external_die hook.
14431 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14432 (dwarf2_debug_hooks): Use them.
14433 (dwarf2_lineno_debug_hooks): Adjust.
14434 (struct die_struct): Add with_offset flag.
14435 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14436 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14437 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14438 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14439 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14440 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14441 defining section names for the early LTO debug variants.
14442 (reset_indirect_string): New helper.
14443 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14444 (print_dw_val): Add support for offsetted symbol references.
14445 (get_ultimate_context): Split out from is_cxx.
14446 (is_cxx): Use get_ultimate_context.
14447 (is_fortran): Add decl overload.
14448 (compute_comp_unit_symbol): Split out worker from
14449 compute_section_prefix.
14450 (compute_section_prefix): Call compute_comp_unit_symbol and
14451 set comdat_type_p here.
14452 (output_die): Skip DIE symbol output for the LTO added one.
14453 Handle DIE symbol references with offset.
14454 (output_comp_unit): Guard section name mangling properly.
14455 For LTO debug sections emit a symbol at the section beginning
14456 which we use to refer to its DIEs.
14457 (add_abstract_origin_attribute): For DIEs registered via
14458 dwarf2out_register_external_die directly refer to the early
14459 DIE rather than indirectly through the shadow one we created.
14460 Remove obsolete call to dwarf2out_abstract_function for
14461 non-function/block origins.
14462 (gen_array_type_die): When generating early LTO debug do
14463 not emit DW_AT_string_length.
14464 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14465 late when in LTO. As suggested place a gcc_unreachable for
14466 the DECL_ABSTRACT_P case.
14467 (gen_subprogram_die): Avoid another specification DIE
14468 for early built declarations/definitions for the late LTO case.
14469 (gen_variable_die): Add type references for late duplicated VLA dies
14471 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14472 we have the abstract instance already.
14473 (process_scope_var): Adjust decl DIE contexts in LTO which
14474 first puts them in limbo.
14475 (gen_decl_die): Do not generate type DIEs late apart from
14476 types for VLAs or for decls we do not yet have a DIE. Do not
14477 call dwarf2out_abstract_function late.
14478 (dwarf2out_early_global_decl): Make sure to create DIEs
14479 for abstract instances of a decl first.
14480 (dwarf2out_late_global_decl): Adjust comment.
14481 (output_macinfo_op): With multiple macro sections use
14482 macinfo_label_base to distinguish labels.
14483 (output_macinfo): Likewise. Update macinfo_label_base.
14484 Pass in the line info label.
14485 (note_variable_value_in_expr): When generating LTO resolve
14486 all variable values here by generating DIEs as needed.
14487 (init_sections_and_labels): Add early LTO debug flag parameter
14488 and generate different sections and names if set. Add generation
14489 counter for the labels so we can have multiple of them.
14490 (reset_dies): Helper to allow DIEs to be output multiple times.
14491 (dwarf2out_finish): When outputting DIEs to the fat part of an
14492 LTO object first reset DIEs.
14493 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14494 (modified_type_die): Check for decl_ultimate_origin being self
14496 (gen_type_die_with_usage): Likewise.
14497 (gen_typedef_die): Allow decl_ultimate_origin being self.
14498 (set_decl_abstract_flags): Remove.
14499 (set_block_abstract_flags): Likewise.
14500 (dwarf2out_abstract_function): Treat the early generated DIEs
14501 as the abstract copy and only add DW_AT_inline and
14502 DW_AT_artificial here and call set_decl_origin_self.
14503 If the DIE has an abstract origin don't do anything.
14504 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14505 if we have none yet (Go fails to build one, PR78628).
14506 (variably_modified_type_p): Prevent endless recursion for Ada
14507 cyclic pointer types.
14508 * lto-streamer-in.c: Include debug.h.
14509 (dref_queue): New global.
14510 (lto_read_tree_1): Stream in DIE references.
14511 (lto_input_tree): Register DIE references.
14512 (input_function): Stream DECL_DEBUG_ARGS.
14513 * lto-streamer-out.c: Include debug.h.
14514 (lto_write_tree_1): Output DIE references.
14515 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14516 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14517 (output_function): Stream DECL_DEBUG_ARGS.
14518 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14519 Stream DECL_ABSTRACT_ORIGIN.
14520 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14521 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14522 DECL_CONTEXT for file-scope decls.
14523 * lto-streamer.h (struct dref_entry): Declare.
14524 (dref_queue): Likewise.
14525 * cfgexpand.c (pass_expand::execute): Do not call the
14526 outlining_inline_function hook here.
14527 * lto-wrapper.c (debug_obj): New global.
14528 (tool_cleanup): Unlink it if required.
14529 (debug_objcopy): New function.
14530 (run_gcc): Handle early debug sections in the IL files by
14531 extracting them to separate files, partially linkin them and
14532 feeding the result back as result to the linker.
14533 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14534 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14535 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14536 sections into a separate segment.
14537 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14539 (darwin_asm_dwarf_section): Likewise.
14540 (darwin_asm_output_dwarf_offset): Likewise.
14541 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14543 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14544 Alan Hayward <alan.hayward@arm.com>
14545 David Sherwood <david.sherwood@arm.com>
14547 * read-md.h (md_reader::record_potential_iterator_use): Replace
14548 pointer argument with an rtx and an index.
14549 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14550 (apply_mode_iterator): Likewise.
14551 (apply_code_iterator): Likewise.
14552 (apply_int_iterator): Likewise.
14553 (apply_subst_iterator): Likewise.
14554 (record_iterator_use): Likewise.
14555 (record_attribute_use): Likewise.
14556 (md_reader::record_potential_iterator_use): Likewise. Update calls
14557 to record_iterator_use and apply_iterator.
14558 (iterator_use): Replace ptr with x and index.
14559 (attribute_use): Likewise.
14560 (apply_attribute_uses): Update calls to apply_iterator.
14561 (apply_iterators): Likewise. Update initialization of iterator_use.
14562 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14563 and record_potential_iterator_use.
14564 (rtx_reader::read_rtx_operand): Likewise.
14566 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14567 Alan Hayward <alan.hayward@arm.com>
14568 David Sherwood <david.sherwood@arm.com>
14570 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14573 2017-08-21 Richard Biener <rguenther@suse.de>
14575 PR middle-end/81884
14576 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14577 at struct end conservatively when comparing common bases.
14579 2017-08-21 Richard Biener <rguenther@suse.de>
14581 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14582 (mem_ref_in_stmt): Remove.
14583 (determine_max_movement): Use ref index to get at the reference.
14584 (invariantness_dom_walker::before_dom_children): Deal with
14585 lim data already initialized.
14586 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14588 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14590 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14591 (TARGET_ISA_ROUND): Ditto.
14592 (TARGET_ROUND): Ditto.
14593 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14594 * config/i386/i386.md: Ditto.
14595 * config/i386/sse.md: Ditto.
14596 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14597 with OPTION_MASK_ISA_SSE4_1.
14599 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14602 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14603 __builtin_ia32_lzcnt_u16.
14605 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14608 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14609 (rs6000_set_current_function): Rewrite function to use it.
14611 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14614 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14615 and TYPE_WARN_IF_NOT_ALIGN.
14616 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14617 (handle_warn_if_not_align): New.
14618 (place_union_field): Call handle_warn_if_not_align.
14619 (place_field): Call handle_warn_if_not_align.
14620 Copy TYPE_WARN_IF_NOT_ALIGN.
14621 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14622 (layout_type): Likewise.
14623 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14625 (tree_decl_common): Add warn_if_not_align.
14626 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14627 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14628 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14629 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14630 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14631 * doc/extend.texi: Document warn_if_not_aligned attribute.
14632 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14634 2017-08-17 Martin Liska <mliska@suse.cz>
14637 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14638 (get_data_dependence): Use it as pointer type.
14639 (distribute_loop): Likewise.
14641 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14643 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14644 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14645 (p8_vmrgow_<mode>_direct): New define_insn.
14646 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14647 handle endianness for vmrgew and vmrgow permute patterns.
14649 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14651 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14652 * config/rs6000/rs6000-cpus.def: Remove comment.
14653 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14654 (POWERPC_MASKS): Likewise.
14655 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14656 use of TARGET_VSX_TIMODE.
14657 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14658 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14659 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14660 (rs6000_option_override_internal): Remove dead code.
14661 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14662 (rs6000_legitimize_reload_address): Likewise.
14663 (rs6000_legitimate_address_p): Likewise.
14664 (rs6000_opt_masks): Delete "vsx-timode".
14665 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14666 from function comment.
14667 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14668 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14669 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14671 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14672 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14673 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14674 (VSX_TI): Likewise.
14676 (define_peephole2): Likewise.
14678 2017-08-17 Martin Sebor <msebor@redhat.com>
14681 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14682 past the end of an array.
14683 (test_pp_format): Add test cases.
14685 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14687 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14688 missing ECF_NOTHROW flags.
14690 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14693 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14694 operands residing in integer registers.
14695 (*vsx_le_perm_load_<mode>): Likewise.
14696 (*vsx_le_perm_store_<mode>): Likewise.
14697 (define_peephole2): Add peepholes to optimize the above.
14699 2017-08-17 Marek Polacek <polacek@redhat.com>
14701 PR middle-end/81814
14702 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14703 to mimic what shorten_compare did. Change the return type to bool.
14704 (fold_cond_expr_with_comparison): Update call to
14705 operand_equal_for_comparison_p.
14706 (fold_ternary_loc): Likewise.
14708 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14710 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14712 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14714 2017-08-17 Richard Biener <rguenther@suse.de>
14716 * tree-ssa-structalias.c (solve_graph): When propagating
14717 to successors update the graphs succ edges and avoid duplicate work.
14719 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14722 * config/i386/i386.c (ix86_option_override_internal): Save target
14723 specific options after ix86_stack_protector_guard_reg was changed.
14725 2017-08-17 Richard Biener <rguenther@suse.de>
14727 PR tree-optimization/81827
14728 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14729 (new_var_info): Initialize it conservatively.
14730 (get_call_vi): Mark register vars.
14731 (new_scalar_tmp_constraint_exp): Likewise.
14732 (handle_rhs_call): Likewise.
14733 (handle_const_call): Likewise.
14734 (create_function_info_for): Likewise.
14735 (solve_constraints): Sort varinfos to separate register from
14736 non-register vars to pack points-to solution bitmaps during
14739 2017-08-17 Marek Polacek <polacek@redhat.com>
14741 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14743 2017-08-17 Richard Biener <rguenther@suse.de>
14745 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14746 to true when overflow is undefined and we saturated the result.
14748 2017-08-17 Alan Modra <amodra@gmail.com>
14751 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14752 Don't use store multiple if only one reg needs saving.
14753 (interesting_frame_related_regno): New function.
14754 (rs6000_frame_related): Don't emit frame info for regs that
14756 (rs6000_emit_epilogue): Likewise.
14758 2017-08-16 Nathan Sidwell <nathan@acm.org>
14760 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14761 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14762 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14763 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14764 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14765 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14766 (verify_type): Adjust for TYPE_BINFO move.
14767 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14768 process TYPE_BINFO directly.
14769 (hash_tree): Likewise.
14770 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14772 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14775 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14777 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14779 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14782 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14783 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14784 Add ix86_binary_operator_ok to insn constraint.
14785 (*iordi_1_bts): Ditto.
14786 (*xordi_1_btc): Ditto.
14787 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14788 Update corresponding peephole2 pattern.
14792 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14794 PR tree-optimization/81832
14795 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14796 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14798 2017-08-16 Marek Polacek <polacek@redhat.com>
14801 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14802 regarding size_zero_node.
14804 2017-08-16 Martin Liska <mliska@suse.cz>
14807 * config.gcc: Respect previously set extra_objs in case
14810 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14812 PR tree-optimization/81835
14813 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14814 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14815 not depend on the phi.
14817 2017-08-16 Alan Modra <amodra@gmail.com>
14819 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14822 2017-08-16 Alan Modra <amodra@gmail.com>
14824 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14825 (save_reg_p): ..into this. Update all callers.
14826 (first_reg_to_save): Simplify.
14828 2017-08-16 Alan Modra <amodra@gmail.com>
14830 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14833 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14837 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14838 constant count if that count is less than 32.
14840 2017-08-15 Nathan Sidwell <nathan@acm.org>
14842 * gcc.c (execute): Emit friendlier message if inferior is killed
14843 by an external cause.
14845 2017-08-15 Richard Biener <rguenther@suse.de>
14847 PR tree-optimization/81790
14848 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14849 CONSTRUCTORs from simplifying and VN.
14851 2017-08-14 Martin Sebor <msebor@redhat.com>
14853 * builtin-attrs.def: Add comments.
14855 2017-08-14 Martin Sebor <msebor@redhat.com>
14858 * doc/extend.texi (attribute nonstring): Document new attribute.
14860 2017-08-14 Martin Sebor <msebor@redhat.com>
14863 * tree-diagnostic.c (default_tree_printer): Handle %G.
14864 * gimple-pretty-print.h (percent_G_format): Declare new function.
14865 * gimple-pretty-print.c (percent_G_format): Define.
14866 * tree-pretty-print.c (percent_K_format): Add argument.
14868 2017-08-14 Martin Sebor <msebor@redhat.com>
14870 PR translation/79998
14871 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14872 Remove a stray space.
14874 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14877 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14878 (*iordi_1_bts): Ditto.
14879 (*xordi_1_btc): Ditto.
14881 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14884 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14885 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14887 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14888 quoted strings, and make more translator-friendly.
14889 (darwin_rs6000_override_options): Likewise.
14890 (rs6000_option_override_internal): Likewise.
14891 (rs6000_return_in_memory): Fix overlong line.
14892 (init_cmulative_args): Use quoted strings, and make more
14893 translator-friendly.
14894 (rs6000_pass_by_reference): Fix overlong line.
14895 (def_builtin): Use quoted strings.
14896 (altivec_expand_predicate_builtin): Use quoted strings, and make
14897 more translator-friendly.
14898 (htm_expand_builtin): Use quoted strings.
14899 (cpu_expand_builtin): Use quoted strings, and make more
14900 translator-friendly.
14901 (altivec_expand_builtin): Likewise.
14902 (paired_expand_predicate_builtin): Likewise.
14903 (rs6000_invalid_builtin): Likewise.
14904 (builtin_function_type): Use quoted strings.
14905 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14906 more translator-friendly.
14907 (rs6000_trampoline_init): Likewise.
14908 (rs6000_handle_altivec_attribute): Likewise.
14909 (rs6000_inner_target_options): Use quoted strings.
14910 (rs6000_disable_incompatible_switches): Likewise.
14911 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14912 strings, and make more translator-friendly.
14913 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14915 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14917 PR tree-optimization/81799
14918 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14919 cond_expr to simple gimple operand.
14921 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14923 PR middle-end/46932
14924 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14926 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14930 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14931 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14932 TARGET_GASISR_PROLOGUES.
14933 * config/avr/avr.c (avr_option_override): Same.
14934 (avr_pass_pre_proep::execute): Same.
14936 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14939 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14940 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14942 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14944 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14945 define_insn_and_split. Split to a memory load from 0 in
14946 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14947 using PTR mode iterator.
14948 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14949 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14950 (*add_tp_<mode>): Redefine as define_insn_and_split.
14951 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14952 address space. Merge with *add_tp_x32 using PTR mode iterator.
14953 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14954 Split to an add with a memory load from 0 in
14955 DEFAULT_TLS_SEG_REG address space.
14957 2017-08-12 Andrew Pinski <apinski@cavium.com>
14959 * config/aarch64/aarch64-option-extensions.def (rdma):
14960 Fix feature string to what Linux prints out in /proc/cpuinfo.
14962 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14965 * dwarf2out.c (modified_type_die): For C typedef types that have
14966 an ultimate origin, process the ultimate origin instead of the
14968 (gen_typedef_die): Assert that input DECLs have no ultimate
14970 (gen_type_die_with_usage): For typedef variants that have an
14971 ultimate origin, just call gen_decl_die on the original DECL.
14972 (process_scope_var): Avoid creating DIEs for local typedefs and
14973 concrete static variables.
14975 2017-08-12 Alan Modra <amodra@gmail.com>
14979 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14980 match gnu-user.h startfile.
14981 (ENDFILE_LINUX_SPEC): Similarly.
14983 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14986 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14988 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14990 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14991 * config/aarch64/aarch64.md (mov<mode>): Change.
14992 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14993 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14994 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14996 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14998 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14999 for storage order barriers.
15001 2017-08-11 Martin Liska <mliska@suse.cz>
15004 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
15005 variables of void type.
15007 2017-08-11 Martin Liska <mliska@suse.cz>
15009 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
15010 TARGET_SUPPORTS_ALIASES.
15011 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
15012 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
15013 (optimize_weakref): Likewise.
15014 * symtab.c (symtab_node::noninterposable_alias): Likewise.
15015 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
15016 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
15018 2017-08-11 Martin Liska <mliska@suse.cz>
15021 * config/i386/i386.c (make_resolver_func): Do complete
15022 refactoring of the function.
15024 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15027 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
15028 * config/i386/i386.c (ix86_stack_protect_guard): Use
15029 ix86_stack_protect_guard_symbol_str to generate varible declaration.
15030 * doc/invoke.texi (x86 Options): Document
15031 -mstack-protector-guard-symbol= option.
15033 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
15035 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
15036 * config/i386/i386.c (ix86_split_stack_guard): New function.
15037 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
15038 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
15039 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
15040 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
15041 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
15042 (split_stack_space_check): Call ix86_split_stack_guard.
15044 2017-08-10 Martin Sebor <msebor@redhat.com>
15046 * print-tree.c (print_node): Print location using the established
15048 Replace spaces with colons.
15049 (debug_raw, debug): Ditto.
15051 2017-08-10 Martin Sebor <msebor@redhat.com>
15054 * pretty-print.c (pp_format): Correct the handling of %s precision.
15056 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
15059 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15061 (ix86_finalize_stack_frame_flags): This. Also clear
15062 frame_pointer_needed if -fno-omit-frame-pointer is used without
15064 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15065 with ix86_finalize_stack_frame_flags.
15066 (ix86_expand_epilogue): Likewise.
15067 (ix86_expand_split_stack_prologue): Likewise.
15068 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
15070 2017-08-10 Martin Liska <mliska@suse.cz>
15073 * c-attribs.c (handle_target_attribute):
15074 Report warning for an empty string argument of target attribute.
15076 2017-08-09 Jakub Jelinek <jakub@redhat.com>
15079 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
15081 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
15082 or DECL_NONLOCAL labels.
15083 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
15084 or DECL_NONLOCAL labels here.
15086 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
15088 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
15089 to indicate when early gimple folding has been disabled.
15090 (rs6000_gimple_fold_builtin): Add debug content.
15091 (rs6000_invalid_builtin): Fix whitespace.
15092 (rs6000_expand_builtin): Fix whitespace.
15093 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
15095 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
15098 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
15099 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15101 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15103 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15104 (qdf24xx): Likewise.
15105 * config/aarch64/aarch64.md: Include falkor.md.
15106 * config/aarch64/falkor.md: New.
15108 2017-08-09 Marek Polacek <polacek@redhat.com>
15111 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15112 * diagnostic.c (emit_diagnostic): Add a comment.
15113 (emit_diagnostic_valist): New function.
15115 2017-08-09 Marek Polacek <polacek@redhat.com>
15118 * input.c (make_location): New overload.
15119 * input.h (make_location): Declare.
15121 2017-08-08 Alan Modra <amodra@gmail.com>
15122 H.J. Lu <hongjiu.lu@intel.com>
15125 * gcc.c (NO_PIE_SPEC): Delete.
15126 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15128 (LINK_PIE_SPEC): ..to here.
15129 (LINK_COMMAND_SPEC): Support -no-pie.
15130 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15131 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15133 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15134 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15135 (ENDFILE_CRTEND_SPEC): Similarly.
15137 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15140 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15141 (mstack-protector-guard-offset=): Ditto.
15142 * config/i386/i386.c (ix86_option_override): Handle
15143 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15145 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15146 ix86_stack_protect_guard_offset variables.
15147 (TARGET_STACK_PROTECT_GUARD): Always define.
15148 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15149 and -mstack-protector-guard-offset= options.
15151 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15153 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15154 boundary case for the last candidate.
15156 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15158 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15159 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15161 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15163 PR middle-end/19706
15164 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15165 * config/aarch64/aarch64-builtins.c
15166 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15167 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15168 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15170 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15171 Andrew Pinski <pinskia@gmail.com>
15173 PR middle-end/19706
15174 * internal-fn.def (XORSIGN): New.
15175 * optabs.def (xorsign_optab): New.
15176 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15177 (convert_expand_mult_copysign): New.
15178 (pass_optimize_widening_mul::execute): Call
15179 convert_expand_mult_copysign.
15181 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15183 PR tree-optimization/81354
15184 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15185 Insert on edges rather than explicitly creating landing pads.
15186 (analyze_candidates_and_replace): Commit edge inserts.
15188 2017-08-08 Richard Biener <rguenther@suse.de>
15190 PR middle-end/81719
15191 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15192 (expand_simple_operations): Also look through ADDR_EXPRs with
15193 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15195 2017-08-08 Richard Biener <rguenther@suse.de>
15197 PR tree-optimization/81723
15198 * tree-vect-slp.c (struct bst_traits): New hash traits.
15199 (bst_fail): New global.
15200 (vect_build_slp_tree_2): New worker, split out from ...
15201 (vect_build_slp_tree): ... this now wrapping it with using
15202 bst_fail set to cache SLP tree build fails. Properly handle
15204 (vect_analyze_slp_instance): Allocate and free bst_fail.
15206 2017-08-08 Martin Liska <mliska@suse.cz>
15209 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15210 LABEL_DECLs that can be from a different function.
15212 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15214 PR tree-optimization/81744
15215 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15216 loop's number of iterations.
15218 2017-08-08 Martin Liska <mliska@suse.cz>
15220 * asan.c: Include header files.
15221 * attribs.c (build_decl_attribute_variant): New function moved
15223 (build_type_attribute_qual_variant): Likewise.
15224 (cmp_attrib_identifiers): Likewise.
15225 (simple_cst_list_equal): Likewise.
15226 (omp_declare_simd_clauses_equal): Likewise.
15227 (attribute_value_equal): Likewise.
15228 (comp_type_attributes): Likewise.
15229 (build_type_attribute_variant): Likewise.
15230 (lookup_ident_attribute): Likewise.
15231 (remove_attribute): Likewise.
15232 (merge_attributes): Likewise.
15233 (merge_type_attributes): Likewise.
15234 (merge_decl_attributes): Likewise.
15235 (merge_dllimport_decl_attributes): Likewise.
15236 (handle_dll_attribute): Likewise.
15237 (attribute_list_equal): Likewise.
15238 (attribute_list_contained): Likewise.
15239 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15240 (lookup_attribute_by_prefix): Likewise.
15241 * bb-reorder.c: Include header files.
15242 * builtins.c: Likewise.
15243 * calls.c: Likewise.
15244 * cfgexpand.c: Likewise.
15245 * cgraph.c: Likewise.
15246 * cgraphunit.c: Likewise.
15247 * convert.c: Likewise.
15248 * dwarf2out.c: Likewise.
15249 * final.c: Likewise.
15250 * fold-const.c: Likewise.
15251 * function.c: Likewise.
15252 * gimple-expr.c: Likewise.
15253 * gimple-fold.c: Likewise.
15254 * gimple-pretty-print.c: Likewise.
15255 * gimple.c: Likewise.
15256 * gimplify.c: Likewise.
15257 * hsa-common.c: Likewise.
15258 * hsa-gen.c: Likewise.
15259 * internal-fn.c: Likewise.
15260 * ipa-chkp.c: Likewise.
15261 * ipa-cp.c: Likewise.
15262 * ipa-devirt.c: Likewise.
15263 * ipa-fnsummary.c: Likewise.
15264 * ipa-inline.c: Likewise.
15265 * ipa-visibility.c: Likewise.
15267 * lto-cgraph.c: Likewise.
15268 * omp-expand.c: Likewise.
15269 * omp-general.c: Likewise.
15270 * omp-low.c: Likewise.
15271 * omp-offload.c: Likewise.
15272 * omp-simd-clone.c: Likewise.
15273 * opts-global.c: Likewise.
15274 * passes.c: Likewise.
15275 * predict.c: Likewise.
15276 * sancov.c: Likewise.
15277 * sanopt.c: Likewise.
15278 * symtab.c: Likewise.
15279 * toplev.c: Likewise.
15280 * trans-mem.c: Likewise.
15281 * tree-chkp.c: Likewise.
15282 * tree-eh.c: Likewise.
15283 * tree-into-ssa.c: Likewise.
15284 * tree-object-size.c: Likewise.
15285 * tree-parloops.c: Likewise.
15286 * tree-profile.c: Likewise.
15287 * tree-ssa-ccp.c: Likewise.
15288 * tree-ssa-live.c: Likewise.
15289 * tree-ssa-loop.c: Likewise.
15290 * tree-ssa-sccvn.c: Likewise.
15291 * tree-ssa-structalias.c: Likewise.
15292 * tree-ssa.c: Likewise.
15293 * tree-streamer-in.c: Likewise.
15294 * tree-vectorizer.c: Likewise.
15295 * tree-vrp.c: Likewise.
15296 * tsan.c: Likewise.
15297 * ubsan.c: Likewise.
15298 * varasm.c: Likewise.
15299 * varpool.c: Likewise.
15300 * tree.c: Remove functions moved to attribs.[ch].
15301 * tree.h: Likewise.
15302 * config/aarch64/aarch64.c: Add attrs.h header file.
15303 * config/alpha/alpha.c: Likewise.
15304 * config/arc/arc.c: Likewise.
15305 * config/arm/arm.c: Likewise.
15306 * config/avr/avr.c: Likewise.
15307 * config/bfin/bfin.c: Likewise.
15308 * config/c6x/c6x.c: Likewise.
15309 * config/cr16/cr16.c: Likewise.
15310 * config/cris/cris.c: Likewise.
15311 * config/darwin.c: Likewise.
15312 * config/epiphany/epiphany.c: Likewise.
15313 * config/fr30/fr30.c: Likewise.
15314 * config/frv/frv.c: Likewise.
15315 * config/ft32/ft32.c: Likewise.
15316 * config/h8300/h8300.c: Likewise.
15317 * config/i386/winnt.c: Likewise.
15318 * config/ia64/ia64.c: Likewise.
15319 * config/iq2000/iq2000.c: Likewise.
15320 * config/lm32/lm32.c: Likewise.
15321 * config/m32c/m32c.c: Likewise.
15322 * config/m32r/m32r.c: Likewise.
15323 * config/m68k/m68k.c: Likewise.
15324 * config/mcore/mcore.c: Likewise.
15325 * config/microblaze/microblaze.c: Likewise.
15326 * config/mips/mips.c: Likewise.
15327 * config/mmix/mmix.c: Likewise.
15328 * config/mn10300/mn10300.c: Likewise.
15329 * config/moxie/moxie.c: Likewise.
15330 * config/msp430/msp430.c: Likewise.
15331 * config/nds32/nds32-isr.c: Likewise.
15332 * config/nds32/nds32.c: Likewise.
15333 * config/nios2/nios2.c: Likewise.
15334 * config/nvptx/nvptx.c: Likewise.
15335 * config/pa/pa.c: Likewise.
15336 * config/pdp11/pdp11.c: Likewise.
15337 * config/powerpcspe/powerpcspe.c: Likewise.
15338 * config/riscv/riscv.c: Likewise.
15339 * config/rl78/rl78.c: Likewise.
15340 * config/rx/rx.c: Likewise.
15341 * config/s390/s390.c: Likewise.
15342 * config/sh/sh.c: Likewise.
15343 * config/sol2.c: Likewise.
15344 * config/sparc/sparc.c: Likewise.
15345 * config/spu/spu.c: Likewise.
15346 * config/stormy16/stormy16.c: Likewise.
15347 * config/tilegx/tilegx.c: Likewise.
15348 * config/tilepro/tilepro.c: Likewise.
15349 * config/v850/v850.c: Likewise.
15350 * config/vax/vax.c: Likewise.
15351 * config/visium/visium.c: Likewise.
15352 * config/xtensa/xtensa.c: Likewise.
15354 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15357 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15358 constraints since the -mupper-regs-* switches have been
15360 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15361 into a vector from a double word element that was extracted from
15362 another vector, and eliminate extra XXPERMDI instructions.
15363 (vsx_concat_<mode>_2): Likewise.
15364 (vsx_concat_<mode>_3): Likewise.
15365 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15366 concat to allow optimizing inserts from previous extracts.
15368 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15370 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15371 memory reference to a SSP offset in TLS address space.
15372 (ix86_print_operand) <case '@'>: Remove.
15373 (ix86_print_operand_punct_valid_p): Remove '@' code.
15374 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15375 UNSPEC_SP_TLS_TEST.
15376 (stack_tls_protect_set_<mode>): Remove.
15377 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15378 (stack_tls_protect_test_<mode>): Remove.
15379 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15381 2017-08-07 Olivier Hainque <hainque@adacore.com>
15384 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15386 2017-08-07 Douglas Rupp <rupp@adacore.com>
15388 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15389 variable was referenced as multidir in command.
15391 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15394 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15397 2017-08-07 Martin Liska <mliska@suse.cz>
15399 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15400 * config/rl78/rl78.c: Add include of attribs.h.
15401 * config/sh/sh.c: Likewise.
15402 * config/v850/v850.c: Likewise.
15404 2017-08-07 Tom de Vries <tom@codesourcery.com>
15406 PR middle-end/78266
15407 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15409 2017-08-07 Martin Liska <mliska@suse.cz>
15411 * config/mips/mips.c: Include attribs.h.
15413 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15416 * doc/invoke.texi: Document change in behvaior for -Ofast for
15419 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15421 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15423 (aarch64_pop_regs): Likewise.
15424 (aarch64_gen_load_pair): Likewise.
15425 (aarch64_save_callee_saves): Likewise.
15426 (aarch64_restore_callee_saves): Likewise.
15428 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15430 * config/i386/i386.c: Revert the last change.
15432 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15435 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15437 (ix86_finalize_stack_frame_flags): This. Also clear
15438 frame_pointer_needed if -fno-omit-frame-pointer is used without
15440 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15441 with ix86_finalize_stack_frame_flags.
15442 (ix86_expand_epilogue): Likewise.
15443 (ix86_expand_split_stack_prologue): Likewise.
15445 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15448 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15449 to P_AES for Westmere.
15451 2017-08-07 Jonathan Yong <10walls@gmail.com>
15453 * config/i386/mingw.opt (fset-stack-executable): Removed.
15454 * config/i386/cygming.opt (fset-stack-executable): Moved
15456 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15458 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15460 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15462 2017-08-07 Marek Polacek <polacek@redhat.com>
15464 PR middle-end/81737
15465 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15467 2017-08-07 Martin Liska <mliska@suse.cz>
15469 * attribs.h (canonicalize_attr_name): New function.
15470 (cmp_attribs): Move from c-format.c and adjusted.
15471 (is_attribute_p): Moved from tree.h.
15472 * tree-inline.c: Add new includes.
15473 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15474 (private_is_attribute_p): Remove.
15475 (private_lookup_attribute): Likewise.
15476 (private_lookup_attribute_by_prefix): Simplify.
15477 (remove_attribute): Use is_attribute_p.
15478 * tree.h: Remove removed declarations.
15480 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15482 PR middle-end/81698
15483 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15484 instead of computing it in the function. Formatting fix.
15485 (expand_case): Don't rely on default_edge being the first edge,
15486 clear it if removing it, pass default_edge to
15487 emit_case_dispatch_table.
15488 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15491 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15493 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15494 insn in the function, emit NOP after the insn.
15496 2017-08-06 Tom de Vries <tom@codesourcery.com>
15498 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15501 2017-08-06 Tom de Vries <tom@codesourcery.com>
15503 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15506 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15508 PR tree-optimization/57371
15509 * match.pd: New pattern.
15511 2017-08-04 Marek Polacek <polacek@redhat.com>
15513 PR middle-end/81695
15514 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15515 perform the computation in offset_int.
15517 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15519 PR tree-optimization/81136
15520 * tree-vectorizer.h: Include tree-hash-traits.h.
15521 (vec_base_alignments): New typedef.
15522 (vec_info): Add a base_alignments field.
15523 (vect_record_base_alignments): Declare.
15524 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15526 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15527 (create_data_ref): Add an is_conditional_in_stmt argument.
15528 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15529 the is_conditional_in_stmt field.
15530 (data_ref_loc): Add an is_conditional_in_stmt field.
15531 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15532 (find_data_references_in_stmt): Update call to create_data_ref.
15533 (graphite_find_data_references_in_stmt): Likewise.
15534 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15535 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15536 (vect_record_base_alignment): New function.
15537 (vect_record_base_alignments): Likewise.
15538 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15539 for nested statements even if we fail to compute a misalignment.
15540 Use pooled base alignments for unconditional references.
15541 (vect_find_same_alignment_drs): Compare base addresses instead
15543 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15544 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15546 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15548 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15549 Add an explicit name for the enum. Use auto_vec for slp_instances
15550 and grouped_stores.
15551 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15553 (_bb_vec_info): Add a constructor and destructor.
15554 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15555 (destroy_loop_vec_info): Delete.
15556 (vect_destroy_datarefs): Likewise.
15557 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15558 (vec_info::vec_info): New function.
15559 (vec_info::~vec_info): Likewise.
15560 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15561 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15562 destroy_loop_vec_info.
15563 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15564 (_loop_vec_info::_loop_vec_info): ...this.
15565 (destroy_loop_vec_info): Replace with...
15566 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15567 the stmt_vec_infos. Leave handling of vec_info information to its
15568 destructor. Remove explicit vector releases.
15569 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15570 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15571 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15572 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15573 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15574 (destroy_bb_vec_info): Replace with...
15575 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15576 information to its destructor.
15577 (vect_slp_analyze_bb_1): Use new and delete instead of
15578 new_bb_vec_info and destroy_bb_vec_info.
15579 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15582 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15584 * tree-data-ref.h (subscript): Add access_fn field.
15585 (data_dependence_relation): Add could_be_independent_p.
15586 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15587 (same_access_functions): Move to tree-data-ref.c.
15588 * tree-data-ref.c (ref_contains_union_access_p): New function.
15589 (access_fn_component_p): Likewise.
15590 (access_fn_components_comparable_p): Likewise.
15591 (dr_analyze_indices): Add a reference to access_fn_component_p.
15592 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15594 (constant_access_functions): Likewise.
15595 (add_other_self_distances): Likewise.
15596 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15597 (initialize_data_dependence_relation): Use XCNEW and remove
15598 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15599 of access functions that have the same type. Allow the
15600 subsequence to end with different bases in some circumstances.
15601 Record the chosen access functions in SUB_ACCESS_FN.
15602 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15603 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15604 (subscript_dependence_tester_1): Likewise dra and drb.
15605 (build_classic_dist_vector): Update calls accordingly.
15606 (subscript_dependence_tester): Likewise.
15607 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15608 DDR_COULD_BE_INDEPENDENT_P.
15609 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15610 comp_alias_ddrs instead of may_alias_ddrs.
15611 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15613 (vect_analyze_data_ref_dependence): Use it if
15614 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15615 distance vectors if that fails.
15616 (dependence_distance_ge_vf): New function.
15617 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15618 LOOP_VINFO_MAY_ALIAS_DDRS.
15620 2017-08-04 Richard Biener <rguenther@suse.de>
15622 PR middle-end/81705
15623 * fold-const.c (fold_binary_loc): Properly restrict
15624 minus_var0 && minus_var1 case when associating undefined overflow
15627 2017-08-04 Tom de Vries <tom@codesourcery.com>
15629 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15631 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15633 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15634 Don't start diagnostic messages with a capital letter.
15635 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15637 (rs6000_invalid_builtin): Likewise.
15638 (rs6000_trampoline_init): Likewise.
15640 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15643 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15644 after setting changeable df flags.
15646 2017-08-03 Richard Biener <rguenther@suse.de>
15648 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15649 up if the use is in USE - X.
15651 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15653 * toplev.c (dumpfile.h): New include.
15654 (internal_error_reentered): New static function. Use it...
15655 (internal_error_function): ...here to handle reentered internal_error.
15657 2017-08-03 Richard Biener <rguenther@suse.de>
15659 PR middle-end/81148
15660 * fold-const.c (split_tree): Add minus_var and minus_con
15661 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15662 here but always use minus_*.
15663 (associate_trees): Assert we never associate with MINUS_EXPR
15664 and NULL first operand. Do not recurse for PLUS_EXPR operands
15665 when associating as MINUS_EXPR either.
15666 (fold_binary_loc): Track minus_var and minus_con.
15668 2017-08-03 Tom de Vries <tom@codesourcery.com>
15671 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15672 ACCEL_COMPILER, apply finish_options on
15673 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15675 2017-08-03 Tom de Vries <tom@codesourcery.com>
15678 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15679 function_entry_patch_area_size > 0.
15681 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15684 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15685 instead of 10??LU, perform unit multiplication in wide_int,
15686 don't change alloc_object_size_limit if the limit is larger
15689 PR tree-optimization/81655
15690 PR tree-optimization/81588
15691 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15692 the case when ranges[i].low and high are 1 for unsigned type with
15695 PR middle-end/81052
15696 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15697 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15699 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15701 * tree-vrp.h: Add include guard.
15703 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15706 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15707 (ud2): New insn pattern.
15708 * config/i386/i386.c (ix86_expand_epilogue):
15709 For naked functions, generate ud2 instead of trap insn.
15711 2017-08-02 Marek Polacek <polacek@redhat.com>
15714 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15716 2017-08-02 Tom de Vries <tom@codesourcery.com>
15717 Cesar Philippidis <cesar@codesourcery.com>
15719 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15720 Add missing edge probabilities.
15722 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15724 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15725 Correct endianness.
15727 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15729 PR middle-end/79499
15730 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15731 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15732 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15733 prologue_seq sequences - if any.
15735 2017-08-02 Richard Biener <rguenther@suse.de>
15737 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15738 via vectors if supported, integer extracts via punning if supported
15739 or otherwise vector extracts.
15741 2017-08-02 Richard Biener <rguenther@suse.de>
15743 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15745 (bitmap_insert_into_set): ... this.
15747 2017-08-02 Richard Biener <rguenther@suse.de>
15749 PR tree-optimization/81633
15751 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15753 PR tree-optimization/71752
15754 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15756 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15758 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15759 (machine_function::call_ms2sysv_pad_out): Remove field.
15760 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15761 (ix86_compute_frame_layout): Likewise.
15763 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15766 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15767 attribute with interrupt attribute.
15769 2017-08-01 Andrew Pinski <apinski@cavium.com>
15771 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15772 BIT_INSERT_EXPR's operand 1
15773 to see if the types precision matches.
15775 2017-08-01 Martin Liska <mliska@suse.cz>
15777 PR middle-end/70140
15778 * builtins.c (expand_builtin_memcpy_args): Remove.
15779 (expand_builtin_memcpy): Call newly added function
15780 expand_builtin_memory_copy_args.
15781 (expand_builtin_memcpy_with_bounds): Likewise.
15782 (expand_builtin_mempcpy): Remove last argument.
15783 (expand_builtin_mempcpy_with_bounds): Likewise.
15784 (expand_builtin_memory_copy_args): New function created from
15785 expand_builtin_mempcpy_args with small modifications.
15786 (expand_builtin_mempcpy_args): Remove.
15787 (expand_builtin_stpcpy): Remove unused argument.
15788 (expand_builtin): Likewise.
15789 (expand_builtin_with_bounds): Likewise.
15791 2017-08-01 Martin Liska <mliska@suse.cz>
15794 Make mempcpy more optimal (PR middle-end/70140).
15796 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15799 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15800 __builtin_vec_cmpne verify both arguments are compatible vectors
15801 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15802 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15803 move computation of aligned to after checking the argument types.
15807 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15808 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15811 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15812 Doug Rupp <rupp@adacore.com>
15813 Olivier Hainque <hainque@adacore.com>
15815 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15816 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15818 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15819 for TARGET_OS_CPP_BUILTIN.
15820 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15821 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15823 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15824 passing required abi options to the assembler for EABI configurations.
15825 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15826 of .text.hot and .text.unlikely sections for kernel modules when
15827 using ARM style exceptions.
15828 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15829 options. Add EXTRA_CC1_SPEC.
15830 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15832 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15834 (ARM_TARGET2_DWARF_FORMAT): Define.
15835 * config/arm/t-vxworks: Adjust multilib control to removal of the
15836 Diab command line options.
15838 2017-08-01 Martin Liska <mliska@suse.cz>
15840 PR gcov-profile/81561
15841 * gcov.c (unblock): Make unblocking safe as we need to preserve
15842 index correspondence of blocks and block_lists.
15844 2017-08-01 Richard Biener <rguenther@suse.de>
15846 PR tree-optimization/81181
15847 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15848 (compute_antic): ... end of iteration here.
15850 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15852 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15853 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15854 (ftree-slp-vectorize): Likewise.
15855 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15856 can no longer be set independent of flag_tree_loop_vectorize.
15857 * omp-general.c (emp_max_vf): Likewise.
15858 * opts.c (enable_fdo_optimizations): Remove references to
15859 flag_tree_vectorize, these are now implicit.
15860 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15861 and leave it for the options machinery.
15863 2017-08-01 Martin Liska <mliska@suse.cz>
15865 PR middle-end/70140
15866 * builtins.c (expand_builtin_memcpy_args): Remove.
15867 (expand_builtin_memcpy): Call newly added function
15868 expand_builtin_memory_copy_args.
15869 (expand_builtin_memcpy_with_bounds): Likewise.
15870 (expand_builtin_mempcpy): Remove last argument.
15871 (expand_builtin_mempcpy_with_bounds): Likewise.
15872 (expand_builtin_memory_copy_args): New function created from
15873 expand_builtin_mempcpy_args with small modifications.
15874 (expand_builtin_mempcpy_args): Remove.
15875 (expand_builtin_stpcpy): Remove unused argument.
15876 (expand_builtin): Likewise.
15877 (expand_builtin_with_bounds): Likewise.
15879 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15882 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15883 print "ds:" only for immediates in generic address space.
15885 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15888 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15889 (ix86_function_ok_for_sibcall): Return false for naked functions.
15891 2017-08-01 Richard Biener <rguenther@suse.de>
15893 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15894 (compute_antic): Seed worklist with exit block predecessors.
15895 * cfganal.c (dfs_find_deadend): For a cycle return the source
15896 of the edge closing it.
15898 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15900 * config/aarch64/aarch64.c
15901 (aarch64_can_const_movi_rtx_p): Move 0 check.
15903 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15905 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15906 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15908 * match.pd: Ditto in address comparison pattern.
15910 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15912 PR tree-optimization/81627
15913 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15914 closed ssa form for store-store chain.
15916 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15918 PR tree-optimization/81620
15919 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15920 for store-store chain.
15922 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15924 PR tree-optimization/81588
15925 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15926 ranges[i].in_p, invert comparison code ccode. For >/>=,
15927 swap rhs1 and rhs2 and comparison code unconditionally,
15928 for </<= don't do that. Don't swap rhs1/rhs2 again if
15929 ranges[i].in_p, instead invert comparison code ccode if
15930 opcode or oe->rank is BIT_IOR_EXPR.
15933 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15934 a direct optab to conversion optab.
15935 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15936 with GET_MODE_INNER as last argument instead of optab_handler.
15937 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15938 vector extraction if possible and optab is available.
15939 * expr.c (store_constructor): Use convert_optab_handler instead
15940 of optab_handler. Use vector initialization from smaller
15941 vectors if possible and optab is available.
15942 * tree-vect-stmts.c (vectorizable_load): Likewise.
15943 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15944 now have two modes.
15945 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15946 of vec_init from half-sized vectors with the same element mode.
15947 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15948 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15949 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15950 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15951 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15952 after mode in gen_vec_extract* calls.
15953 (vec_extract<mode>): Renamed to ...
15954 (vec_extract<mode><ssescalarmodelower>): ... this.
15955 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15956 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15957 element mode after mode in gen_vec_init* calls.
15958 (VEC_INIT_HALF_MODE): New mode iterator.
15959 (vec_init<mode>): Renamed to ...
15960 (vec_init<mode><ssescalarmodelower>): ... this.
15961 (vec_init<mode><ssehalfvecmodelower>): New expander.
15962 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15963 (vec_extractv2sfsf): ... this.
15964 (vec_initv2sf): Renamed to ...
15965 (vec_initv2sfsf): ... this.
15966 (vec_extractv2si): Renamed to ...
15967 (vec_extractv2sisi): ... this.
15968 (vec_initv2si): Renamed to ...
15969 (vec_initv2sisi): ... this.
15970 (vec_extractv4hi): Renamed to ...
15971 (vec_extractv4hihi): ... this.
15972 (vec_initv4hi): Renamed to ...
15973 (vec_initv4hihi): ... this.
15974 (vec_extractv8qi): Renamed to ...
15975 (vec_extractv8qiqi): ... this.
15976 (vec_initv8qi): Renamed to ...
15977 (vec_initv8qiqi): ... this.
15978 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15979 (vec_init<mode>): Renamed to ...
15980 (vec_init<mode><VEC_base_l>): ... this.
15981 (vec_extract<mode>): Renamed to ...
15982 (vec_extract<mode><VEC_base_l>): ... this.
15983 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15984 (vec_initv2sfsf): ... this.
15985 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15986 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15987 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15988 element mode after mode in gen_vec_init* calls.
15989 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15990 (vec_init<mode><Vel>): ... this.
15991 (vec_extract<mode>): Renamed to ...
15992 (vec_extract<mode><Vel>): ... this.
15993 * config/aarch64/iterators.md (Vel): New mode attribute.
15994 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15995 Add element mode after mode in gen_vec_extract* calls.
15996 * config/s390/vector.md (non_vec_l): New mode attribute.
15997 (vec_extract<mode>): Renamed to ...
15998 (vec_extract<mode><non_vec_l>): ... this.
15999 (vec_init<mode>): Renamed to ...
16000 (vec_init<mode><non_vec_l>): ... this.
16001 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
16002 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
16004 * config/arm/iterators.md (V_elem_l): New mode attribute.
16005 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
16006 (vec_extract<mode><V_elem_l>): ... this.
16007 (vec_extractv2di): Renamed to ...
16008 (vec_extractv2didi): ... this.
16009 (vec_init<mode>): Renamed to ...
16010 (vec_init<mode><V_elem_l>): ... this.
16011 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
16012 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
16013 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
16014 Add element mode after gen_vec_extract* calls.
16015 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
16016 (vec_init<mode><unitmode>): ... this.
16017 (vec_extract<mode>): Renamed to ...
16018 (vec_extract<mode><unitmode>): ... this.
16019 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
16020 (vec_init<mode><unitmode>): ... this.
16021 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
16022 (vec_initv2sfsf): ... this.
16023 (vec_extractv2sf): Renamed to ...
16024 (vec_extractv2sfsf): ... this.
16025 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
16026 Add element mode after gen_vec_extract* calls.
16027 * config/mips/mips.md (unitmode): New mode iterator.
16028 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
16029 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
16030 * config/spu/spu.md (inner_l): New mode attribute.
16031 (vec_init<mode>): Renamed to ...
16032 (vec_init<mode><inner_l>): ... this.
16033 (vec_extract<mode>): Renamed to ...
16034 (vec_extract<mode><inner_l>): ... this.
16035 * config/sparc/sparc.md (veltmode): New mode iterator.
16036 (vec_init<VMALL:mode>): Renamed to ...
16037 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
16038 * config/ia64/vect.md (vec_initv2si): Renamed to ...
16039 (vec_initv2sisi): ... this.
16040 (vec_initv2sf): Renamed to ...
16041 (vec_initv2sfsf): ... this.
16042 (vec_extractv2sf): Renamed to ...
16043 (vec_extractv2sfsf): ... this.
16044 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
16045 (vec_init<mode>): Renamed to ...
16046 (vec_init<mode><VEC_base_l>): ... this.
16047 (vec_extract<mode>): Renamed to ...
16048 (vec_extract<mode><VEC_base_l>): ... this.
16049 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
16050 (vec_initv2sfsf): ... this.
16051 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
16052 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
16053 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
16054 gen_vec_init* calls.
16056 2017-08-01 Richard Biener <rguenther@suse.de>
16058 PR tree-optimization/81297
16059 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
16060 TREE_OVERFLOW from INTEGER_CSTs.
16062 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
16064 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
16066 2017-07-31 Carl Love <cel@us.ibm.com>
16068 * config/rs6000/rs6000-c: Add support for built-in functions
16069 vector signed char vec_xl_be (signed long long, signed char *);
16070 vector unsigned char vec_xl_be (signed long long, unsigned char *);
16071 vector signed int vec_xl_be (signed long long, signed int *);
16072 vector unsigned int vec_xl_be (signed long long, unsigned int *);
16073 vector signed long long vec_xl_be (signed long long, signed long long *);
16074 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
16075 vector signed short vec_xl_be (signed long long, signed short *);
16076 vector unsigned short vec_xl_be (signed long long, unsigned short *);
16077 vector double vec_xl_be (signed long long, double *);
16078 vector float vec_xl_be (signed long long, float *);
16079 * config/rs6000/altivec.h (vec_xl_be): Add #define.
16080 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
16081 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
16083 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
16084 (altivec_expand_builtin): Add switch statement to call
16085 altivec_expand_xl_be for each builtin.
16086 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
16087 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
16088 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
16089 __builtin_vsx_le_be_v16qi.
16090 * doc/extend.texi: Update the built-in documentation file for the
16091 new built-in functions.
16093 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16096 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
16098 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16100 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16102 * config.gcc: Add z14.
16103 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16104 CPU model numbers for z13s and z14.
16105 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16107 * config/s390/s390-opts.h (enum processor_type): Rename
16108 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16109 * config/s390/s390.c (processor_table): Add field for CPU name to
16110 be passed to Binutils.
16111 (s390_asm_output_machine_for_arch): Use the new field in
16112 processor_table for Binutils.
16113 (s390_expand_builtin): Replace arch12 with z14.
16114 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16115 (s390_get_sched_attrmask): Likewise.
16116 (s390_get_unit_mask): Likewise.
16117 * config/s390/s390.opt: Add z14 to processor_type enum.
16119 2017-07-31 Martin Jambor <mjambor@suse.cz>
16122 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16123 regardless of optimization level.
16125 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16126 Martin Liska <mliska@suse.cz>
16128 * predict.def: Remove old comment and adjust probability.
16129 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16130 PREDICT statements.
16132 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16135 * config/i386/i386.c (ix86_function_naked): New function.
16136 (ix86_can_use_return_insn_p): Return false for naked functions.
16137 (ix86_expand_prologue): Skip prologue for naked functions.
16138 (ix86_expand_epilogue): Skip epilogue for naked functions
16139 and emit trap instruction.
16140 (ix86_warn_func_return): New function.
16141 (ix86_attribute_table): Add "naked" attribute specification.
16142 (TARGET_WARN_FUNC_RETURN): Define.
16143 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16145 2017-07-31 Martin Liska <mliska@suse.cz>
16147 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16148 (dump_gimple_bb_header): Always dump BB info.
16149 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16151 2017-07-31 Martin Liska <mliska@suse.cz>
16154 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16155 also with current_function_decl non-null equality.
16157 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16160 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16161 change type to the element type, instead add eltype variable and
16162 use it where we are interested in the element type.
16164 PR tree-optimization/81603
16165 * ipa-polymorphic-call.c
16166 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16167 offset arithmetic in offset_int, bail out if the resulting bit offset
16168 doesn't fit into shwi.
16170 2017-07-31 Martin Liska <mliska@suse.cz>
16172 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16173 (gimplify_save_expr): Fix comment.
16175 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16178 * config/i386/i386.c (ix86_function_arg): Update arguments for
16180 (ix86_compute_frame_layout): Set the initial stack offset to
16181 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16182 INCOMING_FRAME_SP_OFFSET.
16183 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16184 stack before exception handler returns.
16185 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16186 the 'ERROR_CODE' for exception handler.
16188 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16190 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16191 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16192 (ASM_OUTPUT_REG_POP): Ditto.
16193 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16194 instead of asm_fprintf to output pure string.
16196 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16198 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16199 to imported_module_or_decl hook.
16200 (debug_nothing_tree_tree_tree_bool): Remove.
16201 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16202 * debug.c (do_nothing_debug_hooks): Use
16203 debug_nothing_tree_tree_tree_bool_bool instead of
16204 debug_nothing_tree_tree_tree_bool.
16205 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16206 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16207 * sdbout.c (sdb_debug_hooks): Likewise.
16208 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16209 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16211 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16212 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16213 attribute, don't add anything.
16215 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16217 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16218 (fold_build2_stat_loc): Likewise.
16219 (fold_build3_stat_loc): Likewise.
16220 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16221 (fold_build1_loc): Remove macro.
16222 (fold_build2_loc): Likewise.
16223 (fold_build3_loc): Likewise.
16225 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16227 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16228 (gimple_build_debug_bind_source_stat): Likewise.
16229 * gimple.h (gimple_build_debug_bind): Remove macro.
16230 (gimple_build_debug_bind_source): Likewise.
16232 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16234 * bitmap.c (bitmap_alloc): Adjust.
16235 (bitmap_gc_alloc): Likewise.
16236 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16238 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16240 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16241 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16242 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16243 (bitmap_gc_alloc_stat): Likewise.
16244 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16246 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16248 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16249 * rtl.h (shallow_copy_rtx): Remove macro.
16251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16253 * emit-rtl.c (gen_raw_REG): Adjust.
16254 * gengenrtl.c (gendef): Likewise.
16255 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16256 * rtl.h (rtx_alloc): Remove macro.
16258 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16260 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16261 (build_tree_list_stat): Likewise.
16262 * tree.h (build_tree_list): Remove macro.
16263 (build_tree_list_vec): Likewise.
16265 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16267 * tree.c (make_vector_stat): Remove _stat from name.
16268 (build_vector_stat): Likewise.
16269 * tree.h (make_vector_stat): Remove macro.
16270 (build_vector_stat): Likewise.
16272 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16274 * tree.h (build_var_debug_value): Remove prototype.
16276 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16278 * tree.c (tree_cons_stat): Remove _stat from name.
16279 * tree.h (tree_cons): Remove macro.
16281 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16283 * tree.c (build_vl_exp_stat): Remove _stat from name.
16284 * tree.h (build_vl_exp): Remove macro.
16286 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16288 * tree.c (build_decl_stat): Remove _stat from name.
16289 * tree.h (build_decl): Remove macro.
16291 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16293 * gimple.c (gimple_build_with_ops_stat): Adjust.
16294 (gimple_alloc_stat): Remove _stat from name.
16295 * gimple.h (gimple_alloc): Remove macro.
16297 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16299 * tree.c (make_tree_vec_stat): Remove _stat from name.
16300 (grow_tree_vec_stat): Likewise.
16301 * tree.h (make_tree_vec_stat): Adjust prototype.
16302 (grow_tree_vec_stat): Likewise.
16303 (make_tree_vec): Remove macro.
16304 (grow_tree_vec): Likewise.
16306 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16308 * fold-const.c (fold_build1_stat_loc): Adjust.
16309 (fold_build2_stat_loc): Likewise.
16310 (fold_build3_stat_loc): Likewise.
16311 * tree.c (build0_stat): Remove _stat from name.
16312 (build1_stat): Likewise.
16313 (build2_stat): Likewise.
16314 (build3_stat): Likewise.
16315 (build4_stat): Likewise.
16316 (build5_stat): Likewise.
16317 * tree.h (build1_loc): Remove macro, and rename _stat function
16319 (build2_loc): Likewise.
16320 (build3_loc): Likewise.
16321 (build4_loc): Likewise.
16322 (build5_loc): Likewise.
16324 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16326 * tree.c (make_int_cst_stat): Remove _stat from name.
16327 * tree.h (make_int_cst_stat): Adjust prototype.
16328 (make_int_cst): Remove macro.
16330 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16332 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16333 * tree.h (make_tree_binfo_stat): Adjust prototype.
16334 (make_tree_binfo): Remove.
16336 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16338 * tree.c (copy_node_stat): Rename to copy_node.
16339 (build_distinct_type_copy): Adjust.
16340 * tree.h (copy_node_stat): Adjust prototype.
16341 (copy_node): Remove macro.
16343 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16345 * tree.c (make_node_stat): rename to make_node.
16346 (build_tree_list_stat): Adjust.
16347 (build0_stat): Likewise.
16348 (build2_stat): Likewise.
16349 (build3_stat): Likewise.
16350 (build4_stat): Likewise.
16351 (build5_stat): Likewise.
16352 (build_decl_stat): Likewise.
16353 * tree.h (make_node_stat): Adjust prototype.
16354 (make_node): remove macro.
16356 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16358 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16359 (PPC_FEATURE2_SCV): Likewise.
16360 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16362 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16364 * config/aarch64/aarch64.c
16365 (aarch64_internal_mov_immediate): Add new special pattern.
16366 * config/aarch64/aarch64.md (*movdi_aarch64):
16367 Add reg/32bit const mov case.
16369 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16370 Richard Sandiford <richard.sandiford@linaro.org>
16372 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16373 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16374 Add integer and movi cases.
16375 (movi-split-hf-df-sf split, fp16): New.
16376 (enabled): Added TARGET_FP_F16INST.
16377 * config/aarch64/iterators.md (GPF_HF): New.
16378 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16380 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16382 * config/aarch64/aarch64.c
16383 (aarch64_simd_container_mode): Add prototype.
16384 (aarch64_expand_mov_immediate): Add HI support.
16385 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16386 (aarch64_can_const_movi_rtx_p): New.
16387 (aarch64_preferred_reload_class):
16388 Remove restrictions of using FP registers for certain SIMD operations.
16389 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16390 (aarch64_valid_floating_const): Add integer move validation.
16391 (aarch64_simd_imm_scalar_p): Remove.
16392 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16393 (aarch64_legitimate_constant_p): Expand list of supported cases.
16394 * config/aarch64/aarch64-protos.h
16395 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16396 (aarch64_reinterpret_float_as_int): New.
16397 (aarch64_simd_imm_scalar_p): Remove.
16398 * config/aarch64/constraints.md (Uvi): New.
16399 (Dd): Split into Ds and new Dd.
16400 * config/aarch64/aarch64.md (*movsi_aarch64):
16402 (*movdi_aarch64): Add SIMD mov case.
16404 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16406 * tree-predcom.c: (struct chain): Handle store-store chain in which
16407 stores for elimination only store loop invariant values.
16408 (execute_pred_commoning_chain): Ditto.
16409 (prepare_initializers_chain_store_elim): Ditto.
16410 (prepare_finalizers): Ditto.
16411 (is_inv_store_elimination_chain): New function.
16412 (initialize_root_vars_store_elim_1): New function.
16414 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16416 * tree-predcom.c: Revise general description of the pass.
16417 (enum chain_type): New enum type for store elimination.
16418 (struct chain): New field supporting store elimination.
16419 (struct component): Ditto.
16420 (dump_chain): Dump store-stores chain.
16421 (release_chain): Release resources.
16422 (split_data_refs_to_components): Compute and create component
16423 contains only stores for elimination.
16424 (get_chain_last_ref_at): New function.
16425 (make_invariant_chain): Initialization.
16426 (make_rooted_chain): Specify chain type in parameter and record it.
16427 (add_looparound_copies): Skip for store-stores chain.
16428 (determine_roots_comp): Compute type of chain and pass it to
16430 (initialize_root_vars_store_elim_2): New function.
16431 (finalize_eliminated_stores): New function.
16432 (remove_stmt): Handle store for elimination.
16433 (execute_pred_commoning_chain): Execute predictive commoning on
16434 store-store chains.
16435 (determine_unroll_factor): Skip unroll for store-stores chain.
16436 (prepare_initializers_chain_store_elim): New function.
16437 (prepare_initializers_chain): Hanlde store-store chain.
16438 (prepare_finalizers_chain, prepare_finalizers): New function.
16439 (tree_predictive_commoning_loop): Return integer value indicating
16440 if loop is unrolled or lcssa form is corrupted.
16441 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16443 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16445 * tree-predcom.c (initialize_root): Delete.
16446 (execute_pred_commoning_chain): Initialize root vars and replace
16447 reference of non-combined chain directly, rather than call above
16450 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16452 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16453 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16455 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16457 * tree-predcom.c (struct chain): New field init_seq.
16458 (release_chain): Release init_seq.
16459 (prepare_initializers_chain): Record intialization stmts in above
16461 (insert_init_seqs): New function.
16462 (tree_predictive_commoning_loop): Call insert_init_seqs.
16464 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16466 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16468 2017-07-28 Richard Biener <rguenther@suse.de>
16470 * match.pd: Remove superfluous :c.
16471 * genmatch.c (simplify::id): Add member.
16472 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16474 (current_id): New global.
16475 (dt_node::parent): Move from ...
16476 (dt_operand::parent): ... here. Add for_id member.
16477 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16478 (decision_tree::find_node): Relax order requirement when
16479 merging DT_TRUE nodes to ones inbetween the current simplify
16480 and the one we try to merge with. Add diagnostic whenever
16481 we need to enforce pattern order by not merging.
16482 (decision_tree::insert): Set current_id.
16483 (decision_tree::print_node): Dump parent node and for_id.
16484 (parser::last_id): Add member.
16485 (parser::push_simplify): Assign unique id.
16486 (parser::parser): Initialize last_id.
16488 2017-07-28 Martin Liska <mliska@suse.cz>
16491 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16492 gimple_build_debug_bind.
16494 2017-07-28 Richard Biener <rguenther@suse.de>
16496 PR tree-optimization/81502
16497 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16499 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16501 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16502 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16503 for BIT_FIELD_REF args.
16504 * fold-const.c (make_bit_field_ref): Likewise.
16505 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16507 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16510 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16511 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16512 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16513 Or it into SANITIZER_UNDEFINED.
16514 * ubsan.c: Include gimple-fold.h and varasm.h.
16515 (ubsan_expand_ptr_ifn): New function.
16516 (instrument_pointer_overflow): New function.
16517 (maybe_instrument_pointer_overflow): New function.
16518 (instrument_object_size): Formatting fix.
16519 (pass_ubsan::execute): Call instrument_pointer_overflow
16520 and maybe_instrument_pointer_overflow.
16521 * internal-fn.c (expand_UBSAN_PTR): New function.
16522 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16523 * sanitizer.def (__ubsan_handle_pointer_overflow,
16524 __ubsan_handle_pointer_overflow_abort): New builtins.
16525 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16526 * internal-fn.def (UBSAN_PTR): New internal function.
16527 * opts.c (sanitizer_opts): Add pointer-overflow.
16528 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16529 * fold-const.c (build_range_check): Compute pointer range check in
16530 integral type if pointer arithmetics would be needed. Formatting
16533 2017-07-28 Martin Liska <mliska@suse.cz>
16536 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16537 parameters that are of a variable-length.
16539 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16541 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16543 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16544 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16545 (CRT_CALL_STATIC_FUNCTION): Likewise.
16546 (ASM_DEFAULT_SPEC): New define.
16547 (ASM_SPEC32): Likewise.
16548 (ASM_SPEC64): Likewise.
16549 (ASM_SPEC_COMMON): Likewise.
16550 (ASM_SPEC): Likewise.
16551 (INVALID_64BIT): Likewise.
16552 (LINK_OS_DEFAULT_SPEC): Likewise.
16553 (LINK_OS_SPEC32): Likewise.
16554 (LINK_OS_SPEC64): Likewise.
16555 (POWERPC_LINUX): Likewise.
16556 (PTRDIFF_TYPE): Likewise.
16557 (RESTORE_FP_PREFIX): Likewise.
16558 (RESTORE_FP_SUFFIX): Likewise.
16559 (SAVE_FP_PREFIX): Likewise.
16560 (SAVE_FP_SUFFIX): Likewise.
16561 (SIZE_TYPE): Likewise.
16562 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16563 (TARGET_64BIT): Likewise.
16564 (TARGET_64BIT): Likewise.
16565 (TARGET_AIX): Likewise.
16566 (WCHAR_TYPE_SIZE): Likewise.
16567 (WCHAR_TYPE): Undefine.
16568 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16569 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16570 (CPP_OS_RTEMS_SPEC): Delete.
16571 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16572 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16574 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16576 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16578 PR tree-optimization/81578
16579 * tree-parloops.c (build_new_reduction): Bail out if
16580 reduction_code isn't one of the standard OpenMP reductions.
16581 Move the details printing after that decision.
16583 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16585 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16586 related to reload_in_progress.
16587 (splat_input_operand): Likewise.
16588 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16590 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16592 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16593 (TARGET_INSTANTIATE_DECLS): Likewise.
16594 (legitimate_indexed_address_p): Delete reload_in_progress code.
16595 (rs6000_debug_legitimate_address_p): Likewise.
16596 (rs6000_eliminate_indexed_memrefs): Likewise.
16597 (rs6000_emit_le_vsx_store): Likewise.
16598 (rs6000_emit_move_si_sf_subreg): Likewise.
16599 (rs6000_emit_move): Likewise.
16600 (register_to_reg_type): Likewise.
16601 (rs6000_pre_atomic_barrier): Likewise.
16602 (rs6000_machopic_legitimize_pic_address): Likewise.
16603 (rs6000_allocate_stack_temp): Likewise.
16604 (rs6000_address_for_fpconvert): Likewise.
16605 (rs6000_address_for_altivec): Likewise.
16606 (rs6000_secondary_memory_needed_rtx): Delete function.
16607 (rs6000_check_sdmode): Likewise.
16608 (rs6000_alloc_sdmode_stack_slot): Likewise.
16609 (rs6000_instantiate_decls): Likewise.
16610 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16611 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16612 Delete reload_in_progress.
16613 (*vec_reload_and_plus_<mptrsize>): Likewise.
16614 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16615 (vsx_div_v2di): Likewise.
16616 (vsx_udiv_v2di): Likewise.
16618 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16620 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16621 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16622 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16623 (rs6000_debug_reg_global): Delete print of LRA status.
16624 (rs6000_option_override_internal): Delete dead LRA related code.
16625 (rs6000_lra_p): Delete function.
16626 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16628 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16630 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16631 * config/riscv/rtems.h: New file.
16633 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16634 Sudakshina Das <sudi.das@arm.com>
16636 * config/aarch64/aarch64.md
16637 (define_split for and<mode>3nr_compare): Move
16638 non aarch64_logical_operand to a register.
16639 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16640 register immediate operand to a register.
16641 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16643 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16645 PR middle-end/81564
16646 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16648 2017-07-27 Richard Biener <rguenther@suse.de>
16650 PR tree-optimization/81573
16651 PR tree-optimization/81494
16652 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16653 multi defuse cycle case.
16655 2017-07-27 Richard Biener <rguenther@suse.de>
16657 PR tree-optimization/81571
16658 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16661 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16663 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16664 earlier and only if MASK_FPU is set. Adjust formatting.
16666 2017-07-27 Martin Liska <mliska@suse.cz>
16668 * opt-functions.awk: Add validation of value of Init.
16669 * optc-gen.awk: Pass new argument.
16671 2017-07-27 Martin Liska <mliska@suse.cz>
16673 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16674 Fix wrong condition.
16676 2017-07-27 Martin Liska <mliska@suse.cz>
16678 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16679 BBs and edges seen by autoFDO.
16681 2017-07-27 Richard Biener <rguenther@suse.de>
16683 PR tree-optimization/81502
16684 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16685 with incompatible but same sized type.
16686 (execute_update_addresses_taken): Likewise.
16688 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16690 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16691 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16693 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16696 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16697 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16698 Change s_operand to memory_operand.
16700 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16702 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16703 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16704 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16705 Emit instructions rather than returning an expression. Handle TFmode
16706 and KFmode by casting to TImode.
16707 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16708 (rs6000_emit_le_vsx_store): Likewise.
16709 * config/rs6000/vsx.md (VSX_TI): New iterator.
16710 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16711 (*vsx_le_undo_permute_<mode>): Likewise.
16712 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16713 emit the split sequence.
16714 (*vsx_le_perm_store_<mode>): Likewise.
16716 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16718 PR tree-optimization/81555
16719 PR tree-optimization/81556
16720 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16721 if true, force CHANGED for the recursive invocation.
16722 (reassociate_bb): Remember original length of ops array, pass
16723 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16725 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16726 attributes for noipa attribute. For naked attribute use
16727 lookup_attribute first before lookup_attribute_spec.
16728 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16730 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16731 for functions with noipa attribute.
16732 (cgraph_externally_visible_p): Return true for functions with noipa
16734 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16735 for functions with noipa attribute.
16736 * doc/extend.texi: Document noipa function attribute.
16737 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16738 also for functions with noipa attribute.
16739 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16741 2017-07-26 Andrew Pinski <apinski@cavium.com>
16743 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16744 vec_unalign_load_cost and vec_unalign_store_cost.
16746 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16748 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16749 -mvsx-small-integer option.
16750 (ISA_3_0_MASKS_IEEE): Likewise.
16751 (OTHER_VSX_VECTOR_MASKS): Likewise.
16752 (POWERPC_MASKS): Likewise.
16753 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16754 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16755 code, only testing for DImode being allowed in non-VSX floating
16757 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16758 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16760 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16761 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16762 TARGET_P8_VECTOR test.
16763 (rs6000_secondary_reload_simple_move): Likewise.
16764 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16765 since TARGET_P9_VECTOR was already tested.
16766 (rs6000_opt_masks): Remove -mvsx-small-integer.
16767 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16768 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16770 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16771 test for TARGET_VEXTRACTUB was used, and that uses
16773 (p9 extract splitter): Likewise.
16774 (vsx_extract_<mode>_di_p9): Likewise.
16775 (vsx_extract_<mode>_store_p9): Likewise.
16776 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16777 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16778 the elimination of TARGET_VSX_SMALL_INTEGER.
16779 (vsx_extract_<mode>_p8): Likewise.
16780 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16781 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16782 (vsx_set_<mode>_p9): Likewise.
16783 (vsx_set_v4sf_p9): Likewise.
16784 (vsx_set_v4sf_p9_zero): Likewise.
16785 (vsx_insert_extract_v4sf_p9): Likewise.
16786 (vsx_insert_extract_v4sf_p9_2): Likewise.
16787 * config/rs6000/rs6000.md (sign extend splitter): Change
16788 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16789 (floatsi<mode>2_lfiwax_mem): Likewise.
16790 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16791 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16792 since a test for TARGET_P9_VECTOR was used.
16793 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16794 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16795 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16796 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16797 TARGET_P8_VECTOR test.
16798 (fix_trunc<mode>si2_stfiwx): Likewise.
16799 (fix_trunc<mode>si2_internal): Likewise.
16800 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16801 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16803 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16804 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16805 TARGET_P8_VECTOR test.
16806 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16807 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16808 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16810 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16811 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16812 since a test for TARGET_P9_VECTOR was used.
16813 (splitter for loading small constants): Likewise.
16815 2017-07-26 Andrew Pinski <apinski@cavium.com>
16817 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16820 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16823 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16824 (fp_valid_at): Likewise.
16826 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16828 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16829 (qdf24xx_addrcost_table): Likewise.
16830 (cortexa57_tunings): Update to use generic_branch_cost.
16831 (cortexa72_tunings): Likewise.
16832 (cortexa73_tunings): Likewise.
16833 (qdf24xx_tunings): Likewise.
16835 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16837 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16838 (thunderx2t99_branch_cost): Likewise.
16839 (cortexa35_tunings): Update to use generic_branch_cost.
16840 (cortexa53_tunings): Likewise.
16841 (cortexa57_tunings): Likewise.
16842 (cortexa72_tunings): Likewise.
16843 (cortexa73_tunings): Likewise.
16844 (thunderx2t99_tunings): Likewise.
16846 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16848 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16849 (sparc_option_override): Honour MASK_FSMULD.
16850 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16851 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16852 * config/sparc/sparc.opt (mfsmuld): New option.
16853 * doc/invoke.texi (mfsmuld): Document option.
16855 2017-07-26 Marek Polacek <polacek@redhat.com>
16857 PR middle-end/70992
16858 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16860 2017-07-26 Richard Biener <rguenther@suse.de>
16862 * gimple-match-head.c (do_valueize): Return OP if valueize
16864 (get_def): New helper to get at the def stmt of a SSA name
16865 if valueize allows.
16866 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16867 do_valueize to get at the def stmt.
16868 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16870 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16872 PR middle-end/46932
16873 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16875 2017-07-26 Martin Liska <mliska@suse.cz>
16878 * function.c (expand_function_start): Make expansion of
16879 nonlocal_goto_save_area after parm_birth_insn.
16881 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16883 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16884 from all CPU target flags enable members.
16886 2017-07-26 Richard Biener <rguenther@suse.de>
16888 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16889 (decision_tree::gen): Make 'type' const.
16890 (write_predicate): Likewise.
16892 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16894 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16895 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16896 (rs6000_option_override_internal): Likewise.
16897 (rs6000_expand_vector_set): Likewise.
16898 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16899 (TARGET_UPPER_REGS_SF): Likewise.
16900 (TARGET_UPPER_REGS_DI): Likewise.
16901 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16902 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16903 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16904 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16905 (Splitters for DI constants in Altivec registers): Likewise.
16906 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16907 (vsx_set_v4sf_p9): Likewise.
16908 (vsx_set_v4sf_p9_zero): Likewise.
16909 (vsx_insert_extract_v4sf_p9): Likewise.
16910 (vsx_insert_extract_v4sf_p9_2): Likewise.
16912 2017-07-25 Carl Love <cel@us.ibm.com>
16914 * doc/extend.texi: Update the built-in documentation file for the
16915 existing built-in functions
16916 vector signed char vec_cnttz (vector signed char);
16917 vector unsigned char vec_cnttz (vector unsigned char);
16918 vector signed short vec_cnttz (vector signed short);
16919 vector unsigned short vec_cnttz (vector unsigned short);
16920 vector signed int vec_cnttz (vector signed int);
16921 vector unsigned int vec_cnttz (vector unsigned int);
16922 vector signed long long vec_cnttz (vector signed long long);
16923 vector unsigned long long vec_cnttz (vector unsigned long long);
16925 2017-07-25 Andrew Pinski <apinski@cavium.com>
16927 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16928 accesses where the use is for the first operand of a BIT_INSERT.
16930 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16933 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16934 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16936 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16938 * config/i386/gstabs.h: Delete.
16939 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16941 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16943 * config/i386/i386.c (ix86_decompose_address): Do not check for
16944 register RTX when looking at index_reg or base_reg.
16945 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16947 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16949 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16950 to update EH info here.
16952 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16954 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16956 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16958 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16960 2017-07-25 Torsten Duwe <duwe@suse.de>
16962 * common.opt: Introduce -fpatchable-function-entry
16963 command line option, and its variables function_entry_patch_area_size
16964 and function_entry_patch_area_start.
16965 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16966 including a two-value parser.
16967 * target.def (print_patchable_function_entry): New target hook.
16968 * targhooks.h (default_print_patchable_function_entry): New function.
16969 * targhooks.c (default_print_patchable_function_entry): Likewise.
16970 * toplev.c (process_options): Switch off IPA-RA if
16971 patchable function entries are being generated.
16972 * varasm.c (assemble_start_function): Look at the
16973 patchable-function-entry command line switch and current
16974 function attributes and maybe generate NOP instructions by
16975 calling the print_patchable_function_entry hook.
16976 * doc/extend.texi: Document patchable_function_entry attribute.
16977 * doc/invoke.texi: Document -fpatchable_function_entry
16978 command line option.
16979 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16981 * doc/tm.texi: Re-generate.
16983 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16986 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16987 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16989 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16991 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16993 2017-07-25 Richard Biener <rguenther@suse.de>
16995 PR tree-optimization/81455
16996 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16997 not walk in cycles when looking for guards.
16999 2017-07-25 Richard Biener <rguenther@suse.de>
17001 PR tree-optimization/81529
17002 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
17003 when optimizing backedge uses.
17005 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
17007 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
17009 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
17010 to dl_section_ref. On AIX, append an expression to subtract
17011 the size of the section length to dl_section_ref.
17013 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
17015 * configure.ac: If any of the config.* scripts fail, exit 1.
17016 * configure: Regenerate.
17018 2017-07-25 Richard Biener <rguenther@suse.de>
17020 PR middle-end/81546
17021 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
17022 of immediate uses, be more verbose on errors.
17024 2017-07-25 Richard Biener <rguenther@suse.de>
17026 PR tree-optimization/81510
17027 * tree-vect-loop.c (vect_is_simple_reduction): When the
17028 reduction stmt is not inside the loop bail out.
17030 2017-07-25 Richard Biener <rguenther@suse.de>
17032 PR tree-optimization/81303
17033 * tree-vect-loop-manip.c (vect_loop_versioning): Build
17034 profitability check against LOOP_VINFO_NITERSM1.
17036 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
17038 * domwalk.c (cmp_bb_postorder): Simplify.
17039 (sort_bbs_postorder): New function. Use it...
17040 (dom_walker::walk): ...here to optimize common cases.
17042 2017-07-25 Martin Liska <mliska@suse.cz>
17045 * ipa-visibility.c (function_and_variable_visibility): Make the
17046 redirection just on target that supports aliasing.
17047 Fix GNU coding style.
17049 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
17052 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
17054 * config/arm/rtems.h: Likewise.
17055 * config/bfin/rtems.h: Likewise.
17056 * config/i386/rtemself.h: Likewise.
17057 * config/lm32/rtems.h: Likewise.
17058 * config/m32c/rtems.h: Likewise.
17059 * config/m68k/rtemself.h: Likewise.
17060 * config/microblaze/rtems.h: Likewise.
17061 * config/mips/rtems.h: Likewise.
17062 * config/moxie/rtems.h: Likewise.
17063 * config/nios2/rtems.h: Likewise.
17064 * config/powerpcspe/rtems.h: Likewise.
17065 * config/rs6000/rtems.h: Likewise.
17066 * config/rtems.h: Likewise.
17067 * config/sh/rtems.h: Likewise.
17068 * config/sh/rtemself.h: Likewise.
17069 * config/sparc/rtemself.h: Likewise.
17071 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17074 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
17075 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
17076 * tree-ssa-structalias.c (alias_get_name): Same.
17078 2017-07-25 Bin Cheng <bin.cheng@arm.com>
17081 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
17082 instructions if no du chain is found.
17084 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
17086 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
17088 2017-07-25 Richard Biener <rguenther@suse.de>
17090 PR middle-end/81505
17091 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
17094 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
17096 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
17097 upper-regs options.
17098 (ISA_2_7_MASKS_SERVER): Likewise.
17099 (ISA_3_0_MASKS_IEEE): Likewise.
17100 (OTHER_P8_VECTOR_MASKS): Likewise.
17101 (OTHER_VSX_VECTOR_MASKS): Likewise.
17102 (POWERPC_MASKS): Likewise.
17103 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17104 duplicate list of options.
17105 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17106 explicit -mupper-regs options.
17107 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17108 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17109 alias for -mupper-regs-df.
17110 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17111 (rs6000_init_hard_regno_mode_ok): Likewise.
17112 (rs6000_option_override_internal): Likewise.
17113 (rs6000_opt_masks): Likewise.
17114 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17115 options in terms of whether -mvsx or -mpower8-vector was used.
17116 (TARGET_UPPER_REGS_DI): Likewise.
17117 (TARGET_UPPER_REGS_SF): Likewise.
17118 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17119 -mupper-regs-* options.
17121 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17123 * passes.c (emergency_dump_function): Print some empty lines and a
17124 header before the RTL dump.
17126 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17128 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17130 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17133 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17134 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17136 2017-07-24 Carl Love <cel@us.ibm.com>
17138 * config/rs6000/rs6000-c.c: Add support for built-in functions
17139 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17140 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17141 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17142 vec_extract_fp_from_shortl): Add defines for the two builtins.
17143 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17144 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17146 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17147 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17148 * doc/extend.texi: Update the built-in documentation file for the
17149 new built-in function.
17151 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17154 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17156 * doc/generic.texi: Likewise.
17157 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17158 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17160 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17162 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17163 (aarch64_mls_elt_merge<mode>): Likewise.
17165 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17167 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17168 having __cxa_atexit.
17170 2017-07-23 Michael Collison <michael.collison@arm.com>
17172 * config/arm/arm.c (arm_option_override): Deprecate
17173 use of -mstructure-size-boundary.
17174 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17175 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17177 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17180 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17181 Reduce cost estimate for direct moves.
17183 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17186 * config/i386/i386.c (ix86_option_override_internal): Disable
17187 BMI, BMI2 and TBM instructions for -m16.
17189 2017-07-21 Carl Love <cel@us.ibm.com>
17191 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17192 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17193 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17194 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17195 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17196 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17197 VMULOSW): New enum "unspec" values.
17198 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17199 altivec_vmulosw): New patterns.
17200 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17201 VMULOSW): Add definitions.
17203 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17205 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17206 (qdf24xx): Likewise.
17207 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17208 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17209 (AARCH64_FL_V8_1): Renumber.
17210 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17211 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17212 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17213 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17214 rdma to feature modifiers list.
17216 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17218 PR middle-end/56727
17219 * ipa-visibility (function_and_variable_visibility): Convert
17220 recursive PLT call to direct call if appropriate.
17222 2017-07-21 Andrew Pinski <apinski@cavium.com>
17224 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17225 operand 1 to see if the types precision matches.
17226 * fold-const.c (operand_equal_p): Likewise.
17228 2017-07-21 Richard Biener <rguenther@suse.de>
17230 PR tree-optimization/81303
17231 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17232 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17233 (vect_peeling_hash_get_lowest_cost): Adjust.
17234 (vect_enhance_data_refs_alignment): Likewise. Use
17235 vect_get_peeling_costs_all_drs to compute the penalty for no
17236 peeling to match up costs.
17238 2017-07-21 Richard Biener <rguenther@suse.de>
17240 PR tree-optimization/81500
17241 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17242 we didn't identify a reduction path.
17244 2017-07-21 Tom de Vries <tom@codesourcery.com>
17245 Cesar Philippidis <cesar@codesourcery.com>
17247 PR gcov-profile/81442
17248 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17251 2017-07-21 Tom de Vries <tom@codesourcery.com>
17254 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17256 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17257 nvptx_override_options_after_change.
17259 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17261 * dwarf2out.c (output_file_names): Avoid double testing for
17262 dwarf_version >= 5.
17264 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17266 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17268 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17270 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17272 (try_crossjump_to_edge): Do not punt on partitioned functions.
17274 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17276 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17277 Put all BBs reachable only via paths crossing cold region to cold
17279 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17281 2016-07-21 Richard Biener <rguenther@suse.de>
17283 PR tree-optimization/81303
17284 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17285 into account prologue and epilogue iterations when raising
17286 min_profitable_iters to sth at least covering one vector iteration.
17288 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17290 * config/arm/arm.c (arm_test_cpu_arch_dat):
17293 2017-07-20 Nathan Sidwell <nathan@acm.org>
17295 Remove TYPE_METHODS.
17296 * tree.h (TYPE_METHODS): Delete.
17297 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17298 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17299 (dbxout_type_methods): Scan TYPE_FIELDS.
17300 (dbxout_type): Don't check TYPE_METHODS here.
17301 * function.c (use_register_for_decl): Always ignore register for
17302 class types when not optimizing.
17303 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17304 * tree.c (free_lang_data_in_type): Stitch out member functions and
17305 templates from TYPE_FIELDS.
17306 (build_distinct_type_copy, verify_type_variant,
17307 verify_type): Member fns are on TYPE_FIELDS.
17308 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17309 * tree-pretty-print.c (dump_generic_node): Likewise.
17311 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17314 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17315 V2TImode and V4TImode.
17316 (ix86_expand_vector_extract): Likewise.
17317 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17318 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17319 (ssescalarmode): Handle V4TImode and V2TImode.
17320 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17321 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17322 (VEXTRACTI128_MODE): New mode iterator.
17323 (splitter for *vec_extractv?ti first element): New.
17324 (VEC_INIT_MODE): New mode iterator.
17325 (vec_init<mode>): Consolidate 3 expanders into one using
17326 VEC_INIT_MODE mode iterator.
17328 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17330 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17331 non_spilled_static_chain_regno_p.
17333 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17335 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17337 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17339 * bb-reorder.c (connect_traces): Allow copying of blocks within
17342 2017-07-20 Richard Biener <rguenther@suse.de>
17344 * gimple.h (gimple_phi_result): Add gphi * overload.
17345 (gimple_phi_result_ptr): Likewise.
17346 (gimple_phi_arg): Likewise. Adjust index assert to only
17347 allow actual argument accesses rather than all slots available
17349 (gimple_phi_arg_def): Add gphi * overload.
17350 * tree-phinodes.c (make_phi_node): Initialize only actual
17352 (resize_phi_node): Clear memory not covered by old node,
17353 do not initialize excess argument slots.
17354 (reserve_phi_args_for_new_edge): Initialize new argument slot
17357 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17359 PR tree-optimization/81388
17361 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17363 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17364 by removing computation of may_be_zero.
17366 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17367 Tom de Vries <tom@codesourcery.com>
17369 PR middle-end/81030
17370 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17371 when gimple level profile disagrees with what RTL expander did.
17373 2017-07-20 Richard Biener <rguenther@suse.de>
17375 PR tree-optimization/61171
17376 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17377 (vect_analyze_stmt): Add slp instance parameter.
17378 (vectorizable_reduction): Likewise.
17379 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17380 (vect_is_simple_reduction): Deal with chains not detected
17381 as SLP reduction chain, specifically not properly associated
17382 chains containing a mix of plus/minus.
17383 (get_reduction_op): Remove.
17384 (get_initial_defs_for_reduction): Simplify, pass in whether
17385 this is a reduction chain, pass in the SLP node for the PHIs.
17386 (vect_create_epilog_for_reduction): Get the SLP instance as
17388 (vectorizable_reduction): Get the SLP instance as arg.
17389 During analysis remember the SLP node with the PHIs in the
17390 instance. Simplify getting at the vectorized reduction PHIs.
17391 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17392 through SLP instance.
17393 (vect_slp_analyze_operations): Likewise.
17394 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17395 (vect_transform_stmt): Likewise.
17397 2017-07-20 Tom de Vries <tom@codesourcery.com>
17399 PR tree-optimization/81489
17400 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17401 read of phi arg location to before loop that modifies phi.
17403 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17405 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17408 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17410 PR middle-end/81331
17411 * except.c (execute): Fix ordering issue.
17413 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17415 PR rtl-optimization/81423
17416 * combine.c (make_compound_operation_int): Don't try to optimize
17417 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17419 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17421 PR rtl-optimization/81423
17422 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17423 with a constant that is -1 in the truncated to mode.
17425 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17427 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17428 (determine_unlikely_bbs): ... here.
17429 * predict.h (propagate_unlikely_bbs_forward): Declare.
17430 * cfgexpand.c (pass_expand::execute): Use it.
17431 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17433 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17434 propagate_unlikely_bbs_forward.
17436 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17438 PR middle-end/81331
17439 * except.c (maybe_add_nop_after_section_switch): New function.
17442 2017-07-19 Tom de Vries <tom@codesourcery.com>
17444 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17446 2017-07-19 Tom de Vries <tom@codesourcery.com>
17448 * gimple.h (gimple_phi_arg): Make assert more strict.
17450 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17452 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17453 * config/rs6000/mmintrin.h: New file.
17454 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17456 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17458 PR tree-optimization/81346
17459 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17461 2017-07-19 Tom de Vries <tom@codesourcery.com>
17463 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17465 2017-07-19 Tom de Vries <tom@codesourcery.com>
17467 * config/nvptx/nvptx-modes.def: Add V2DImode.
17468 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17469 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17470 (nvptx_output_mov_insn): Handle lack of mov.b128.
17471 (nvptx_print_operand): Handle 'H' and 'L' codes.
17472 (nvptx_vector_mode_supported): Allow V2DImode.
17473 (nvptx_preferred_simd_mode): New function.
17474 (nvptx_data_alignment): New function.
17475 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17476 nvptx_preferred_simd_mode.
17477 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17479 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17481 2017-07-19 Tom de Vries <tom@codesourcery.com>
17483 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17484 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17485 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17486 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17487 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17488 (mov<VECIM>_insn): New define_insn.
17489 (define_expand "mov<VECIM>): New define_expand.
17491 2017-07-19 Tom de Vries <tom@codesourcery.com>
17493 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17495 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17497 PR tree-optimization/81346
17498 * fold-const.h (fold_div_compare, range_check_type): Declare.
17499 * fold-const.c (range_check_type): New function.
17500 (build_range_check): Use range_check_type.
17501 (fold_div_compare): No longer static, rewritten into
17502 a match.pd helper function.
17503 (fold_comparison): Don't call fold_div_compare here.
17504 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17505 as helper function.
17507 2017-07-19 Nathan Sidwell <nathan@acm.org>
17509 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17510 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17511 * tree.c (find_decls_types_r, verify_type): Use
17512 TYPE_{MIN,MAX}_VALUE_RAW.
17513 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17514 (hash_tree): Likewise.
17515 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17517 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17520 2017-07-18 Tom de Vries <tom@codesourcery.com>
17522 PR middle-end/81464
17523 * omp-expand.c (expand_omp_for_static_chunk): Handle
17524 equal-argument loop exit phi.
17526 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17529 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17530 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17531 operand 2 predicate.
17532 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17533 operand 2 predicate.
17534 (ror,rol -> rorx splitters): Use const_int_operand as
17535 operand 2 predicate.
17537 2017-06-18 Richard Biener <rguenther@suse.de>
17539 PR tree-optimization/81410
17540 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17541 the gap in the ! slp_perm SLP case after each group.
17543 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17545 PR middle-end/81463
17546 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17549 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17551 PR middle-end/81462
17552 * predict.c (set_even_probabilities): Cleanup; do not affect
17553 probabilities that are already known.
17554 (combine_predictions_for_bb): Call even when count is set.
17556 2017-07-18 Nathan Sidwell <nathan@acm.org>
17558 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17561 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17564 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17565 optimization for loop niter analysis.
17567 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17570 * config/avr/avr.c (avr_optimize_casesi): Don't use
17571 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17573 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17575 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17576 body_cost_vec from _vect_peel_extended_info.
17577 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17578 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17581 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17583 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17585 2017-07-18 Richard Biener <rguenther@suse.de>
17587 PR tree-optimization/80620
17588 PR tree-optimization/81403
17589 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17590 info when re-using a VN table entry.
17592 2017-07-18 Richard Biener <rguenther@suse.de>
17594 PR tree-optimization/81418
17595 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17596 vectype_in. Verify that with lane-reducing reduction operations
17597 we have a single def-use cycle.
17599 2017-07-17 Carl Love <cel@us.ibm.com>
17601 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17603 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17604 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17605 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17606 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17607 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17608 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17609 VMULOSW): New enum "unspec" values.
17610 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17611 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17612 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17613 altivec_vmulosw): New patterns.
17614 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17615 VMULOSW): Add definitions.
17616 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17618 * config/alpha/alpha.c: Include predict.h.
17620 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17622 * tree-vrp.c (compare_assert_loc): Fix comparison function
17623 to return predictable results.
17625 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17627 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17629 (subdi3): Likewise.
17630 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17631 * doc/invoke.texi (mexpand-adddi): Update text.
17633 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17635 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17636 that also clobbers the CC register. The old expand code is moved
17638 (*arc_clzsi2): ... here.
17639 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17640 the CC register. The old expand code is moved to ...
17641 (arc_ctzsi2): ... here.
17643 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17645 * config/arc/arc.opt (mindexed-loads): Use initial value
17646 TARGET_INDEXED_LOADS_DEFAULT.
17647 (mauto-modify-reg): Use initial value
17648 TARGET_AUTO_MODIFY_REG_DEFAULT.
17649 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17650 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17651 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17652 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17654 2017-07-17 Martin Liska <mliska@suse.cz>
17657 * opts.c (finish_options): Do not allow -fgnu-tm
17658 w/ -fsanitize={kernel-,}address. Say sorry.
17660 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17663 * tree-loop-distribution.c (classify_partition): Only assert on
17664 numer of iterations.
17665 (merge_dep_scc_partitions): Delete prameter. Update function call.
17666 (distribute_loop): Remove code handling loop with unknown niters.
17667 (pass_loop_distribution::execute): Skip loop with unknown niters.
17669 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17672 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17673 function sort_partitions_by_post_order.
17675 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17677 PR tree-optimization/81374
17678 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17679 the max index of basic blocks, rather than number of basic blocks.
17681 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17683 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17685 (arc_legitimate_pic_operand_p): Likewise.
17686 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17688 (arc_needs_pcl_p): Likewise.
17689 (arc_legitimate_pc_offset_p): Likewise.
17690 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17691 function is also used in constrains.md.
17692 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17693 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17694 PLUS. Only return true/false in known cases, otherwise assert.
17695 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17696 is already called in arc_legitimate_constant_p.
17697 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17699 (LEGITIMATE_PIC_OPERAND_P): Use
17700 arc_raw_symbolic_reference_mentioned_p function.
17701 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17706 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17707 Andrew Burgess <andrew.burgess@embecosm.com>
17709 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17710 (arc_return_address_register): New function.
17711 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17712 (arc_handle_fndecl_attribute): Add naked attribute.
17713 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17714 (TARGET_WARN_FUNC_RETURN): Likewise.
17715 (arc_allocate_stack_slots_for_args): New function.
17716 (arc_warn_func_return): Likewise.
17717 (machine_function): Change type fn_type.
17718 (arc_compute_function_type): Consider new naked function type,
17719 change function return type.
17720 (arc_must_save_register): Adapt to handle new
17721 arc_compute_function_type's return type.
17722 (arc_expand_prologue): Likewise.
17723 (arc_expand_epilogue): Likewise.
17724 (arc_return_address_regs): Delete.
17725 (arc_return_address_register): New function.
17726 (arc_epilogue_uses): Use above function.
17727 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17728 (arc_function_type): Change encoding, add naked type.
17729 (ARC_INTERRUPT_P): Change to handle the new encoding.
17730 (ARC_FAST_INTERRUPT_P): Likewise.
17731 (ARC_NORMAL_P): Define.
17732 (ARC_NAKED_P): Likewise.
17733 (arc_compute_function_type): Delete prototype.
17734 * config/arc/arc.md (in_ret_delay_slot): Use
17735 arc_return_address_register function.
17736 (simple_return): Likewise.
17737 (p_return_i): Likewise.
17739 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17741 PR tree-optimization/81428
17742 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17743 can't be built for those types.
17745 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17747 Remove stuff dead since r239246.
17749 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17750 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17751 (avr_inform_devices): Remove dead stuff.
17753 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17755 * config/arm/arm_neon.h: Fix softp typo.
17757 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17759 PR tree-optimization/81365
17760 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17761 aggregate moves onto bb predecessor edges, make sure there are no
17762 loads that could alias the lhs in between the start of bb and the
17765 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17768 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17769 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17770 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17772 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17774 PR tree-optimization/81396
17775 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17776 (init_symbolic_number): Initialize it to 1.
17777 (perform_symbolic_merge): Add n_ops from both operands into the new
17779 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17780 without base_addr as useless if they need more than one operation.
17781 (bswap_replace): Handle !bswap case for NULL base_addr.
17783 2017-07-17 Tom de Vries <tom@codesourcery.com>
17786 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17789 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17791 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17792 conditional builtin define __FIX_LEON3FT_B2BST.
17794 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17796 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17797 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17800 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17802 PR rtl-optimization/81424
17803 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17804 to remove potential trapping from operands if -fnon-call-exceptions.
17806 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17808 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17809 profile_proability for scalling.
17810 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17812 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17814 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17816 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17818 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17819 fixpoint arithmetics.
17821 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17823 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17824 fixpoint arithmetics.
17826 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17828 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17829 fixpoint arithmetics.
17831 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17833 * profile-count.h (profile_probability::from_reg_br_prob_note,
17834 profile_probability::to_reg_br_prob_note): New functions.
17835 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17836 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17837 * predict.c (probability_reliable_p): Update.
17838 (edge_probability_reliable_p): Update.
17839 (br_prob_note_reliable_p): Update.
17840 (invert_br_probabilities): Update.
17841 (add_reg_br_prob_note): New function.
17842 (combine_predictions_for_insn): Update.
17843 * asan.c (asan_clear_shadow): Update.
17844 * cfgbuild.c (compute_outgoing_frequencies): Update.
17845 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17846 (update_br_prob_note): Update.
17847 (rtl_verify_edges): Update.
17848 (purge_dead_edges): Update.
17849 (fixup_reorder_chain): Update.
17850 * emit-rtl.c (try_split): Update.
17851 * ifcvt.c (cond_exec_process_insns): Update.
17852 (cond_exec_process_if_block): Update.
17853 (dead_or_predicable): Update.
17854 * internal-fn.c (expand_addsub_overflow): Update.
17855 (expand_neg_overflow): Update.
17856 (expand_mul_overflow): Update.
17857 * loop-doloop.c (doloop_modify): Update.
17858 * loop-unroll.c (compare_and_jump_seq): Update.
17859 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17860 * predict.h: Update.
17861 * reorg.c (mostly_true_jump): Update.
17863 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17864 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17865 * config/arc/arc.c: (emit_unlikely_jump): Update.
17866 * config/arm/arm.c: (emit_unlikely_jump): Update.
17867 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17868 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17869 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17870 (ix86_print_operand): Update.
17871 (ix86_split_fp_branch): Update.
17872 (predict_jump): Update.
17873 * config/ia64/ia64.c (ia64_print_operand): Update.
17874 * config/mmix/mmix.c (mmix_print_operand): Update.
17875 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17876 (rs6000_expand_split_stack_prologue): Update.
17877 * config/rs6000/rs6000.c: Update.
17878 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17879 (s390_expand_vec_movstr): Update.
17880 (s390_expand_cs_tdsi): Update.
17881 (s390_expand_split_stack_prologue): Update.
17882 * config/sh/sh.c (sh_print_operand): Update.
17883 (expand_cbranchsi4): Update.
17884 (expand_cbranchdi4): Update.
17885 * config/sparc/sparc.c (output_v9branch): Update.
17886 * config/spu/spu.c (get_branch_target): Update.
17887 (ea_load_store_inline): Update.
17888 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17889 * config/tilepro/tilepro.c: Update.
17891 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17893 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17894 (gimplify_save_expr): Likewise.
17896 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17898 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17900 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17902 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17904 * timevar.def (TV_IPA_FNSUMMARY): Define.
17906 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17908 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17909 to back store errata sensitive sequence from being generated.
17910 (sqrtdf2_fix): Likewise.
17912 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17914 * tree-ssa-threadupdate.c (compute_path_counts,
17915 update_joiner_offpath_counts): Use profile_probability.
17917 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17920 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17922 * config/arm/arm-c.c (arm_cpu_builtins): Define
17923 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17925 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17927 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17928 array entries to represent __ieee128 versions of the
17929 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17930 scalar_extract_sig, and scalar_insert_exp built-in functions.
17931 (altivec_resolve_overloaded_builtin): Add special case handling
17932 for the __builtin_scalar_insert_exp function, as represented by
17933 the P9V_BUILTIN_VEC_VSIEDP constant.
17934 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17935 exponent support for __ieee128 argument.
17936 (VSESQP): Add scalar extract signature support for __ieee128
17938 (VSTDCNQP): Add scalar test negative support for __ieee128
17940 (VSIEQP): Add scalar insert exponent support for __int128 argument
17941 with __ieee128 result.
17942 (VSIEQPF): Add scalar insert exponent support for __ieee128
17943 argument with __ieee128 result.
17944 (VSTDCQP): Add scalar test data class support for __ieee128
17946 (VSTDCNQP): Add overload support for scalar test negative with
17947 __ieee128 argument.
17948 (VSTDCQP): Add overload support for scalar test data class
17949 __ieee128 argument.
17950 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17951 UNSPEC_VSX_SXSIGDP.
17952 (UNSPEC_VSX_SIEXPQP): New constant.
17953 (xsxexpqp): New insn for VSX scalar extract exponent quad
17955 (xsxsigqp): New insn for VSX scalar extract significand quad
17957 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17958 precision with floating point argument.
17959 (xststdcqp): New expand for VSX scalar test data class quad
17961 (xststdcnegqp): New expand for VSX scalar test negative quad
17963 (xststdcqp): New insn to match expansions for VSX scalar test data
17964 class quad precision and VSX scalar test negative quad precision.
17965 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17966 special case operand checking to enforce that second operand of
17967 VSX scalar test data class with quad precision argument is a 7-bit
17969 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17970 prototypes and descriptions of __ieee128 versions of
17971 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17972 scalar_test_data_class, and scalar_test_neg built-in functions.
17974 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17976 PR tree-optimization/81162
17977 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17978 replace a negate with an add.
17980 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17982 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17984 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17986 * config/arm/arm-c.c (arm_cpu_builtins): Define
17987 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17989 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17991 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17992 (armv8-r): Set ARM Cortex-R52 as default CPU.
17993 * config/arm/arm-tables.opt: Regenerate.
17994 * config/arm/arm-tune.md: Regenerate.
17995 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17997 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17998 extension for -mcpu=cortex-r52.
18000 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
18002 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
18003 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
18004 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
18005 (fp-armv8): Define it as FP_ARMv8 only.
18006 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
18007 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
18009 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
18010 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
18011 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
18012 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
18013 than TARGET_FPU_ARMV8.
18014 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
18015 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
18016 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
18018 * config/arm/neon.md (neon_vrint): Likewise.
18019 (neon_vcvt): Likewise.
18020 (neon_<fmaxmin_op><mode>): Likewise.
18021 (<fmaxmin><mode>3): Likewise.
18022 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
18023 * config/arm/predicates.md (arm_cond_move_operator): Check against
18024 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
18026 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
18028 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
18029 to top of function.
18031 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18033 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
18034 loop in comment with memset.
18036 2017-07-14 Martin Liska <mliska@suse.cz>
18038 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
18039 * dwarf2out.c (is_java): Remove the function.
18040 (output_pubname): Remove usage of the function.
18041 (lower_bound_default): Remove usage of DW_LANG_Java.
18042 (gen_compile_unit_die): Likewise.
18043 * gcc.c: Remove compiler defaults for .java and .zip files.
18044 * gimple-expr.c (remove_suffix): Change as there's no longer
18045 extension than 4-letter one.
18046 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
18047 (gimplify_save_expr): Likewise.
18048 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
18049 as it's possible even for other languages than Java.
18050 * langhooks.h (struct lang_hooks): Remove Java from a comment.
18051 * lto-opts.c (lto_write_options): Remove reference to Java.
18052 * opts.c (strip_off_ending): Update file extension handling.
18053 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
18054 * tree-eh.c (lower_resx): Likewise.
18055 * tree.c (free_lang_data_in_type): Remove dead code.
18056 (find_decls_types_r): Likewise.
18057 (build_common_builtin_nodes): Remove Java from a comment.
18058 (verify_type): Remove dead code.
18059 * varasm.c (assemble_external): Remove Java from a comment.
18061 2017-07-14 Martin Liska <mliska@suse.cz>
18063 * opts.c (finish_options): Add quotes.
18064 (common_handle_option): Likewise.
18066 2017-07-14 Martin Liska <mliska@suse.cz>
18068 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
18069 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
18070 Remove N_SO_PASCAL.
18071 * dwarf2out.c (lower_bound_default): Do not handle
18073 (gen_compile_unit_die): Likewise.
18074 * gcc.c: Remove default extension binding for GNU Pascal.
18075 * stmt.c: Remove Pascal language from a comment.
18076 * xcoffout.c: Likewise.
18078 2017-07-13 David Malcolm <dmalcolm@redhat.com>
18081 * diagnostic-show-locus.c (fixit_cmp): New function.
18082 (layout::layout): Sort m_fixit_hints.
18083 (column_range::column_range): Assert that the values are valid.
18084 (struct char_span): New struct.
18085 (correction::overwrite): New method.
18086 (struct source_line): New struct.
18087 (line_corrections::add_hint): Add assertions. Reimplement memcpy
18088 calls in terms of classes source_line and char_span, and
18089 correction::overwrite.
18090 (selftest::test_overlapped_fixit_printing_2): New function.
18091 (selftest::diagnostic_show_locus_c_tests): Call it.
18093 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
18095 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
18096 early if there is no lhs.
18098 2017-07-13 Martin Liska <mliska@suse.cz>
18100 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18101 (gen_reference_type_die): Likewise.
18102 * stor-layout.c: Remove Pascal-related comment.
18104 2017-07-13 Martin Liska <mliska@suse.cz>
18106 * opts.c (finish_options): Add quotes to error messages.
18107 (parse_sanitizer_options): Likewise.
18109 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18111 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18113 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18115 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18117 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18119 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18121 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18123 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18126 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18127 provides the hardware capability bits, define the macro
18128 __BUILTIN_CPU_SUPPORTS__.
18129 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18130 if GLIBC does not provide the hardware capability bits. Add a
18131 gcc_unreachable call if the built-in cpu function is neither
18132 __builtin_cpu_is nor __builtin_cpu_supports.
18133 (rs6000_get_function_versions_dispatcher): Change the warning
18134 that an old GLIBC is used which does not export the capability
18135 bits to be an error.
18136 * doc/extend.texi (target_clones attribute): Document the
18137 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18138 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18139 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18140 the macros defined by GCC if the newer GLIBC is available.
18142 2017-07-12 Jeff Law <law@redhat.com>
18144 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18145 remaining includes slightly.
18146 * config/riscv/riscv-builtins.c: Include profile-count.h.
18148 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18151 * config/avr/avr.c (avr_set_current_function): In diagnostic
18152 messages: Quote keywords and (parts of) identifiers.
18153 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18156 2017-07-12 Carl Love <cel@us.ibm.com>
18158 * config/rs6000/rs6000-c.c: Add support for built-in functions
18159 vector bool char vec_revb (vector bool char);
18160 vector bool short vec_revb (vector short char);
18161 vector bool int vec_revb (vector bool int);
18162 vector bool long long vec_revb (vector bool long long);
18163 * doc/extend.texi: Update the built-in documentation file for the
18164 new built-in functions.
18166 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18168 * config/s390/s390.md: Remove movcc splitter.
18170 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18172 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18173 load/store on condition.
18175 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18178 * config/avr/avr.c (avr_encode_section_info)
18179 [progmem && !TREE_READONLY]: Error if progmem object needs
18182 2017-07-11 Michael Collison <michael.collison@arm.com>
18184 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18187 2017-07-11 Carl Love <cel@us.ibm.com>
18189 * config/rs6000/rs6000-c.c: Add support for builtins
18190 vector unsigned int vec_parity_lsbb (vector signed int);
18191 vector unsigned int vec_parity_lsbb (vector unsigned int);
18192 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18193 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18194 vector unsigned long long vec_parity_lsbb (vector signed long long);
18195 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18196 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18197 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18198 * doc/extend.texi: Update the built-in documentation file for the
18199 new built-in functions.
18201 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18203 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18204 (layout::m_primary_loc): New field.
18205 (layout::layout): Initialize new field. Move location filtering
18206 logic from here to...
18207 (layout::maybe_add_location_range): ...this new method. Add
18208 support for filtering to just the lines already specified by other
18210 (layout::will_show_line_p): New method.
18211 (gcc_rich_location::add_location_if_nearby): New method.
18212 (selftest::test_add_location_if_nearby): New test function.
18213 (selftest::diagnostic_show_locus_c_tests): Call it.
18214 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18217 2017-07-11 Tom de Vries <tom@codesourcery.com>
18219 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18220 (bb_first_real_insn): New function.
18221 (nvptx_single): Add extra initialization of broadcasted condition
18224 2017-07-11 Nathan Sidwell <nathan@acm.org>
18226 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18228 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18230 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18231 Binutils doc as TEXI will mess them up.
18232 * doc/invoke.texi (AVR Options): Same here.
18234 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18236 * config/sparc/sparc.opt (mfix-ut700): New option.
18237 (mfix-gr712rc): Likewise.
18238 (sparc_fix_b2bst): New variable.
18239 * doc/invoke.texi (SPARC options): Document them.
18240 (ARM options): Fix warnings.
18241 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18242 instructions to prevent sequences that can trigger the store-store
18243 errata for certain LEON3FT processors.
18244 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18245 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18246 * config/sparc/sparc.md (fix_b2bst): New attribute.
18247 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18249 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18252 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18254 (*rsqrtsf2_sse): Ditto.
18256 (div<mode>3): Macroize insn from divdf3 and divsf3
18257 using MODEF mode iterator.
18259 2017-07-10 Martin Sebor <msebor@redhat.com>
18261 PR tree-optimization/80397
18262 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18263 instead of testing for equality to INTEGER_TYPE.
18265 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18267 * config.gcc: Remove uclibc from arc target spec.
18269 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18271 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18273 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18276 * lto-wrapper.c (remove_option): New function.
18277 (merge_and_complain): Merge PIC/PIE options more realistically.
18279 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18281 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18285 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18286 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18287 * config.in: Regenerate.
18288 * configure: Regenerate.
18289 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18290 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18291 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18292 (TARGET_GASISR_PROLOGUES): ...target mask.
18293 * common/config/avr/avr-common.c
18294 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18295 Set -mgas-isr-prologues.
18296 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18297 INSERT_PASS_BEFORE for it.
18298 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18299 * config/avr/avr.c (avr_option_override)
18300 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18301 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18302 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18303 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18304 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18306 (make_avr_pass_pre_proep): New function.
18307 (emit_push_sfr) <treg>: Add argument to function and use it
18308 instead of TMP_REG.
18309 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18310 and set machine->gasisr.yes.
18311 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18312 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18313 __gcc_isr.n_pushed to .L__stack_usage.
18314 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18315 (avr_asm_final_postscan_insn): ...this new static function.
18316 * config/avr/avr.h (machine_function)
18317 <is_no_gccisr, use_L__stack_usage>: New fields.
18318 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18319 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18320 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18321 (gasisr, *gasisr): New expander and insn.
18322 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18323 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18324 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18326 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18328 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18331 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18333 Move jump-tables out of .text again.
18336 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18337 (ASM_OUTPUT_ADDR_VEC): New function.
18338 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18339 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18340 INSN_ADDRESSes as asm comment.
18341 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18342 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18343 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18344 * config/avr/avr.md (*tablejump): Adjust comment.
18345 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18346 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18348 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18349 (avr_output_addr_vec): New proto.
18350 (avr_log_t) <insn_addresses>: New field.
18352 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18355 * config/i386/i386.c (ix86_function_arg_advance): Set
18356 outgoing_args_on_stack to true if there are outgoing arguments
18358 (ix86_function_arg): Likewise.
18359 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18360 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18361 * config/i386/i386.h (machine_function): Add
18362 outgoing_args_on_stack.
18364 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18366 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18367 supporting pthreds.
18368 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18370 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18372 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18373 (REAL_H): Remove $(MACHMODE_H).
18374 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18376 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18377 $(MACHMODE_H) and double-int.h.
18378 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18380 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18383 2017-07-07 Andrew Pinski <apinski@cavium.com>
18385 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18386 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18388 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18390 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18391 Add warning if GCC was not configured to link against a GLIBC that
18392 exports the hardware capability bits.
18393 (make_resolver_func): Make resolver function private and not a
18394 COMDAT function. Create the name with clone_function_name instead
18395 of make_unique_name.
18398 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18399 correct operand in doing the split.
18401 2017-07-07 Carl Love <cel@us.ibm.com>
18403 * config/rs6000/rs6000-c: Add support for built-in function
18404 vector unsigned short vec_pack_to_short_fp32 (vector float,
18406 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18407 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18408 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18409 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18410 (convert_4f32_8i16): Add define_expand.
18411 * doc/extend.texi: Update the built-in documentation file for the
18412 new built-in function.
18414 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18416 * config/sparc/m8.md: New file.
18417 * config/sparc/sparc.md: Include m8.md.
18419 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18421 * config/sparc/sparc.opt: New option -mvis4b.
18422 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18423 (sparc_option_override): Handle VIS4B.
18424 (enum sparc_builtins): Define
18425 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18426 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18427 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18428 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18429 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18430 (check_constant_argument): New function.
18431 (sparc_vis_init_builtins): Define builtins
18432 __builtin_vis_dictunpack{8,16,32},
18433 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18434 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18435 __builtin_vis_fpcmpde{8,16,32}shl and
18436 __builtin_vis_fpcmpur{8,16,32}shl.
18437 (sparc_expand_builtin): Check that the constant operands to
18438 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18439 constant and in range.
18440 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18442 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18443 (SPARC_IMM5_P): Likewise.
18444 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18445 (enabled): Handle vis4b.
18446 (UNSPEC_DICTUNPACK): New unspec.
18447 (UNSPEC_FPCMPSHL): Likewise.
18448 (UNSPEC_FPUCMPSHL): Likewise.
18449 (UNSPEC_FPCMPDESHL): Likewise.
18450 (UNSPEC_FPCMPURSHL): Likewise.
18451 (cpu_feature): New CPU feature `vis4b'.
18452 (dictunpack{8,16,32}): New insns.
18453 (FPCSMODE): New mode iterator.
18454 (fpcscond): New code iterator.
18455 (fpcsucond): Likewise.
18456 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18457 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18458 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18459 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18460 * config/sparc/constraints.md: Define constraints `q' for unsigned
18461 2-bit integer constants and `t' for unsigned 5-bit integer
18463 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18465 (imm5_operand_dictunpack16): Likewise.
18466 (imm5_operand_dictunpack32): Likewise.
18467 (imm2_operand): Likewise.
18468 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18469 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18470 ditunpack* and fpcmp*shl builtins.
18472 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18474 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18475 * config.in: Add HAVE_AS_SPARC6 define.
18476 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18478 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18480 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18481 (CPP_CPU_SPEC): Handle m8.
18482 (ASM_CPU_SPEC): Likewise.
18483 * config/sparc/sparc-opts.h (enum processor_type): Add
18485 * config/sparc/sparc.c (m8_costs): New struct.
18486 (sparc_option_override): Handle TARGET_CPU_m8.
18487 (sparc32_initialize_trampoline): Likewise.
18488 (sparc64_initialize_trampoline): Likewise.
18489 (sparc_issue_rate): Likewise.
18490 (sparc_register_move_cost): Likewise.
18491 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18492 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18493 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18494 (CPP_CPU_SPEC): Handle M8.
18495 (ASM_CPU_SPEC): Likewise.
18496 (AS_M8_FLAG): Define.
18497 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18498 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18499 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18501 * configure: Regenerate.
18502 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18505 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18507 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18509 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18510 ("*movdi_insn_sp32"): Do not set v3pipe.
18511 ("*movsi_insn"): Likewise.
18512 ("*movdi_insn_sp64"): Likewise.
18513 ("*movsf_insn"): Likewise.
18514 ("*movdf_insn_sp32"): Likewise.
18515 ("*movdf_insn_sp64"): Likewise.
18516 ("*zero_extendsidi2_insn_sp64"): Likewise.
18517 ("*sign_extendsidi2_insn"): Likewise.
18518 ("*mov<VM32:mode>_insn"): Likewise.
18519 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18520 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18521 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18522 ("<vlop:code><VL:mode>3"): Likewise.
18523 ("*not_<vlop:code><VL:mode>3"): Likewise.
18524 ("*nand<VL:mode>_vis"): Likewise.
18525 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18526 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18527 ("one_cmpl<VL:mode>2"): Likewise.
18528 ("faligndata<VM64:mode>_vis"): Likewise.
18529 ("alignaddrsi_vis"): Likewise.
18530 ("alignaddrdi_vis"): Likweise.
18531 ("alignaddrlsi_vis"): Likewise.
18532 ("alignaddrldi_vis"): Likewise.
18533 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18534 ("bmaskdi_vis"): Likewise.
18535 ("bmasksi_vis"): Likewise.
18536 ("bshuffle<VM64:mode>_vis"): Likewise.
18537 ("cmask8<P:mode>_vis"): Likewise.
18538 ("cmask16<P:mode>_vis"): Likewise.
18539 ("cmask32<P:mode>_vis"): Likewise.
18540 ("pdistn<P:mode>_vis"): Likewise.
18541 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18543 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18545 * config/sparc/sparc.md ("subtype"): New insn attribute.
18546 ("*wrgsr_sp64"): Set insn subtype.
18547 ("*rdgsr_sp64"): Likewise.
18548 ("alignaddrsi_vis"): Likewise.
18549 ("alignaddrdi_vis"): Likewise.
18550 ("alignaddrlsi_vis"): Likewise.
18551 ("alignaddrldi_vis"): Likewise.
18552 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18553 ("fexpand_vis"): Likewise.
18554 ("fpmerge_vis"): Likewise.
18555 ("faligndata<VM64:mode>_vis"): Likewise.
18556 ("bshuffle<VM64:mode>_vis"): Likewise.
18557 ("cmask8<P:mode>_vis"): Likewise.
18558 ("cmask16<P:mode>_vis"): Likewise.
18559 ("cmask32<P:mode>_vis"): Likewise.
18560 ("fchksm16_vis"): Likewise.
18561 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18562 ("fmean16_vis"): Likewise.
18563 ("fp<plusminus_insn>64_vis"): Likewise.
18564 ("<plusminus_insn>v8qi3"): Likewise.
18565 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18566 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18567 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18568 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18569 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18570 ("*movqi_insn"): Likewise.
18571 ("*movhi_insn"): Likewise.
18572 ("*movsi_insn"): Likewise.
18573 ("movsi_pic_gotdata_op"): Likewise.
18574 ("*movdi_insn_sp32"): Likewise.
18575 ("*movdi_insn_sp64"): Likewise.
18576 ("movdi_pic_gotdata_op"): Likewise.
18577 ("*movsf_insn"): Likewise.
18578 ("*movdf_insn_sp32"): Likewise.
18579 ("*movdf_insn_sp64"): Likewise.
18580 ("*zero_extendhisi2_insn"): Likewise.
18581 ("*zero_extendqihi2_insn"): Likewise.
18582 ("*zero_extendqisi2_insn"): Likewise.
18583 ("*zero_extendqidi2_insn"): Likewise.
18584 ("*zero_extendhidi2_insn"): Likewise.
18585 ("*zero_extendsidi2_insn_sp64"): Likewise.
18586 ("ldfsr"): Likewise.
18587 ("prefetch_64"): Likewise.
18588 ("prefetch_32"): Likewise.
18589 ("tie_ld32"): Likewise.
18590 ("tie_ld64"): Likewise.
18591 ("*tldo_ldub_sp32"): Likewise.
18592 ("*tldo_ldub1_sp32"): Likewise.
18593 ("*tldo_ldub2_sp32"): Likewise.
18594 ("*tldo_ldub_sp64"): Likewise.
18595 ("*tldo_ldub1_sp64"): Likewise.
18596 ("*tldo_ldub2_sp64"): Likewise.
18597 ("*tldo_ldub3_sp64"): Likewise.
18598 ("*tldo_lduh_sp32"): Likewise.
18599 ("*tldo_lduh1_sp32"): Likewise.
18600 ("*tldo_lduh_sp64"): Likewise.
18601 ("*tldo_lduh1_sp64"): Likewise.
18602 ("*tldo_lduh2_sp64"): Likewise.
18603 ("*tldo_lduw_sp32"): Likewise.
18604 ("*tldo_lduw_sp64"): Likewise.
18605 ("*tldo_lduw1_sp64"): Likewise.
18606 ("*tldo_ldx_sp64"): Likewise.
18607 ("*mov<VM32:mode>_insn"): Likewise.
18608 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18609 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18611 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18613 * config/sparc/sparc.md ("type"): New insn type viscmp.
18614 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18616 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18617 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18618 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18619 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18621 ("n7_vis_logical_11cycle"): Likewise.
18622 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18623 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18624 * config/sparc/niagara.md ("niag_vis"): Likewise.
18625 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18626 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18628 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18630 * config/sparc/sparc.md: New instruction type `bmask'.
18631 (bmaskdi_vis): Use the `bmask' type.
18632 (bmasksi_vis): Likewise.
18633 * config/sparc/ultra3.md (us3_array): Likewise.
18634 * config/sparc/niagara7.md (n7_array): Likewise.
18635 * config/sparc/niagara4.md (n4_array): Likewise.
18636 * config/sparc/niagara2.md (niag2_vis): Likewise.
18637 (niag3_vis): Likewise.
18638 * config/sparc/niagara.md (niag_vis): Likewise.
18640 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18642 * ipa-comdats.c: Remove optimize check from gate.
18643 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18644 for functions not optimized.
18645 (ipa_fn_summary_read): Skip optimize check.
18646 (ipa_fn_summary_write): Likewise.
18647 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18649 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18651 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18652 (check_callers): Give up on caller which is not optimized.
18653 (inline_small_functions): Likewise.
18654 (ipa_inline): Do not give up when not optimizing.
18655 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18656 away unoptimizes cdtors.
18657 (whole_program_function_and_variable_visibility): Do
18658 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18659 * ipa.c (process_references): Do not check optimize.
18660 (symbol_table::remove_unreachable_nodes): Update optimize check.
18661 (set_writeonly_bit): Update optimize check.
18662 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18663 (pass_ipa_single_use::gate): Remove.
18665 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18667 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18668 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18669 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18670 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18671 permute_load, permute_store, adjust_extract, adjust_splat,
18672 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18673 replace_swap_with_copy, dump_swap_insn_table,
18674 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18675 recombine_lvx_pattern, recombine_stvx_pattern,
18676 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18677 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18678 to file rs6000-p8swap.c.
18679 * config/rs6000/rs6000-p8swap.c: New file.
18680 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18681 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18682 and rs6000*-*-* targets.
18684 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18686 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18688 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18690 * lto-wrapper.c (merge_and_complain): Do not merge
18691 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18692 fsigned_zeros, ftrapping_math, fwrapv.
18693 (append_compiler_options): Do not track these options.
18694 (append_linker_options): Likewie
18696 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18698 * cgraphunit.c (cgraph_node::finalize_function): When
18699 !flag_toplevel_reorde set no_reorder flag.
18700 (varpool_node::finalize_decl): Likewise.
18701 (symbol_table::compile): Drop no toplevel reorder path.
18703 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18705 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18706 edges; zero probability is not better than uninitialized.
18708 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18710 * asan.h (asan_sanitize_allocas_p): Declare.
18711 * asan.c (asan_sanitize_allocas_p): New function.
18712 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18713 (handle_builtin_alloca): Likewise.
18714 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18715 if !asan_sanitize_allocas_p.
18716 * params.def (asan-instrument-allocas): Add new option.
18717 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18718 * opts.c (common_handle_option): Disable allocas sanitization for
18721 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18723 * asan.c: Include gimple-fold.h.
18724 (get_last_alloca_addr): New function.
18725 (handle_builtin_stackrestore): Likewise.
18726 (handle_builtin_alloca): Likewise.
18727 (asan_emit_allocas_unpoison): Likewise.
18728 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18729 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18730 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18731 (instrument_builtin_call): Pass gimple iterator to
18732 get_mem_refs_of_builtin_call.
18733 (last_alloca_addr): New global.
18734 * asan.h (asan_emit_allocas_unpoison): Declare.
18735 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18736 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18737 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18738 if function calls alloca.
18739 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18740 * gimple-fold.h (replace_call_with_value): Declare.
18741 * internal-fn.c: Include asan.h.
18742 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18743 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18745 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18747 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18748 (C_SELFTEST_FLAGS): New.
18749 (CPP_SELFTEST_FLAGS): New.
18750 (SELFTEST_DEPS): New, from deps of s-selftest.
18751 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18752 (CPP_SELFTEST_DEPS): New.
18753 (selftest): Add dependency on s-selftest-c++.
18754 (s-selftest): Rename to...
18755 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18756 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18757 than SELFTEST_FLAGS.
18758 (selftest-gdb): Rename to...
18759 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18761 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18762 (selftest-valgrind): Rename to...
18763 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18765 (selftest-valgrind): Reintroduce as an alias for
18766 selftest-c-valgrind.
18767 (s-selftest-c++): New.
18768 (selftest-c++-gdb): New.
18769 (selftest-c++-valgrind): New.
18771 2017-07-06 Olivier Hainque <hainque@adacore.com>
18773 * gcc.c (process_command): When deciding if undefined variables
18774 should be ignored when processing specs, accept "gcc -v" as well.
18776 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18778 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18779 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18781 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18783 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18784 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18785 * config/arm/arm-tables.opt: Regenerate.
18786 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18788 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18790 2017-07-06 Carl Love <cel@us.ibm.com>
18792 * ChangeLog: Clean up from mid air collision
18794 2017-07-06 Carl Love <cel@us.ibm.com>
18796 * config/rs6000/rs6000-c.c: Add support for built-in functions
18797 vector signed int vec_subc (vector signed int, vector signed int);
18798 vector signed __int128 vec_subc (vector signed __int128,
18799 vector signed __int128);
18800 vector unsigned __int128 vec_subc (vector unsigned __int128,
18801 vector unsigned __int128);
18802 vector signed int vec_sube (vector signed int, vector signed int,
18803 vector signed int);
18804 vector unsigned int vec_sube (vector unsigned int,
18805 vector unsigned int,
18806 vector unsigned int);
18807 vector signed __int128 vec_sube (vector signed __int128,
18808 vector signed __int128,
18809 vector signed__int128);
18810 vector unsigned __int128 vec_sube (vector unsigned __int128,
18811 vector unsigned __int128,
18812 vector unsigned __int128);
18813 vector signed int vec_subec (vector signed int, vector signed int,
18814 vector signed int);
18815 vector unsigned int vec_subec (vector unsigned int,
18816 vector unsigned int,
18817 vector unsigned int);
18818 vector signed __int128 vec_subec (vector signed __int128,
18819 vector signed __int128,
18820 vector signed__int128);
18821 vector unsigned __int128 vec_subec (vector unsigned __int128,
18822 vector unsigned __int128,
18823 vector unsigned __int128);
18824 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18825 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18826 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18827 BU_ALTIVEC_OVERLOAD_X definitions.
18828 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18829 * doc/extend.texi: Update the built-in documentation file for the new
18830 built-in functions.
18832 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18835 * diagnostic-show-locus.c (layout::layout): Use start and finish
18836 spelling location for the start and finish of each range.
18837 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18838 Add unused aspect param.
18839 * input.c (expand_location_1): Add "aspect" param, and use it
18840 to access the correct part of the location.
18841 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18843 (expand_location_to_spelling_point): Likewise.
18844 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18845 param, and pass it to expand_location_1.
18847 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18849 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18850 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18851 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18852 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18853 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18854 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18855 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18856 _mm_maskz_getmant_ss): New intrinsics.
18857 (__builtin_ia32_getexpss128_mask): Changed to ...
18858 __builtin_ia32_getexpss128_round ... this.
18859 (__builtin_ia32_getexpsd128_mask): Changed to ...
18860 __builtin_ia32_getexpsd128_round ... this.
18861 * config/i386/i386-builtin-types.def
18862 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18863 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18864 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18865 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18866 __builtin_ia32_getmantss_mask_round): New builtins.
18867 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18868 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18869 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18870 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18871 * config/i386/sse.md
18872 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18873 avx512f_sgetexp<mode><mask_scalar_name>
18874 <round_saeonly_scalar_name> ... this.
18875 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18876 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18877 vgetexp<ssescalarmodesuffix>
18878 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18879 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18880 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18881 avx512f_vgetmant<mode><mask_scalar_name>
18882 <round_saeonly_scalar_name> ... this.
18883 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18884 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18885 vgetmant<ssescalarmodesuffix>
18886 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18887 %0<mask_scalar_operand4>, %1, %2
18888 <round_saeonly_scalar_mask_op4>, %3} ... this.
18889 * config/i386/subst.md (mask_scalar_operand4,
18890 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18891 round_saeonly_scalar_nimm_predicate): New subst attributes.
18893 2017-07-06 Julia Koval <julia.koval@intel.com>
18895 * config/i386/i386.c (ix86_erase_embedded_rounding):
18896 Remove code for old rounding pattern.
18898 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18900 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18902 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18904 * doc/sourcebuild.texi (Test Directives, Variants of
18905 dg-require-support): Add documentation for dg-require-stack-check.
18907 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18909 * config/i386/subst.md (mask_scalar, round_scalar,
18910 round_saeonly_scalar): New meta-templates.
18911 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18912 round_scalar_mask_operand3, round_scalar_mask_op3,
18913 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18914 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18915 round_saeonly_scalar_constraint,
18916 round_saeonly_scalar_prefix): New subst attribute.
18917 * config/i386/sse.md
18918 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18919 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18920 <round_scalar_name> ... this.
18921 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18922 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18923 <round_scalar_name> ... this.
18924 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18925 <sse>_vm<code><mode>3<mask_scalar_name>
18926 <round_saeonly_scalar_name> ... this.
18927 (v<plusminus_mnemonic><ssescalarmodesuffix>
18928 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18929 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18930 v<plusminus_mnemonic><ssescalarmodesuffix>
18931 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18932 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18933 (v<multdiv_mnemonic><ssescalarmodesuffix>
18934 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18935 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18936 v<multdiv_mnemonic><ssescalarmodesuffix>
18937 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18938 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18939 (v<maxmin_float><ssescalarmodesuffix>
18940 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18941 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18942 v<maxmin_float><ssescalarmodesuffix>
18943 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18944 %0<mask_scalar_operand3>, %1, %<iptr>2
18945 <round_saeonly_scalar_mask_op3>} ... this.
18947 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18949 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18950 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18952 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18953 Alan Hayward <alan.hayward@arm.com>
18954 David Sherwood <david.sherwood@arm.com>
18956 * combine.c (simplify_if_then_else): Remove "enum" before
18958 * compare-elim.c (can_eliminate_compare): Likewise.
18959 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18961 (aarch64_lookup_simd_builtin_type): Likewise.
18962 (aarch64_simd_builtin_type): Likewise.
18963 (aarch64_init_simd_builtin_types): Likewise.
18964 (aarch64_simd_expand_args): Likewise.
18965 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18967 (aarch64_reverse_mask): Likewise.
18968 (aarch64_simd_emit_reg_reg_move): Likewise.
18969 (aarch64_gen_adjusted_ldpstp): Likewise.
18970 (aarch64_ccmp_mode_to_code): Likewise.
18971 (aarch64_operands_ok_for_ldpstp): Likewise.
18972 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18973 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18975 (aarch64_min_divisions_for_recip_mul): Likewise.
18976 (aarch64_reassociation_width): Likewise.
18977 (aarch64_get_condition_code_1): Likewise.
18978 (aarch64_simd_emit_reg_reg_move): Likewise.
18979 (aarch64_simd_attr_length_rglist): Likewise.
18980 (aarch64_reverse_mask): Likewise.
18981 (aarch64_operands_ok_for_ldpstp): Likewise.
18982 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18983 (aarch64_gen_adjusted_ldpstp): Likewise.
18984 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18986 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18987 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18988 (arm_lookup_simd_builtin_type): Likewise.
18989 (arm_simd_builtin_type): Likewise.
18990 (arm_init_simd_builtin_types): Likewise.
18991 (arm_expand_builtin_args): Likewise.
18992 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18993 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18994 (ft32_setup_incoming_varargs): Likewise.
18995 (ft32_function_arg): Likewise.
18996 (ft32_function_arg_advance): Likewise.
18997 (ft32_pass_by_reference): Likewise.
18998 (ft32_arg_partial_bytes): Likewise.
18999 (ft32_valid_pointer_mode): Likewise.
19000 (ft32_addr_space_pointer_mode): Likewise.
19001 (ft32_addr_space_legitimate_address_p): Likewise.
19002 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
19004 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
19005 (ix86_emit_outlined_ms2sysv_restore): Likewise.
19006 (iamcu_alignment): Likewise.
19007 (canonicalize_vector_int_perm): Likewise.
19008 (ix86_noce_conversion_profitable_p): Likewise.
19009 (ix86_mpx_bound_mode): Likewise.
19010 (ix86_operands_ok_for_move_multiple): Likewise.
19011 * config/microblaze/microblaze-protos.h
19012 (microblaze_expand_conditional_branch_reg): Likewise.
19013 * config/microblaze/microblaze.c
19014 (microblaze_expand_conditional_branch_reg): Likewise.
19015 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19017 (rs6000_reassociation_width): Likewise.
19018 (rs6000_invalid_binary_op): Likewise.
19019 (fusion_p9_p): Likewise.
19020 (emit_fusion_p9_load): Likewise.
19021 (emit_fusion_p9_store): Likewise.
19022 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
19024 (riscv_hard_regno_mode_ok_p): Likewise.
19025 (riscv_address_insns): Likewise.
19026 (riscv_split_symbol): Likewise.
19027 (riscv_legitimize_move): Likewise.
19028 (riscv_function_value): Likewise.
19029 (riscv_hard_regno_nregs): Likewise.
19030 (riscv_expand_builtin): Likewise.
19031 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
19032 (riscv_build_integer): Likewise.
19033 (riscv_split_integer): Likewise.
19034 (riscv_legitimate_constant_p): Likewise.
19035 (riscv_cannot_force_const_mem): Likewise.
19036 (riscv_regno_mode_ok_for_base_p): Likewise.
19037 (riscv_valid_base_register_p): Likewise.
19038 (riscv_valid_offset_p): Likewise.
19039 (riscv_valid_lo_sum_p): Likewise.
19040 (riscv_classify_address): Likewise.
19041 (riscv_legitimate_address_p): Likewise.
19042 (riscv_address_insns): Likewise.
19043 (riscv_load_store_insns): Likewise.
19044 (riscv_force_binary): Likewise.
19045 (riscv_split_symbol): Likewise.
19046 (riscv_force_address): Likewise.
19047 (riscv_legitimize_address): Likewise.
19048 (riscv_move_integer): Likewise.
19049 (riscv_legitimize_const_move): Likewise.
19050 (riscv_legitimize_move): Likewise.
19051 (riscv_address_cost): Likewise.
19052 (riscv_subword): Likewise.
19053 (riscv_output_move): Likewise.
19054 (riscv_canonicalize_int_order_test): Likewise.
19055 (riscv_emit_int_order_test): Likewise.
19056 (riscv_function_arg_boundary): Likewise.
19057 (riscv_pass_mode_in_fpr_p): Likewise.
19058 (riscv_pass_fpr_single): Likewise.
19059 (riscv_pass_fpr_pair): Likewise.
19060 (riscv_get_arg_info): Likewise.
19061 (riscv_function_arg): Likewise.
19062 (riscv_function_arg_advance): Likewise.
19063 (riscv_arg_partial_bytes): Likewise.
19064 (riscv_function_value): Likewise.
19065 (riscv_pass_by_reference): Likewise.
19066 (riscv_setup_incoming_varargs): Likewise.
19067 (riscv_print_operand): Likewise.
19068 (riscv_elf_select_rtx_section): Likewise.
19069 (riscv_save_restore_reg): Likewise.
19070 (riscv_for_each_saved_reg): Likewise.
19071 (riscv_register_move_cost): Likewise.
19072 (riscv_hard_regno_mode_ok_p): Likewise.
19073 (riscv_hard_regno_nregs): Likewise.
19074 (riscv_class_max_nregs): Likewise.
19075 (riscv_memory_move_cost): Likewise.
19076 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
19077 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
19078 (rl78_addr_space_address_mode): Likewise.
19079 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19081 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19082 (rs6000_reassociation_width): Likewise.
19083 (rs6000_invalid_binary_op): Likewise.
19084 (fusion_p9_p): Likewise.
19085 (emit_fusion_p9_load): Likewise.
19086 (emit_fusion_p9_store): Likewise.
19087 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
19088 (ok_for_simple_move_operands): Likewise.
19089 (ok_for_simple_move_strict_operands): Likewise.
19090 (ok_for_simple_arith_logic_operands): Likewise.
19091 (visium_legitimize_reload_address): Likewise.
19092 (visium_select_cc_mode): Likewise.
19093 (output_cbranch): Likewise.
19094 (visium_split_double_move): Likewise.
19095 (visium_expand_copysign): Likewise.
19096 (visium_expand_int_cstore): Likewise.
19097 (visium_expand_fp_cstore): Likewise.
19098 * config/visium/visium.c (visium_pass_by_reference): Likewise.
19099 (visium_function_arg): Likewise.
19100 (visium_function_arg_advance): Likewise.
19101 (visium_libcall_value): Likewise.
19102 (visium_setup_incoming_varargs): Likewise.
19103 (visium_legitimate_constant_p): Likewise.
19104 (visium_legitimate_address_p): Likewise.
19105 (visium_legitimize_address): Likewise.
19106 (visium_secondary_reload): Likewise.
19107 (visium_register_move_cost): Likewise.
19108 (visium_memory_move_cost): Likewise.
19109 (prepare_move_operands): Likewise.
19110 (ok_for_simple_move_operands): Likewise.
19111 (ok_for_simple_move_strict_operands): Likewise.
19112 (ok_for_simple_arith_logic_operands): Likewise.
19113 (visium_function_value_1): Likewise.
19114 (rtx_ok_for_offset_p): Likewise.
19115 (visium_legitimize_reload_address): Likewise.
19116 (visium_split_double_move): Likewise.
19117 (visium_expand_copysign): Likewise.
19118 (visium_expand_int_cstore): Likewise.
19119 (visium_expand_fp_cstore): Likewise.
19120 (visium_split_cstore): Likewise.
19121 (visium_select_cc_mode): Likewise.
19122 (visium_split_cbranch): Likewise.
19123 (output_cbranch): Likewise.
19124 (visium_print_operand_address): Likewise.
19125 * expmed.c (flip_storage_order): Likewise.
19126 * expmed.h (emit_cstore): Likewise.
19127 (flip_storage_order): Likewise.
19128 * genrecog.c (validate_pattern): Likewise.
19129 * hsa-gen.c (gen_hsa_addr): Likewise.
19130 * internal-fn.c (expand_arith_overflow): Likewise.
19131 * ira-color.c (allocno_copy_cost_saving): Likewise.
19132 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19133 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19134 (process_invariant_for_inheritance): Likewise.
19135 * lra-eliminations.c (move_plus_up): Likewise.
19136 * omp-low.c (lower_oacc_reductions): Likewise.
19137 * simplify-rtx.c (simplify_subreg): Likewise.
19138 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19139 (TARGET_CHKP_BOUND_MODE): Likewise..
19140 * targhooks.c (default_chkp_bound_mode): Likewise.
19141 (default_setup_incoming_vararg_bounds): Likewise.
19142 * targhooks.h (default_chkp_bound_mode): Likewise.
19143 (default_setup_incoming_vararg_bounds): Likewise.
19144 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19145 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19146 (have_whole_vector_shift): Likewise.
19147 * tree-vect-stmts.c (vectorizable_load): Likewise.
19148 * doc/tm.texi: Regenerate.
19150 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19152 Graceful degrade if Binutils PR21472 is not available.
19155 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19156 .rodata in flash test fails.
19157 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19158 * confgure: Regenerate.
19159 * config.in: Regenerate.
19160 * config/avr/avr.c (avr_asm_named_section)
19161 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19162 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19163 (avr_asm_init_sections): Same.
19165 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19167 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19168 (fma<VH:mode>4_intrinsic): Likewise.
19169 (*fmsub<VCVTF:mode>4): Likewise.
19170 (*fmsub<VH:mode>4_intrinsic): Likewise.
19172 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19175 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19176 Don't depend on "optimize > 0".
19177 (out_movhi_r_mr, out_movqi_mr_r): Same.
19178 (out_movhi_mr_r, out_movqi_r_mr): Same.
19179 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19180 io_address_operand on "optimize > 0".
19182 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19184 * tree-loop-distribution.c: Add general explanantion on the pass.
19185 (generate_loops_for_partition): Mark distributed loop.
19186 (pg_add_dependence_edges): New parameter. Handle alias data
19187 dependence specially and record it in the parameter if asked.
19188 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19189 (init_partition_graph_vertices, add_partition_graph_edge): New.
19190 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19191 (free_partition_graph_vdata, build_partition_graph): New.
19192 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19193 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19194 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19195 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19196 (version_for_distribution_p, finalize_partitions): New.
19197 (distribute_loop): Handle alias data dependence specially. Factor
19198 out loop fusion code as functions and call these functions.
19200 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19202 * tree-loop-distribution.c (classify_partition): New parameter and
19203 better handle reduction statement.
19204 (rdg_build_partitions): Revise comment.
19205 (distribute_loop): Compute statements in all partitions and pass it
19206 to classify_partition.
19208 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19210 * tree-loop-distribution.c (enum partition_type): New.
19211 (struct partition): New field type.
19212 (partition_merge_into): Add parameter. Update partition type.
19213 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19214 (build_rdg_partition_for_vertex): Compute partition type.
19215 (rdg_build_partitions): Dump partition type.
19216 (distribute_loop): Update calls to partition_merge_into.
19218 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19220 * tree-loop-distribution.c (struct ddr_hasher): New.
19221 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19223 (classify_partition): Call get_data_dependence.
19224 (pg_add_dependence_edges): Ditto.
19225 (distribute_loop): Release data dependence hash table.
19227 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19229 * tree-loop-distribution.c (ref_base_address): Delete.
19230 (similar_memory_accesses): Rename ...
19231 (share_memory_accesses): ... to this. Check if partitions access
19232 the same memory reference.
19233 (distribute_loop): Call share_memory_accesses.
19235 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19237 * tree-loop-distribution.c (struct partition): New field recording
19238 its data reference.
19239 (partition_alloc, partition_free): Init and release data refs.
19240 (partition_merge_into): Merge data refs.
19241 (build_rdg_partition_for_vertex): Collect data refs for partition.
19242 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19244 (distribute_loop): Remve data refs from vertice data of partition
19247 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19249 * tree-loop-distribution.c (params.h): Include header file.
19250 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19251 (datarefs_vec): New global var.
19252 (create_rdg_vertices): Use datarefs_vec directly.
19253 (free_rdg): Don't free data references.
19254 (build_rdg): Update use. Don't free data references.
19255 (distribute_loop): Compute global variable for data references.
19256 Bail out if there are too many data references.
19258 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19260 * tree-loop-distribution.c (loop_nest): New global var.
19261 (build_rdg): Use loop directly, rather than loop nest.
19262 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19264 (distribute_loop): Compute global variable loop nest. Update use.
19266 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19268 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19269 (partition_merge_into): New parameter. Dump reason for fusion.
19270 (distribute_loop): Update use of partition_merge_into.
19272 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19274 * tree-loop-distribution.c (bb_top_order_index): New.
19275 (bb_top_order_index_size, bb_top_order_cmp): New.
19276 (stmts_from_loop): Use topological order.
19277 (pass_loop_distribution::execute): Compute and release topological
19278 order for basic blocks.
19280 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19282 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19285 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19287 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19288 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19289 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19290 * internal-fn.def (LOOP_DIST_ALIAS): New.
19291 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19292 (fold_loop_internal_call): ... this.
19293 (vect_loop_dist_alias_call): New function.
19294 (set_uid_loop_bbs): Call fold_loop_internal_call.
19295 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19298 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19301 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19302 Require dead FLAGS_REG at the beginning of a peephole.
19304 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19307 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19308 arguments in the call to __builtin_ia32_sbb_u32.
19309 (_subborrow_u64): Swap _X and _Y arguments in the call to
19310 __builtin_ia32_sbb_u64.
19312 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19315 * tree-vrp.c (compare_assert_loc): Turn into a function template
19316 with stable template parameter. Only test if a->e is NULL,
19317 !a->e == !b->e has been verified already. Use e == NULL or
19318 e != NULL instead of e or ! e tests. If stable is true, don't use
19319 iterative_hash_expr, on the other side allow a or b or both NULL
19320 and sort the NULLs last.
19321 (process_assert_insertions): Sort using compare_assert_loc<false>
19322 instead of compare_assert_loc, later sort using
19323 compare_assert_loc<true> before calling process_assert_insertions_for
19324 in a loop. Use break instead of continue once seen NULL pointer.
19326 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19328 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19329 Cortex-R7 and Cortex-R8 processors.
19331 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19333 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19334 uninitialized while src is not.
19336 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19338 * common/config/arm/arm-common.c: Adjust include path for
19340 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19341 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19342 (arm-cpu-data.h): Likewise.
19343 (arm-cpu-cdata.h): Likewise.
19344 * config/arm/arm-cpu.h: Delete.
19345 * config/arm/arm-cpu-cdata.h: Delete.
19346 * config/arm/arm-cpu-data.h: Delete.
19348 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19350 * config/arm/arm-cpus.in (cortex-a55): New.
19351 (cortex-a75): Likewise.
19352 (cortex-a75.cortex-a55): Likewise.
19353 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19355 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19356 * config/arm/arm-cpu-cdata.h: Regenerate.
19357 * config/arm/arm-cpu-data.h: Regenerate.
19358 * config/arm/arm-cpu.h: Regenerate.
19359 * config/arm/arm-tables.opt: Regenerate.
19360 * config/arm/arm-tune.md: Regenerate.
19362 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19364 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19366 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19368 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19371 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19373 PR tree-optimization/81292
19374 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19375 full_string_p, also call adjust_related_strinfos if the adjustment
19376 is simple, otherwise invalidate related strinfos.
19378 2017-07-04 Martin Liska <mliska@suse.cz>
19381 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19382 newly created variable as DECL_IGNORED_P.
19384 2017-07-04 Martin Liska <mliska@suse.cz>
19387 * ipa-inline.c (inline_small_functions):
19388 Use xstrdup_for_dump.
19390 2017-07-04 Tom de Vries <tom@codesourcery.com>
19392 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19394 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19397 * config/darwin.c (darwin_function_switched_text_sections):
19400 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19402 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19404 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19406 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19408 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19410 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19411 min_profitable_iters, and th as inclusive lower bounds.
19412 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19413 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19414 for min_profitable_iters and min_profitable_estimate.
19415 (vect_transform_loop): Treat th as an inclusive lower bound.
19416 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19418 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19421 * config/darwin.c (darwin_function_switched_text_sections):
19422 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19423 in two pieces, and suppress the use of buf.
19425 2017-07-03 Nathan Sidwell <nathan@acm.org>
19427 * hash-table.h (hash_table_mod1): Fix indentation.
19429 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19431 PR middle-end/81290
19432 * predict.c (force_edge_cold): Be more careful about propagation
19434 * profile-count.h (profile_probability::guessed,
19435 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19437 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19439 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19441 * doc/invoke.texi (rcpc architecture extension): Document it.
19443 2017-07-03 Richard Biener <rguenther@suse.de>
19445 PR tree-optimization/60510
19446 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19447 the scalar reduction PHI and use it.
19448 (vectorizable_reduction): Properly guard the single_defuse_cycle
19449 path for non-SLP reduction chains where we cannot use it.
19450 Rework reduc_def/index and vector type deduction. Rework
19451 vector operand gathering during reduction op code-gen.
19452 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19453 chains dissolve the chain and leave it to non-SLP reduction
19456 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19458 * tree-data-ref.h (dr_alignment): Declare.
19459 * tree-data-ref.c (dr_alignment): New function.
19460 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19461 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19463 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19465 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19467 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19468 and base_misalignment fields.
19469 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19470 * tree-data-ref.c: Include builtins.h.
19471 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19472 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19473 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19474 * tree-vect-data-refs.c: Include tree-cfg.h.
19475 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19476 fields instead of calculating an alignment here.
19477 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19478 innermost_loop_behavior fields.
19480 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19482 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19484 (DR_STEP_ALIGNMENT): New macro.
19485 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19486 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19487 (create_data_ref): Print it.
19488 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19489 to tell whether the step preserves vector (mis)alignment.
19490 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19491 Move the check for an integer step and generalise to all INTEGER_CST.
19492 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19493 Print the outer step alignment.
19495 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19497 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19498 with offset_alignment.
19499 (DR_ALIGNED_TO): Delete.
19500 (DR_OFFSET_ALIGNMENT): New macro.
19501 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19502 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19503 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19504 (create_data_ref): Likewise.
19505 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19506 (vect_analyze_data_refs): Likewise.
19507 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19508 creating dummy innermost behavior.
19510 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19512 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19513 with a "innermost_loop_behavior *" and refeence tree.
19514 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19515 (create_data_ref): Update call accordingly.
19516 * tree-predcom.c (find_looparound_phi): Likewise.
19518 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19520 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19521 fields with dr_wrt_vec_loop.
19522 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19523 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19524 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19525 (vect_dr_behavior): New function.
19526 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19527 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19528 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19529 track whether the step preserves the misalignment.
19530 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19531 Use vect_dr_behavior.
19532 (vect_setup_realignment): Update call accordingly.
19533 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19534 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19535 call to vect_create_addr_base_for_vector_ref.
19536 (vect_create_cond_for_align_checks): Likewise.
19537 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19538 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19539 (vect_recog_mask_conversion_pattern): Likewise.
19540 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19541 (new_stmt_vec_info): Remove redundant zeroing.
19543 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19545 * common/config/arm/arm-common.c (arm_be8_option): New function.
19546 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19547 (ISA_ARMv6): Add isa_bit_be8.
19548 * config/arm/arm.h (arm_be8_option): Add prototype.
19549 (BE8_SPEC_FUNCTION): New define.
19550 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19551 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19552 (mlittle-endian): Similarly.
19553 (mbe8, mbe32): New options.
19554 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19555 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19557 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19559 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19561 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19563 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19564 (cleanup_tree_cfg_bb): Use it.
19565 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19567 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19569 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19572 * loop-doloop.c (add_test): Update profile.
19574 2017-07-03 Martin Liska <mliska@suse.cz>
19577 * sanopt.c (rewrite_usage_of_param): New function.
19578 (sanitize_rewrite_addressable_params): Likewise.
19579 (pass_sanopt::execute): Call rewrite_usage_of_param.
19581 2017-07-03 Richard Biener <rguenther@suse.de>
19583 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19584 back to using VIEW_CONVERT_EXPR.
19586 2017-07-03 Martin Liska <mliska@suse.cz>
19589 * doc/extend.texi: Document when a resolver function is
19590 generated for target_clones.
19592 2017-07-03 Martin Liska <mliska@suse.cz>
19594 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19595 and shadow memory of auto variables which are subject of
19596 use-after-scope sanitization.
19597 (asan_expand_mark_ifn): Add do set only when is_poison.
19599 2016-07-03 Richard Biener <rguenther@suse.de>
19601 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19603 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19604 (vectorizable_reduction): Perform reduction PHI creation when
19605 visiting a reduction PHI and adjust and simplify code generation
19606 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19607 (vect_transform_loop): Visit reduction PHIs.
19608 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19609 defs into the SLP tree.
19610 (vect_build_slp_tree): Reduction defs terminate the recursion.
19611 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19613 (vect_get_vec_defs_for_stmt_copy): Export.
19614 (vect_get_vec_defs): Likewise.
19615 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19617 (vect_get_vec_defs_for_stmt_copy): Declare.
19618 (vect_get_vec_defs): Likewise.
19620 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19622 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19623 parameter with a "loop" parameter and use it instead of the
19624 loop containing DR_STMT. Don't check simple_iv when doing
19625 BB analysis. Describe the two analysis modes in the comment.
19627 2017-07-03 Tom de Vries <tom@codesourcery.com>
19629 PR tree-optimization/69468
19630 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19631 (find_same_succ_bb): Handle ignore_edge_flags.
19633 2017-07-03 Tom de Vries <tom@codesourcery.com>
19635 PR tree-optimization/81192
19636 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19638 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19640 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19642 2017-07-03 Tom de Vries <tom@codesourcery.com>
19644 PR tree-optimization/81192
19645 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19646 BB_SAME_SUCC (bb) == NULL.
19648 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19650 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19653 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19655 * dumpfile.c: Include profile-count.h
19656 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19658 (insert_cond_bb): Update profile.
19659 * tree-cfg.h (insert_cond_bb): Update prototype.
19660 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19661 * tree-dump.c: Do not include tree-cfg.
19663 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19665 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19667 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19669 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19672 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19674 * tree-complex.c (expand_complex_div_wide): update profile.
19676 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19677 Alan Hayward <alan.hayward@arm.com>
19678 David Sherwood <david.sherwood@arm.com>
19680 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19681 (CORETYPES_H): New define.
19682 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19683 (insn-modes-inline.h, s-modes-inline-h): New rules.
19684 (generated_files): Add insn-modes-inline.h.
19685 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19686 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19687 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19688 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19689 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19690 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19691 (build/gencodes.o, build/genconditions.o): Likewise.
19692 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19693 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19694 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19695 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19696 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19697 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19698 * coretypes.h: Include everything up to real.h for generators.
19699 Include insn-modes.h first. Include wide-int-print.h after
19700 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19701 * machmode.h: Don't include insn-modes.h here.
19702 * function-tests.c: Remove includes of signop.h, machmode.h,
19703 double-int.h and wide-int.h.
19705 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19707 * optc-save-gen.awk: Likewise.
19708 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19709 * godump.c: Remove include of wide-int-print.h.
19710 * pretty-print.h: Likewise.
19711 * wide-int-print.cc: Likewise.
19712 * wide-int.cc: Likewise.
19713 * hash-map-tests.c: Remove include of signop.h.
19714 * hash-set-tests.c: Likewise.
19715 * rtl-tests.c: Likewise.
19716 * mkconfig.sh: Remove include of machmode.h.
19717 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19719 (emit_insn_modes_inline_h): ...this new function. Emit the code
19720 into an insn-modes-inline.h header file, adding appropriate
19721 include guards and end comments.
19722 (emit_insn_modes_c_header): Remove include of machmode.h.
19723 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19725 (main): Handle -i flag and call emit_insn_modes_inline_h when
19728 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19730 * tree-ssa-strlen.c (strinfo): Rename the length field to
19731 nonzero_chars. Add a full_string_p field.
19732 (compare_nonzero_chars, zero_length_string_p): New functions.
19733 (get_addr_stridx): Add an offset_out parameter.
19734 Use compare_nonzero_chars.
19735 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19736 (new_strinfo): Update after above changes to strinfo.
19737 (set_endptr_and_length): Set full_string_p.
19738 (get_string_length): Update after above changes to strinfo.
19739 (unshare_strinfo): Update call to new_strinfo.
19740 (maybe_invalidate): Likewise.
19741 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19742 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19743 as a uhwi instead of an shwi. Update after above changes to
19744 strinfo and new_strinfo.
19745 (zero_length_string): Assert that chainsi contains full strings.
19746 Use zero_length_string_p. Update call to new_strinfo.
19747 (adjust_related_strinfos): Update after above changes to strinfo.
19748 Copy full_string_p from origsi.
19749 (adjust_last_stmt): Use zero_length_string_p.
19750 (handle_builtin_strlen): Update after above changes to strinfo and
19751 new_strinfo. Install the lhs as the string length if the previous
19752 entry didn't describe a full string.
19753 (handle_builtin_strchr): Update after above changes to strinfo
19755 (handle_builtin_strcpy): Likewise.
19756 (handle_builtin_strcat): Likewise.
19757 (handle_builtin_malloc): Likewise.
19758 (handle_pointer_plus): Likewise.
19759 (handle_builtin_memcpy): Likewise. Track nonzero characters
19760 that aren't necessarily followed by a nul terminator.
19761 (handle_char_store): Likewise.
19763 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19765 PR tree-optimization/80769
19766 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19767 for malloc and calloc. Document the new invariant that all related
19768 strinfos have delayed lengths or none do.
19769 (verify_related_strinfos): Move earlier in file.
19770 (set_endptr_and_length): New function, split out from...
19771 (get_string_length): ...here. Also set the lengths of related
19773 (zero_length_string): Assert that chainsi has known (rather than
19775 (adjust_related_strinfos): Likewise.
19777 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19779 PR tree-optimization/81136
19780 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19781 assert that two references with the same misalignment have the same
19782 compile-time misalignment if those compile-time misalignments
19785 2017-07-01 Andi Kleen <ak@linux.intel.com>
19787 * print-tree.c (print_node): Print all attributes.
19789 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19791 * cfg.c (scale_bbs_frequencies): New function.
19792 * cfg.h (scale_bbs_frequencies): Declare it.
19793 * cfgloopanal.c (single_likely_exit): Cleanup.
19794 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19796 (scale_loop_profile): Likewise.
19797 (loop_version): Likewise.
19798 (create_empty_loop_on_edge): Update.
19799 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19800 scale_loop_frequencies, scale_loop_profile, loopify,
19801 loop_version): Update prototypes.
19802 * modulo-sched.c (sms_schedule): Update.
19803 * predict.c (unlikely_executed_edge_p): Also check probability.
19804 (probably_never_executed_edge_p): Fix typo.
19805 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19806 * tree-parloops.c (gen_parallel_loop): Update.
19807 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19808 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19809 * tree-ssa-loop-split.c (split_loop): Update.
19810 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19811 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19812 (vect_loop_versioning): Update.
19813 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19815 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19817 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19819 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19821 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19822 to keep profile consistent.
19824 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19826 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19827 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19828 * profile-count.h (max_safe_multiplier): Make unsigned.
19829 (profile_count::guessed_zero): New.
19831 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19833 * bb-reorder.c (fix_up_crossing_landing_pad,
19834 fix_crossing_conditional_branches): Use make_single_succ_edge
19835 to keep profile consistent.
19837 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19839 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19842 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19845 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19846 the right scopes, make sure cond_jump isn't preserved between multiple
19847 iterations. Search for fallthru edge whenever there are 3+ edges and
19848 use find_fallthru_edge for it.
19850 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19852 Patch by Alexander Monakov <amonakov@ispras.ru>
19853 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19854 probabilities consistently.
19856 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19858 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19859 emit_cmp_and_jump_insns.
19861 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19864 * tree-inline.c (expand_call_inline): Combine profile statuses.
19866 2017-06-30 Andrew Pinski <apinski@cavium.com>
19868 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19869 fold_stmt returned true.
19871 2017-06-30 Nathan Sidwell <nathan@acm.org>
19873 * ggc.h (empty_string): Delete.
19874 * cfgexpand.c (expand_asm_stmt): Use plain "".
19875 * optabs.c (expand_asm_memory_barrier): Likewise.
19876 * stringpool.c (empty_string): Delete.
19877 (digit_vector, digit_string): Delete.
19878 (ggc_alloc_string): Use plain "", don't optimize single digit
19879 strings. Use ggc_alloc_atomic.
19881 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19883 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19884 comparison set and one other set, use the cost of the non-comparison
19887 2017-06-30 Nathan Sidwell <nathan@acm.org>
19889 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19892 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19894 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19895 loops. Remove now unneeded calls to gimple_switch_set_label() that
19896 just set removed labels to NULL_TREE.
19898 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19900 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19901 (set_range_info): ...here. Only call set_range_info_raw if domain
19903 (set_nonzero_bits): Call set_range_info_raw.
19904 * tree-ssanames.h (set_range_info_raw): New.
19906 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19909 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19910 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19911 of nonimmediate_operand and <store_mask_constraint> instead of m
19912 for the input operand. For V8FI iterator, always split if input
19913 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19914 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19915 <store_mask_predicate> instead of register_operand and
19916 <store_mask_constraint> instead of v for the input operand. Make
19917 sure both operands aren't MEMs for if not <mask_applied>.
19919 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19921 * lto-wrapper.c (copy_file) Close both file descriptors before
19924 2017-06-30 Martin Liska <mliska@suse.cz>
19927 * multiple_target.c (create_dispatcher_calls): Make ifunc
19928 also for function that don't have calls or are not referenced.
19930 2017-06-30 Richard Biener <rguenther@suse.de>
19932 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19933 analyze the first scalar stmt. Move vector type computation
19934 for the BB case here from ...
19935 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19936 live operation processing in the SLP case properly.
19938 2017-06-30 Richard Biener <rguenther@suse.de>
19940 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19942 2017-06-30 Martin Liska <mliska@suse.cz>
19945 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19946 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19948 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19950 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19953 2017-06-30 Richard Biener <rguenther@suse.de>
19955 PR tree-optimization/81249
19956 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19957 condition reduction result to original scalar type.
19959 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19961 * profile-count.h (enum profile_quality): Fix typos and whitespace
19964 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19966 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19967 type for branch probabilities.
19969 2017-06-29 Julian Brown <julian@codesourcery.com>
19970 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19972 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19973 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19974 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19975 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19977 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19979 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19980 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19981 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19982 CC usage from generic code to here.
19983 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19984 CC usage into the target macros.
19986 2017-06-29 Maya Rashish <coypu@sdf.org>
19988 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19991 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19993 * arm/arm-builtins.c: Include profile-count.h
19994 * except.c (sjlj_emit_function_enter): Use
19995 profile_probability::unlikely.
19997 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19999 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
20000 and tocrel_offset be pointer args rather than implicitly using
20002 (legitimate_constant_pool_address_p, rs6000_emit_move,
20003 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
20004 tocrel_offset and use in toc_relative_expr_p call.
20005 (print_operand, print_operand_address): Use static tocrel_base_oac
20006 and tocrel_offset_oac.
20007 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
20010 2017-06-29 Maya Rashish <coypu@sdf.org>
20012 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
20014 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
20016 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
20017 objects, take into account only the alignment of 'op0' and 'mode1' if
20020 2017-06-29 Steve Ellcey <sellcey@cavium.com>
20022 * ccmp.c (ccmp_tree_comparison_p): New function.
20023 (ccmp_candidate_p): Update to use above function.
20024 (get_compare_parts): New function.
20025 (expand_ccmp_next): Update to use new functions.
20026 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
20028 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
20029 take mode as argument.
20030 * ccmp.h (expand_ccmp_expr): Add mode as argument.
20031 * expr.c (expand_expr_real_1): Pass mode as argument.
20033 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
20035 * combine.c (combine_instructions): Print insns to dump_file, together
20038 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
20040 * asan.c (asan_emit_stack_protection): Update.
20041 (create_cond_insert_point): Update.
20042 * auto-profile.c (afdo_propagate_circuit): Update.
20043 * basic-block.h (struct edge_def): Turn probability to
20044 profile_probability.
20045 (EDGE_FREQUENCY): Update.
20046 * bb-reorder.c (find_traces_1_round): Update.
20047 (better_edge_p): Update.
20048 (sanitize_hot_paths): Update.
20049 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
20050 (make_single_succ_edge): Update.
20051 (check_bb_profile): Update.
20052 (dump_edge_info): Update.
20053 (update_bb_profile_for_threading): Update.
20054 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
20055 probabilitycount to 0.
20056 * cfgbuild.c (compute_outgoing_frequencies): Update.
20057 * cfgcleanup.c (try_forward_edges): Update.
20058 (outgoing_edges_match): Update.
20059 (try_crossjump_to_edge): Update.
20060 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
20061 (expand_gimple_tailcall): Update.
20062 (construct_init_block): Use make_single_succ_edge.
20063 (construct_exit_block): Use make_single_succ_edge.
20064 * cfghooks.c (verify_flow_info): Update.
20065 (redirect_edge_succ_nodup): Update.
20066 (split_edge): Update.
20067 (account_profile_record): Update.
20068 * cfgloopanal.c (single_likely_exit): Update.
20069 * cfgloopmanip.c (scale_loop_profile): Update.
20070 (set_zero_probability): Remove.
20071 (duplicate_loop_to_header_edge): Update.
20072 * cfgloopmanip.h (loop_version): Update prototype.
20073 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
20074 (force_nonfallthru_and_redirect): Update.
20075 (update_br_prob_note): Update.
20076 (rtl_verify_edges): Update.
20077 (purge_dead_edges): Update.
20078 (rtl_lv_add_condition_to_bb): Update.
20079 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
20080 * cgraphunit.c (init_lowered_empty_function): Update.
20081 (cgraph_node::expand_thunk): Update.
20082 * cilk-common.c: Include profile-count.h
20083 * dojump.c (inv): Remove.
20084 (jumpifnot): Update.
20085 (jumpifnot_1): Update.
20086 (do_jump_1): Update.
20088 (do_jump_by_parts_greater_rtx): Update.
20089 (do_compare_rtx_and_jump): Update.
20090 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
20091 do_jump_1. do_compare_rtx_and_jump): Update prototype.
20092 * dwarf2cfi.c: Include profile-count.h
20093 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
20094 (sjlj_emit_dispatch_table): Likewise.
20095 * explow.c: Include profile-count.h
20096 * expmed.c (emit_store_flag_force): Update.
20097 (do_cmp_and_jump): Update.
20098 * expr.c (compare_by_pieces_d::generate): Update.
20099 (compare_by_pieces_d::finish_mode): Update.
20100 (emit_block_move_via_loop): Update.
20101 (store_expr_with_bounds): Update.
20102 (store_constructor): Update.
20103 (expand_expr_real_2): Update.
20104 (expand_expr_real_1): Update.
20105 * expr.h (try_casesi, try_tablejump): Update prototypes.
20106 * gimple-pretty-print.c (dump_probability): Update.
20107 (dump_profile): New.
20108 (dump_gimple_label): Update.
20109 (dump_gimple_bb_header): Update.
20110 * graph.c (draw_cfg_node_succ_edges): Update.
20111 * hsa-gen.c (convert_switch_statements): Update.
20112 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20113 (find_if_case_1): Update.
20114 (find_if_case_2): Update.
20115 * internal-fn.c (expand_arith_overflow_result_store): Update.
20116 (expand_addsub_overflow): Update.
20117 (expand_neg_overflow): Update.
20118 (expand_mul_overflow): Update.
20119 (expand_vector_ubsan_overflow): Update.
20120 * ipa-cp.c (good_cloning_opportunity_p): Update.
20121 * ipa-split.c (split_function): Use make_single_succ_edge.
20122 * ipa-utils.c (ipa_merge_profiles): Update.
20123 * loop-doloop.c (add_test): Update.
20124 (doloop_modify): Update.
20125 * loop-unroll.c (compare_and_jump_seq): Update.
20126 (unroll_loop_runtime_iterations): Update.
20127 * lra-constraints.c (lra_inheritance): Update.
20128 * lto-streamer-in.c (input_cfg): Update.
20129 * lto-streamer-out.c (output_cfg): Update.
20130 * mcf.c (adjust_cfg_counts): Update.
20131 * modulo-sched.c (sms_schedule): Update.
20132 * omp-expand.c (expand_omp_for_init_counts): Update.
20133 (extract_omp_for_update_vars): Update.
20134 (expand_omp_ordered_sink): Update.
20135 (expand_omp_for_ordered_loops): Update.
20136 (expand_omp_for_generic): Update.
20137 (expand_omp_for_static_nochunk): Update.
20138 (expand_omp_for_static_chunk): Update.
20139 (expand_cilk_for): Update.
20140 (expand_omp_simd): Update.
20141 (expand_omp_taskloop_for_outer): Update.
20142 (expand_omp_taskloop_for_inner): Update.
20143 * omp-simd-clone.c (simd_clone_adjust): Update.
20144 * optabs.c (expand_doubleword_shift): Update.
20145 (expand_abs): Update.
20146 (emit_cmp_and_jump_insn_1): Update.
20147 (expand_compare_and_swap_loop): Update.
20148 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20149 * predict.c (predictable_edge_p): Update.
20150 (edge_probability_reliable_p): Update.
20151 (set_even_probabilities): Update.
20152 (combine_predictions_for_insn): Update.
20153 (combine_predictions_for_bb): Update.
20154 (propagate_freq): Update.
20155 (estimate_bb_frequencies): Update.
20156 (force_edge_cold): Update.
20157 * profile-count.c (profile_count::dump): Add missing space into dump.
20158 (profile_count::debug): Add newline.
20159 (profile_count::differs_from_p): Explicitly convert to unsigned.
20160 (profile_count::stream_in): Update.
20161 (profile_probability::dump): New member function.
20162 (profile_probability::debug): New member function.
20163 (profile_probability::differs_from_p): New member function.
20164 (profile_probability::differs_lot_from_p): New member function.
20165 (profile_probability::stream_in): New member function.
20166 (profile_probability::stream_out): New member function.
20167 * profile-count.h (profile_count_quality): Rename to ...
20168 (profile_quality): ... this one.
20169 (profile_probability): New.
20170 (profile_count): Update.
20171 * profile.c (compute_branch_probabilities): Update.
20172 * recog.c (peep2_attempt): Update.
20173 * sched-ebb.c (schedule_ebbs): Update.
20174 * sched-rgn.c (find_single_block_region): Update.
20175 (compute_dom_prob_ps): Update.
20176 (schedule_region): Update.
20177 * sel-sched-ir.c (compute_succs_info): Update.
20178 * stmt.c (struct case_node): Update.
20179 (do_jump_if_equal): Update.
20180 (get_outgoing_edge_probs): Update.
20181 (conditional_probability): Update.
20182 (emit_case_dispatch_table): Update.
20183 (expand_case): Update.
20184 (expand_sjlj_dispatch_table): Update.
20185 (emit_case_nodes): Update.
20186 * targhooks.c: Update.
20187 * tracer.c (better_p): Update.
20188 (find_best_successor): Update.
20189 * trans-mem.c (expand_transaction): Update.
20190 * tree-call-cdce.c: Update.
20191 * tree-cfg.c (gimple_split_edge): Upate.
20192 (move_sese_region_to_fn): Upate.
20193 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20194 * tree-eh.c (lower_resx): Upate.
20195 (cleanup_empty_eh_move_lp): Upate.
20196 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20197 * tree-inline.c (copy_edges_for_bb): Update.
20198 (copy_cfg_body): Update.
20199 * tree-parloops.c (gen_parallel_loop): Update.
20200 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20201 (gimple_gen_time_profiler): Update.
20202 * tree-ssa-dce.c (remove_dead_stmt): Update.
20203 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20204 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20205 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20206 (unloop_loops): Update.
20207 (try_peel_loop): Update.
20208 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20209 * tree-ssa-loop-split.c (connect_loops): Update.
20210 (split_loop): Update.
20211 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20212 (hoist_guard): Update.
20213 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20214 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20215 (value_replacement): Update.
20216 * tree-ssa-reassoc.c (branch_fixup): Update.
20217 * tree-ssa-tail-merge.c (replace_block_by): Update.
20218 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20219 (create_edge_and_update_destination_phis): Update.
20220 (compute_path_counts): Update.
20221 (recompute_probabilities): Update.
20222 (update_joiner_offpath_counts): Update.
20223 (freqs_to_counts_path): Update.
20224 (duplicate_thread_path): Update.
20225 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20226 (struct switch_conv_info): Update.
20227 (gen_inbound_check): Update.
20228 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20229 (vect_do_peeling): Update.
20230 (vect_loop_versioning): Update.
20231 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20232 (optimize_mask_stores): Update.
20233 * ubsan.c (ubsan_expand_null_ifn): Update.
20234 * value-prof.c (gimple_divmod_fixed_value): Update.
20235 (gimple_divmod_fixed_value_transform): Update.
20236 (gimple_mod_pow2): Update.
20237 (gimple_mod_pow2_value_transform): Update.
20238 (gimple_mod_subtract): Update.
20239 (gimple_mod_subtract_transform): Update.
20240 (gimple_ic): Update.
20241 (gimple_stringop_fixed_value): Update.
20242 (gimple_stringops_transform): Update.
20243 * value-prof.h: Update.
20245 2017-06-29 Carl Love <cel@us.ibm.com>
20247 * config/rs6000/rs6000-c.c: Add support for built-in functions
20248 vector signed int vec_signed (vector float);
20249 vector signed long long vec_signed (vector double);
20250 vector signed int vec_signed2 (vector double, vector double);
20251 vector signed int vec_signede (vector double);
20252 vector signed int vec_signedo (vector double);
20253 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20254 instruction generator.
20255 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20256 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20257 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20259 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20260 vunsignede_v2df): Add define_expands.
20261 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20262 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20263 VEC_UNSIGNEDO): Add definitions.
20264 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20265 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20266 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20267 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20268 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20269 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20270 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20271 vec_unsignede, vec_unsignedo): Add builtin defines.
20272 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20274 * doc/extend.texi: Update the built-in documentation file for the
20275 new built-in functions.
20277 2017-06-29 Richard Biener <rguenther@suse.de>
20279 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20280 reduction chains to LOOP_VINFO_REDUCTIONS.
20281 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20282 SLP reductions after processing reduction chains.
20284 2017-06-29 Nathan Sidwell <nathan@acm.org>
20286 * builtins.c (fold_builtin_FUNCTION): Use
20287 lang_hooks.decl_printable_name.
20289 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20291 PR middle-end/81194
20292 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20293 with only one label.
20294 * stmt.c (expand_case): Assert NCASES is greater than one.
20296 2017-06-29 Richard Biener <rguenther@suse.de>
20298 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20300 (group_case_labels): Likewise.
20301 (find_taken_edge): Push sanity checking on val to workers...
20302 (find_taken_edge_cond_expr): ... here
20303 (find_taken_edge_switch_expr): ... and here, handle cases
20304 with just a default label.
20305 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20306 (group_case_labels): Likewise.
20307 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20308 group_case_labels does anything cleanup the CFG again.
20310 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20312 PR tree-optimization/81196
20313 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20314 exit condition comparing two IVs.
20316 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20318 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20319 profile to the dummy entry at the end of the list of architectures.
20320 * config/arm/arm-cpu-cdata.h: Regenerated.
20322 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20323 Michael Collison <michael.collison@arm.com>
20326 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20328 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20329 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20330 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20331 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20332 of shift when the shift amount is masked with constant equal to
20333 the size of the mode.
20334 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20337 2017-06-29 Martin Liska <mliska@suse.cz>
20339 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20341 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20344 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20347 (edge_growth_cache_entry::edge_growth_cache_entry): New
20349 (reset_edge_growth_cache): Update to use constructor.
20351 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20353 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20354 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20355 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20357 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20359 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20360 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20362 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20364 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20365 (*-linux-uclibc*): Add t-uclibc tmake_file.
20366 * config/t-musl: New.
20367 * config/t-uclibc: New.
20369 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20371 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20373 (gen_comm_data): Emit architectural setting of arch_prof.
20374 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20376 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20377 (armv8-m.base, armv8-m.main): Likewise.
20378 * arm-protos.h (arm_build_target): Add profile field.
20379 (arch_option): Likewise.
20380 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20382 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20383 arm_active_target.profile.
20385 2017-06-28 Richard Biener <rguenther@suse.de>
20387 PR middle-end/81227
20388 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20389 TYPE_OVERFLOW_WRAPS.
20390 * match.pd (negate_expr_p): Likewise.
20391 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20392 fold_build2, not fold_binary.
20394 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20396 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20397 Convert memory address to Pmode.
20398 (aarch64_print_operand): Assert MEM operands are always Pmode.
20400 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20403 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20404 Remove redundant if.
20405 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20406 * config/arm/aarch-common-protos.h
20407 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20408 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20410 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20413 * multiple_target.c (create_dispatcher_calls): Set the default
20414 clone to be static, not public.
20416 2017-06-28 Richard Biener <rguenther@suse.de>
20418 * tree-vect-loop.c (vectorizable_reduction): Move special
20419 cond reduction IV var creation ...
20420 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20421 parameter. Use STMT_VINFO_VECTYPE.
20422 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20425 2017-06-28 Martin Liska <mliska@suse.cz>
20428 * ipa-visibility.c (non_local_p): Handle visibility.
20430 2017-06-28 Martin Liska <mliska@suse.cz>
20433 * common.opt: Add IntegerRange to various options.
20434 * opt-functions.awk (integer_range_info): New function.
20435 * optc-gen.awk: Add integer_range_info to cl_options struct.
20436 * opts-common.c (decode_cmdline_option): Handle
20437 CL_ERR_INT_RANGE_ARG.
20438 (cmdline_handle_error): Likewise.
20439 * opts.c (print_filtered_help): Show valid interval in
20440 when --help is provided.
20441 * opts.h (struct cl_option): Add range_min and range_max fields.
20442 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20444 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20446 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20447 (x * C EQ/NE y * C): New transformation.
20449 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20451 * genmultilib (combination_space): Accept '+' in option names.
20453 2017-06-28 Martin Liska <mliska@suse.cz>
20456 * asan.c (instrument_derefs): Bail out inner references
20457 that are hard register variables.
20459 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20462 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20463 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20465 2017-06-28 Richard Biener <rguenther@suse.de>
20467 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20468 (vect_get_slp_defs): Adjust.
20469 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20471 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20473 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20474 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20475 (vectorizable_reduction): Adjust.
20476 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20478 (vect_get_slp_defs): Likewise.
20479 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20480 (vectorizable_bswap): Adjust.
20481 (vectorizable_call): Likewise.
20482 (vectorizable_conversion): Likewise.
20483 (vectorizable_assignment): Likewise.
20484 (vectorizable_shift): Likewise.
20485 (vectorizable_operation): Likewise.
20486 (vectorizable_store): Likewise.
20487 (vectorizable_condition): Likewise.
20488 (vectorizable_comparison): Likewise.
20490 2017-06-28 Michael Collison <michael.collison@arm.com>
20493 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20495 (arm_gen_movmemqi): Removed unused variable 'i'.
20496 Convert 'for' loop into 'while' loop.
20497 (arm_expand_prologue): Remove last unnecessary set of insn.
20498 (thumb_pop): Remove unused variable 'pushed_words'.
20499 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20501 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20503 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20504 * config/s390/s390-protos.h: Add prototype of
20505 s390_rel_address_ok_p.
20506 * config/s390/s390.c (s390_got_symbol): New function.
20507 (s390_rel_address_ok_p): New function.
20508 (legitimize_pic_address): Use s390_rel_address_ok_p.
20509 (s390_load_got): Use s390_got_symbol.
20510 (s390_option_override): Issue error if
20511 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20512 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20514 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20516 2017-06-27 Andrew Pinski <apinski@cavium.com>
20518 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20519 (X * copysign (1.0, X)): New pattern.
20520 (X * copysign (1.0, -X)): New pattern.
20521 (copysign (-1.0, CST)): New pattern.
20523 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20525 * genmultilib (combination_space): Remove variable.
20526 Validate reuse rules against regular expression for any sequence
20527 of multilib options in any order.
20529 2017-06-27 Michael Collison <michael.collison@arm.com>
20531 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20532 call aarch64_split_simd_combine.
20533 * (aarch64_combine_internal<mode>): Delete pattern.
20534 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20535 Allow register and subreg operands.
20537 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20539 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20540 specific need, just fallback on defaults.
20541 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20543 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20544 Olivier Hainque <hainque@adacore.com>
20546 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20548 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20549 targets. Pick a default if no particular attempt applied.
20550 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20553 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20555 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20556 (x86_64-wrs-vxworks7): Likewise.
20558 2017-06-27 Marek Polacek <polacek@redhat.com>
20561 * ubsan.c (instrument_null): Check get_base_address's result for null.
20563 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20565 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20567 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20569 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20570 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20571 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20572 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20573 New function types.
20574 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20575 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20576 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20577 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20578 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20579 BUILT_IN_FEUPDATEENV): New builtins.
20580 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20581 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20582 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20583 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20585 (builtin_structptr_types): Adjust size.
20586 * tree.c (builtin_structptr_types): Add four entries.
20588 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20589 Olivier Hainque <hainque@adacore.com>
20591 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20592 (TLS_SYM): New local macro, forcing reference to __tls__ on
20593 link command lines for VxWorks 7 RTPs, triggering initialization
20595 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20596 OS features TLS support, true for RTPs on VxWorks 7.
20597 * config/vxworks.c (vxworks_override_options): Setup emutls
20600 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20602 * predict.c (test_prediction_value_range): Use -1U instead of -1
20603 to avoid narrowing conversion warning.
20604 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20605 to avoid narrowing conversion warning.
20606 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20608 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20610 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20612 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20613 64bit configurations.
20614 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20615 (SIZE_TYPE): Likewise.
20616 * config/vxworks.c (vxworks_emutls_var_fields): Use
20617 long_unsigned_type_node instead of unsigned_type_node as the offset
20618 field type, which is "pointer" mode in emutls.c.
20620 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20623 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20625 PR middle-end/81207
20626 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20627 gimple_vuse copying separately from gimple_vdef copying.
20629 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20631 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20634 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20635 Olivier Hainque <hainque@adacore.com>
20637 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20638 for all vxworks7 targets.
20639 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20640 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20641 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20642 variations for VX6/VX7 and 32/64bits later on in ...
20643 (VXWORKS_LIB_SPEC): Leverage new macros.
20644 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20645 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20647 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20649 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20650 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20652 2017-06-26 Carl Love <cel@us.ibm.com>
20654 * config/rs6000/rs6000-c.c: Add support for built-in functions
20655 vector bool char vec_reve (vector bool char);
20656 vector signed char vec_reve (vector signed char);
20657 vector unsigned char vec_reve (vector unsigned char);
20658 vector bool int vec_reve (vector bool int);
20659 vector signed int vec_reve (vector signed int);
20660 vector unsigned int vec_reve (vector unsigned int);
20661 vector bool long long vec_reve (vector bool long long);
20662 vector signed long long vec_reve (vector signed long long);
20663 vector unsigned long long vec_reve (vector unsigned long long);
20664 vector bool short vec_reve (vector bool short);
20665 vector signed short vec_reve (vector signed short);
20666 vector double vec_reve (vector double);
20667 vector float vec_reve (vector float);
20668 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20669 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20670 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20671 (altivec_vreve): New pattern.
20672 * config/rs6000/altivec.h (vec_reve): New define.
20673 * doc/extend.texi (vec_rev): Update the built-in documentation file
20674 for the new built-in functions.
20676 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20678 PR tree-optimization/71815
20679 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20681 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20683 (slsr_process_phi): Likewise.
20684 (replace_uncond_cands_and_profitable_phis): Don't replace a
20685 multiply candidate with a stride of 1 (copy or cast).
20686 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20688 (lowest_cost_path): Likewise.
20689 (total_savings): Likewise.
20691 2017-06-26 Richard Biener <rguenther@suse.de>
20694 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20695 Use def_builtin_pure for all gather builtins.
20697 2017-06-26 Richard Biener <rguenther@suse.de>
20699 PR tree-optimization/81203
20700 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20701 non-dominating BBs.
20703 2017-06-26 Marek Polacek <polacek@redhat.com>
20706 * doc/invoke.texi: Document -Wmultistatement-macros.
20708 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20710 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20711 arm_neon_ok_no_float_abi effective target.
20713 2017-06-26 Richard Biener <rguenther@suse.de>
20715 PR tree-optimization/80928
20716 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20717 (copy_bbs): Set BB_DUPLICATED flag early.
20718 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20720 (execute_on_shrinking_pred): Likewise.
20721 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20722 BB_DUPLICATED blocks.
20723 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20724 iterate over all PHIs considering removal of *gsi.
20726 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20728 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20731 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20733 * config/rs6000/rs6000-string.c: (expand_block_clear,
20734 do_load_for_compare, select_block_compare_mode,
20735 compute_current_alignment, expand_block_compare,
20736 expand_strncmp_align_check, expand_strn_compare,
20737 expand_block_move, rs6000_output_load_multiple)
20738 Move functions related to string/block move/compare
20739 to a separate file.
20740 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20741 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20742 for this function which is now used in two files.
20743 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20744 * config.gcc: Add rs6000-string.o to extra_objs for
20745 targets powerpc*-*-* and rs6000*-*-*.
20747 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20750 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20751 32-bit, since indexed is not valid for DImode.
20752 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20753 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20754 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20755 (define_peephole2 for Altivec d-form store): Likewise.
20758 * multiple_target.c (create_dispatcher_calls): Only create the
20759 dispatcher call if the function is the default clone of a
20760 versioned function.
20762 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20764 PR middle-end/80902
20765 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20766 a call, force the call to not be a tail call.
20768 2017-06-23 Jeff Law <law@redhat.com>
20770 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20773 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20775 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20776 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20777 handling for early expansion of vector shifts (sl,sr,sra,rl).
20778 (builtin_function_type): Add vector shift right instructions
20779 to the unsigned argument list.
20781 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20783 rtl-optimizatoin/79286
20784 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20785 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20786 trap. PIC register plus a const unspec without offset can never trap.
20788 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20790 * tree.h (builtin_structptr_type): New type.
20791 (builtin_structptr_types): Declare new array.
20792 * tree.c (builtin_structptr_types): New array.
20793 (free_lang_data, build_common_tree_nodes): Use it.
20795 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20798 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20801 2017-06-22 Matt Turner <mattst88@gmail.com>
20803 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20804 Lake models to skylake case. Assume skylake for unknown
20805 models with clflushopt.
20807 2017-06-22 Jeff Law <law@redhat.com>
20809 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20810 frame sizes that do not satisfy aarch64_uimm12_shift.
20812 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20814 * profile-count.h (apply_probability,
20815 apply_scale, probability_in): Fix checks for zero.
20817 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20819 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20820 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20822 2016-06-22 Richard Biener <rguenther@suse.de>
20824 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20825 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20826 REDUC_MAX_EXPR support.
20827 (vectorizable_reduction): Likewise.
20828 (vect_create_epilog_for_reduction): Likewise.
20830 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20832 * match.pd (A / (1 << B) -> A >> B): New.
20833 * generic-match-head.c: Include optabs-tree.h.
20834 * gimple-match-head.c: Likewise.
20835 * optabs-tree.h (target_supports_op_p): New.
20836 * optabs-tree.c (target_supports_op_p): New.
20838 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20840 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20841 $gcc_cv_ld --help output.
20842 (gcc_cv_ld_demangle): Likewise.
20843 (gcc_cv_ld_eh_frame_hdr): Likewise.
20844 (gcc_cv_ld_pie): Likewise.
20845 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20846 (gcc_cv_ld_buildid): Likewise.
20847 (gcc_cv_ld_sysroot): Likewise.
20848 (ld_bndplt_support): Likewise.
20849 (ld_pushpopstate_support): Likewise.
20850 * configure: Regenerate.
20851 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20853 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20856 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20857 operands indexes to avoid gap between operands and match_dups.
20859 2017-06-21 Andrew Pinski <apinski@cavium.com>
20861 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20862 Increment Arith_shift and Arith_shift_reg by 1.
20863 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20865 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20866 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20867 (aarch64_strip_extend): Add new argument and test for it.
20868 (aarch64_cheap_mult_shift_p): New function.
20869 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20870 add a cost if it is true.
20871 Update calls to aarch64_strip_extend.
20872 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20874 2017-06-21 Andrew Pinski <apinski@cavium.com>
20876 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20878 (thunderxt88): Likewise.
20879 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20880 (thunderx_prefetch_tune): New variable.
20881 (thunderx2t99_prefetch_tune): Update for the correct values.
20882 (thunderxt88_tunings): New variable.
20883 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20884 generic_prefetch_tune.
20885 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20887 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20889 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20890 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20891 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20892 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20893 (aarch64_atomic_cas<mode>, GPI): Likewise.
20895 2017-06-21 Martin Liska <mliska@suse.cz>
20897 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20898 statements on cold and hot labels.
20899 * predict.c (tree_estimate_probability_bb): Remove the
20900 prediction from this place.
20902 2017-06-21 Martin Liska <mliska@suse.cz>
20904 PR tree-optimization/79489
20905 * gimplify.c (maybe_add_early_return_predict_stmt): New
20907 (gimplify_return_expr): Call the function.
20908 * predict.c (tree_estimate_probability_bb): Remove handling
20910 * predict.def: Update comment about early return predictor.
20911 * gimple-predict.h (is_gimple_predict): New function.
20912 * predict.def: Change default value of early return to 66.
20913 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20915 * passes.def: Put pass_strip_predict_hints to the beginning of
20918 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20920 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20921 FUNCTION_DECL declarations.
20922 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20924 (dwaf2out_decl): Likewise.
20925 * godump.c (go_early_global_decl): Skip call to the real debug hook
20926 for FUNCTION_DECL declarations.
20927 * passes.c (rest_of_decl_compilation): Skip call to the
20928 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20929 -fdump-go-spec is passed.
20931 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20933 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20934 Reorder for compactness.
20935 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20936 (def_builtin_pure, def_builtin_pure2): New functions.
20937 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20939 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20941 * match.pd (nop_convert): New predicate.
20942 ((A +- CST1) +- CST2): Allow some NOP conversions.
20944 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20947 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20948 with ctors/dtors if GOVD_SHARED is set.
20950 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20952 * config/aarch64/aarch64.md (movti_aarch64):
20953 Emit mov rather than orr.
20954 (movtf_aarch64): Likewise.
20955 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20956 Emit mov rather than orr.
20958 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20960 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20961 Swap alternatives, make integer dup more expensive.
20963 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20965 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20966 Return true for non-tls symbols.
20968 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20970 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20971 (cortex-a75): Likewise.
20972 (cortex-a75.cortex-a55): Likewise.
20973 * config/aarch64/aarch64-tune.md: Regenerate.
20974 * doc/invoke.texi (-mtune): Document new values for -mtune.
20976 2017-06-21 Tom de Vries <tom@codesourcery.com>
20978 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20979 stack_size feature.
20980 (Effective-Target Keywords, Other attributes): Suggest using
20981 dg-add-options stack_size feature to get stack limit in stack_size
20982 effective target documentation.
20984 2017-06-21 Julian Brown <julian@codesourcery.com>
20985 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20987 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20988 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20989 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20991 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20992 attribute type list for neon_multiply.
20993 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20994 attribute type list for neon_multiply.
20995 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20996 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20997 attribute type list for neon_multiply.
20998 * config/arm/types.md (crypto_pmull): Add.
20999 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
21000 attribute type list.
21002 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
21004 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
21007 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21009 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
21010 to make sure not to dereference a NULL cost_classes_ptr pointer.
21012 2017-06-20 Carl Love <cel@us.ibm.com>
21014 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21015 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21016 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21017 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21018 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21019 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21020 VMULOSW): New enum "unspec" values.
21021 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21022 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21023 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21024 altivec_vmulosw): New patterns.
21025 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21026 VMULOSW): Add definitions.
21028 2017-06-20 Julia Koval <julia.koval@intel.com>
21030 * config/i386/i386.c: Fix rounding expand for new pattern.
21031 * config/i386/subst.md: Fix pattern (parallel -> unspec).
21033 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21035 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
21036 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
21038 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21040 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
21043 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
21045 * config/aarch64/aarch64-cores.def: Rearrange to sort by
21046 architecture, then by implementer ID.
21047 * config/aarch64/aarch64-tune.md: Regenerate.
21049 2017-06-20 Richard Biener <rguenther@suse.de>
21051 PR middle-end/81097
21052 * fold-const.c (split_tree): Fold to type before negating.
21054 2017-06-20 David Malcolm <dmalcolm@redhat.com>
21056 * diagnostic-show-locus.c
21057 (selftest::test_fixit_deletion_affecting_newline): New function.
21058 (selftest::diagnostic_show_locus_c_tests): Call it.
21060 2017-06-20 Andreas Schwab <schwab@suse.de>
21063 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
21066 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21068 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
21069 __ARM_FEATURE_COPROC according to support.
21071 2017-06-20 Jakub Jelinek <jakub@redhat.com>
21073 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21074 Rewritten to avoid overflow for > 32-bit pointers.
21077 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
21078 by removing enum keyword.
21079 (ubsan_type_descriptor): Likewise. Formatting fix.
21082 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
21083 splitter): Require TARGET_SSE2 in the condition.
21085 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
21088 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
21089 for doing vector set of SFmode on ISA 3.0.
21090 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
21091 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
21093 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
21094 SFmode value into a V4SF variable that was extracted from another
21095 V4SF variable without converting the element to double precision
21096 and back to single precision vector format.
21097 (vsx_insert_extract_v4sf_p9_2): Likewise.
21099 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21101 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21102 in UWHI to avoid undefined overflow.
21105 * ubsan.h (enum ubsan_encode_value_phase): New.
21106 (ubsan_encode_value): Change second argument to
21107 enum ubsan_encode_value_phase with default value of
21108 UBSAN_ENCODE_VALUE_GENERIC.
21109 * ubsan.c (ubsan_encode_value): Change second argument to
21110 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21111 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21112 create_tmp_var_raw instead of create_tmp_var and use a
21114 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21115 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21116 ubsan_encode_value callers.
21119 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21120 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21121 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21123 2017-06-19 Richard Biener <rguenther@suse.de>
21125 PR middle-end/81118
21126 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21127 estimates if we changed anything.
21129 2017-06-19 Richard Biener <rguenther@suse.de>
21131 PR tree-optimization/80887
21132 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21133 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21134 simplified lookups, then reset mprts_hook.
21135 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21137 (try_to_simplify): Likewise.
21139 2017-06-19 Martin Liska <mliska@suse.cz>
21142 * gimplify.c (gimplify_switch_expr):
21143 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21145 2017-06-19 Martin Liska <mliska@suse.cz>
21147 * doc/install.texi: Document that PGO runs in 4 stages.
21149 2017-06-19 Martin Liska <mliska@suse.cz>
21152 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21153 to dispatcher function name.
21154 * multiple_target.c (replace_function_decl): New function.
21155 (create_dispatcher_calls): Redirect both edges and references.
21157 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21159 * profile-count.c (profile_count::dump): Dump quality.
21160 (profile_count::differs_from_p): Update for unsigned val.
21161 * profile-count.h (profile_count_quality): New enum.
21162 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21164 2017-06-19 Richard Biener <rguenther@suse.de>
21166 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21167 struct function as arg.
21168 (estimate_numbers_of_iterations): Export overload with loop arg.
21169 (free_numbers_of_iterations_estimates_loop): Use an overload of
21170 free_numbers_of_iterations_estimates instead.
21171 * tree-cfg.c (remove_bb): Adjust.
21172 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21173 * tree-parloops.c (gen_parallel_loop): Likewise.
21174 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21176 (tree_unroll_loops_completely): Likewise.
21177 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21178 Use an overload instead and export.
21179 (estimated_loop_iterations): Adjust.
21180 (max_loop_iterations): Likewise.
21181 (likely_max_loop_iterations): Likewise.
21182 (estimate_numbers_of_iterations): Take struct function as arg
21184 (loop_exits_before_overflow): Adjust.
21185 (free_numbers_of_iterations_estimates_loop): Use an overload.
21186 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21187 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21189 2017-06-19 Richard Biener <rguenther@suse.de>
21192 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21193 RANGE_EXPR conservatively.
21195 2017-06-16 Carl Love <cel@us.ibm.com>
21197 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21198 definitions for vec_float, vec_float2, vec_floato,
21199 vec_floate built-ins.
21200 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21201 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21203 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21204 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21205 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21206 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21207 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21208 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21209 vec_floato): Add builtin defines.
21210 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21211 Update the built-in documentation file for the new built-in
21214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21216 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21217 (mthumb): Mark as the negative of -marm.
21219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21221 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21223 (ARM Options, -mtune): Document that this accepts the same
21224 extension options as -mcpu.
21225 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21229 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21230 permitted extensions.
21232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21234 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21235 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21236 (armv8-m.main): Add option +nodsp.
21237 * config/arm/arm-cpu-cdata.h: Regenerated.
21239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21241 * config/arm/t-fuchsia: New file.
21242 * config.gcc (arm*-*-fuchsia*): Use it.
21244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21246 * config/arm/t-symbian: Rewrite for new option infrastructure.
21248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21250 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21251 (MULTILIB_REQUIRED): Likewise.
21253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21255 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21256 (MULTILIB_RESUE): Likewise.
21257 (MULTILIB_MATCHES): Likewise.
21258 (MULTLIB_REQUIRED): Likewise.
21260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21262 * config/arm/t-rtems: Rewrite for new option framework.
21264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21266 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21267 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21268 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21269 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21270 * config/arm/t-multilib: ... here.
21271 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21272 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21273 armv7-a and armv8*-a when A-profile libraries have not been built.
21274 * config/arm/t-rmprofile: Rewrite.
21276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21278 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21279 with a backslash. Remove the backslash after substituting unescaped
21281 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21283 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21285 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21286 the driver to pass through the default setting of -mfloat-abi.
21287 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21289 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21290 (all_feat_combs): New rule.
21291 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21293 * config/arm/t-aprofile: Rewrite.
21295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21297 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21298 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21299 fpu is not specified by the user/command-line.
21300 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21301 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21302 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21303 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21304 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21305 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21309 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21310 * config/arm/t-arm-elf: Rewritten.
21312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21314 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21315 have some floating-point instructions.
21316 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21317 (TARGET_MAYBE_HARD_FLOAT): New macro.
21318 * config/arm/arm-builtins.c (arm_init_builtins): Use
21319 TARGET_MAYBE_HARD_FLOAT.
21320 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21322 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21324 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21325 (configargs.h): Include it.
21326 (arm_print_hint_for_fpu_option): New function.
21327 (arm_parse_fpu_option): New function.
21328 (candidate_extension): New class.
21329 (arm_canon_for_multilib): New function.
21330 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21331 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21332 (ARCH_CANONICAL_SPECS): New macro.
21333 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21335 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21337 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21338 are set after handling multilib fragments. Set target_cpu_default2
21341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21343 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21345 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21346 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21348 (check_arch): Likewise.
21349 * config/arm/arm.c (arm_configure_build_target): Handle
21350 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21351 options in the default.
21353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21355 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21356 when an option is an alias of another.
21357 * config/arm/parsecpu.awk (optalias): New parser token.
21358 (gen_comm_data): Mark non-alias options as such. Emit entries
21359 for extension aliases.
21360 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21361 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21362 (armv6kz, armv6zk, armv6t2): Likewise.
21363 (armv7): Make vfpv3-d16 an alias.
21364 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21366 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21367 Sort in canonical order.
21368 (armv8-a): Sort in canonical order.
21369 (armv8.1-a, armv8.2-a): Likewise.
21370 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21372 (cortex-a9): Sort in canonical order.
21373 * config/arm/arm.c (selftests.h): Include it.
21374 (arm_test_cpu_arch_data): New function.
21375 (arm_run_self_tests): New function.
21376 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21377 (targetm): Move declaration to the end of the file.
21378 * arm-cpu-cdata.h: Regenerated.
21380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21382 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21383 call to target_mode_check describing the type of option passed.
21384 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21385 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21386 arm_parse_cpu_option_name to match parameters against list of
21388 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21389 arm_arch_core_flags data structure.
21390 * config/arm/arm-cpu_cdata.h: Regenerated.
21392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21394 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21396 (arm_print_hint_for_cpu_option): Likewise.
21397 (arm_print_hint_for_arch_option): Likewise.
21398 (arm_parse_cpu_option_name): Likewise.
21399 (arm_parse_arch_option_name): Likewise.
21400 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21401 of entries in the all_fpus list.
21402 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21403 (arm_parse_cpu_option_name): Declare.
21404 (arm_parse_arch_option_name): Declare.
21405 (arm_parse_option_features): Declare.
21406 (arm_intialize_isa): Declare.
21407 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21409 (gen_comm_data): ... here. Make definitions non-static.
21410 * config/arm/arm-cpu-data.h: Regenerated.
21411 * config/arm/arm-cpu-cdata.h: Regenerated.
21413 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21415 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21416 (cpu_arch_extension): New structure.
21417 (cpu_arch_option, arch_option, cpu_option): New structures.
21418 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21419 architecture types.
21420 (gen_data): Generate new format data tables.
21421 * config/arm/arm.c (cpu_tune): New structure.
21422 (cpu_option, processors): Delete.
21423 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21424 (arm_print_hint_for_cpu_option): ... this and ...
21425 (arm_print_hint_for_arch_option): ... this.
21426 (arm_parse_arch_cpu_name): Delete. Replace with ...
21427 (arm_parse_cpu_option_name): ... this and ...
21428 (arm_parse_arch_option_name): ... this.
21429 (arm_unrecognized_feature): Change type of target parameter to
21431 (arm_parse_arch_cpu_features): Delete. Replace with ...
21432 (arm_parse_option_features): ... this.
21433 (arm_configure_build_target): Rework to use new configuration data
21435 (arm_print_tune_info): Rework for new configuration data tables.
21436 * config/arm/arm-cpu-data.h: Regenerated.
21437 * config/arm/arm-cpu.h: Regenerated.
21439 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21441 * Makefile.in (OBJS): Move sbitmap.o from here ...
21442 (OBJS-libcommon): ... to here.
21444 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21446 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21447 (ISA_ALL_CRYPTO): New macro.
21448 (ISA_ALL_SIMD): New macro
21449 (ISA_ALL_FP): New macro.
21450 * config/arm/arm.c (fpu_bitlist): Update initializer.
21451 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21453 (arm9e): Add fpu. Add option for nofp
21454 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21455 (arm926ej-s, arm1026ej-s): Likewise.
21456 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21457 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21458 neon-fp16, neon-vfpv4, nofp and nosimd.
21459 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21460 (cortex-a8): Add fpu. Add option for nofp.
21461 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21462 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21463 (cortex-r4f): Add fpu.
21464 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21465 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21467 (cortex-r8): Likewise.
21468 (cortex-m4): Add fpu. Add option for nofp.
21469 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21470 (cortex-a17.cortex-a7): Likewise.
21471 (cortex-a32): Add fpu. Add options for crypto and nofp.
21472 (cortex-a35, cortex-a53): Likewise.
21473 (cortex-a57): Add fpu. Add option for crypto.
21474 (cortex-a72, cortex-a73): Likewise.
21475 (exynos-m1): Likewise.
21476 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21477 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21478 (cortex-m33): Add fpu. Add option for nofp.
21479 * config/arm/arm-cpu-cdata.h: Regenerated
21480 * config/arm/arm-cpu-data.h: Regenerated.
21482 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21484 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21485 (armv5te, armv5tej): Likewise.
21486 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21487 (armv7): Add options fp and vfpv3-d16.
21488 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21489 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21491 (armv7ve): Likewise.
21492 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21493 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21494 (armv8-a): Add nocrypto option.
21495 (armv8.1-a, armv8.2-a): Likewise.
21496 (armv8-m.main): add options fp, fp.dp and nofp.
21498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21500 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21502 (armv8-a+crc): Delete.
21503 (armv8.1-a): Add options simd, crypto and nofp.
21504 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21505 (armv8.2-a+fp16): Delete.
21506 (armv8-m.main): Add option dsp.
21507 (armv8-m.main+dsp): Delete.
21508 (cortex-a8): Add fpu. Add nofp option.
21509 (cortex-a9): Add fpu. Add nofp and nosimd options.
21510 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21511 link to main cpu and architecture data structures.
21512 (gen_comm_data): Only put isa attributes from the main architecture
21514 (option): New statement for architecture and CPU entries.
21515 * arm.c (struct cpu_option): New structure.
21516 (struct processors): Add entry for options.
21517 (arm_unrecognized_feature): New function.
21518 (arm_parse_arch_cpu_name): Ignore any characters after the first
21520 (arm_parse_arch_cpu_feature): New function.
21521 (arm_configure_build_target): Separate out any CPU and architecture
21522 features and parse separately. Don't error out if -mfpu=auto is
21523 used with only an architecture string.
21524 (arm_print_asm_arch_directives): New function.
21525 (arm_file_start): Call it.
21526 * config/arm/arm-cpu-cdata.h: Regenerated.
21527 * config/arm/arm-cpu-data.h: Likewise.
21528 * config/arm/arm-tables.opt: Likewise.
21530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21532 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21533 assembler when it is not -mfpu=auto.
21535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21537 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21538 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21539 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21540 (ASM_CPU_SPEC): Rewrite.
21541 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21542 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21543 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21544 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21545 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21546 copied string is NUL-terminated. Also strip any characters prefixed
21548 (arm_rewrite_selected_arch): New function.
21549 (arm_rewrite_march): New function.
21551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21553 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21554 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21555 (march, mcpu, mtune): Convert to string-based options.
21556 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21557 (arm_parse_arch_cpu_name): New function.
21558 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21559 identify selected architecture or CPU.
21560 (arm_option_save): New function.
21561 (TARGET_OPTION_SAVE): Redefine.
21562 (arm_option_restore): Restore string options.
21563 (arm_option_print): Print string options.
21565 2017-06-16 Martin Sebor <msebor@redhat.com>
21567 PR tree-optimization/80933
21568 PR tree-optimization/80934
21569 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21570 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21571 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21572 (gimple_fold_builtin): Call them.
21574 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21576 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21577 as unlikely; update profile.
21579 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21581 * predict.c (force_edge_cold): Handle declaring edges impossible
21584 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21586 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21588 (try_unroll_loop_completely): Fix reporting.
21590 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21592 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21594 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21597 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21598 if given a non-constant argument for an intrinsic which requires a
21601 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21603 * profile.c (compare_freqs): New function.
21604 (branch_prob): Sort edge list.
21605 (find_spanning_tree): Assume that the list is priority sorted.
21607 2017-06-16 Richard Biener <rguenther@suse.de>
21609 PR tree-optimization/81090
21610 * passes.def (pass_record_bounds): Remove.
21611 * tree-pass.h (make_pass_record_bounds): Likewise.
21612 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21613 make_pass_record_bounds): Likewise.
21614 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21615 not free niter estimates at the beginning but at the end.
21616 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21618 2017-06-16 Richard Biener <rguenther@suse.de>
21620 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21621 initializer to workaround ICE in host GCC 4.8.
21623 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21625 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21627 (clone_inlined_nodes): Update.
21629 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21631 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21632 prefetch settings, and enable prefetching by default at -O3.
21634 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21636 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21637 Set flag_prefetch_loop_arrays according to tuning data.
21639 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21641 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21642 New tune structure.
21643 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21644 [Unrelated to main purpose of the patch] Place the pointer field last
21645 to enable type checking errors when tune structure are wrongly merged.
21646 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21647 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21648 (thunderx2t99_prefetch_tune): New tune constants.
21649 (tune_params *_tunings): Update all tunings (no functional change).
21650 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21651 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21652 from tunings structures.
21654 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21657 * ubsan.c (instrument_null): Add T argument, use it instead
21658 of computing it based on IS_LHS.
21659 (instrument_object_size): Likewise.
21660 (pass_ubsan::execute): Adjust instrument_null and
21661 instrument_object_size callers to pass gimple_get_lhs or
21662 gimple_assign_rhs1 result to it. Use instrument_null instead of
21663 calling get_base_address and instrument_mem_ref. Handle
21664 aggregate call arguments for object-size sanitization.
21666 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21668 PR tree-optimization/81089
21669 * tree-vrp.c (is_masked_range_test): Validate operands of
21672 2017-06-15 Martin Sebor <msebor@redhat.com>
21675 * dumpfile.c (dump_register): Avoid calling memset to initialize
21676 a class with a default ctor.
21677 * gcc.c (struct compiler): Remove const qualification.
21678 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21679 * hash-table.h: Ditto.
21680 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21682 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21683 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21685 * params.h (struct param_info): Make struct members non-const.
21686 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21687 with default initialization.
21688 * vec.h (vec_copy_construct, vec_default_construct): New helper
21690 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21691 with vec_copy_construct.
21692 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21693 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21694 * doc/invoke.texi (-Wclass-memaccess): Document.
21696 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21698 * emit-rtl.h (is_leaf): Update comment about local
21699 register allocator.
21701 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21704 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21705 for a variable to have a section before checking if the section has a
21707 Set section to.persistent if persistent attribute is set.
21708 Warn if .persistent attribute is used on an automatic variable.
21710 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21712 PR rtl-optimization/80474
21713 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21715 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21717 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21720 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21722 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21723 (casesi): Emit bounds checking as RTL.
21724 (casesi_internal_mips16_<mode>): Remove bounds checking.
21726 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21728 * config/xtensa/xtensa.c (xtensa_option_override): Append
21729 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21730 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21731 xtensa_doloop_hooks): Define unconditionally.
21732 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21734 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21735 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21736 for it in xtensa_option_override.
21737 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21738 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21740 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21742 * doc/cppopts.texi: Document '-' special value to -MF.
21744 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21746 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21747 (cortex_a53_fconst): Likewise.
21748 (cortex_a53_fpmul): Likewise.
21749 (cortex_a53_f_load_64): Likewise.
21750 (cortex_a53_f_load_many): Likewise.
21751 (cortex_a53_advsimd_alu): Likewise.
21752 (cortex_a53_advsimd_alu_q): Likewise.
21753 (cortex_a53_advsimd_mul): Likewise.
21754 (cortex_a53_advsimd_mul_q): Likewise.
21755 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21756 Add missing fmul, r2f_cvt and fconst cases.
21758 2017-06-14 Richard Biener <rguenther@suse.de>
21760 PR middle-end/81088
21761 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21763 (fold_binary_loc): When associating do not treat pre-existing
21764 TREE_OVERFLOW on literal constants as a reason to allow
21765 TREE_OVERFLOW on associated literal constants.
21767 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21769 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21770 (MASK_FEATURES): New macro.
21771 * config/sparc/sparc.c (sparc_option_override): Remove the special
21772 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21774 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21776 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21777 a division of 0 if non-call exceptions are enabled.
21779 2017-06-14 Andrew Pinski <apinski@cavium.com>
21780 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21783 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21784 Improve vector initialization code gen for only variable case.
21786 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21788 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21790 2017-06-14 Richard Biener <rguenther@suse.de>
21792 PR tree-optimization/81083
21793 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21796 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21798 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21799 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21800 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21801 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21802 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21803 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21805 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21807 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21808 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21810 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21812 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21814 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21816 * config/rs6000/t-rtems: Don't handle SPE.
21818 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21820 * config/rs6000/t-linux: Don't handle SPE.
21822 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21824 * config/rs6000/eabispe.h: Delete file.
21826 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21828 * config/rs6000/t-spe: Delete file.
21830 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21832 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21833 (rs6000_legitimate_offset_address_p): Return false for anything in
21834 V2SImode or V2SFmode.
21836 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21838 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21839 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21840 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21842 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21843 (rs6000_legitimate_offset_address_p): Ditto.
21844 (rs6000_emit_move): Ditto.
21845 (rs6000_init_builtins): Remove V4HI_type_node.
21847 2017-06-13 Martin Liska <mliska@suse.cz>
21850 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21851 (gate_asan): Likewise.
21852 * asan.h (asan_no_sanitize_address_p): Remove the function.
21853 (sanitize_flags_p): New function.
21854 * builtins.def: Fix coding style.
21855 * common.opt: Use renamed enum value.
21856 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21857 * doc/extend.texi: Document no_sanitize attribute.
21858 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21859 to SANITIZE_UNDEFINED_NONDEFAULT.
21860 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21861 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21862 Use sanitize_flags_p.
21863 * gimplify.c (gimplify_function_tree): Likewise.
21864 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21865 * opts.c (parse_no_sanitize_attribute): New function.
21866 (common_handle_option): Use renamed enum value.
21867 * opts.h (parse_no_sanitize_attribute): Declare.
21868 * tree.c (sanitize_flags_p): New function.
21869 * tree.h: Declared here.
21870 * tsan.c: Use sanitize_flags_p.
21871 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21872 (instrument_mem_ref): Likewise.
21873 (instrument_bool_enum_load): Likewise.
21874 (do_ubsan_in_current_function): Remove the function.
21875 (pass_ubsan::execute): Use sanitize_flags_p.
21876 * ubsan.h: Remove do_ubsan_in_current_function
21877 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21878 (dump_function_to_file): Use it here.
21880 2017-06-13 Martin Jambor <mjambor@suse.cz>
21882 PR tree-optimization/80803
21883 PR tree-optimization/81063
21884 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21885 (propagate_subaccesses_across_link): Enqueue subtree whenever
21886 necessary instead of relying on the caller.
21888 2017-06-13 Martin Jambor <mjambor@suse.cz>
21890 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21891 that have a first_link.
21892 (sort_and_splice_var_accesses): Do not check first_link before
21894 (subtree_mark_written_and_enqueue): Likewise.
21895 (propagate_all_subaccesses): Likewise and do not stop at first
21896 parent with a first_link.
21898 2017-06-13 Martin Jambor <mjambor@suse.cz>
21900 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21903 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21905 * match.pd: New pattern.
21907 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21909 * tree-vrp.c (is_masked_range_test): New function.
21910 (register_edge_assert_for): Determine ranges for
21913 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21915 PR tree-optimization/67328
21916 * fold-const.c (maskable_range_p): New function.
21917 (build_range_check): Generate bittests if possible.
21919 2017-06-13 Martin Liska <mliska@suse.cz>
21921 * gimple-pretty-print.c (dump_probability): Add new argument.
21922 (dump_edge_probability): Dump both probability and count.
21923 (dump_gimple_label): Likewise.
21924 (dump_gimple_bb_header): Likewise.
21926 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21929 * config/avr/avr-devices.c: Fix indentation.
21930 * config/avr/gen-avr-mmcu-specs.c: Dito.
21932 2017-06-13 Richard Biener <rguenther@suse.de>
21934 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21935 instead get vector type from stmt_info.
21936 (vectorizable_reduction): Adjust. Remove dead code.
21938 2017-06-13 Richard Biener <rguenther@suse.de>
21940 PR middle-end/81065
21941 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21942 case of C * (x * C2 + C3).
21943 (fold_addr_of_array_ref_difference): Properly fold index difference.
21945 2017-06-12 David S. Miller <davem@davemloft.net>
21948 * config/sparc/sparc.md (return expander): Emit frame blockage if
21949 function uses alloca.
21951 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21953 * combine.c (make_field_assignment): Check len rather than the mode
21954 precision when calling force_to_mode.
21956 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21958 Support multilibs and devices that see flash in RAM address range.
21961 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21962 (avr_mcu_t) <flash_pm_offset>: New field.
21963 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21964 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21965 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21966 (AVR_TINY_PM_OFFSET): Remove macro.
21967 * config/avr/avr.opt (-mshort-calls): New option.
21968 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21969 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21970 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21971 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21972 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21973 instead of avr_arch->have_jmp_call.
21974 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21975 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21976 avr_arch->flash_pm_offset to define.
21977 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21978 new field flash_pm_offset. Add entry for avrxmega3.
21979 (avr_texinfo): Add entry for avrxmega3.
21980 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21981 attiny212, attiny214,
21982 attiny412, attiny414, attiny416, attiny417,
21983 attiny814, attiny816, attiny817,
21984 attiny1614, attiny1616, attiny1617,
21985 attiny3214, attiny3216, attiny3217.
21986 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21987 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21988 (avr_print_operand_address) [AVR_TINY]: Same.
21989 (avr_asm_init_sections) <readonly_data_section>: Only patch
21990 callback if avr_arch->flash_pm_offset = 0.
21991 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21992 for rodata if avr_arch->flash_pm_offset != 0.
21993 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21994 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21995 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21996 (m_options): Append opt_rcall.
21997 (m_dirnames): Append dir_rcall.
21998 * config/avr/t-multilib: Regenerate.
22000 * configure.ac [target=avr]: Check whether avrxmega3 default
22001 linker description file works as needed.
22002 * configure: Regenerate.
22003 * doc/avr-mmcu.texi: Regenerate.
22004 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
22005 <__AVR_ARCH__>: Document avrxmega3 and 103.
22006 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
22007 <__AVR_SHORT_CALLS__>: Document it.
22008 <__AVR_PM_BASE_ADDRESS__>: Document it.
22009 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
22010 (AVR Variable Attributes) <progmem>: Document this is
22011 not needed for avrxmega3.
22012 (AVR Named Address Spaces) <__flash>: Dito.
22014 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
22016 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
22018 2017-06-12 Doug Rupp <rupp@adacore.com>
22020 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
22021 Append vxworks-stdint.h to the tm_file list.
22022 * config/vxworks-stdint.h: New file.
22024 2017-06-12 Martin Liska <mliska@suse.cz>
22026 PR tree-optimization/81041
22027 * tree-profile.c (gimple_gen_ic_func_profiler):
22028 Create an extra BB in profile-generate
22029 (gimple_gen_time_profiler): Likewise.
22031 2017-06-12 Jakub Jelinek <jakub@redhat.com>
22033 PR tree-optimization/81003
22034 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
22035 (update_range_test): Use it instead of force_gimple_operand_gsi.
22037 2017-06-12 Richard Biener <rguenther@suse.de>
22039 PR tree-optimization/81053
22040 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
22041 with backedge value not defined in loop. Simplify def stmt
22044 2017-06-11 Tom de Vries <tom@codesourcery.com>
22047 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
22049 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
22050 nvptx_cannot_force_const_mem.
22052 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22054 * opts.c (finish_options): Move test for flag_split_stack after
22055 it has been initialized.
22057 2017-06-11 Jason Merrill <jason@redhat.com>
22059 * tree.h (id_equal): New.
22060 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
22061 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
22062 instead of strcmp of IDENTIFIER_POINTER.
22064 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22066 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
22067 (mark_all_inlined_calls_cdtor): Fix formating.
22068 (inline_transform): Rescale profile before inlining.
22070 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22072 * cgraph.h (cgraph_edge::clone): Update prototype.
22073 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
22074 (cgraph_node::create_clone): Update.
22075 (cgraph_node::create_version_clone): Update.
22076 * tree-inline.c (copy_bb): Update.
22077 (expand_call_inline): Update.
22079 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22081 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
22082 factored out from ...
22083 (rs6000_emit_prologue): ... here.
22085 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
22087 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
22088 factored out from ...
22089 (rs6000_emit_prologue): ... here.
22091 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22093 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
22095 (handle_missing_profiles): Fix computation of tp_first_run.
22096 (counts_to_freqs): Do not touch freqs when count is 0.
22098 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
22100 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22103 2017-06-10 Tom de Vries <tom@codesourcery.com>
22105 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22106 attributes): Document signal effective target.
22108 2017-06-10 Tom de Vries <tom@codesourcery.com>
22110 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22111 Document effective target stack_size.
22113 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22115 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22116 to the edit_context if they can be auto-applied.
22118 2017-06-9 Ian Lance Taylor <iant@golang.org>
22120 * opts.c (finish_options): If -fsplit-stack, disable implicit
22121 -forder-blocks-and-partition.
22122 * doc/invoke.texi (Optimize Options): Document that when using
22123 -fsplit-stack -forder-blocks-and-partition is not implicitly
22126 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22128 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22129 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22130 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22131 * builtins.def (abort, trap, unreachable): Declare cold.
22132 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22133 * tree-core.h (ECF_COLD): New.
22134 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22135 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22137 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22139 * predict.c (unlikely_executed_stmt_p): Cleanup.
22141 2017-06-09 Richard Biener <rguenther@suse.de>
22143 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22144 model if the ref is always written to.
22146 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22148 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22150 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22152 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22155 2017-06-09 Tom de Vries <tom@codesourcery.com>
22158 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22159 "target cannot support label values" when encountering LABEL_REF.
22161 2017-06-09 Martin Liska <mliska@suse.cz>
22163 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22164 (gimple_gen_ic_func_profiler): Emit direct comparison
22165 of __gcov_indirect_call_callee with NULL.
22166 (gimple_gen_time_profiler): Change probability from
22167 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22169 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22171 * profile.c (edge_gcov_counts): Turn to pointer.
22172 (compute_branch_probabilities, compute_branch_probabilities): Update.
22173 (branch_prob): Do not clear edge_gcov_count.
22174 * profile.h (edge_gcov_counts): Turn to pointer.
22175 (edge_gcov_count): Update.
22177 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22179 * gimple.h (gimple_check_failed): Mark cold.
22181 2017-06-09 Richard Biener <rguenther@suse.de>
22183 PR tree-optimization/66623
22184 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22185 refactor check_reduction into two parts, properly computing
22186 whether we have to check reduction validity for outer loop
22189 2017-06-09 Richard Biener <rguenther@suse.de>
22191 PR tree-optimization/79483
22192 * graphite-scop-detection.c (order): New global.
22193 (get_order): Compute bb to order mapping that satisfies code
22194 generation constraints.
22195 (cmp_pbbs): New helper.
22196 (build_scops): Start domwalk at entry block, sort generated
22199 2017-06-09 Richard Biener <rguenther@suse.de>
22201 PR middle-end/81007
22202 * ipa-polymorphic-call.c
22203 (ipa_polymorphic_call_context::restrict_to_inner_class):
22204 Skip FIELD_DECLs with error_mark_node type.
22205 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22208 2017-06-09 Martin Liska <mliska@suse.cz>
22210 * predict.c (struct branch_predictor): New struct.
22211 (test_prediction_value_range): New test.
22212 (predict_c_tests): New function.
22213 * selftest-run-tests.c (selftest::run_tests): Run the function.
22214 * selftest.h: Declare new tests.
22216 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22219 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22220 gen_add3_insn did not fail.
22221 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22222 r0, construct that number in a temporary reg and add that reg to r0.
22223 If asked to put the result in r0 as well, fail.
22225 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22227 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22228 for early expansion of vec_eqv.
22230 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22232 PR middle-end/81005
22233 * ubsan.c (instrument_null): Avoid pointless code temporary.
22234 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22236 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22240 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22243 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22246 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22248 * predict.c (unlikely_executed_edge_p): Move ahead.
22249 (probably_never_executed_edge_p): Use it.
22251 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22253 PR middle-end/79988
22254 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22255 gimple_call_builtin_p call.
22257 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22259 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22260 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22261 rtl_check_failed_type2, rtl_check_failed_code1,
22262 rtl_check_failed_code2, rtl_check_failed_code_mode,
22263 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22264 rtvec_check_failed_bounds, rtl_check_failed_flag,
22265 _fatal_insn_not_found, _fatal_insn): Likewise.
22266 * tree.h (tree_contains_struct_check_failed,
22267 tree_check_failed, tree_not_check_failed,
22268 tree_class_check_failed, tree_range_check_failed,
22269 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22270 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22271 tree_operand_check_failed, omp_clause_check_failed,
22272 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22275 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22277 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22278 flag_branch_probabilities.
22279 * ipa-inline.c (edge_badness): Likewise.
22280 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22281 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22282 * predict.c (maybe_hot_frequency_p): Likewise.
22283 (probably_never_executed): Likewise.
22284 * sched-ebb.c (schedule_ebbs): Likewise.
22285 * sched-rgn.c (find_single_block_region): Likewise.
22286 * tracer.c (tail_duplicate): Likewise.
22288 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22290 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22291 longer requires x_flag_profile_use.
22293 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22295 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22296 instead of flag_reorder_blocks_and_partition.
22297 * dbxout.c (dbxout_function_end): Likewise.
22298 * dwarf2out.c (gen_subprogram_die): Likewise.
22299 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22300 * hw-doloop.c (reorg_loops): Likewise.
22301 * varasm.c (assemble_start_function,
22302 assemble_end_function): Likewise.
22303 (decide_function_section): Do not check for
22304 flag_reorder_blocks_and_partition.
22306 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22308 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22310 (chkp_get_hard_register_fake_addr_expr): Ditto.
22311 (chkp_build_addr_expr): Add check for hard reg case.
22312 (chkp_parse_array_and_component_ref): Ditto.
22313 (chkp_find_bounds_1): Ditto.
22314 (chkp_process_stmt): Don't generate bounds store for
22317 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22319 * predict.c (maybe_hot_bb_p): Do not check profile status.
22320 (maybe_hot_edge_p): Likewise.
22321 (probably_never_executed): Check for zero counts even if profile
22323 (unlikely_executed_edge_p): New function.
22324 (unlikely_executed_stmt_p): New function.
22325 (unlikely_executed_bb_p): New function.
22326 (set_even_probabilities): Use unlikely predicates.
22327 (combine_predictions_for_bb): Likewise.
22328 (predict_paths_for_bb): Likewise.
22329 (predict_paths_leading_to_edge): Likewise.
22330 (determine_unlikely_bbs): New function.
22331 (estimate_bb_frequencies): Use it.
22332 (compute_function_frequency): Use zero counts even if profile is
22334 * profile-count.h: Fix typo.
22336 2017-08-08 Julia Koval <julia.koval@intel.com>
22338 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22339 _mm512_mask_cvtsepi16_storeu_epi8,
22340 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22341 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22342 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22343 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22344 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22345 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22346 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22347 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22348 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22349 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22350 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22351 __builtin_ia32_pmovuswb256mem_mask,
22352 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22353 __builtin_ia32_pmovwb512mem_mask): New builtins.
22355 2017-08-08 Julia Koval <julia.koval@intel.com>
22357 PR target/73350,80862
22358 * config/i386/subst.md (round): Fix round pattern.
22359 * config/i386/i386.c (ix86_erase_embedded_rounding):
22360 Fix erasing rounding for the fixed pattern.
22362 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22364 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22366 2017-06-08 Martin Liska <mliska@suse.cz>
22368 PR gcov-profile/80911
22369 * gcov.c (block_info::block_info): New constructor.
22371 2017-06-07 Carl Love <cel@us.ibm.com>
22373 * config/rs6000/rs6000-c: The return type of the following
22374 built-in functions was implemented as int not long long. Fix sign
22375 of return value for the unsigned version of vec_mulo and vec_mule.
22376 vector unsigned long long vec_bperm (vector unsigned long long,
22377 vector unsigned char)
22378 vector signed long long vec_mule (vector signed int,
22380 vector unsigned long long vec_mule (vector unsigned int,
22381 vector unsigned int)
22382 vector signed long long vec_mulo (vector signed int,
22384 vector unsigned long long vec_mulo (vector unsigned int,
22385 vector unsigned int)
22386 * doc/extend.texi: Fix the documentation for the built-in
22389 2017-06-07 Carl Love <cel@us.ibm.com>
22392 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22395 2017-06-07 Carl Love <cel@us.ibm.com>
22397 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22398 support, Generate doublehv for signed int/float for BE case only.
22400 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22402 * doc/invoke.texi (mcx16): Rewrite.
22404 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22406 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22407 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22408 *mov<mode>_softfloat, and an anonymous splitter): Use
22409 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22411 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22413 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22415 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22416 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22417 (rs6000_debug_reg_global): Adjust.
22418 (rs6000_init_hard_regno_mode_ok): Adjust.
22419 (rs6000_dbx_register_number): Adjust.
22420 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22421 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22422 Remove SPE_ACC and SPEFSCR.
22423 (REG_ALLOC_ORDER): Ditto.
22424 (FRAME_POINTER_REGNUM): Change to 111.
22425 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22426 (REG_CLASS_NAMES): Ditto.
22427 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22428 (REGISTER_NAMES): Ditto.
22429 (ADDITIONAL_REG_NAMES): Ditto.
22430 (rs6000_reg_names): Ditto.
22431 * config/rs6000/rs6000.md: Renumber some register number
22434 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22436 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22438 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22439 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22441 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22442 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22443 Delete the SPE high registers.
22444 (REG_ALLOC_ORDER): Ditto.
22445 (enum reg_class): Remove SPE_HIGH_REGS.
22446 (REG_CLASS_NAMES): Ditto.
22447 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22448 (REGISTER_NAMES): Ditto.
22449 (rs6000_reg_names): Ditto.
22450 * doc/tm.texi.in: Remove SPE as example.
22451 * doc/tm.texi: Regenerate.
22453 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22455 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22456 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22457 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22458 * config/rs6000/rs6000.md (type): Remove "brinc".
22460 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22462 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22463 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22464 * config/rs6000/linuxspe.h: Delete file.
22465 * config/rs6000/rs6000.md: Don't include spe.md.
22466 * config/rs6000/spe.h: Delete file.
22467 * config/rs6000/spe.md: Delete file.
22468 * config/rs6000/t-rs6000: Remove spe.md.
22470 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22472 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22473 (reg_or_none500mem_operand): Delete.
22474 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22475 instead of reg_or_none500mem_operand.
22477 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22480 handling of SPE flags.
22481 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22483 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22485 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22487 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22488 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22489 paired_divv2sf3): Similar.
22490 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22491 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22492 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22494 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22495 Rename the paired_* instruction patterns.
22496 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22498 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22499 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22500 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22501 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22502 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22503 PAIRED_VECTOR_MODE.
22504 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22505 (spe_func_has_64bit_regs_p): Delete.
22506 (spe_expand_predicate_builtin): Delete.
22507 (spe_expand_evsel_builtin): Delete.
22508 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22509 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22510 (invalid_e500_subreg): Delete.
22511 (rs6000_legitimize_address): Always force_reg op2 as well, for
22512 paired single memory accesses.
22513 (rs6000_member_type_forces_blk): Delete.
22514 (rs6000_spe_function_arg): Delete.
22515 (rs6000_expand_unop_builtin): Delete SPE handling.
22516 (rs6000_expand_binop_builtin): Ditto.
22517 (spe_expand_stv_builtin): Delete.
22518 (bdesc_2arg_spe): Delete.
22519 (spe_expand_builtin): Delete.
22520 (spe_expand_predicate_builtin): Delete.
22521 (spe_expand_evsel_builtin): Delete.
22522 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22523 (spe_init_builtins): Delete.
22524 (spe_func_has_64bit_regs_p): Delete.
22525 (savres_routine_name): Delete "info" parameter. Adjust callers.
22526 (rs6000_emit_stack_reset): Ditto.
22527 (rs6000_dwarf_register_span): Delete.
22528 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22529 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22530 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22532 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22534 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22535 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22536 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22537 mulv2sf3, divv2sf3): Delete expanders.
22539 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22541 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22543 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22545 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22546 * config/rs6000/rs6000.c: Ditto.
22548 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22550 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22551 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22552 comparison_operator.
22554 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22556 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22557 * config/rs6000/rs6000.opt: Ditto.
22558 * config/rs6000/t-rtems: Ditto.
22560 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22562 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22563 TARGET_E500_SINGLE by 0, simplify.
22564 * config/rs6000/rs6000.c: Ditto.
22565 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22566 (spe_build_register_parallel): Delete.
22567 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22568 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22569 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22570 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22571 (E500_CONVERT): Delete.
22572 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22574 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22576 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22577 * config/rs6000/dfp.md: Ditto.
22578 (negdd2, *negdd2_fpr): Merge.
22579 (absdd2, *absdd2_fpr): Merge.
22580 (negtd2, *negtd2_fpr): Merge.
22581 (abstd2, *abstd2_fpr): Merge.
22582 * config/rs6000/e500.h: Delete file.
22583 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22584 TARGET_FPRS by 1 and simplify.
22585 * config/rs6000/rs6000-c.c: Ditto.
22586 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22587 TARGET_DF_SPE by 0.
22588 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22590 * config/rs6000/rs6000.md: Ditto.
22591 (floatdidf2, *floatdidf2_fpr): Merge.
22592 (move_from_CR_gt_bit): Delete.
22593 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22594 (E500_CR_IOR_COMPARE): Delete.
22595 (All patterns that require !TARGET_FPRS): Delete.
22596 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22598 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22600 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22602 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22604 * graphds.c (add_edge): Intitialize edge's attached data.
22605 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22606 pointer parameter. Call pointed function on each edge during
22607 graph traversing. Skip traversing the edge when the function
22609 (graphds_dfs, graphds_scc): Ditto.
22610 (for_each_edge): New parameter. Pass the new parameter to callback
22612 * graphds.h (skip_edge_callback): New function pointer type.
22613 (graphds_dfs, graphds_scc): New function pointer parameter.
22614 (graphds_edge_callback, for_each_edge): New parameter.
22616 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22618 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22619 out code checking if runtime alias check is possible to below ...
22620 Call the new function.
22621 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22622 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22624 2017-06-07 Marek Polacek <polacek@redhat.com>
22627 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22628 TYPE_OVERFLOW_WRAPS check.
22630 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22632 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22633 if versioning is required.
22634 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22635 peeling with the check for versioning.
22637 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22639 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22640 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22641 Set true to new parameter if new ssa variable is defined.
22642 (vect_gen_vector_loop_niters): Refactor. Set range information
22643 for the new vector loop bound variable.
22644 (vect_do_peeling): Ditto.
22646 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22648 * tree-affine.c (ssa.h): Include header file.
22649 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22650 has wrapping overflow behavior.
22652 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22654 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22656 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22658 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22659 (tree_to_aff_combination): ... here.
22661 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22663 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22664 reg_pressure model function.
22665 (ivopts_global_cost_for_size): Delete.
22666 (determine_set_costs, iv_ca_recount_cost): Call new model function
22667 ivopts_estimate_reg_pressure.
22669 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22671 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22672 expensive than udiv. Remove floating point cases from mod.
22674 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22676 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22677 Increase idiv cost.
22679 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22681 * config/aarch64/aarch64.md
22682 (copysignsf3): Fix mask generation.
22684 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22686 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22688 (class dump_manager): Add register_dumps method.
22689 * dumpfile.c: Include langhooks.h.
22690 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22691 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22692 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22693 (dump_manager::dump_register): Start with 512 entries instead of 32.
22694 (dump_manager::register_dumps): New method.
22695 * toplev.c (general_init): Instead of invoking register_dumps
22696 langhook, invoke register_dumps method on the dump manager.
22697 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22700 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22702 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22703 Say that # requires an associated define_split to exist, and that
22704 the define_split must be suitable for use after register allocation.
22706 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22708 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22709 (compute_outgoing_frequencies): Also initialize zero counts.
22710 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22711 around loops; preserve more of profile when nothing changes.
22713 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22715 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22717 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22718 * config/arm/arm-cpu-cdata.h: Regenerate.
22719 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22720 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22721 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22722 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22723 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22725 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22727 * config/arm/t-rmprofile: Likewise.
22728 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22730 2017-06-06 David S. Miller <davem@davemloft.net>
22733 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22734 blockage if function uses alloca.
22736 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22738 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22739 New "uid" fields to hold pretty-print IDs of group and ref.
22740 Memory references are now identified as <group_id>:<ref_id>
22741 instead of using [random] addresses.
22742 (dump_mem_details): Simplify, no functional change.
22743 (dump_mem_ref): Simplify and make output more concise.
22744 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22745 (find_or_create_group): Initialize group uid.
22746 (record_ref): Initialize ref uid. Improve debug output.
22747 (prune_group_by_reuse, should_issue_prefetch_p,)
22748 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22749 (mark_nontemporal_store, determine_loop_nest_reuse):
22750 Improve debug output.
22752 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22754 * dbgcnt.def (prefetch): New debug counter.
22755 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22756 (schedule_prefetches): Stop issueing prefetches if debug counter
22759 2017-06-06 Tom de Vries <tom@codesourcery.com>
22761 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22762 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22764 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22766 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22767 Use aarch64_reg_or_zero predicate for operand 4.
22768 (aarch64_compare_and_swap<mode> define_insn_and_split):
22769 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22770 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22772 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22774 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22775 (arm_compute_save_core_reg_mask): This.
22776 (thumb1_compute_save_reg_mask): Rename into ...
22777 (thumb1_compute_save_core_reg_mask): This.
22778 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22779 (arm_compute_frame_layout): Likewise.
22781 2017-06-06 Richard Biener <rguenther@suse.de>
22783 PR tree-optimization/80974
22784 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22785 keep or clear leaders SSA info.
22787 2017-06-06 Tom de Vries <tom@codesourcery.com>
22789 * config/nvptx/nvptx.c (split_mode_p): New function.
22790 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22792 2017-06-06 Tom de Vries <tom@codesourcery.com>
22794 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22796 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22799 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22802 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22804 * shrink-wrap.c (handle_simple_exit): Update profile.
22805 (try_shrink_wrapping): Upate profile.
22807 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22809 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22810 (tree_guess_outgoing_edge_probabilities): New.
22811 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22812 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22814 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22816 * ipa-split.c (split_function): Initialize return bb profile.
22818 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22820 * profile.c (compute_branch_probabilities): Also initialize
22821 EXIT_BLOCK profile.
22823 2017-06-06 Richard Biener <rguenther@suse.de>
22825 PR tree-optimization/80928
22826 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22827 (vect_analyze_loop_operations): Properly guard analysis for
22829 (vect_transform_loop): Likewise.
22830 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22831 (vect_model_induction_cost): Do not cost for pure SLP.
22832 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22833 of induction in inner loop vectorization.
22834 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22835 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22836 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22838 (vect_analyze_slp_cost_1): Cost induction.
22839 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22840 (vect_get_slp_vect_defs): Likewise.
22841 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22842 (vect_transform_stmt): Handle SLP reductions.
22843 * tree-vectorizer.h (vectorizable_induction): Adjust.
22845 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22847 * config/rs6000/rs6000.c (make_resolver_func): Update
22848 init_lowered_empty_function call.
22850 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22852 * doc/invoke.texi: Document the -fprofile-abs-path option.
22853 * common.opt (fprofile-abs-path): New option.
22854 * gcov-io.h (gcov_write_filename): Declare.
22855 * gcov-io.c (gcov_write_filename): New function.
22856 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22857 * profile.c (output_location): Likewise.
22859 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22861 * shring-wrap.c: Revert accidental commit.
22863 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22865 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22867 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22869 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22871 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22872 profile in callgraph edge.
22873 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22874 (apply_scale): Likewise.
22875 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22876 Also scale profile when inlining function with zero profile.
22877 (initialize_cfun): Update exit block profile even when it is zero.
22878 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22879 when profile is read.
22881 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22883 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22884 (CLONE_*): New constants to define the processors we can generate
22885 code for with the target_clone attribute.
22886 (rs6000_clone_map): New array to identify which clone processors
22887 the current program is running on.
22888 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22889 target_clone attribute.
22890 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22891 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22892 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22893 (cpu_expand_builtin): Add support for target_clone attribute.
22894 (rs6000_valid_attribute_p): Allow "default" attribute.
22895 (get_decl_name): New debug function to simplify printing the
22896 current function name in debugging statements.
22897 (rs6000_clone_priority): New functions to support the target_clone
22898 attribute, and be able to generate code to switch between ISA 2.05
22899 through ISA 3.0 (power6 through power9).
22900 (rs6000_compare_version_priority): Likewise.
22901 (rs6000_get_function_versions_dispatcher): Likewise.
22902 (make_resolver_func): Likewise.
22903 (add_condition_to_bb): Likewise.
22904 (dispatch_function_versions): Likewise.
22905 (rs6000_generate_version_dispatcher_body): Likewise.
22906 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22907 (fusion_gpr_load_p): Fix a spacing issue.
22908 * doc/extend.texi (Common Function Attributes): Document that the
22909 PowerPC supports the target_clone attribute.
22911 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22913 * config/arm/arm.h: explain F symbol found in description of ARM
22914 register allocation in its legend.
22916 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22918 * config/mips/frame-header-opt.c: Include profile-count.h.
22919 * config/riscv/riscv.c: Include profile-count.h
22921 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22923 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22925 (sm_set_flag_if_changed): Add bbs field.
22926 (execute_sm_if_changed_flag_set): Pass BBS.
22927 (execute_sm): Update.
22929 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22931 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22934 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22936 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22937 (peephole2): New peephole2 to emit the above.
22938 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22940 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22942 * config/aarch64/aarch64.c (define_peephole2 above
22943 *sub_<shift>_<mode>): New peephole.
22945 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22947 * config/i386/i386.c (make_resolver_func): Update.
22948 * Makefile.in: Add profile-count.h and profile-count.o
22949 * auto-profile.c (afdo_indirect_call): Update to new API.
22950 (afdo_set_bb_count): Update.
22951 (afdo_propagate_edge): Update.
22952 (afdo_propagate_circuit): Update.
22953 (afdo_calculate_branch_prob): Update.
22954 (afdo_annotate_cfg): Update.
22955 * basic-block.h: Include profile-count.h
22956 (struct edge_def): Turn count to profile_count.
22957 (struct basic_block_def): Likewie.
22958 (REG_BR_PROB_BASE): Move to profile-count.h
22959 (RDIV): Move to profile-count.h
22960 * bb-reorder.c (max_entry_count): Turn to profile_count.
22961 (find_traces): Update.
22962 (rotate_loop):Update.
22963 (connect_traces):Update.
22964 (sanitize_hot_paths):Update.
22965 * bt-load.c (migrate_btr_defs): Update.
22966 * cfg.c (RDIV): Remove.
22967 (init_flow): Use alloc_block.
22968 (alloc_block): Uninitialize count.
22969 (unchecked_make_edge): Uninitialize count.
22970 (check_bb_profile): Update.
22971 (dump_edge_info): Update.
22972 (dump_bb_info): Update.
22973 (update_bb_profile_for_threading): Update.
22974 (scale_bbs_frequencies_int): Update.
22975 (scale_bbs_frequencies_gcov_type): Update.
22976 (scale_bbs_frequencies_profile_count): New.
22977 * cfg.h (update_bb_profile_for_threading): Update.
22978 (scale_bbs_frequencies_profile_count): Declare.
22979 * cfgbuild.c (compute_outgoing_frequencies): Update.
22980 (find_many_sub_basic_blocks): Update.
22981 * cfgcleanup.c (try_forward_edges): Update.
22982 (try_crossjump_to_edge): Update.
22983 * cfgexpand.c (expand_gimple_tailcall): Update.
22984 (construct_exit_block): Update.
22985 * cfghooks.c (verify_flow_info): Update.
22986 (dump_bb_for_graph): Update.
22987 (split_edge): Update.
22988 (make_forwarder_block): Update.
22989 (duplicate_block): Update.
22990 (account_profile_record): Update.
22991 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22992 (get_estimated_loop_iterations): Update.
22993 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22994 (single_likely_exit): Update.
22995 * cfgloopmanip.c (scale_loop_profile): Update.
22997 (set_zero_probability): Update.
22998 (lv_adjust_loop_entry_edge): Update.
22999 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23000 (purge_dead_edges): Update.
23001 (rtl_account_profile_record): Update.
23002 * cgraph.c (cgraph_node::create): Uninitialize count.
23003 (symbol_table::create_edge): Uninitialize count.
23004 (cgraph_update_edges_for_call_stmt_node): Update.
23005 (cgraph_edge::dump_edge_flags): Update.
23006 (cgraph_node::dump): Update.
23007 (cgraph_edge::maybe_hot_p): Update.
23008 * cgraph.h: Include profile-count.h
23009 (create_clone), create_edge, create_indirect_edge): Update.
23010 (cgraph_node): Turn count to profile_count.
23011 (cgraph_edge0: Likewise.
23012 (make_speculative, clone): Update.
23013 (create_edge): Update.
23014 (init_lowered_empty_function): Update.
23015 * cgraphclones.c (cgraph_edge::clone): Update.
23016 (duplicate_thunk_for_node): Update.
23017 (cgraph_node::create_clone): Update.
23018 * cgraphunit.c (cgraph_node::analyze): Update.
23019 (cgraph_node::expand_thunk): Update.
23020 * final.c (dump_basic_block_info): Update.
23021 * gimple-streamer-in.c (input_bb): Update.
23022 * gimple-streamer-out.c (output_bb): Update.
23023 * graphite.c (print_global_statistics): Update.
23024 (print_graphite_scop_statistics): Update.
23025 * hsa-brig.c: Include basic-block.h.
23026 * hsa-dump.c: Include basic-block.h.
23027 * hsa-gen.c (T sum_slice): Update.
23028 (convert_switch_statements):Update.
23029 * hsa-regalloc.c: Include basic-block.h.
23030 * ipa-chkp.c (chkp_produce_thunks): Update.
23031 * ipa-cp.c (struct caller_statistics): Update.
23032 (init_caller_stats): Update.
23033 (gather_caller_stats): Update.
23034 (ipcp_cloning_candidate_p): Update.
23035 (good_cloning_opportunity_p): Update.
23036 (get_info_about_necessary_edges): Update.
23037 (dump_profile_updates): Update.
23038 (update_profiling_info): Update.
23039 (update_specialized_profile): Update.
23040 (perhaps_add_new_callers): Update.
23041 (decide_about_value): Update.
23042 (ipa_cp_c_finalize): Update.
23043 * ipa-devirt.c (struct odr_type_warn_count): Update.
23044 (struct decl_warn_count): Update.
23045 (struct final_warning_record): Update.
23046 (possible_polymorphic_call_targets): Update.
23047 (ipa_devirt): Update.
23048 * ipa-fnsummary.c (redirect_to_unreachable): Update.
23049 * ipa-icf.c (sem_function::merge): Update.
23050 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
23051 * ipa-inline.c (compute_uninlined_call_time): Update.
23052 (compute_inlined_call_time): Update.
23053 (want_inline_small_function_p): Update.
23054 (want_inline_self_recursive_call_p): Update.
23055 (edge_badness): Update.
23056 (lookup_recursive_calls): Update.
23057 (recursive_inlining): Update.
23058 (inline_small_functions): Update.
23059 (dump_overall_stats): Update.
23060 (dump_inline_stats): Update.
23061 * ipa-profile.c (ipa_profile_generate_summary): Update.
23062 (ipa_propagate_frequency): Update.
23063 (ipa_profile): Update.
23064 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23065 * ipa-utils.c (ipa_merge_profiles): Update.
23066 * loop-doloop.c (doloop_modify): Update.
23067 * loop-unroll.c (report_unroll): Update.
23068 (unroll_loop_runtime_iterations): Update.
23069 * lto-cgraph.c (lto_output_edge): Update.
23070 (lto_output_node): Update.
23071 (input_node): Update.
23072 (input_edge): Update.
23073 (merge_profile_summaries): Update.
23074 * lto-streamer-in.c (input_cfg): Update.
23075 * lto-streamer-out.c (output_cfg): Update.
23076 * mcf.c (create_fixup_graph): Update.
23077 (adjust_cfg_counts): Update.
23078 (sum_edge_counts): Update.
23079 * modulo-sched.c (sms_schedule): Update.
23080 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
23081 * predict.c (maybe_hot_count_p): Update.
23082 (probably_never_executed): Update.
23083 (dump_prediction): Update.
23084 (combine_predictions_for_bb): Update.
23085 (propagate_freq): Update.
23086 (handle_missing_profiles): Update.
23087 (counts_to_freqs): Update.
23088 (rebuild_frequencies): Update.
23089 (force_edge_cold): Update.
23090 * predict.h: Include profile-count.h
23091 (maybe_hot_count_p, counts_to_freqs): UPdate.
23092 * print-rtl-function.c: Do not include cfg.h
23093 * print-rtl.c: Include basic-block.h
23094 * profile-count.c: New file.
23095 * profile-count.h: New file.
23096 * profile.c (is_edge_inconsistent): Update.
23097 (correct_negative_edge_counts): Update.
23098 (is_inconsistent): Update.
23099 (set_bb_counts): Update.
23100 (read_profile_edge_counts): Update.
23101 (compute_frequency_overlap): Update.
23102 (compute_branch_probabilities): Update; Initialize and deinitialize
23104 (branch_prob): Update.
23105 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23106 (edge_gcov_count): New.
23107 (bb_gcov_count): New.
23108 * shrink-wrap.c (try_shrink_wrapping): Update.
23109 * tracer.c (better_p): Update.
23110 * trans-mem.c (expand_transaction): Update.
23111 (ipa_tm_insert_irr_call): Update.
23112 (ipa_tm_insert_gettmclone_call): Update.
23113 * tree-call-cdce.c: Update.
23114 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23115 (gimple_duplicate_sese_tail): Update.
23116 (gimple_account_profile_record): Update.
23117 (execute_fixup_cfg): Update.
23118 * tree-inline.c (copy_bb): Update.
23119 (copy_edges_for_bb): Update.
23120 (initialize_cfun): Update.
23121 (freqs_to_counts): Update.
23122 (copy_cfg_body): Update.
23123 (expand_call_inline): Update.
23124 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23125 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23126 (try_unroll_loop_completely): Update.
23127 (try_peel_loop): Update.
23128 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23129 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23130 * tree-ssa-loop-split.c (connect_loops): Update.
23131 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23132 * tree-ssa-reassoc.c (branch_fixup): Update.
23133 * tree-ssa-tail-merge.c (replace_block_by): Update.
23134 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23135 (compute_path_counts): Update.
23136 (update_profile): Update.
23137 (recompute_probabilities): Update.
23138 (update_joiner_offpath_counts): Update.
23139 (estimated_freqs_path): Update.
23140 (freqs_to_counts_path): Update.
23141 (clear_counts_path): Update.
23142 (ssa_fix_duplicate_block_edges): Update.
23143 (duplicate_thread_path): Update.
23144 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23145 (struct switch_conv_info): Update.
23146 * tree-tailcall.c (decrease_profile): Update.
23147 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23148 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23149 * value-prof.c (check_counter): Update.
23150 (gimple_divmod_fixed_value): Update.
23151 (gimple_mod_pow2): Update.
23152 (gimple_mod_subtract): Update.
23153 (gimple_ic_transform): Update.
23154 (gimple_stringop_fixed_value): Update.
23155 * value-prof.h (gimple_ic): Update.
23157 2017-06-02 Carl Love <cel@us.ibm.com>
23159 * config/rs6000/rs6000-c: Add support for built-in functions
23160 vector double vec_doublee (vector signed int);
23161 vector double vec_doublee (vector unsigned int);
23162 vector double vec_doublee (vector float);
23163 vector double vec_doubleh (vector signed int);
23164 vector double vec_doubleh (vector unsigned int);
23165 vector double vec_doubleh (vector float);
23166 vector double vec_doublel (vector signed int);
23167 vector double vec_doublel (vector unsigned int);
23168 vector double vec_doublel (vector float);
23169 vector double vec_doubleo (vector signed int);
23170 vector double vec_doubleo (vector unsigned int);
23171 vector double vec_doubleo (vector float);.
23172 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23173 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23175 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23176 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23177 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23179 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23180 vec_doublel, vec_doubleh.
23181 * doc/extend.texi: Update the built-in documentation file for the
23182 new built-in functions.
23184 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23187 * ipa-inline-analysis.c (free_growth_caches): Set
23188 edge_removal_hook_holder to NULL after removing it.
23190 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23192 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23193 comparision with zero.
23195 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23196 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23197 for early expansion of vec_min and vec_max builtins.
23198 (builtin_function_type): Add min/max unsigned variants to those
23199 identified as having unsigned arguments.
23201 2017-06-02 Olivier Hainque <hainque@adacore.com>
23203 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23205 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23207 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23208 Use VALL_F16 iterator rather than VALL.
23210 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23212 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23213 Emit CBNZ inside loop when doing a strong exchange and comparing
23214 against zero. Generate the CC flags after the loop.
23216 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23218 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23219 (dl_section_ref): New.
23220 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23221 On AIX, append an expression to subtract the size of the
23222 section length to dl_section_ref.
23224 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23226 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23227 for early expansion of vector absolute builtins.
23229 2017-06-02 Richard Biener <rguenther@suse.de>
23231 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23232 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23234 2017-06-02 Richard Biener <rguenther@suse.de>
23236 PR tree-optimization/80948
23237 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23238 stmt order as well.
23240 2017-06-02 Richard Biener <rguenther@suse.de>
23242 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23244 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23245 for inductions as relevant.
23247 2017-06-02 Richard Biener <rguenther@suse.de>
23249 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23250 (vectorizable_induction): ... this. Remove dead code.
23252 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23254 * builtins. (expand_builtin_alloca): Remove second parameter and
23255 infer its value from the first parameter instead.
23256 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23258 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23260 PR rtl-optimization/80903
23261 * loop-doloop.c (add_test): Unshare sequence.
23263 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23265 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23267 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23269 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23271 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23272 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23273 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23274 (xlogue_layout::get_stub_rtx): Make static.
23275 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23276 (xlogue_layout::compute_stub_managed_regs): Rename to...
23277 (xlogue_layout::count_stub_managed_regs): ...this.
23278 (xlogue_layout::is_stub_managed_reg): New function.
23279 (xlogue_layout::m_stub_names): Rename to...
23280 (xlogue_layout::s_stub_names): ...this, make static.
23281 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23282 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23283 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23284 xlogue_layout::s_stub_names): Instantiate statics.
23285 (stub_managed_regs): Remove.
23286 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23287 (disable_call_ms2sysv_xlogues): Rename to...
23288 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23289 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23291 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23292 change after reload_completed.
23293 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23295 (ix86_expand_prologue): Likewise.
23296 (ix86_expand_epilogue): Likewise.
23297 (ix86_expand_split_stack_prologue): Likewise.
23298 (ix86_compute_frame_layout): Remove frame parameter ...
23299 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23300 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23302 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23303 (ix86_frame): Move from here ...
23304 * config/i386/i386.h (ix86_frame): ... to here.
23305 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23306 complete ix86_frame data structure instead. Remove some_ld_name.
23308 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23310 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23311 symbols that hold a DECL_VALUE_EXPR.
23313 2017-06-01 Martin Jambor <mjambor@suse.cz>
23315 PR tree-optimization/80898
23316 * tree-sra.c (process_subtree_disqualification): Removed.
23317 (disqualify_candidate): Do not acll
23318 process_subtree_disqualification.
23319 (subtree_mark_written_and_enqueue): New function.
23320 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23321 RHS has been disqualified and re-queue LHS if necessary. Apart
23322 from that, ignore disqualified RHS.
23324 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23326 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23327 address fetch for z10 or later.
23329 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23331 * config/arc/arc.md (tst_movb): Add guard when splitting.
23333 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23335 * config/arc/arc.c (arc_can_eliminate): Test against
23336 arc_frame_pointer_needed.
23338 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23340 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23341 to prevent store reordering.
23342 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23343 (type): Add block type.
23344 (stack_tie): Define special instruction to be used in
23347 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23349 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23350 constraint. It is not valid for the pattern.
23351 (noncommutative_binary_comparison): Likewise.
23353 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23355 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23358 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23360 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23361 be used by the reg-alloc.
23363 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23365 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23366 reg-alloc when having mul64 or mul32x16 instructions.
23367 (mulsidi3): Likewise.
23368 (umulsidi3): Likewise.
23369 (mulsi32x16): New pattern.
23370 (mulsi64): Likewise.
23371 (mulsidi64): Likewise.
23372 (umulsidi64): Likewise.
23373 (MUL32x16_REG): Define.
23374 (mul64_600): Use MUL32x16_REG.
23375 (mac64_600): Likewise.
23376 (umul64_600): Likewise.
23377 (umac64_600): Likewise.
23379 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23381 * config/arc/arc.md (mulsi3_700): Make it commutative.
23383 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23385 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23387 (*sign_extendsidi2_insn): Likewise for movstosw.
23389 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23391 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23392 the type of the input discriminant value. Convert the
23393 discriminant value of signedness vary.
23395 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23397 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23398 Add to -Wall section.
23400 2017-06-01 Richard Biener <rguenther@suse.de>
23402 PR middle-end/66313
23403 * fold-const.c (fold_plusminus_mult_expr): If the factored
23404 factor may be zero use a wrapping type for the inner operation.
23405 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23406 and handle moved defs.
23407 (process_assignment): Properly guard the unary op case. Return a
23408 tri-state indicating that moving the stmt before the call may allow
23409 to continue. Pass through to_move.
23410 (find_tail_calls): Handle moving unrelated defs before
23413 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23416 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23417 splitter result in the canonical way.
23419 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23421 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23422 also for 32bit target. Update insn attributes.
23423 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23425 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23427 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23428 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23429 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23430 (_mm_maskz_min_ss): New intrinsics.
23432 2017-05-31 Martin Liska <mliska@suse.cz>
23434 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23435 Change comment style to one we normally use.
23436 (vectorizable_reduction): Likewise.
23437 (vectorizable_induction): Likewise.
23438 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23439 (vectorizable_call): Likewise.
23440 (vectorizable_simd_clone_call): Likewise.
23441 (vectorizable_conversion): Likewise.
23442 (vectorizable_assignment): Likewise.
23443 (vectorizable_shift): Likewise.
23444 (vectorizable_operation): Likewise.
23445 (vectorizable_store): Likewise.
23446 (vectorizable_load): Likewise.
23447 * tree-vectorizer.h: Likewise.
23449 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23451 * passes.c (emergency_dump_function): New.
23452 * tree-pass.h (emergency_dump_function): Declare.
23453 * plugin.c (plugins_internal_error_function): Remove.
23454 * plugin.h (plugins_internal_error_function): Remove declaration.
23455 * toplev.c (internal_error_function): New static function. Use it...
23456 (general_init): ...here.
23458 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23460 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23461 (arc_rtx_costs): Add costs for new patterns.
23462 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23463 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23465 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23467 * tree-ssa-strlen.c (get_next_strinfo): New function.
23468 (get_stridx_plus_constant): Use it.
23469 (zero_length_string): Likewise.
23470 (adjust_related_strinfos): Likewise.
23471 (adjust_last_stmt): Likewise.
23473 2017-05-31 Richard Biener <rguenther@suse.de>
23476 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23477 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23479 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23481 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23482 loop_vinfo argument and use of dependence distance vectors.
23483 Check instead whether the two references differ only in their
23484 initial value and assume that they have the same alignment if the
23485 difference is a multiple of the vector alignment.
23486 (vect_analyze_data_refs_alignment): Update call accordingly.
23488 2017-05-31 Martin Liska <mliska@suse.cz>
23491 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23493 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23495 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23496 (create_intersect_range_checks): Move from ...
23497 * tree-data-ref.c (create_intersect_range_checks_index)
23498 (create_intersect_range_checks): ... to here.
23499 (create_runtime_alias_checks): New function factored from ...
23500 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23501 here. Call above function.
23502 * tree-data-ref.h (create_runtime_alias_checks): New function.
23504 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23506 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23507 segment length for dr_b and compute it in wide_int.
23509 2017-05-31 Richard Biener <rguenther@suse.de>
23511 PR tree-optimization/80906
23512 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23513 and pass through iv_map.
23514 (copy_bb_and_scalar_dependences): Adjust.
23515 (translate_pending_phi_nodes): Likewise.
23516 (copy_loop_close_phi_args): Handle code-generating IVs instead
23519 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23521 * diagnostic-color.c (color_dict): Add "type-diff".
23522 (parse_gcc_colors): Update comment.
23523 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23524 -fdiagnostics-show-template-tree and -fno-elide-type.
23525 (GCC_COLORS): Add type-diff to example.
23527 (-fdiagnostics-show-template-tree): New.
23528 (-fno-elide-type): New.
23529 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23530 the pp_format_decoder callback. Call any m_format_postprocessor's
23532 (pretty_printer::pretty_printer): Initialize
23533 m_format_postprocessor.
23534 (pretty_printer::~pretty_printer): Delete any
23535 m_format_postprocessor.
23536 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23537 (class format_postprocessor): New class.
23538 (struct pretty_printer::format_decoder): Document the new parameters.
23539 (struct pretty_printer::m_format_postprocessor): New field.
23540 * tree-diagnostic.c (default_tree_printer): Update for new
23541 bool and const char ** params.
23542 * tree-diagnostic.h (default_tree_printer): Likewise.
23544 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23546 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23547 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23548 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23549 rs6000_gen_cell_microcode code.
23550 (rs6000_final_prescan_insn): Delete.
23551 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23552 "warn-cell-microcode" entries.
23553 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23554 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23555 throughout. Change cc_reg_not_micro_cr0_operand to
23556 cc_reg_not_cr0_operand throughout.
23557 (*extendhi<mode>2_noload): Delete.
23558 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23559 (mwarn-cell-microcode): Delete.
23560 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23561 -mgen-cell-microcode and -mwarn-cell-microcode.
23563 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23566 * config/i386/constraints.md (Yd): New constraint.
23568 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23569 and (?Yd, r) alternatives. Update insn attributes.
23570 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23571 and (?*Yd, r) alternatives. Update insn attributes.
23572 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23574 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23576 * gimplify.c (gimplify_modify_expr): Don't create a
23577 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23580 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23582 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23584 2017-05-30 Richard Biener <rguenther@suse.de>
23586 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23587 and reduc_def fields.
23588 (STMT_VINFO_REDUC_TYPE): New define.
23589 (STMT_VINFO_REDUC_DEF): Likewise.
23590 (vect_force_simple_reduction): Adjust prototype.
23591 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23592 (vect_is_simple_reduction): Remove check_reduction argument.
23593 (vect_force_simple_reduction): Adjust and set
23594 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23595 (vectorizable_reduction): Do not re-do reduction analysis
23596 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23597 * tree-parloops.c (gather_scalar_reductions): Adjust.
23599 2017-05-30 Richard Biener <rguenther@suse.de>
23601 PR middle-end/80901
23602 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23605 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23607 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23608 Introduce unknown_misalignment parameter and remove vf.
23609 (vect_peeling_hash_get_lowest_cost):
23610 Pass unknown_misalignment parameter.
23611 (vect_enhance_data_refs_alignment):
23612 Fix unsupportable data ref treatment.
23614 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23616 * tree-vect-data-refs.c (vect_get_data_access_cost):
23617 Workaround for SLP handling.
23618 (vect_enhance_data_refs_alignment):
23619 Compute costs for doing no peeling at all, compare to the best
23620 peeling costs so far and avoid peeling if cheaper.
23622 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23624 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23625 Return peeling info and set costs to zero for unlimited cost
23627 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23628 with unknown misalignment. Compute and costs for unknown
23629 misalignment, compare them to the costs for known misalignment
23630 and choose the cheapest for peeling.
23632 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23634 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23635 (vect_get_peeling_costs_all_drs): Create function.
23636 (vect_peeling_hash_get_lowest_cost):
23637 Use vect_get_peeling_costs_all_drs.
23638 (vect_peeling_supportable): Create function.
23639 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23641 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23643 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23644 DR_HAS_NEGATIVE_STEP.
23645 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23646 (vect_enhance_data_refs_alignment): Use.
23647 (vect_duplicate_ssa_name_ptr_info): Use.
23648 * tree-vectorizer.h (dr_misalignment): Use.
23649 (known_alignment_for_access_p): Use.
23651 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23654 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23656 (has_section_name): New function.
23658 2017-05-30 Martin Liska <mliska@suse.cz>
23661 * auto-profile.c (get_function_decl_from_block): Fix
23664 2017-05-30 Richard Biener <rguenther@suse.de>
23666 PR middle-end/80876
23667 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23669 2017-05-30 Martin Liska <mliska@suse.cz>
23671 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23672 * dumpfile.h (struct dump_file_info): Remove ctors.
23674 2017-05-30 Martin Liska <mliska@suse.cz>
23676 * predict.def: Fix GNU coding style.
23678 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23680 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23681 Mark 'to' argument with ATTRIBUTE_UNUSED.
23683 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23685 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23686 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23687 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23690 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23692 * doc/install.texi (Options specification): Restore entry of
23693 --enable-sjlj-exceptions.
23695 2017-05-27 Michael Eager <eager@eagercon.com>
23698 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23700 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23702 * config/microblaze/microblaze.h
23703 (FIXED_REGISTERS): Update in macro.
23704 (CALL_USED_REGISTERS): Update in macro.
23706 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23708 * doc/install.texi: Add links to macOS binary distributions.
23710 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23714 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23716 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23718 2017-05-26 Martin Liska <mliska@suse.cz>
23720 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23722 2017-05-26 Martin Liska <mliska@suse.cz>
23724 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23725 always leading ';; '.
23726 (dump_bb_info): Likewise.
23727 (brief_dump_cfg): Likewise.
23728 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23729 * dumpfile.c: Remove usage of TDF_VERBOSE.
23730 * dumpfile.h (enum dump_kind): Likewise.
23731 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23732 * print-tree.c (debug_verbose): Remove.
23733 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23734 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23735 * tree-diagnostic.c (default_tree_printer): Replace
23736 TDF_DIAGNOSTIC with TDF_SLIM.
23738 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23740 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23741 in parameter loop, rather than loop_vinfo.
23742 (create_intersect_range_checks): Ditto.
23743 (vect_create_cond_for_alias_checks): Update call to above functions.
23745 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23747 PR tree-optimization/80815
23748 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23749 for merging runtime alias checks. Handle negative DR_STEPs.
23751 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23753 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23755 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23756 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23757 out code pruning runtime alias checks.
23758 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23759 factored out from above.
23760 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23762 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23764 (prune_runtime_alias_test_list): New decalaration.
23766 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23768 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23769 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23770 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23771 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23772 (operator==, comp_dr_with_seg_len_pair): Ditto.
23773 (vect_prune_runtime_alias_test_list): Ditto.
23775 2017-05-26 Martin Liska <mliska@suse.cz>
23778 * params.def: Bound partial-inlining-entry-probability param.
23780 2017-05-26 Marek Polacek <polacek@redhat.com>
23783 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23786 2017-05-26 Richard Biener <rguenther@suse.de>
23788 PR tree-optimization/80842
23789 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23792 2017-05-26 Richard Biener <rguenther@suse.de>
23794 PR tree-optimization/80844
23795 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23797 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23799 * doc/md.texi (Machine Constraints): Update x86 family
23800 machine constraints section to match 'config/i386/constraints.md'.
23802 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23804 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23806 2017-05-25 Nathan Sidwell <nathan@acm.org>
23808 * doc/invoke.texi (--enable-languages): Update documentation.
23810 2017-05-25 Martin Liska <mliska@suse.cz>
23812 * dumpfile.c: Add TDF_FOLDING.
23813 * dumpfile.h (enum dump_kind): Likewise.
23814 * genmatch.c (dt_simplify::gen_1): Use it.
23816 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23818 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23820 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23822 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23823 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23825 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23827 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23828 * match.pd (X == C): Rewrite it here.
23829 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23830 with_certain_nonzero_bits2): New predicates.
23831 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23833 2017-05-24 Nathan Sidwell <nathan@acm.org>
23835 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23838 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23841 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23843 * config/powerpcspe: New port. Files are copied from the rs6000
23844 port, with "rs6000" in filenames replaced by "powerpcspe".
23846 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23848 PR rtl-optimization/80754
23849 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23851 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23853 * config/sparc/sparc.md (length): Return the correct value for -mflat
23854 sibcalls to match output_sibcall.
23856 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23860 * config/rs6000/rs6000.c (struct machine_function): Add new field
23862 (rs6000_get_separate_components): Init that field, use it.
23863 (rs6000_components_for_bb): Use the field.
23865 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23867 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23869 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23871 PR middle-end/80823
23872 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23874 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23877 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23878 address against address_operand predicate.
23879 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23881 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23883 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23884 parameters passed indirectly.
23886 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23888 * config/i386/i386.md (*movdi_internal): Remove SSE4
23889 alternative 18 (?r, *v). Update insn attributes.
23890 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23891 Update insn attributes.
23892 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23893 Update insn attributes.
23894 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23895 alternative 1 (r, v). Remove isa attribute.
23896 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23897 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23898 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23900 2017-05-23 Tom de Vries <tom@codesourcery.com>
23902 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23905 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23907 * cgraphunit.c (symbol_table::process_new_functions): Update.
23908 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23909 (inline_generate_summary): Rename to ...
23910 (ipa_fn_summary_generate): ... this one.
23911 (inline_read_summary): Rename to ...
23912 (ipa_fn_summary_read): ... this one.
23913 (inline_write_summary): Rename to ...
23914 (ipa_fn_summary_write): ... this one.
23915 (inline_free_summary): Rename to ...
23916 (ipa_free_fn_summary): ... this one.
23917 (pass_data_local_fn_summary, pass_local_fn_summary,
23918 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23919 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23920 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23921 make_pass_ipa_fn_summary): New.
23922 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23923 inline_write_summary, inline_free_summary): Remove.
23924 (ipa_free_fn_summary) : New.
23925 * ipa-inline.c (ipa_inline): Update.
23926 (pass_ipa_inline): Do not generate summaries.
23927 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23929 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23930 and add pass_ipa_fn_summary.
23931 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23933 (make_pass_inline_parameters): Remove.
23935 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23937 * omp-low.c (struct omp_context): Remove "default_kind" member.
23940 * omp-offload.c (execute_oacc_device_lower): Remove the
23941 parallelism dimensions function attributes for unparallelized
23942 OpenACC kernels constructs.
23944 2017-05-23 Martin Liska <mliska@suse.cz>
23946 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23948 (cgraph_edge::make_speculative): Likewise.
23949 (cgraph_edge::resolve_speculation): Likewise.
23950 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23951 (cgraph_node::dump): Likewise.
23952 * cgraph.h: Likewise.
23953 * cgraphunit.c (analyze_functions): Likewise.
23954 (symbol_table::compile): Likewise.
23955 * ipa-cp.c (print_all_lattices): Likewise.
23956 (determine_versionability): Likewise.
23957 (initialize_node_lattices): Likewise.
23958 (ipcp_verify_propagated_values): Likewise.
23959 (estimate_local_effects): Likewise.
23960 (update_profiling_info): Likewise.
23961 (create_specialized_node): Likewise.
23962 (perhaps_add_new_callers): Likewise.
23963 (decide_about_value): Likewise.
23964 (decide_whether_version_node): Likewise.
23965 (identify_dead_nodes): Likewise.
23966 (ipcp_store_bits_results): Likewise.
23967 * ipa-devirt.c (dump_targets): Likewise.
23968 (ipa_devirt): Likewise.
23969 * ipa-icf.c (sem_item::dump): Likewise.
23970 (sem_function::equals): Likewise.
23971 (sem_variable::equals): Likewise.
23972 (sem_item_optimizer::read_section): Likewise.
23973 (sem_item_optimizer::execute): Likewise.
23974 (congruence_class::dump): Likewise.
23975 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23976 (dump_inline_summary): Likewise.
23977 (estimate_node_size_and_time): Likewise.
23978 (inline_analyze_function): Likewise.
23979 * ipa-inline-transform.c (inline_call): Likewise.
23980 * ipa-inline.c (report_inline_failed_reason): Likewise.
23981 (want_early_inline_function_p): Likewise.
23982 (edge_badness): Likewise.
23983 (update_edge_key): Likewise.
23984 (inline_small_functions): Likewise.
23985 * ipa-profile.c (ipa_profile): Likewise.
23986 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23987 (ipa_make_edge_direct_to_target): Likewise.
23988 (remove_described_reference): Likewise.
23989 (ipa_impossible_devirt_target): Likewise.
23990 (propagate_controlled_uses): Likewise.
23991 (ipa_print_node_params): Likewise.
23992 (ipcp_transform_function): Likewise.
23993 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23994 (propagate_pure_const): Likewise.
23995 * ipa-reference.c (generate_summary): Likewise.
23996 (read_write_all_from_decl): Likewise.
23997 (propagate): Likewise.
23998 (ipa_reference_read_optimization_summary): Likewise.
23999 * ipa-utils.c (ipa_merge_profiles): Likewise.
24000 * ipa.c (walk_polymorphic_call_targets): Likewise.
24001 (symbol_table::remove_unreachable_nodes): Likewise.
24002 (ipa_single_use): Likewise.
24003 * passes.c (execute_todo): Likewise.
24004 * predict.c (drop_profile): Likewise.
24005 * symtab.c (symtab_node::get_dump_name): New function.
24006 (symtab_node::dump_name): Likewise.
24007 (symtab_node::dump_asm_name): Likewise.
24008 (symtab_node::dump_references): Likewise.
24009 (symtab_node::dump_referring): Likewise.
24010 (symtab_node::dump_base): Likewise.
24011 (symtab_node::debug_symtab): Likewise.
24012 * tree-sra.c (convert_callers_for_node): Likewise.
24013 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
24014 * value-prof.c (init_node_map): Likewise.
24016 2017-05-23 Martin Liska <mliska@suse.cz>
24018 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
24019 and symtab_node::debug_symtab to symbol_table::debug.
24020 * cgraphunit.c (analyze_functions): Use the renamed function.
24021 (symbol_table::compile): Likewise.
24022 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
24023 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
24024 * passes.c (execute_todo): Likewise.
24025 * symtab.c (symbol_table::dump): New function.
24026 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
24028 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24030 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
24031 that nonconst implies exec.
24033 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
24035 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24036 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24037 (inline_edge_summary_vec): Turn into ...
24038 (ipa_call_summaries): ... this one.
24039 (redirect_to_unreachable, edge_set_predicate,
24040 evaluate_properties_for_edge, inline_summary_alloc,
24041 reset_ipa_call_summary, reset_inline_summary,
24042 inline_summary_t::duplicate): Update.
24043 (inline_edge_duplication_hook): Turn to ...
24044 (ipa_call_summary_t::duplicate): ... this one.
24045 (inline_edge_removal_hook): Turn to ...
24046 (ipa_call_summary_t::remove): ... this one.
24047 (dump_inline_edge_summary): Turn to ...
24048 (dump_ipa_call_summary): ... this one.
24049 (estimate_function_body_sizes): Update.
24050 (inline_update_callee_summaries): Update.
24051 (remap_edge_change_prob): Update.
24052 (remap_edge_summaries): Update.
24053 (inline_merge_summary): Update.
24054 (do_estimate_edge_time): Update.
24055 (inline_generate_summary): Update.
24056 (inline_read_section): Update.
24057 (inline_read_summary): Update.
24058 (inline_free_summary): Update.
24059 * ipa-inline.c (can_inline_edge_p): Update.
24060 (compute_inlined_call_time): Update.
24061 (want_inline_small_function_p): Update.
24062 (edge_badness): Update.
24063 (early_inliner): Update.
24064 * ipa-inline.h (inline_edge_summary): Turn to ...
24065 (ipa_call_summary): ... this one.
24066 (ipa_call_summary_t): New class.
24067 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24068 (ipa_call_summaries): New.
24069 (inline_edge_summary): Remove.
24070 (estimate_edge_growth): Update.
24071 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24072 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24073 * ipa-split.c (execute_split_functions): Update.
24074 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24076 2017-05-23 Tom de Vries <tom@codesourcery.com>
24078 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24079 attributes): Document rdrand effective target.
24081 2017-05-23 Tom de Vries <tom@codesourcery.com>
24083 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
24084 attributes): Sort alphabetically.
24086 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
24088 * config/avr/genmultilib.awk: Use gsub instead of gensub.
24090 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24093 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
24094 V2DF/V2DI splat into two separate patterns, one that handles
24095 registers, and the other that only handles memory. Drop support
24096 for splatting from a GPR on ISA 2.07 and then splitting the
24097 splat into direct move and splat.
24098 (vsx_splat_<mode>_reg): Likewise.
24099 (vsx_splat_<mode>_mem): Likewise.
24101 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24103 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24105 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24107 PR middle-end/80809
24108 * omp-low.c (finish_taskreg_remap): New function.
24109 (finish_taskreg_scan): If unit size of ctx->record_type
24110 is non-constant, unshare the size expression and replace
24111 decls in it with possible outer var refs.
24113 PR middle-end/80809
24114 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24115 GOVD_SHARED rather than GOVD_PRIVATE with it.
24116 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24117 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24119 PR middle-end/80853
24120 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24121 as last argument to build_outer_var_ref for pointer bases of array
24122 section reductions.
24124 2017-05-19 Martin Sebor <msebor@redhat.com>
24126 * print-tree.c (print_node): Print DECL_READ_P flag.
24128 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24130 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24131 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24132 * cgraph.c: Likewise.
24133 * cgraphunit.c: Likewise.
24134 * gengtype.c: Likewise.
24135 * ipa-cp.c: Likewise.
24136 * ipa-devirt.c: Likewise.
24137 * ipa-icf.c: Likewise.
24138 * ipa-predicate.c: Likewise.
24139 * ipa-profile.c: Likewise.
24140 * ipa-prop.c: Likewise.
24141 * ipa-split.c: Likewise.
24143 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24144 edge_predicate_pool, dump_inline_hints,
24145 inline_summary::account_size_time, redirect_to_unreachable,
24146 edge_set_predicate, set_hint_predicate,
24147 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24148 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24149 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24150 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24151 ipa_call_summary_t::remove, initialize_growth_caches,
24152 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24153 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24154 mark_modified, unmodified_parm_1, unmodified_parm,
24155 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24156 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24157 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24158 will_be_nonconstant_predicate, record_modified_bb_info,
24159 get_minimal_bb, record_modified, param_change_prob,
24160 phi_result_unknown_predicate, predicate_for_phi_result,
24161 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24162 estimate_function_body_sizes, compute_inline_parameters,
24163 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24164 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24165 inline_update_callee_summaries, remap_edge_change_prob,
24166 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24167 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24168 inline_analyze_function, inline_summary_t::insert,
24169 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24170 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24171 inline_free_summary): Move to ipa-fnsummary.h
24172 (predicate_t): Remove.
24173 * ipa-fnsummary.c: New file.
24174 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24175 (enum inline_hints_vals, inline_hints, agg_position_info,
24176 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24177 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24178 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24179 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24180 inline_read_summary, inline_write_summary, inline_free_summary,
24181 inline_analyze_function, initialize_inline_failed,
24182 inline_merge_summary, inline_update_overall_summary,
24183 compute_inline_parameters): Move to ipa-fnsummary.h
24184 * ipa-fnsummary.h: New file.
24185 * ipa-inline-transform.h: Include ipa-inline.h.
24186 * ipa-inline.c: LIkewise.
24188 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24190 * ipa-inline.c (edge_badness): Use inlined_time instead of
24191 inline_summaries->get.
24193 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24195 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24197 2017-05-22 Nathan Sidwell <nathan@acm.org>
24199 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24200 (fdump-lang): Document 'raw' option.
24201 * dumpfile.h (TDI_tu): Delete.
24202 * dumpfile.c (dump_files): Remove translation-unit.
24203 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24205 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24207 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24208 command option from $(AWK) call.
24209 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24210 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24211 [FORMAT]: Remove handling of variable.
24212 * config/avr/t-multilib: Regenerate.
24214 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24216 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24218 (dump_inline_summary): Do not print self_time.
24219 (estimate_function_body_sizes): Do not set self_time.
24220 (compute_inline_parameters): Likewise.
24221 (inline_read_section, inline_write_summary): Do not stream self_time.
24222 * ipa-inline.h (inline_summary): Drop self_time.
24224 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24226 * ipa-inline-analysis.c (account_size_time): Rename to ...
24227 (inline_summary::account_size_time): ... this one.
24228 (reset_ipa_call_summary): Turn to ...
24229 (ipa_call_summary::reset): ... this one.
24230 (reset_inline_summary): Turn to ...
24231 (inline_summary::reset): ... this one.
24232 (inline_summary_t::remove): Update.
24233 (inline_summary_t::duplicate): Update.
24234 (ipa_call_summary_t::remove): Update.
24235 (dump_inline_summary): Update.
24236 (estimate_function_body_sizes): Update.
24237 (compute_inline_parameters): Update.
24238 (estimate_node_size_and_time): Update.
24239 (inline_merge_summary): Update.
24240 (inline_update_overall_summary): Update.
24241 (inline_read_section): Update.
24242 (inline_write_summary): Update.
24243 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24244 add account_size_time and reset member functions.
24245 (ipa_call_summary): Add reset function.
24246 * ipa-predicate.h (predicate::operator &): Constify.
24248 2017-05-22 Richard Biener <rguenther@suse.de>
24250 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24252 2017-05-19 Jason Merrill <jason@redhat.com>
24254 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24256 2017-05-19 Marek Polacek <polacek@redhat.com>
24259 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24260 TYPE_OVERFLOW_WRAPS checks.
24262 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24264 * tree-core.h (enum omp_clause_default_kind): Add
24265 "OMP_CLAUSE_DEFAULT_PRESENT".
24266 * tree-pretty-print.c (dump_omp_clause): Handle it.
24267 * gimplify.c (enum gimplify_omp_var_data): Add
24268 "GOVD_MAP_FORCE_PRESENT".
24269 (gimplify_adjust_omp_clauses_1): Map it to
24270 "GOMP_MAP_FORCE_PRESENT".
24271 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24273 * gimplify.c (oacc_default_clause): Clarify.
24275 2017-05-19 Nathan Sidwell <nathan@acm.org>
24277 LANG_HOOK_REGISTER_DUMPS
24278 * toplev.c (general_init): Call register dump lang hook.
24279 * doc/invoke.texi: Document -fdump-lang option family.
24280 * dumpfile.c (dump_files): Remove class dump here.
24281 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24282 * dumpfile.h (tree_dump_index): Remove TDI_class.
24283 * langhooks-def.h (lhd_register_dumps): Declare.
24284 (LANG_HOOKS_REGISTER_DUMPS): Define.
24285 (LANG_HOOKS_INITIALIZER): Add it.
24286 * langhooks.c (lhd_register_dumps): Define.
24287 * langhooks.h (struct lang_hooks): Add register_dumps.
24289 2017-05-19 Nathan Sidwell <nathan@acm.org>
24291 * context.h (context::set_passes): New.
24292 * context.c (context::context): Do not create pass manager.
24293 * toplev.c (general_init): Create pass manager here.
24295 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24297 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24298 use this splitter if two add or or instructions would also work for
24299 the constant we want to generate.
24301 2017-05-19 Richard Biener <rguenther@suse.de>
24304 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24305 predicate evaluation.
24307 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24309 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24311 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24314 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24316 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24317 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24318 (inline_edge_summary_vec): Turn into ...
24319 (ipa_call_summaries): ... this one.
24320 (redirect_to_unreachable, edge_set_predicate,
24321 evaluate_properties_for_edge, inline_summary_alloc,
24322 reset_ipa_call_summary, reset_inline_summary,
24323 inline_summary_t::duplicate): Update.
24324 (inline_edge_duplication_hook): Turn to ...
24325 (ipa_call_summary_t::duplicate): ... this one.
24326 (inline_edge_removal_hook): Turn to ...
24327 (ipa_call_summary_t::remove): ... this one.
24328 (dump_inline_edge_summary): Turn to ...
24329 (dump_ipa_call_summary): ... this one.
24330 (estimate_function_body_sizes): Update.
24331 (inline_update_callee_summaries): Update.
24332 (remap_edge_change_prob): Update.
24333 (remap_edge_summaries): Update.
24334 (inline_merge_summary): Update.
24335 (do_estimate_edge_time): Update.
24336 (inline_generate_summary): Update.
24337 (inline_read_section): Update.
24338 (inline_read_summary): Update.
24339 (inline_free_summary): Update.
24340 * ipa-inline.c (can_inline_edge_p): Update.
24341 (compute_inlined_call_time): Update.
24342 (want_inline_small_function_p): Update.
24343 (edge_badness): Update.
24344 (early_inliner): Update.
24345 * ipa-inline.h (inline_edge_summary): Turn to ...
24346 (ipa_call_summary): ... this one.
24347 (ipa_call_summary_t): New class.
24348 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24349 (ipa_call_summaries): New.
24350 (inline_edge_summary): Remove.
24351 (estimate_edge_growth): Update.
24352 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24353 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24354 * ipa-split.c (execute_split_functions): Update.
24355 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24357 2017-05-19 Richard Biener <rguenther@suse.de>
24359 PR middle-end/80764
24360 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24362 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24364 * config/rs6000/rs6000.c (struct machine_function): Add field
24365 fpr_is_wrapped_separately.
24366 (rs6000_get_separate_components): Use 64 components. Handle the
24367 new FPR components.
24368 (rs6000_components_for_bb): Handle the FPR components.
24369 (rs6000_emit_prologue_components): Handle the FPR components.
24370 (rs6000_emit_epilogue_components): Handle the FPR components.
24371 (rs6000_set_handled_components): Handle the FPR components.
24372 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24373 that are already separately shrink-wrapped.
24374 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24375 that are already separately shrink-wrapped.
24377 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24380 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24383 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24384 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24385 cases where the register allocator uses a move and an offsettable
24386 memory operation to/from a FPR register on ISA 2.06/2.07.
24387 (define_peephole2 for Altivec d-form store): Likewise.
24389 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24392 * config/i386/mmx.md (*mov<mode>_internal): Enable
24393 alternatives 11, 12, 13 and 14 also for 32bit targets.
24394 Remove alternatives 15, 16, 17 and 18.
24395 * config/i386/sse.md (vec_concatv2di): Change
24396 alternative (!x, *y) to (x, ?!*Yn).
24398 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24400 * dumpfile.h (enum dump_kind): Remove stray comma.
24402 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24404 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24405 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24406 predicate::num_conditions
24407 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24408 (CHANGED): turn into predicate::changed.
24409 (agg_position_info): Move to ipa-predicate.h
24410 (add_condition, predicate::add_clause, predicate::operator &=,
24411 predicate::or_with, predicate::evaluate, predicate::probability,
24412 dump_condition, dump_clause, predicate::dump,
24413 predicate::remap_after_duplication, predicate::remap_after_inlining,
24414 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24415 (evaluate_conditions_for_known_args): Update.
24416 (set_cond_stmt_execution_predicate): Update.
24417 * ipa-inline.h: Include ipa-predicate.h
24418 (condition, inline_param_summary, conditions, agg_position_info,
24419 predicate): Move to ipa-predicate.h
24420 * ipa-predicate.c: New file.
24421 * ipa-predicate.h: New file.
24423 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24425 * final.c (leaf_function_p): Check we are not in a sequence.
24427 2017-05-18 Martin Liska <mliska@suse.cz>
24429 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24430 * dumpfile.c (dump_register): Use new enum dump_kind.
24431 (get_dump_file_name): Likewise.
24432 (dump_enable_all): Likewise.
24433 (dump_switch_p_1): Likewise.
24434 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24435 * dumpfile.h (enum dump_kind): New enum type.
24436 (struct dump_file_info): Create constructor and
24437 format fields and comments.
24438 * passes.c (pass_manager::register_one_dump_file):
24440 * statistics.c (statistics_early_init): Likewise.
24441 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24442 TDF_TREE with TDF_SLIM.
24443 (gather_memory_references_ref): Likewise.
24445 2017-05-18 Martin Liska <mliska@suse.cz>
24447 * vec.h (struct vnull): Use it.
24449 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24451 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24452 (true_predicate, false_predicate, true_predicate_p,
24453 false_predicate_p): Remove.
24454 (single_cond_predicate, not_inlined_predicate): Turn to member function
24456 (add_condition): Update.
24457 (add_clause): Turn to...
24458 (predicate::add_clause): ... this one; update; allow passing NULL
24460 (and_predicates): Turn to ...
24461 (predicate::operator &=): ... this one.
24462 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24463 (or_predicates): Turn to ...
24464 (predicate::or_with): ... this one.
24465 (evaluate_predicate): Turn to ...
24466 (predicate::evaluate): ... this one.
24467 (predicate_probability): Turn to ...
24468 (predicate::probability): ... this one.
24469 (dump_condition): Update.
24470 (dump_predicate): Turn to ...
24471 (predicate::dump): ... this one.
24472 (account_size_time): Update.
24473 (edge_set_predicate): Update.
24474 (set_hint_predicate): UPdate.
24475 (evaluate_conditions_for_known_args): Update.
24476 (evaluate_properties_for_edge): Update.
24477 (remap_predicate_after_duplication): Turn to...
24478 (predicate::remap_after_duplication): ... this one.
24479 (remap_hint_predicate_after_duplication): Update.
24480 (inline_summary_t::duplicate): UPdate.
24481 (dump_inline_edge_summary): Update.
24482 (dump_inline_summary): Update.
24483 (set_cond_stmt_execution_predicate): Update.
24484 (set_switch_stmt_execution_predicate): Update.
24485 (compute_bb_predicates): Update.
24486 (will_be_nonconstant_expr_predicate): Update.
24487 (will_be_nonconstant_predicate): Update.
24488 (phi_result_unknown_predicate): Update.
24489 (predicate_for_phi_result): Update.
24490 (array_index_predicate): Update.
24491 (estimate_function_body_sizes): Update.
24492 (estimate_node_size_and_time): Update.
24493 (estimate_ipcp_clone_size_and_time): Update.
24494 (remap_predicate): Rename to ...
24495 (predicate::remap_after_inlining): ... this one.
24496 (remap_hint_predicate): Update.
24497 (inline_merge_summary): Update.
24498 (inline_update_overall_summary): Update.
24499 (estimate_size_after_inlining): Update.
24500 (read_predicate): Rename to ...
24501 (predicate::stream_in): ... this one.
24502 (read_inline_edge_summary): Update.
24503 (write_predicate): Rename to ...
24504 (predicate::stream_out): ... this one.
24505 (write_inline_edge_summary): Update.
24506 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24507 (clause_t): Turn to uint32_t
24508 (predicate): Turn to class; implement constructor and operators
24510 (size_time_entry): Update.
24511 (inline_summary): Update.
24512 (inline_edge_summary): Update.
24514 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24516 * fold-const.c (fold_binary_loc): Move transformation...
24517 * match.pd (C - X CMP X): ... here.
24519 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24521 * config/sparc/sparc.c (sparc_option_override): Set function
24522 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24523 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24525 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24527 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24529 2017-05-18 Marek Polacek <polacek@redhat.com>
24532 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24533 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24534 gimple_assign_load_p.
24536 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24538 PR middle-end/80692
24539 * real.c (do_compare): Give decimal_do_compare preference over
24540 comparing just the signs.
24542 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24544 * doc/md.texi (Canonicalization of Instructions): Describe the
24545 canonical form of instructions that inherently set a condition
24548 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24550 PR middle-end/80775
24551 * tree-cfg.c: Move deletion of unreachable case statements to after
24552 the merging of consecutive case labels.
24554 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24556 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24557 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24558 restoring of callee-saved registers.
24560 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24562 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24563 * config/visium/visium.c (single_set_and_flags): Likewise.
24564 * config/visium/visium.md (Substitutions): Likewise.
24566 2017-05-17 Martin Liska <mliska@suse.cz>
24568 * cfg.c: Introduce dump_flags_t type and
24569 use it instead of int type.
24571 * cfghooks.c: Likewise.
24572 * cfghooks.h (struct cfg_hooks): Likewise.
24573 * cfgrtl.c: Likewise.
24574 * cfgrtl.h: Likewise.
24575 * cgraph.c (cgraph_node::get_body): Likewise.
24576 * coretypes.h: Likewise.
24577 * domwalk.c: Likewise.
24578 * domwalk.h: Likewise.
24579 * dumpfile.c (struct dump_option_value_info): Likewise.
24580 (dump_enable_all): Likewise.
24581 (dump_switch_p_1): Likewise.
24582 (opt_info_switch_p): Likewise.
24583 * dumpfile.h (enum tree_dump_index): Likewise.
24584 (struct dump_file_info): Likewise.
24585 * genemit.c: Likewise.
24586 * generic-match-head.c: Likewise.
24587 * gengtype.c (open_base_files): Likewise.
24588 * gimple-pretty-print.c: Likewise.
24589 * gimple-pretty-print.h: Likewise.
24590 * graph.c (print_graph_cfg): Likewise.
24591 * graphite-scop-detection.c (dot_all_sese): Likewise.
24592 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24593 * loop-unroll.c (report_unroll): Likewise.
24594 * passes.c (pass_manager::register_one_dump_file): Likewise.
24595 * print-tree.c: Likewise.
24596 * statistics.c: Likewise.
24597 * tree-cfg.c: Likewise.
24598 * tree-cfg.h: Likewise.
24599 * tree-dfa.c: Likewise.
24600 * tree-dfa.h: Likewise.
24601 * tree-dump.c (dump_function): Likewise.
24602 * tree-dump.h (struct dump_info): Likewise.
24603 * tree-pretty-print.c: Likewise.
24604 * tree-pretty-print.h: Likewise.
24605 * tree-ssa-live.c: Likewise.
24606 * tree-ssa-live.h: Likewise.
24607 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24608 * tree-vect-loop.c: Likewise.
24609 * tree-vect-slp.c: Likewise.
24611 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24612 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24614 PR tree-optimization/80457
24615 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24616 of all arguments to a statement as scalar_to_vec operations.
24617 (vectorizable_call): Adjust call to vect_model_simple_cost for
24619 (vectorizable_conversion): Likewise.
24620 (vectorizable_assignment): Likewise.
24621 (vectorizable_shift): Likewise.
24622 (vectorizable_operation): Likewise.
24623 (vectorizable_comparison): Likewise.
24624 (vect_is_simple_cond): Record the def types for operands.
24625 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24626 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24627 for statement argument count.
24629 2017-05-16 Carl Love <cel@us.ibm.com>
24631 * config/rs6000/rs6000-c: Add support for built-in functions
24632 vector unsigned long long vec_bperm (vector unsigned long long,
24633 vector unsigned char)
24634 vector signed long long vec_mule (vector signed int,
24636 vector unsigned long long vec_mule (vector unsigned int,
24637 vector unsigned int)
24638 vector signed long long vec_mulo (vector signed int,
24640 vector unsigned long long vec_mulo (vector unsigned int,
24641 vector unsigned int)
24642 vector signed char vec_sldw (vector signed char,
24643 vector signed char,
24645 vector unsigned char vec_sldw (vector unsigned char,
24646 vector unsigned char,
24648 vector signed short vec_sldw (vector signed short,
24649 vector signed short,
24651 vector unsigned short vec_sldw (vector unsigned short,
24652 vector unsigned short,
24654 vector signed int vec_sldw (vector signed int,
24657 vector unsigned int vec_sldw (vector unsigned int,
24658 vector unsigned int,
24660 vector signed long long vec_sldw (vector signed long long,
24661 vector signed long long,
24663 vector unsigned long long vec_sldw (vector unsigned long long,
24664 vector unsigned long long,
24666 * config/rs6000/rs6000-c: Add support for built-in functions
24667 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24668 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24669 * doc/extend.texi: Update the built-in documentation for the
24670 new built-in functions.
24672 2017-05-16 Marek Polacek <polacek@redhat.com>
24676 * tree.c (save_expr): Don't fold the expression.
24678 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24680 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24681 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24682 and (?*y,m). Update insn attributes.
24684 2017-05-16 Martin Liska <mliska@suse.cz>
24686 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24687 flags argument of print_gimple_stmt, print_gimple_expr,
24688 print_generic_stmt and print_generic_expr.
24689 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24690 * coretypes.h: Likewise.
24691 * except.c (dump_eh_tree): Likewise.
24692 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24693 * gimple-pretty-print.h: Likewise.
24694 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24695 (backprop::push_to_worklist): Likewise.
24696 (backprop::pop_from_worklist): Likewise.
24697 (backprop::process_use): Likewise.
24698 (backprop::intersect_uses): Likewise.
24699 (note_replacement): Likewise.
24700 * gimple-ssa-store-merging.c
24701 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24702 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24703 (pass_store_merging::execute): Likewise.
24704 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24705 (ssa_base_cand_dump_callback): Likewise.
24706 (dump_incr_vec): Likewise.
24707 (replace_refs): Likewise.
24708 (replace_mult_candidate): Likewise.
24709 (create_add_on_incoming_edge): Likewise.
24710 (create_phi_basis): Likewise.
24711 (insert_initializers): Likewise.
24712 (all_phi_incrs_profitable): Likewise.
24713 (introduce_cast_before_cand): Likewise.
24714 (replace_one_candidate): Likewise.
24715 * gimplify.c (gimplify_expr): Likewise.
24716 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24717 (set_rename): Likewise.
24718 (rename_uses): Likewise.
24719 (copy_loop_phi_nodes): Likewise.
24720 (add_close_phis_to_merge_points): Likewise.
24721 (copy_loop_close_phi_args): Likewise.
24722 (copy_cond_phi_args): Likewise.
24723 (graphite_copy_stmts_from_block): Likewise.
24724 (translate_pending_phi_nodes): Likewise.
24725 * graphite-poly.c (print_pdr): Likewise.
24726 (dump_gbb_cases): Likewise.
24727 (dump_gbb_conditions): Likewise.
24728 (print_scop_params): Likewise.
24729 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24730 (build_cross_bb_scalars_use): Likewise.
24731 (gather_bbs::before_dom_children): Likewise.
24732 * hsa-dump.c (dump_hsa_immed): Likewise.
24733 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24734 (get_replacement_map): Likewise.
24735 * ipa-inline-analysis.c (dump_condition): Likewise.
24736 (estimate_function_body_sizes): Likewise.
24737 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24738 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24739 * ipa-prop.c (ipa_dump_param): Likewise.
24740 (ipa_print_node_jump_functions_for_edge): Likewise.
24741 (ipa_modify_call_arguments): Likewise.
24742 (ipa_modify_expr): Likewise.
24743 (ipa_dump_param_adjustments): Likewise.
24744 (ipa_dump_agg_replacement_values): Likewise.
24745 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24746 * ipa-pure-const.c (check_stmt): Likewise.
24747 (pass_nothrow::execute): Likewise.
24748 * ipa-split.c (execute_split_functions): Likewise.
24749 * omp-offload.c (dump_oacc_loop_part): Likewise.
24750 (dump_oacc_loop): Likewise.
24751 * trans-mem.c (tm_log_emit): Likewise.
24752 (tm_memopt_accumulate_memops): Likewise.
24753 (dump_tm_memopt_set): Likewise.
24754 (dump_tm_memopt_transform): Likewise.
24755 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24756 (print_loop): Likewise.
24757 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24758 (chkp_gather_checks_info): Likewise.
24759 (chkp_get_check_result): Likewise.
24760 (chkp_remove_check_if_pass): Likewise.
24761 (chkp_use_outer_bounds_if_possible): Likewise.
24762 (chkp_reduce_bounds_lifetime): Likewise.
24763 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24764 (chkp_mark_completed_bounds): Likewise.
24765 (chkp_register_incomplete_bounds): Likewise.
24766 (chkp_mark_invalid_bounds): Likewise.
24767 (chkp_maybe_copy_and_register_bounds): Likewise.
24768 (chkp_build_returned_bound): Likewise.
24769 (chkp_get_bound_for_parm): Likewise.
24770 (chkp_build_bndldx): Likewise.
24771 (chkp_get_bounds_by_definition): Likewise.
24772 (chkp_generate_extern_var_bounds): Likewise.
24773 (chkp_get_bounds_for_decl_addr): Likewise.
24774 * tree-chrec.c (chrec_apply): Likewise.
24775 * tree-data-ref.c (dump_data_reference): Likewise.
24776 (dump_subscript): Likewise.
24777 (dump_data_dependence_relation): Likewise.
24778 (analyze_overlapping_iterations): Likewise.
24779 * tree-inline.c (expand_call_inline): Likewise.
24780 (tree_function_versioning): Likewise.
24781 * tree-into-ssa.c (dump_defs_stack): Likewise.
24782 (dump_currdefs): Likewise.
24783 (dump_names_replaced_by): Likewise.
24784 (dump_update_ssa): Likewise.
24785 (update_ssa): Likewise.
24786 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24787 * tree-parloops.c (build_new_reduction): Likewise.
24788 (try_create_reduction_list): Likewise.
24789 (ref_conflicts_with_region): Likewise.
24790 (oacc_entry_exit_ok_1): Likewise.
24791 (oacc_entry_exit_single_gang): Likewise.
24792 * tree-pretty-print.h: Likewise.
24793 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24794 (get_scalar_evolution): Likewise.
24795 (add_to_evolution): Likewise.
24796 (get_loop_exit_condition): Likewise.
24797 (analyze_evolution_in_loop): Likewise.
24798 (analyze_initial_condition): Likewise.
24799 (analyze_scalar_evolution): Likewise.
24800 (instantiate_scev): Likewise.
24801 (number_of_latch_executions): Likewise.
24802 (gather_chrec_stats): Likewise.
24803 (final_value_replacement_loop): Likewise.
24804 (scev_const_prop): Likewise.
24805 * tree-sra.c (dump_access): Likewise.
24806 (disqualify_candidate): Likewise.
24807 (create_access): Likewise.
24808 (reject): Likewise.
24809 (maybe_add_sra_candidate): Likewise.
24810 (create_access_replacement): Likewise.
24811 (analyze_access_subtree): Likewise.
24812 (analyze_all_variable_accesses): Likewise.
24813 (sra_modify_assign): Likewise.
24814 (initialize_constant_pool_replacements): Likewise.
24815 (find_param_candidates): Likewise.
24816 (decide_one_param_reduction): Likewise.
24817 (replace_removed_params_ssa_names): Likewise.
24818 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24819 * tree-ssa-copy.c (dump_copy_of): Likewise.
24820 (copy_prop_visit_cond_stmt): Likewise.
24821 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24822 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24823 (record_equivalences_from_stmt): Likewise.
24824 * tree-ssa-dse.c (compute_trims): Likewise.
24825 (delete_dead_call): Likewise.
24826 (delete_dead_assignment): Likewise.
24827 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24828 (forward_propagate_into_cond): Likewise.
24829 (pass_forwprop::execute): Likewise.
24830 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24831 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24833 (move_computations_worker): Likewise.
24834 (execute_sm): Likewise.
24835 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24836 (remove_exits_and_undefined_stmts): Likewise.
24837 (remove_redundant_iv_tests): Likewise.
24838 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24839 (adjust_iv_update_pos): Likewise.
24840 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24841 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24842 (value_replacement): Likewise.
24843 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24844 * tree-ssa-pre.c (print_pre_expr): Likewise.
24845 (get_representative_for): Likewise.
24846 (create_expression_by_pieces): Likewise.
24847 (insert_into_preds_of_block): Likewise.
24848 (eliminate_insert): Likewise.
24849 (eliminate_dom_walker::before_dom_children): Likewise.
24850 (eliminate): Likewise.
24851 (remove_dead_inserted_code): Likewise.
24852 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24853 * tree-ssa-reassoc.c (get_rank): Likewise.
24854 (eliminate_duplicate_pair): Likewise.
24855 (eliminate_plus_minus_pair): Likewise.
24856 (eliminate_not_pairs): Likewise.
24857 (undistribute_ops_list): Likewise.
24858 (eliminate_redundant_comparison): Likewise.
24859 (update_range_test): Likewise.
24860 (optimize_range_tests_var_bound): Likewise.
24861 (optimize_vec_cond_expr): Likewise.
24862 (rewrite_expr_tree): Likewise.
24863 (rewrite_expr_tree_parallel): Likewise.
24864 (linearize_expr): Likewise.
24865 (break_up_subtract): Likewise.
24866 (linearize_expr_tree): Likewise.
24867 (attempt_builtin_powi): Likewise.
24868 (attempt_builtin_copysign): Likewise.
24869 (transform_stmt_to_copy): Likewise.
24870 (transform_stmt_to_multiply): Likewise.
24871 (dump_ops_vector): Likewise.
24872 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24873 (print_scc): Likewise.
24874 (set_ssa_val_to): Likewise.
24875 (visit_reference_op_store): Likewise.
24876 (visit_use): Likewise.
24877 (sccvn_dom_walker::before_dom_children): Likewise.
24878 (run_scc_vn): Likewise.
24879 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24881 (expr_hash_elt::print): Likewise.
24882 (const_and_copies::pop_to_marker): Likewise.
24883 (const_and_copies::record_const_or_copy_raw): Likewise.
24884 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24885 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24886 (dump_predicates): Likewise.
24887 (find_uninit_use): Likewise.
24888 (warn_uninitialized_phi): Likewise.
24889 (pass_late_warn_uninitialized::execute): Likewise.
24890 * tree-ssa.c (verify_vssa): Likewise.
24891 (verify_ssa): Likewise.
24892 (maybe_optimize_var): Likewise.
24893 * tree-vrp.c (dump_value_range): Likewise.
24894 (dump_all_value_ranges): Likewise.
24895 (dump_asserts_for): Likewise.
24896 (register_edge_assert_for_2): Likewise.
24897 (vrp_visit_cond_stmt): Likewise.
24898 (vrp_visit_switch_stmt): Likewise.
24899 (vrp_visit_stmt): Likewise.
24900 (vrp_visit_phi_node): Likewise.
24901 (simplify_cond_using_ranges_1): Likewise.
24902 (fold_predicate_in): Likewise.
24903 (evrp_dom_walker::before_dom_children): Likewise.
24904 (evrp_dom_walker::push_value_range): Likewise.
24905 (evrp_dom_walker::pop_value_range): Likewise.
24906 (execute_early_vrp): Likewise.
24908 2017-05-16 Richard Biener <rguenther@suse.de>
24910 * dwarf2out.c (loc_list_from_tree_1): Do not create
24911 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24913 2017-05-16 Richard Biener <rguenther@suse.de>
24915 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24917 (note_variable_value_in_expr): If we resolved the decl ref
24918 do not push to the stack.
24920 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24922 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24923 operations in fast-math mode.
24924 (vaddq_f16): Likewise.
24925 (vmul_f16): Likewise.
24926 (vmulq_f16): Likewise.
24927 (vsub_f16): Likewise.
24928 (vsubq_f16): Likewise.
24929 * config/arm/neon.md (add<mode>3): New.
24931 (fma:<VH:mode>3): New. Also remove outdated comment.
24934 2017-05-16 Martin Liska <mliska@suse.cz>
24938 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24939 (odr_types_equivalent_p): Likewise.
24941 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24943 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24945 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24948 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24949 non-interunit SSE move alternatives with '?'.
24950 (zero-extendsidi peephole2): New peephole to skip intermediate
24951 general register in SSE zero-extend sequence.
24953 2017-05-15 Jeff Law <law@redhat.com>
24955 * reorg.c (relax_delay_slots): Create a new variable to hold
24956 the temporary target rather than clobbering TARGET_LABEL.
24958 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24959 missing argument to extract_bit_field call.
24960 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24962 2017-05-15 Martin Liska <mliska@suse.cz>
24965 * gcc.c (process_command): Do not allow empty argument of -o option.
24967 2017-05-15 Renlin Li <renlin.li@arm.com>
24969 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24970 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24971 * config/aarch64/constraints.md (Usf): Add long call check.
24972 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24973 (call_value): Likewise.
24974 (sibcall): Likewise.
24975 (sibcall_value): Likewise.
24977 (call_value_insn): New.
24978 (sibcall_insn): Update rtx pattern.
24979 (sibcall_value_insn): Likewise.
24980 (call_internal): Remove.
24981 (call_value_internal): Likewise.
24982 (sibcall_internal): Likewise.
24983 (sibcall_value_internal): Likewise.
24984 (call_reg): Likewise.
24985 (call_symbol): Likewise.
24986 (call_value_reg): Likewise.
24987 (call_value_symbol): Likewise.
24989 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24992 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24994 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24996 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24997 compatible with CCGOCmode and with CCZmode.
24999 2017-05-14 Martin Sebor <msebor@redhat.com>
25001 PR middle-end/77671
25002 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
25003 (gimple_fold_builtin_snprintf): Same.
25004 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
25005 (gimple_fold_builtin_snprintf): Same.
25006 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
25007 of character types.
25008 (is_call_safe): New function.
25009 (try_substitute_return_value): Call it.
25010 (try_simplify_call): New function.
25011 (pass_sprintf_length::handle_gimple_call): Call it.
25013 2017-05-14 Martin Sebor <msebor@redhat.com>
25015 PR middle-end/80669
25016 * builtins.c (expand_builtin_stpncpy): Simplify.
25018 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
25020 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
25021 * config/i386/i386.h
25022 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
25023 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
25024 (struct machine_function): Add new members call_ms2sysv,
25025 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
25026 (struct machine_frame_state): New fields sp_realigned and
25027 sp_realigned_offset.
25028 * config/i386/i386.c
25029 (enum xlogue_stub): New enum.
25030 (enum xlogue_stub_sets): New enum.
25031 (class xlogue_layout): New class.
25032 (struct ix86_frame): New fields stack_realign_allocate_offset,
25033 stack_realign_offset and outlined_save_offset. Modify comments to
25034 detail stack layout when using out-of-line stubs.
25035 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
25036 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
25037 -mcall-ms2sysv-xlogues.
25038 (stub_managed_regs): New static variable.
25039 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
25040 registers managed by out-of-line stub.
25041 (disable_call_ms2sysv_xlogues): New function.
25042 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
25043 m->call_ms2sysv when appropriate and compute frame layout for
25045 (sp_valid_at, fp_valid_at): New inline functions.
25046 (choose_basereg): New function.
25047 (choose_baseaddr): Add align parameter, use choose_basereg and modify
25049 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
25050 Use align parameter of choose_baseaddr to generated aligned SSE movs
25052 (pro_epilogue_adjust_stack): Modify to track
25053 machine_frame_state::sp_realigned.
25054 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
25055 (ix86_nsaved_sseregs): Likewise.
25056 (ix86_emit_save_regs): Likewise.
25057 (ix86_emit_save_regs_using_mov): Likewise.
25058 (ix86_emit_save_sse_regs_using_mov): Likewise.
25059 (get_scratch_register_on_entry): Likewise.
25060 (gen_frame_set): New function.
25061 (gen_frame_load): Likewise.
25062 (gen_frame_store): Likewise.
25063 (emit_outlined_ms2sysv_save): Likewise.
25064 (emit_outlined_ms2sysv_restore): Likewise.
25065 (ix86_expand_prologue): Modify stack re-alignment code and call
25066 emit_outlined_ms2sysv_save when appropriate.
25067 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
25068 parameter rtx_insn *insn, which allows the function to be used to only
25069 generate the notes.
25070 (ix86_expand_epilogue): Modify validity checks of frame and stack
25071 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
25072 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
25073 * config/i386/predicates.md
25074 (save_multiple): New predicate.
25075 (restore_multiple): Likewise.
25076 * config/i386/sse.md
25077 (save_multiple<mode>): New pattern.
25078 (save_multiple_realign<mode>): Likewise.
25079 (restore_multiple<mode>): Likewise.
25080 (restore_multiple_and_return<mode>): Likewise.
25081 (restore_multiple_leave_return<mode>): Likewise.
25082 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
25084 2017-05-14 Julia Koval <julia.koval@intel.com>
25086 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
25087 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
25088 (__builtin_ia32_xsetbv): New builtins.
25089 * config/i386/i386.c (ix86_expand_special_args_builtin):
25091 (ix86_expand_builtin): Special expand for new intrinsics.
25092 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
25093 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
25094 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
25096 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25098 * cfganal.c (inverted_post_order_compute): Change argument type
25100 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25101 * df-core.c (rest_of_handle_df_initialize): Adjust.
25102 (rest_of_handle_df_finish): Likewise.
25103 (df_analyze_1): Likewise.
25104 (df_analyze): Likewise.
25105 (loop_inverted_post_order_compute): Change argument to be a vec *.
25106 (df_analyze_loop): Adjust.
25107 (df_get_n_blocks): Likewise.
25108 (df_get_postorder): Likewise.
25109 * df.h (struct df_d): Change field to be a vec.
25110 * lcm.c (compute_laterin): Adjust.
25111 (compute_available): Likewise.
25112 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25113 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25114 * tree-ssa-pre.c (compute_antic): Likewise.
25116 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25118 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25119 (depth_first_search::depth_first_search): Change structure init
25120 function to this constructor.
25121 (depth_first_search::add_bb): Rename function to this member.
25122 (depth_first_search::execute): Likewise.
25123 (flow_dfs_compute_reverse_finish): Adjust.
25125 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25127 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25128 (longest_simple_path): Likewise.
25129 * shrink-wrap.c (spread_components): Likewise.
25130 (disqualify_problematic_components): Likewise.
25131 (emit_common_heads_for_components): Likewise.
25132 (emit_common_tails_for_components): Likewise.
25133 (insert_prologue_epilogue_for_components): Likewise.
25135 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25137 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25140 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25142 * df-core.c (df_set_blocks): Start using auto_bitmap.
25143 (df_compact_blocks): Likewise.
25144 * df-problems.c (df_rd_confluence_n): Likewise.
25145 * df-scan.c (df_insn_rescan_all): Likewise.
25146 (df_process_deferred_rescans): Likewise.
25147 (df_update_entry_block_defs): Likewise.
25148 (df_update_exit_block_uses): Likewise.
25149 (df_entry_block_bitmap_verify): Likewise.
25150 (df_exit_block_bitmap_verify): Likewise.
25151 (df_scan_verify): Likewise.
25152 * lra-constraints.c (lra_constraints): Likewise.
25153 (undo_optional_reloads): Likewise.
25154 (lra_undo_inheritance): Likewise.
25155 * lra-remat.c (calculate_gen_cands): Likewise.
25156 (do_remat): Likewise.
25157 * lra-spills.c (assign_spill_hard_regs): Likewise.
25158 (spill_pseudos): Likewise.
25159 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25160 (bitmap_set_subtract_values): Likewise.
25162 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25164 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25165 management with auto_bitmap.
25166 (fix_inter_tick): Likewise.
25167 (fix_recovery_deps): Likewise.
25168 * ira.c (add_store_equivs): Likewise.
25169 (find_moveable_pseudos): Likewise.
25170 (split_live_ranges_for_shrink_wrap): Likewise.
25171 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25172 (rtx_reuse_manager::seen_def_p): Likewise.
25173 (rtx_reuse_manager::set_seen_def): Likewise.
25174 * print-rtl.h (class rtx_reuse_manager): Likewise.
25176 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25178 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25180 (migrate_btr_def): Likewise.
25181 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25182 * df-core.c (loop_post_order_compute): Likewise.
25183 (loop_inverted_post_order_compute): Likewise.
25184 * hsa-common.h: Likewise.
25185 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25186 * init-regs.c (initialize_uninitialized_regs): Likewise.
25187 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25188 (inline_small_functions): Likewise.
25189 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25190 * ira.c (combine_and_move_insns): Likewise.
25191 (build_insn_chain): Likewise.
25192 * loop-invariant.c (find_invariants): Likewise.
25193 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25194 * predict.c (tree_predict_by_opcode): Likewise.
25195 (predict_paths_leading_to): Likewise.
25196 (predict_paths_leading_to_edge): Likewise.
25197 (estimate_loops_at_level): Likewise.
25198 (estimate_loops): Likewise.
25199 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25200 (spread_components): Likewise.
25201 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25202 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25203 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25204 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25205 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25206 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25207 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25208 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25209 (mark_threaded_blocks): Likewise.
25210 (thread_through_all_blocks): Likewise.
25211 * tree-ssa.c (verify_ssa): Likewise.
25212 (execute_update_addresses_taken): Likewise.
25213 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25215 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25217 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25219 (post_order_compute): Likewise.
25220 (inverted_post_order_compute): Likewise.
25221 (pre_and_rev_post_order_compute_fn): Likewise.
25223 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25225 * genrecog.c (int_set::int_set): Explicitly construct our
25226 auto_vec base class.
25227 * vec.h (auto_vec::auto_vec): New constructor.
25229 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25231 * bitmap.h (class auto_bitmap): New constructor taking
25232 bitmap_obstack * argument.
25234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25236 * bitmap.h (class auto_bitmap): Change type of m_bits to
25237 bitmap_head, and adjust ctor / dtor and member operators.
25239 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25241 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25242 when returned register mode doesn't match original mode.
25244 2017-05-12 Jeff Law <law@redhat.com>
25245 Jakub Jelinek <jakub@redhat.com>
25247 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25248 we look for cc setter after the compare-elim changes.
25249 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25250 within the vector to match what compare-elim now expects.
25251 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25252 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25254 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25255 after the compare-elim changes.
25256 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25257 the vector to match what compare-elim now expects.
25258 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25259 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25260 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25261 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25262 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25264 * config/visium/visium.c (single_set_and_flags): Fix where
25265 we look for cc setter after the compare-elim changes.
25266 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25267 with the vector to match what compare-elim now expects.
25268 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25269 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25270 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25271 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25272 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25273 (neg<mode>2_insn_set_overflow): Likewise.
25275 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25277 PR middle-end/79794
25278 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25279 maybe_expand_insn call, set ops[0].target. If still set after call,
25280 set alt_rtl. Add extra arg to recursive calls.
25281 (extract_bit_field): Add alt_rtl argument. Pass to
25283 * expmed.h (extract_bit_field): Fix prototype.
25284 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25285 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25286 to extract_bit_field_calls.
25287 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25288 Pass alt_rtl to extract_bit_field calls.
25289 * calls.c (store_unaligned_arguments_into_psuedos)
25290 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25291 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25293 * optabs.h (struct expand_operand): Add target bitfield.
25295 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25297 * compare-elim.c (try_eliminate_compare): Canonicalize
25298 operation with embedded compare to
25299 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25300 (set (reg) (operation)].
25302 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25304 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25307 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25308 cost of adding a carry flag for ADC instruction.
25309 [case MINUS]: Ignore the cost of subtracting a carry flag
25310 for SBB instruction.
25312 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25314 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25316 * config/rs6000/bmiintrin.h: New file.
25317 * config/rs6000/bmi2intrin.h: New file.
25318 * config/rs6000/x86intrin.h: New file.
25320 2017-05-12 Jeff Law <law@redhat.com>
25322 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25325 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25327 PR middle-end/80707
25328 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25330 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25332 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25333 early expansion of vector divide builtins.
25334 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25335 builtins identified as having unsigned arguments.
25337 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25339 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25340 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25341 expansion of vector logical operations (and, andc, or, xor,
25344 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25346 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25347 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25349 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25351 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25352 early GIMPLE expansion of vector multiplies.
25354 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25356 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25357 TARGET_HAVE_MOVT conditional.
25358 (movt splitter): Likewise.
25360 2017-05-12 Richard Biener <rguenther@suse.de>
25362 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25363 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25364 Fold all stmts not inplace.
25366 2017-05-12 Richard Biener <rguenther@suse.de>
25368 PR tree-optimization/80713
25369 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25370 inserted_exprs bit for not removed stmts.
25372 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25374 PR middle-end/69921
25375 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25376 parallelized" attribute for parallelized OpenACC kernels.
25377 * omp-offload.c (execute_oacc_device_lower): Use it.
25379 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25380 Set "oacc kernels" attribute.
25381 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25382 parameter. Adjust all users.
25383 (oacc_fn_attrib_kernels_p): Remove function.
25384 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25385 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25386 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25387 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25388 assert "oacc kernels" attribute is set.
25390 2017-05-11 Carl Love <cel@us.ibm.com>
25392 * config/rs6000/rs6000-c: Add support for built-in functions
25393 vector unsigned char vec_popcnt (vector signed char)
25394 vector unsigned char vec_popcnt (vector unsigned char)
25395 vector unsigned short vec_popcnt (vector signed short)
25396 vector unsigned short vec_popcnt (vector unsigned short)
25397 vector unsigned int vec_popcnt (vector signed int)
25398 vector unsigned int vec_popcnt (vector unsigned int)
25399 vector unsigned long long vec_popcnt (vector signed long long)
25400 vector unsigned long long vec_popcnt (vector unsigned long long)
25401 vector signed long long vec_slo (vector signed long long,
25402 vector signed char)
25403 vector signed long long vec_slo (vector signed long long,
25404 vector unsigned char)
25405 vector unsigned long long vec_slo (vector unsigned long long,
25406 vector signed char)
25407 vector unsigned long long vec_slo (vector unsigned long long,
25408 vector unsigned char)
25409 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25410 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25411 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25412 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25413 * doc/extend.texi: Update the built-in documentation file for the
25414 new built-in functions.
25416 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25418 * attribs.h (sorted_attr_string): Move machine independent
25419 functions for target clone support from the i386 port to common
25420 code. Rename ix86_function_versions to common_function_versions.
25421 Rename make_name to make_unique_name.
25422 (common_function_versions): Likewise.
25423 (make_unique_name): Likewise.
25424 (make_dispatcher_decl): Likewise.
25425 (is_function_default_version): Likewise.
25426 * attribs.c (attr_strcmp): Likewise.
25427 (sorted_attr_string): Likewise.
25428 (common_function_versions): Likewise.
25429 (make_unique_name): Likewise.
25430 (make_dispatcher_decl): Likewise.
25431 (is_function_default_version): Likewise.
25432 * config/i386/i386.c (attr_strcmp): Likewise.
25433 (sorted_attr_string): Likewise.
25434 (ix86_function_versions): Likewise.
25435 (make_name): Likewise.
25436 (make_dispatcher_decl): Likewise.
25437 (is_function_default_version): Likewise.
25438 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25440 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25443 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25444 Account for direct move costs for vec_construct of integer
25447 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25450 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25451 (UNSPEC_STX_ATOMIC): Ditto.
25452 (loaddi_via_sse): New insn.
25453 (storedi_via_sse): Ditto.
25454 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25455 Update corresponding peephole2 patterns.
25456 (atomic_storedi_fpu): Ditto.
25458 2017-05-11 Julia Koval <julia.koval@intel.com>
25460 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25461 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25463 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25464 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25465 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25467 2017-05-11 Nathan Sidwell <nathan@acm.org>
25469 * graphite-poly.c: Include dumpfile.h.
25471 * dumpfle.h (dump_function): Declare here ...
25472 * tree-dump.h (dump_function): ... not here.
25473 * dumpfile.c: #include tree-cfg.h.
25474 (dump_function): Move here from ...
25475 * tree-dump.c (dump_function): ... here.
25476 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25477 * graphite-poly.c: Don't include tree-dump.h.
25478 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25479 * print-tree.c: Likewise.
25480 * stor-layout.c: Likewise.
25481 * tree-nested.c: Likewise.
25483 * dumpfile.c (dump_start): Use TDF_FLAGS.
25484 (dump_enable_all): Fix TDF_KIND check thinko.
25486 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25488 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25489 array entries to represent two legal parameterizations of the
25490 overloaded __builtin_cmpb function, as represented by the
25491 P6_OV_BUILTIN_CMPB constant.
25492 (altivec_resolve_overloaded_builtin): Add special case handling
25493 for the __builtin_cmpb function, as represented by the
25494 P6_OV_BUILTIN_CMPB constant.
25495 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25496 (BU_P6_64BIT_2): New macro.
25497 (BU_P6_OVERLOAD_2): New macro
25498 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25499 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25500 (CMPB): Add overload support to represent both 32-bit and 64-bit
25501 compare-bytes function.
25502 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25503 support for TARGET_CMPB.
25504 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25505 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25506 documentation of the __builtin_cmpb overloaded built-in function.
25508 2017-05-11 Richard Biener <rguenther@suse.de>
25510 PR tree-optimization/80705
25511 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25512 bases are not vectorizable.
25514 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25516 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25517 when counting register pressure.
25519 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25521 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25522 (iv_ca_more_deps): Renamed to ...
25523 (iv_ca_compare_deps): ... this.
25524 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25526 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25528 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25530 (determine_group_iv_costs): ... here.
25531 (find_inv_vars_cb): Record inv var if it's not recorded before.
25533 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25535 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25536 (get_shiftadd_cost): Ditto.
25538 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25540 * tree-ssa-address.c: Include header file.
25541 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25543 (add_to_parts): Refactor.
25544 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25545 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25548 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25550 PR tree-optimization/53090
25551 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25553 (extract_cond_operands): Detect condition with IV on both sides
25554 and return COMP_IV_EXPR_2.
25555 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25556 (rewrite_use_compare): Simplify by removing call to function
25557 extract_cond_operands.
25559 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25561 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25562 (extract_cond_operands): Detect condition comparing against non-
25563 invariant bound and return appropriate enum value.
25564 (find_interesting_uses_cond): Update use of extract_cond_operands.
25565 Handle its return value accordingly.
25566 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25568 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25570 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25571 nonlinear iv_use computation in loop invariant sensitive way.
25573 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25575 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25576 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25578 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25580 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25581 (dump_cand): Support iv_cand.inv_exprs.
25582 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25584 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25589 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25591 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25592 as local function. Include necessary header files.
25593 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25597 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25599 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25601 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25602 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25603 RSHIFT_EXPR and BIT_NOT_EXPR.
25605 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25607 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25608 (adjust_setup_cost): New parameter supporting round up adjustment.
25609 (struct address_cost_data): Delete.
25610 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25611 (split_address_cost, ptr_difference_cost): Delete.
25612 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25613 (struct ainc_cost_data): New struct.
25614 (get_address_cost_ainc): New function.
25615 (get_address_cost, get_computation_cost): Reimplement.
25616 (determine_group_iv_cost_address): Record inv_expr for all uses of
25618 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25619 (iv_ca_has_deps): Reimplemented to ...
25620 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25622 (iv_ca_extend): Call iv_ca_more_deps.
25624 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25626 * tree-ssa-address.c (struct mem_address): Move to header file.
25627 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25628 * tree-ssa-address.h (struct mem_address): Move from C file.
25629 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25633 * tree-affine.h (aff_combination_type): New interface.
25634 (aff_combination_zero_p): Remove static.
25635 (aff_combination_const_p): New interface.
25636 (aff_combination_singleton_var_p): New interfaces.
25638 2017-05-11 Richard Biener <rguenther@suse.de>
25640 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25641 Skip unreachable blocks and destinations.
25642 (eliminate): Move stmt removal and fixup ...
25643 (fini_eliminate): ... here. Skip inserted exprs.
25644 (pass_pre::execute): Move fini_pre after fini_eliminate.
25645 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25646 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25647 PRE to get rid of dead code that has invalid SSA form and
25648 split critical edges again.
25650 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25652 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25654 2017-05-11 Richard Biener <rguenther@suse.de>
25656 * passes.c (execute_function_todo): Verify loops if they are
25657 said to be up-to-date.
25658 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25659 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25661 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25664 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25665 handle calling assemble_external ourself.
25668 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25669 modes with zero size. Enhance comment.
25671 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25673 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25674 built-ins for vec_xl and vec_xst with short and char pointer
25677 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25679 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25680 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25681 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25682 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25683 (_mm_maskz_min_round_ss): New intrinsics.
25684 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25685 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25686 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25687 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25688 (__builtin_ia32_minss_mask_round): New builtins.
25689 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25690 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25691 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25693 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25694 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25696 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25699 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25701 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25702 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25703 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25704 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25705 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25706 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25707 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25708 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25709 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25710 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25711 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25712 (__builtin_ia32_mulss_mask_round): New builtins.
25713 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25714 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25715 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25717 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25718 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25720 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25723 2017-05-10 Julia Koval <julia.koval@intel.com>
25725 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25726 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25727 (_mm256_setr_m128i): New intrinsics.
25729 2017-05-10 Julia Koval <julia.koval@intel.com>
25731 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25732 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25733 (_mm_maskz_rcp14_ss): New intrinsics.
25734 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25735 (__builtin_ia32_rcp14ss_mask): New builtins.
25736 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25738 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25740 PR tree-optimization/51513
25741 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25742 (assert_unreachable_fallthru_edge_p): Use it.
25743 (group_case_labels_stmt): Likewise.
25744 * tree-cfg.h: Prototype it.
25745 * stmt.c: Include cfghooks.h and tree-cfg.h.
25746 (emit_case_dispatch_table) <gap_label>: New local variable.
25747 Use it to fill dispatch table gaps.
25748 Test for default_label before updating probabilities.
25749 (expand_case) <default_label>: Remove unneeded initialization.
25750 Test for unreachable default case statement and remove its edge.
25751 Set default_label accordingly.
25752 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25754 2017-05-10 Carl Love <cel@us.ibm.com>
25756 * config/rs6000/rs6000-c: Add support for built-in functions
25757 vector signed char vec_neg (vector signed char)
25758 vector signed short int vec_neg (vector short int)
25759 vector signed int vec_neg (vector signed int)
25760 vector signed long long vec_neg (vector signed long long)
25761 vector float vec_neg (vector float)
25762 vector double vec_neg (vector double)
25763 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25765 * config/rs6000/altivec.h: Add define for vec_neg
25766 * doc/extend.texi: Update the built-in documentation for the
25767 new built-in functions.
25769 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25771 PR tree-optimization/77644
25772 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25774 2017-05-10 Nathan Sidwell <nathan@acm.org>
25776 * dumpfile.h (TDI_lang_all): New.
25777 (TDF_KIND): New. Renumber others
25778 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25780 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25782 (get_dump_file_name): Adjust suffix generation.
25783 (dump_enable_all): Use TDF_KIND.
25784 * doc/invoke.texi (-fdump-lang-all): Document.
25786 * dumpfile.h: Tabify.
25788 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25791 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25792 Move member access before delete.
25794 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25796 * tree-inline.c (expand_call_inline): Split block at stmt
25799 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25802 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25803 are now unused after splitting mov{sf,sd}_hardfloat.
25804 (f32_lr2): Likewise.
25805 (f32_lm): Likewise.
25806 (f32_lm2): Likewise.
25807 (f32_li): Likewise.
25808 (f32_li2): Likewise.
25809 (f32_lv): Likewise.
25810 (f32_sr): Likewise.
25811 (f32_sr2): Likewise.
25812 (f32_sm): Likewise.
25813 (f32_sm2): Likewise.
25814 (f32_si): Likewise.
25815 (f32_si2): Likewise.
25816 (f32_sv): Likewise.
25817 (f32_dm): Likewise.
25818 (f32_vsx): Likewise.
25819 (f32_av): Likewise.
25820 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25821 For movsf, order stores so the VSX stores occur before the GPR
25822 store which encourages the register allocator to use a traditional
25823 FPR instead of a GPR. For movsd, order the stores so that the GPR
25824 store comes before the VSX stores to allow the power6 to work.
25825 This is due to the power6 not having a 32-bit integer store
25826 instruction from a FPR.
25827 (movsf_hardfloat): Likewise.
25828 (movsd_hardfloat): Likewise.
25830 2017-05-09 Martin Sebor <msebor@redhat.com>
25832 PR translation/80280
25833 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25836 PR translation/80280
25837 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25838 data member added in r247778.
25839 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25841 2017-05-09 Nathan Sidwell <nathan@acm.org>
25843 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25845 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25848 2017-05-09 Marek Polacek <polacek@redhat.com>
25850 * doc/invoke.texi: Fix typo.
25852 2017-05-09 Richard Biener <rguenther@suse.de>
25854 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25855 (vrp_val_is_min): Likewise.
25856 (set_value_range_to_value): Likewise.
25857 (set_value_range_to_nonnegative): Likewise.
25858 (gimple_assign_nonzero_p): Likewise.
25859 (gimple_stmt_nonzero_p): Likewise.
25860 (vrp_int_const_binop): Likewise. Remove unreachable case.
25861 (adjust_range_with_scev): Adjust comments.
25862 (compare_range_with_value): Likewise.
25863 (extract_range_from_phi_node): Likewise.
25864 (test_for_singularity): Likewise.
25866 2017-05-09 Richard Biener <rguenther@suse.de>
25868 * tree-vrp.c (get_single_symbol): Add assert that we don't
25869 get overflowed constants as invariant part.
25870 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25871 checks. Use wi::cmp instead of recursing for integer constants.
25872 (compare_values): Just ignore whether we assumed undefined
25873 overflow instead of failing the compare.
25874 (extract_range_for_var_from_comparison_expr): Add comment before the
25875 TREE_NO_WARNING sets.
25876 (test_for_singularity): Likewise.
25877 (extract_range_from_comparison): Do not disable optimization
25878 when we assumed undefined overflow.
25879 (extract_range_basic): Remove init of unused var.
25881 2017-05-09 Richard Biener <rguenther@suse.de>
25883 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25884 (extract_range_from_multiplicative_op_1): Adjust.
25885 (extract_range_from_binary_expr_1): Use int_const_binop.
25887 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25890 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25891 rs6000_store_data_bypass_p in seven define_bypass directives and
25892 in several comments.
25893 * config/rs6000/rs6000-protos.h: Add prototype for
25894 rs6000_store_data_bypass_p function.
25895 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25896 function implements slightly different (rs6000-specific) semantics
25897 than store_data_bypass_p, returning false rather than aborting
25898 with assertion error when arguments do not satisfy the
25899 requirements of store data bypass.
25900 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25901 rs6000_store_data_bypass_p.
25903 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25905 * config/xtensa/xtensa-protos.h
25906 (xtensa_initial_elimination_offset): New declaration.
25907 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25908 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25909 macro definition, add case for FRAME_POINTER_REGNUM when
25910 FRAME_GROWS_DOWNWARD.
25911 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25912 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25913 xtensa_initial_elimination_offset.
25915 2017-05-08 Nathan Sidwell <nathan@acm.org>
25917 * doc/invoke.texi: Alphabetize -fdump options.
25919 2017-05-08 Martin Sebor <msebor@redhat.com>
25921 PR translation/80280
25922 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25924 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25926 * target.def (compute_frame_layout): New optional target hook.
25927 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25928 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25929 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25931 * reload1.c (verify_initial_elim_offsets): Likewise.
25932 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25933 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25934 (arm_get_frame_offsets): Split up into this ...
25935 (arm_compute_frame_layout): ... and this function.
25937 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25939 * config/aarch64/constraints.md (Usa): New constraint.
25940 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25942 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25944 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25945 with_multilib_list after it has been checked.
25947 2017-05-08 Richard Biener <rguenther@suse.de>
25949 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25950 (bitmap_set_subtract_values): Likewise.
25952 2017-05-08 Richard Biener <rguenther@suse.de>
25954 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25955 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25957 (gimple_stmt_nonzero_warnv_p): Rename to ...
25958 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25960 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25961 (extract_range_basic): Adjust, do not disable propagation on
25962 strict overflow sensitive simplification.
25963 (vrp_visit_cond_stmt): Likewise.
25965 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25967 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25968 body size unconditionally.
25970 2017-05-07 Jeff Law <law@redhat.com>
25973 2017-05-06 Jeff Law <law@redhat.com>
25974 PR tree-optimization/78496
25975 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25978 PR tree-optimization/78496
25979 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25980 (simplify_stmt_using_ranges): Call it.
25981 (vrp_dom_walker::before_dom_children): Extract equivalences
25982 from an ASSERT_EXPR with an equality comparison against a
25985 2017-05-06 Jeff Law <law@redhat.com>
25987 PR tree-optimization/78496
25988 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25991 PR tree-optimization/78496
25992 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25993 (simplify_stmt_using_ranges): Call it.
25994 (vrp_dom_walker::before_dom_children): Extract equivalences
25995 from an ASSERT_EXPR with an equality comparison against a
25998 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26000 * lra-constraints.c (lra_copy_reg_equiv): New function.
26001 (split_reg): Use it to copy equivalence information from the
26002 original register to the spill register.
26004 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
26006 PR rtl-optimization/75964
26007 * simplify-rtx.c (simplify_const_relational_operation): Remove
26008 invalid handling of comparisons of integer ABS.
26010 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
26012 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
26013 initialize to zero.
26014 (init_regs): Remove declaration.
26015 (function_arg_advance_32): Initialize error_p as boolean variable.
26017 2017-05-05 Nathan Sidwell <nathan@acm.org>
26019 * store-motion.c (remove_reachable_equiv_notes): Reformat long
26020 lines. Use for (;;).
26022 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26024 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
26025 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
26026 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
26027 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
26028 VF=2 that require versioning.
26030 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26032 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
26035 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26037 * diagnostic.h (diagnostic_override_option_index): Convert from
26038 macro to inline function.
26040 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26042 * diagnostic.c (last_module_changed_p): New function.
26043 (set_last_module): New function.
26044 (diagnostic_report_current_module): Convert macro usage to
26045 the above functions.
26046 * diagnostic.h (diagnostic_context::last_module): Strengthen
26047 from const line_map * to const line_map_ordinary *.
26048 (diagnostic_last_module_changed): Delete macro.
26049 (diagnostic_set_last_module): Delete macro.
26051 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26053 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
26054 with diagnostic_report_diagnostic.
26055 (diagnostic_n_impl_richloc): Likewise.
26056 * diagnostic.h (report_diagnostic): Delete macro.
26057 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
26058 with diagnostic_report_diagnostic.
26059 * substring-locations.c (format_warning_va): Likewise.
26061 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26063 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
26064 save/restor of format_spec. Move option-printing code to...
26065 (print_option_information): ...this new function, and
26066 reimplement by simply printing to the pretty_printer,
26067 rather than appending to the format string.
26069 2017-05-05 David Malcolm <dmalcolm@redhat.com>
26071 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
26072 handling logic into...
26073 (update_effective_level_from_pragmas): ...this new function.
26075 2017-05-04 Andrew Waterman <andrew@sifive.com>
26077 * config/riscv/riscv.opt (mstrict-align): New option.
26078 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
26079 (SLOW_UNALIGNED_ACCESS): Define.
26080 (riscv_slow_unaligned_access): Declare.
26081 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
26083 (riscv_slow_unaligned_access): New variable.
26084 (rocket_tune_info): Set slow_unaligned_access to true.
26085 (optimize_size_tune_info): Set slow_unaligned_access to false.
26086 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
26087 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
26088 (riscv_option_override): Set riscv_slow_unaligned_access.
26089 * doc/invoke.texi: Add -mstrict-align to RISC-V.
26091 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
26093 * config/riscv/riscv.md: Unify indentation.
26095 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26100 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26102 (extendsi<mode>2): Add support for doing sign extension via
26103 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26104 don't have ISA 3.0 instructions.
26105 (extendsi<mode>2 splitter): Likewise.
26106 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26107 generate the normal insns since SImode can now go in vector
26108 registers. Disallow the special UNSPECs needed for previous
26109 machines to hide SImode being used. Add new insns
26110 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26111 (fix_trunc<mode>si2_stfiwx): Likewise.
26112 (fix_trunc<mode>si2_internal): Likewise.
26113 (fixuns_trunc<mode>si2): Likewise.
26114 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26115 (fctiw<u>z_<mode>_smallint): Likewise.
26116 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26117 of floating point to 32-bit integer from doing a direct move to
26118 the GPR registers to do a store.
26119 (fctiwz_<mode>): Break long line.
26121 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26123 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26124 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26125 (addr_list, addr_offset_valid_p): New.
26126 (split_address_groups): Check offset validity with above function.
26127 (gt-tree-ssa-loop-ivopts.h): Include header file.
26129 2017-05-05 Nathan Sidwell <nathan@acm.org>
26131 * config.gcc (arm*-*-*): Add missing 'fi'.
26133 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26135 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26136 included in -fopt-info does not matter.
26137 * doc/optinfo.texi (-fopt-info): Fix description of default
26138 behavour. Explicitly say order of options included in -fopt-info
26141 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26143 * config.gcc: Allow combinations of aprofile and rmprofile values for
26144 --with-multilib-list.
26145 * config/arm/t-multilib: New file.
26146 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26147 variables. Remove setting of ISA and floating-point ABI in
26148 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26149 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26150 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26151 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26153 * config/arm/t-rmprofile: Likewise except for the matches changes.
26154 * doc/install.texi (--with-multilib-list): Document the combination of
26155 aprofile and rmprofile values and warn about pitfalls in doing that.
26157 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26159 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26160 (movdi_aarch64): Likewise.
26162 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26164 PR tree-optimization/80632
26165 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26167 (build_arrays): Initialize it for virtual phis.
26168 (fix_phi_nodes): Use it for virtual phis.
26170 PR tree-optimization/80558
26171 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26172 [x, y] op z into [x op, y op z] for op & or | if conditions
26175 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26176 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26179 * config/arm/arm.md (use_literal_pool): Remove.
26180 (64-bit immediate split): No longer takes cost into consideration
26181 if arm_disable_literal_pool is enabled.
26182 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26183 used when arm_disable_literal_pool is enabled.
26184 (arm_max_const_double_inline_cost): Remove use of
26185 arm_disable_literal_pool.
26186 (push_minipool_fix): Add assert.
26187 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26188 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26189 (no_literal_pool_sf_immediate): New.
26191 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26193 PR tree-optimization/80613
26194 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26195 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26197 2017-05-05 Richard Biener <rguenther@suse.de>
26199 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26201 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26203 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26204 of this flag from insn conditions due to removal from r247495.
26206 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26208 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26210 (arm_early_store_addr_dep_ptr): Likewise.
26211 * config/arm/aarch-common-protos.h
26212 (arm_early_load_addr_dep_ptr): Add prototype.
26213 (arm_early_store_addr_dep_ptr): Likewise.
26214 * config/arm/cortex-a53.md: Add new bypasses.
26216 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26218 * tree.c (next_type_uid): Change type to unsigned.
26219 (type_hash_canon): Decrement back next_type_uid if
26220 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26221 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26224 2017-05-04 Martin Sebor <msebor@redhat.com>
26226 * builtins.c: Fix a trivial typo in a comment.
26228 PR middle-end/79234
26229 * builtins.c (check_sizes): Adjust to handle reading past the end.
26230 Avoid printing excessive upper bound of ranges. Use %E to print
26231 tree nodes instead of converting them to %wu.
26232 (expand_builtin_memchr): New function.
26233 (compute_dest_size): Rename...
26234 (compute_objsize): ...to this.
26235 (expand_builtin_memcpy): Adjust.
26236 (expand_builtin_mempcpy): Adjust.
26237 (expand_builtin_strcat): Adjust.
26238 (expand_builtin_strcpy): Adjust.
26239 (check_strncat_sizes): Adjust.
26240 (expand_builtin_strncat): Adjust.
26241 (expand_builtin_strncpy): Adjust and simplify.
26242 (expand_builtin_memset): Adjust.
26243 (expand_builtin_bzero): Adjust.
26244 (expand_builtin_memcmp): Adjust.
26245 (expand_builtin): Handle memcmp.
26246 (maybe_emit_chk_warning): Check strncat just once.
26248 2017-05-04 Martin Sebor <msebor@redhat.com>
26250 PR preprocessor/79214
26251 PR middle-end/79222
26252 PR middle-end/79223
26253 * builtins.c (check_sizes): Add inlining context and issue
26254 warnings even when -Wno-system-headers is set.
26255 (check_strncat_sizes): Same.
26256 (expand_builtin_strncat): Same.
26257 (expand_builtin_memmove): New function.
26258 (expand_builtin_stpncpy): Same.
26259 (expand_builtin): Handle memmove and stpncpy.
26261 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26263 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26264 which is not used any more.
26266 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26268 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26270 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26272 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26273 (cortexa53_tunings): Likewise.
26274 (cortexa57_tunings): Likewise.
26275 (cortexa72_tunings): Likewise.
26276 (cortexa73_tunings): Likewise.
26278 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26280 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26281 Set loop alignment to 8.
26283 2017-05-04 Martin Sebor <msebor@redhat.com>
26285 PR translation/80280
26286 * builtins.c (expand_builtin_object_size): Add missing quoting to
26287 %D and like directives.
26288 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26289 (hsa_type_for_tree_type): Same.
26290 (verify_function_arguments): Same.
26291 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26292 * varasm.c (get_section): Same.
26295 2017-05-04 Martin Sebor <msebor@redhat.com>
26297 PR translation/80280
26298 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26300 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26302 * config/aarch64/aarch64.c (generic_addrcost_table):
26303 Change HI/TI mode setting.
26305 2017-05-04 Martin Jambor <mjambor@suse.cz>
26307 PR tree-optimization/80622
26308 * tree-sra.c (comes_initialized_p): New function.
26309 (build_accesses_from_assign): Only set write lazily when
26310 comes_initialized_p is false.
26311 (analyze_access_subtree): Use comes_initialized_p.
26312 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26313 instead of testing for PARM_DECL.
26315 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26317 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26318 constraint on operand 0 to allow more general addressing modes.
26319 Adjust output template.
26320 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26322 * config/aarch64/aarch64-protos.h
26323 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26324 * config/aarch64/constraints.md (Dp): New address constraint.
26325 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26328 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26330 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26331 update use of estimate_ipcp_clone_size_and_time.
26332 (estimate_local_effects): Update use of
26333 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26334 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26335 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26336 Return nonspecialized time.
26338 2017-05-04 Richard Biener <rguenther@suse.de>
26340 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26341 for the last VUSE which def dominates the PHI. Directly call
26343 (get_continuation_for_phi_1): Remove.
26345 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26347 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26348 to explain the use of truncating division. Cap the number of
26349 iterations to the maximum given by nb_iterations_upper_bound,
26352 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26354 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26355 * configure: Regenerate.
26356 * config.in: Regenerate.
26357 * config/i386/driver-mingw32.c: new file.
26358 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26359 * config.host: Link driver-mingw32.o on MinGW host.
26360 * doc/install.texi: Document new --enable-mingw-wildcard configure
26363 2017-05-04 Marek Polacek <polacek@redhat.com>
26365 PR tree-optimization/80612
26366 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26368 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26369 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26371 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26372 (movt splitter): Likewise.
26373 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26374 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26375 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26376 block for Thumb-1 with MOVT.
26377 (thumb2_legitimate_address_p): Move code block ...
26378 (can_avoid_literal_pool_for_label_p): ... into this new function.
26379 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26381 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26382 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26383 "M-profile targets with the MOVT instruction".
26385 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26387 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26388 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26389 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26391 2017-05-04 Martin Liska <mliska@suse.cz>
26393 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26394 variable cond_code.
26396 2017-05-04 Richard Biener <rguenther@suse.de>
26398 * tree.c (array_at_struct_end_p): Handle arrays at struct
26399 end with flexarrays more conservatively. Refactor and treat
26400 arrays of arrays or aggregates more strict. Fix
26401 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26402 * tree.c (array_at_struct_end_p): Adjust prototype.
26403 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26404 * gimple-fold.c (get_range_strlen): Likewise.
26405 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26407 2017-05-04 Richard Biener <rguenther@suse.de>
26409 PR tree-optimization/31130
26410 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26412 (supports_overflow_infinity): Likewise.
26413 (is_negative_overflow_infinity): Likewise.
26414 (is_positive_overflow_infinity): Likewise.
26415 (is_overflow_infinity): Likewise.
26416 (stmt_overflow_infinity): Likewise.
26417 (overflow_infinity_range_p): Likewise.
26418 (usable_range_p): Remove as always returning true.
26419 (make_overflow_infinity): Remove.
26420 (negative_overflow_infinity): Likewise.
26421 (positive_overflow_infinity): Likewise.
26422 (avoid_overflow_infinity): Likewise.
26423 (set_value_range): Adjust accordingly.
26424 (set_value_range_to_nonnegative): Likewise, remove now unused
26425 overflow_infinity arg.
26426 (vrp_operand_equal_p): Adjust.
26427 (update_value_range): Likewise.
26428 (range_int_cst_singleton_p): Likewise.
26429 (operand_less_p): Likewise.
26430 (compare_values_warnv): Likewise.
26431 (extract_range_for_var_from_comparison_expr): Likewise.
26432 (vrp_int_const_binop): Likewise.
26433 (zero_nonzero_bits_from_vr): Likewise.
26434 (extract_range_from_multiplicative_op_1): Likewise.
26435 (extract_range_from_binary_expr_1): Likewise.
26436 (extract_range_from_unary_expr): Likewise.
26437 (extract_range_from_comparison): Likewise.
26438 (extract_range_basic): Likewise.
26439 (adjust_range_with_scev): Likewise.
26440 (compare_ranges): Likewise.
26441 (compare_range_with_value): Likewise.
26442 (dump_value_range): Likewise.
26443 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26445 (simplify_cond_using_ranges): Adjust.
26447 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26449 * brig-builtins.def: Added a builtin for class_f64.
26450 * builtin-types.def: Added a builtin type needed by class_f64.
26452 2017-05-03 Jason Merrill <jason@redhat.com>
26454 * timevar.def: Add TV_CONSTEXPR.
26456 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26458 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26460 2017-05-03 Martin Jambor <mjambor@suse.cz>
26462 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26463 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26464 * ipa-cp.c (ipcp_propagate_stage): Do not call
26465 ipa_update_after_lto_read.
26466 * ipa-inline.c (ipa_inline): Likewise.
26468 2017-05-03 Martin Jambor <mjambor@suse.cz>
26470 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26471 tag. Added a default constructor and a destructor.
26472 (ipa_edge_args_sum_t): New class;
26473 (ipa_edge_args_sum): Declare.
26474 (ipa_edge_args_vector): Remove declaration.
26475 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26476 (ipa_free_edge_args_substructures): Remove declaration.
26477 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26478 (ipa_edge_args_info_available_for_edge_p): Likewise.
26479 * ipa-prop.c (ipa_edge_args_vector): Removed.
26480 (edge_removal_hook_holder): Likewise.
26481 (edge_duplication_hook_holder): Likewise.
26482 (ipa_edge_args_sum): New variable.
26483 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26484 ipa_edge_args_vector.
26485 (ipa_free_edge_args_substructures): Likewise.
26486 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26487 ipa_edge_args_vector.
26488 (ipa_edge_removal_hook): Turned into method
26489 ipa_edge_args_sum_t::remove.
26490 (ipa_edge_duplication_hook): Turned into method
26491 ipa_edge_args_sum_t::duplicate.
26492 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26493 registering edge hooks.
26494 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26495 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26496 ipa_edge_args_sum instead of ipa_edge_args_vector.
26497 * ipa-profile.c (ipa_profile): Likewise.
26499 2017-05-03 Martin Jambor <mjambor@suse.cz>
26501 * symbol-summary.h (function_summary): New method exists.
26502 (function_summary::symtab_removal): Deallocate through release.
26503 (call_summary): New class.
26504 (gt_ggc_mx): New overload.
26505 (gt_pch_nx): Likewise.
26506 (gt_pch_nx): Likewise.
26508 2017-05-03 Jeff Law <law@redhat.com>
26510 PR tree-optimization/78496
26511 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26512 from simplify_cond_using_ranges. Split off code to walk
26513 backwards through casts into ...
26514 (simplify_cond_using_ranges_2): New function.
26515 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26516 (execute_vrp): After identifying jump threads, call
26517 simplify_cond_using_ranges_2.
26519 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26522 * ipa-inline.h (inline_summary): Add ctor.
26523 (create_ggc): Do not use ggc_cleared_alloc.
26525 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26526 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26528 * gcc.c (handle_braces): Support escaping in switch matching
26530 * doc/invoke.texi (Spec Files): Document it.
26531 Remove superfluous @code markup in items.
26533 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26535 * diagnostic-show-locus.c (struct column_range): New struct.
26536 (get_affected_columns): New function.
26537 (get_printed_columns): New function.
26538 (struct correction): New struct.
26539 (correction::ensure_capacity): New function.
26540 (correction::ensure_terminated): New function.
26541 (struct line_corrections): New struct.
26542 (line_corrections::~line_corrections): New dtor.
26543 (line_corrections::add_hint): New function.
26544 (layout::print_trailing_fixits): Reimplement in terms of the new
26546 (selftest::test_overlapped_fixit_printing): New function.
26547 (selftest::diagnostic_show_locus_c_tests): Call it.
26549 2017-05-03 Nathan Sidwell <nathan@acm.org>
26551 Canonicalize canonical type hashing
26552 * tree.h (type_hash_canon_hash): Declare.
26553 * tree.c (type_hash_list, attribute_hash_list): Move into
26554 type_hash_canon_hash.
26555 (build_type_attribute_qual_variant): Break out hash code calc into
26556 type_hash_canon_hash.
26557 (type_hash_canon_hash): New. Generic type hash computation.
26558 (build_range_type_1, build_array_type_1, build_function_type,
26559 build_method_type_directly, build_offset_type, build_complex_type,
26560 make_vector_type): Call it.
26562 2017-05-03 Richard Biener <rguenther@suse.de>
26564 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26565 When all DRs have unknown misaligned do not always peel
26566 when there is a store but apply the same costing model as if
26567 there were only loads.
26569 2017-05-03 Richard Biener <rguenther@suse.de>
26572 PR tree-optimization/80492
26573 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26574 compare_base_decls returning dont-know properly.
26576 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26578 * config/arm/iterators.md (CCSI): New mode iterator.
26579 (arch): New mode attribute.
26580 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26581 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26582 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26583 code iterator for success result mode.
26584 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26585 the corresponding new insn generators.
26587 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26590 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26591 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26593 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26595 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26596 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26597 (DDR_A): Wrap DDR argument in brackets.
26598 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26599 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26600 (DDR_REVERSED_P): Likewise.
26602 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26604 PR tree-optimization/79472
26605 * tree-switch-conversion.c (struct switch_conv_info): Add
26606 contiguous_range and default_case_nonstandard fields.
26607 (collect_switch_conv_info): Compute contiguous_range and
26608 default_case_nonstandard fields, don't clear final_bb if
26609 contiguous_range and only the default case doesn't have the required
26611 (check_all_empty_except_final): Set default_case_nonstandard instead
26612 of failing if contiguous_range and the default case doesn't have empty
26614 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26615 and only the default case doesn't have the required constants. Skip
26617 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26618 if default_case_nonstandard.
26619 (build_constructors): Build constant 1 just once. Assert that default
26620 values aren't inserted in between cases if contiguous_range. Skip
26622 (build_arrays): Skip virtual phis.
26623 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26624 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26625 Handle virtual phis.
26626 (gen_inbound_check): Handle default_case_nonstandard case.
26627 (process_switch): Adjust check_final_bb caller. Call
26628 gather_default_values with the first non-default case instead of
26629 default case if default_case_nonstandard.
26631 2017-05-02 Nathan Sidwell <nathan@acm.org>
26633 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26634 check. Fix formatting.
26636 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26638 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26639 errors when comparing specialized and unspecialized times.
26641 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26643 * diagnostic-show-locus.c
26644 (layout::should_print_annotation_line_p): Make private.
26645 (layout::print_annotation_line): Make private.
26646 (layout::annotation_line_showed_range_p): Make private.
26647 (layout::show_ruler): Make private.
26648 (layout::print_source_line): Make private. Pass in line and
26649 line_width, rather than calling location_get_source_line. Drop
26651 (layout::print_leading_fixits): New method.
26652 (layout::print_any_fixits): Rename to...
26653 (layout::print_trailing_fixits): ...this, and make private.
26654 Don't print newline fixits.
26655 (diagnostic_show_locus): Move logic for printing one row into...
26656 (layout::print_line): ...this new function. Move the
26657 location_get_source_line call and error-handling from
26658 print_source_line to here. Call print_leading_fixits, and rename
26659 print_any_fixits to print_trailing_fixits.
26660 (selftest::test_fixit_insert_containing_newline): Update now that
26661 newlines are partially supported.
26662 (selftest::test_fixit_insert_containing_newline_2): New test.
26663 (selftest::test_fixit_replace_containing_newline): Update comments.
26664 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26665 * edit-context.c (class added_line): New class.
26666 (class edited_line): Describe newline handling in comment.
26667 (edited_line::actually_edited_p): New method.
26668 (edited_line::print_content): Delete redundant decl.
26669 (edited_line::m_predecessors): New field.
26670 (edited_file::print_content): Call edited_line::print_content.
26671 (edited_file::print_diff): Update to support newlines.
26672 (edited_file::print_diff_hunk): Likewise.
26673 (edited_file::print_run_of_changed_lines): New function.
26674 (edited_file::print_diff_line): Convert to...
26675 (print_diff_line): ...this.
26676 (edited_file::get_effective_line_count): New function.
26677 (edited_line::edited_line): Initialize new field m_predecessors.
26678 (edited_line::~edited_line): Clean up m_predecessors.
26679 (edited_line::apply_fixit): Handle newlines.
26680 (edited_line::get_effective_line_count): New function.
26681 (edited_line::print_content): New function.
26682 (edited_line::print_diff_lines): New function.
26683 (selftest::test_applying_fixits_insert_containing_newline): New
26685 (selftest::test_applying_fixits_replace_containing_newline): New
26687 (selftest::insert_line): New function.
26688 (selftest::test_applying_fixits_multiple_lines): Add example of
26690 (selftest::edit_context_c_tests): Call the new tests.
26692 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26694 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26695 parameter cand. Update dump information.
26696 (get_computation_cost): Update uses.
26698 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26700 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26701 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26702 (get_computation_at, rewrite_use_address): Update use of
26703 get_computation_aff.
26705 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26707 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26708 (get_computation): Delete.
26709 (get_computation_cost): Implement like get_computation_cost_at.
26710 Use get_computation_at.
26711 (get_computation_cost_at): Delete.
26712 (rewrite_use_nonlinear_expr): Use get_computation_at.
26713 (rewrite_use_compare, remove_unused_ivs): Ditto.
26715 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26717 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26719 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26721 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26722 (ivopts_global_cost_for_size): Rename parameter and update uses.
26723 (iv_ca_recount_cost): Update uses.
26724 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26725 candidates seperately in n_invs and n_cands.
26726 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26728 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26730 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26731 (find_inv_vars_cb): New.
26732 (find_depends): Renamed to ...
26733 (find_inv_vars): ... this.
26734 (add_candidate_1, force_var_cost): Call find_inv_vars.
26735 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26737 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26739 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26740 inv_vars. Add inv_exprs.
26741 (struct iv_cand): Rename depends_on to inv_vars.
26742 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26743 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26744 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26745 (dump_cand): Dump inv_vars.
26746 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26747 (record_invariant, find_depends, add_candidate_1): Ditto.
26748 (set_group_iv_cost, force_var_cost): Ditto.
26749 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26750 (get_computation_cost_at, get_computation_cost): Ditto.
26751 (determine_group_iv_cost_generic): Ditto.
26752 (determine_group_iv_cost_address): Ditto.
26753 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26754 (determine_group_iv_costs): Ditto.
26755 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26756 (iv_ca_set_remove_invariants): Renamed to ...
26757 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26758 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26759 (iv_ca_set_add_invariants): Renamed to ...
26760 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26761 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26762 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26763 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26764 (create_new_ivs): Remove useless dump.
26766 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26768 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26770 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26771 (iv_ca_set_no_cp, create_new_iv): Ditto.
26773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26775 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26777 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26779 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26780 function tree_check2.
26782 2017-05-02 Martin Liska <mliska@suse.cz>
26784 * doc/gcov.texi: Add missing preposition.
26785 * gcov.c (function_info::function_info): Properly fill up
26786 all member variables.
26788 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26790 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26792 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26794 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26796 2017-05-02 Martin Liska <mliska@suse.cz>
26799 * lto-streamer-in.c (lto_read_tree_1): Remove
26800 LTO_STREAMER_DEBUG.
26801 * lto-streamer.c (struct tree_hash_entry): Likewise.
26802 (struct tree_entry_hasher): Likewise.
26803 (tree_entry_hasher::hash): Likewise.
26804 (tree_entry_hasher::equal): Likewise.
26805 (lto_streamer_init): Likewise.
26806 (lto_orig_address_map): Likewise.
26807 (lto_orig_address_get): Likewise.
26808 (lto_orig_address_remove): Likewise.
26809 * lto-streamer.h: Likewise.
26810 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26811 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26813 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26815 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26816 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26817 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26818 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26819 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26820 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26821 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26822 (mm_maskz_sub_ss): New intrinsics.
26823 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26824 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26825 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26826 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26827 (__builtin_ia32_subss_mask_round): New builtins.
26828 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26829 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26830 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26832 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26833 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26835 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26838 2017-05-02 Martin Jambor <mjambor@suse.cz>
26840 PR tree-optimization/78687
26841 * tree-sra.c (access): New field parent.
26842 (process_subtree_disqualification): New function.
26843 (disqualify_candidate): Call it.
26844 (build_accesses_from_assign): Reset write flag if creating an
26846 (build_access_subtree): Fill in parent field and also prpagate
26847 down grp_write flag.
26848 (create_artificial_child_access): New parameter set_grp_write, set
26849 grp_write to its value.
26850 (propagate_subaccesses_across_link): Also propagate grp_write flag
26852 (propagate_all_subaccesses): Push the closest parent back to work
26853 queue if add_access_to_work_queue returned true.
26855 2017-05-02 Richard Biener <rguenther@suse.de>
26857 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26858 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26859 -fstrict-overflow documentation.
26860 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26861 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26862 flag_strict_overflow.
26863 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26864 * lto-opts.c (lto_write_options): Do not stream it.
26865 * lto-wrapper.c (merge_and_complain): Do not handle it.
26866 * opts.c (default_options_table): Do not set -fstrict-overflow.
26867 (finish_options): Likewise do not clear it when sanitizing.
26868 * simplify-rtx.c (simplify_const_relational_operation): Do not
26869 test flag_strict_overflow.
26871 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26873 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26874 using enabled attribute.
26875 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26876 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26877 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26878 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26879 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26880 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26881 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26882 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26883 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26884 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26886 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26888 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26890 2017-05-02 Richard Biener <rguenther@suse.de>
26892 PR tree-optimization/80591
26894 2017-04-10 Richard Biener <rguenther@suse.de>
26896 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26899 2017-05-02 Richard Biener <rguenther@suse.de>
26901 PR tree-optimization/80549
26902 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26903 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26904 headers if they do not have a preheader.
26906 2017-05-02 Martin Liska <mliska@suse.cz>
26909 * common.opt: Fix typo.
26910 * doc/invoke.texi: Likewise.
26912 2017-05-01 Jan Beulich <jbeulich@suse.com>
26914 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26915 swapping, add (x,x,m,x,n) alternative.
26917 2017-05-01 Nathan Sidwell <nathan@acm.org>
26919 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26920 unnecessary unadjusted_alignment check.
26922 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26925 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26927 * gimplify.c (gimplify_cilk_detach): New function.
26928 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26929 * tree-core.h: Document EXPR_CILK_SPAWN.
26930 * tree.h (EXPR_CILK_SPAWN): Define.
26932 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26934 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26935 to use new fixit_hint representation, using the "replace" logic.
26936 (get_line_span_for_fixit_hint): Likewise.
26937 (layout::print_any_fixits): Likewise.
26938 (selftest::test_one_liner_many_fixits): Rename to...
26939 (selftest::test_one_liner_many_fixits_1): ...this, and update
26940 comment and expected output to reflect that the multiple fix-it
26941 hints are now consolidated into one insertion.
26942 (selftest::test_one_liner_many_fixits_2): New test.
26943 (selftest::test_diagnostic_show_locus_one_liner): Update for
26945 (selftest::test_fixit_consolidation): Update for fix-it API
26947 * diagnostic.c (print_parseable_fixits): Likewise.
26948 * edit-context.c (edited_line::m_line_events): Convert from
26949 auto_vec <line_event *> to auto_vec <line_event>.
26950 (class line_event): Convert from abstract base class to a concrete
26951 class, taking over the role of replace_event.
26952 (class insert_event): Delete.
26953 (class replace_event): Rename to class line_event. Convert to
26955 (edit_context::add_fixits): Reimplement.
26956 (edit_context::apply_insert): Delete.
26957 (edit_context::apply_replace): Rename to...
26958 (edit_context::apply_fixit): ...this. Convert to half-open range.
26959 (edited_file::apply_insert): Delete.
26960 (edited_file::apply_replace): Rename to...
26961 (edited_file::apply_fixit): ...this.
26962 (edited_line::~edited_line): Drop deletion of events.
26963 (edited_line::apply_insert): Delete.
26964 (edited_line::apply_replace): Rename to...
26965 (edited_line::apply_fixit): ...this. Convert to half-open range.
26966 Update for change to type of m_line_events.
26967 * edit-context.h (edit_context::apply_insert): Delete.
26968 (edit_context::apply_replace): Rename to...
26969 (edit_context::apply_fixit): ...this.
26971 2017-05-01 Martin Sebor <msebor@redhat.com>
26973 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26976 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26979 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26980 __get_cpuid_max returns 0.
26981 (__get_cpuid_count): Ditto.
26983 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26985 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26986 replacement expression is another instance of one of its arguments.
26988 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26991 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26992 check for stack push/pop autoinc.
26993 * config/i386/i386.c (ix86_agi_dependent): Return false
26994 if the only reason why modified_in_p returned true is that
26995 addr is SP based and set_insn is a push or pop.
26997 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26999 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
27002 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
27005 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
27006 (account_size_time): Use two predicates - exec_pred and
27008 (evaluate_conditions_for_known_args): Compute both clause and
27010 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
27011 (inline_summary_t::duplicate): Update.
27012 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
27014 (compute_inline_parameters): Likewise.
27015 (estimate_edge_size_and_time): Update caluclation of time.
27016 (estimate_node_size_and_time): Compute both time and nonspecialized
27018 (estimate_ipcp_clone_size_and_time): Update.
27019 (inline_merge_summary): Update.
27020 (do_estimate_edge_time): Update.
27021 (do_estimate_edge_size): Update.
27022 (do_estimate_edge_hints): Update.
27023 (inline_read_section, inline_write_summary): Stream both new predicates.
27024 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
27026 (compute_inlined_call_time): Cleanup.
27027 (big_speedup_p): Update.
27028 (edge_badness): Update.
27029 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
27030 (size_time_entry): Replace predicate by exec_predicate and
27031 nonconst_predicate.
27032 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
27033 (estimate_edge_time): Return also nonspec_time.
27034 (reset_edge_growth_cache): Update.
27036 2017-04-29 Jakub Jelinek <jakub@redhat.com>
27038 PR rtl-optimization/80491
27039 * ifcvt.c (noce_process_if_block): When looking for x setter
27040 with missing else_bb, don't check only the insn right before
27041 cond_earliest, but look for the last insn that x is modified in
27042 within the same bb.
27044 PR rtl-optimization/80491
27045 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
27047 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
27049 PR tree-optimization/80487
27050 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
27052 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27054 PR tree-optimization/79697
27055 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
27056 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
27057 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
27059 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
27060 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
27062 2017-04-28 Martin Sebor <msebor@redhat.com>
27064 PR tree-optimization/80523
27065 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
27066 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27068 (maybe_warn, format_directive, parse_directive): Use new functions.
27069 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
27071 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
27073 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
27075 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
27077 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
27078 target_header_dir): Set correctly.
27079 * configure: Regenerated.
27080 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27081 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27082 instead of SYSTEM_HEADER_DIR.
27084 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
27086 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
27087 (estimate_local_effects): Likewise.
27088 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
27089 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
27090 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
27091 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
27092 do_estimate_edge_time, estimate_edge_time): Likewise.
27093 * ipa-inline-analysis.c (estimate_node_size_and_time,
27094 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
27095 (estimate_time_after_inlining): Remove.
27097 2017-04-28 Martin Liska <mliska@suse.cz>
27099 * doc/gcov.texi: Enhance documentation of gcov.
27101 2017-04-28 Martin Liska <mliska@suse.cz>
27103 * doc/gcov.texi: Sort options in alphabetic order.
27104 * doc/gcov-dump.texi: Likewise.
27105 * doc/gcov-tool.texi: Likewise.
27106 * gcov.c (print_usage): Likewise.
27107 * gcov-dump.c (print_usage): Likewise.
27108 * gcov-tool.c (print_merge_usage_message): Likewise.
27109 (print_rewrite_usage_message): Likewise.
27110 (print_overlap_usage_message): Likewise.
27112 2017-04-28 Martin Liska <mliska@suse.cz>
27114 PR gcov-profile/53915
27115 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27117 2017-04-28 Martin Liska <mliska@suse.cz>
27119 PR gcov-profile/79891
27120 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27121 is marked by compiler as living on a line.
27122 (get_cycles_count): Remove usage of the union.
27123 (output_intermediate_file): Likewise.
27124 (find_source): Fix GNU coding style.
27125 (accumulate_line_counts): Remove old non-all block mode.
27126 (output_lines): Remove usage of the union.
27127 * profile.c (output_location): Include all BBs, even if
27128 belonging to a same line (and file) as a previous BB.
27130 2017-04-28 Martin Liska <mliska@suse.cz>
27132 * gcov.c (process_args): Handle new argument 'w'.
27133 (read_graph_file): Assign ID to BBs.
27134 (output_branch_count): Display BB # if verbose flag is set.
27135 (output_lines): Likewise for arcs.
27136 (print_usage): Add '--verbose' option help.
27137 * doc/gcov.texi: Document --verbose (-w) option.
27139 2017-04-28 Martin Liska <mliska@suse.cz>
27141 * gcov.c (struct block_location_info): New struct.
27142 (process_file): Fill up the new structure.
27143 (read_graph_file): Replace usage of encoding by the newly added
27145 (add_line_counts): Likewise.
27146 (accumulate_line_counts): Remove usage of the union.
27147 (function_info::function_info): New function.
27148 (function_info::~function_info): Likewise.
27149 (process_file): Call delete instead of release_function.
27150 (release_function): Release the function.
27151 (release_structures): Call delete instead of release_function.
27152 (solve_flow_graph): Replace usage of num_blocks.
27153 (find_exception_blocks): Likewise.
27154 (output_lines): Fix GNU coding style.
27156 2017-04-28 Martin Liska <mliska@suse.cz>
27159 * coverage.c (coverage_remove_note_file): New function.
27160 * coverage.h: Declare the function.
27161 * toplev.c (finalize): Clean if an error has been seen.
27163 2017-04-28 Martin Liska <mliska@suse.cz>
27165 PR gcov-profile/80031
27166 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27167 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27168 * gcov.c (read_graph_file): Read just number of blocks.
27169 * profile.c (branch_prob): Do not stream 0 flags per a basic
27172 2017-04-28 Martin Liska <mliska@suse.cz>
27174 * gcov-dump.c (tag_*): Add new argument to declarations.
27175 (dump_gcov_file): Likewise.
27176 (tag_blocks): Add and use new argument depth.
27177 (tag_arcs): Likewise.
27178 (tag_lines): Likewise.
27179 (tag_counters): Likewise.
27180 (tag_summary): Likewise.
27181 (dump_working_sets): Use depth to do a proper indentation.
27183 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27186 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27187 * symtab.c (symtab_node::debug_symtab): Move definition here.
27189 2017-04-28 Richard Biener <rguenther@suse.de>
27191 * lto-streamer.h (LTO_major_version): Bump to 7.
27193 2017-04-28 Richard Biener <rguenther@suse.de>
27195 * tree-vrp.c (assert_info): New struct.
27196 (add_assert_info): New helper.
27197 (register_edge_assert_for_2): Refactor to add asserts to a vector
27199 (register_edge_assert_for_1): Likewise.
27200 (register_edge_assert_for): Likewise.
27201 (finish_register_edge_assert_for): New helper actually registering
27202 asserts where live on edge.
27203 (find_conditional_asserts): Adjust.
27204 (find_switch_asserts): Likewise.
27205 (evrp_dom_walker::try_find_new_range): Generalize.
27206 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27208 2017-04-27 Marek Polacek <polacek@redhat.com>
27211 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27212 arg10 and arg11 to itype.
27214 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27216 * doc/extend.texi (Object Size Checking): Improve grammar.
27218 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27221 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27222 that the logic for permitting reciprocal estimates matches that
27225 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27228 * tree.c (type_cache_hasher::equal): Only compare
27229 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27230 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27231 non-aggregate element types.
27232 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27233 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27235 2017-04-27 Richard Biener <rguenther@suse.de>
27237 PR middle-end/80533
27238 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27239 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27240 keeping a reference to a trailing array.
27242 2017-04-27 Richard Biener <rguenther@suse.de>
27244 PR middle-end/80539
27245 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27246 being in loop-closed SSA form conservatively.
27247 (chrec_fold_multiply_poly_poly): Likewise.
27249 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27251 PR middle-end/79665
27252 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27253 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27255 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27258 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27259 (aarch64_function_arg_alignment): Return unsigned int again, but still
27260 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27261 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27262 Don't emit -Wpsabi note.
27263 (aarch64_function_arg_boundary): Likewise.
27264 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27267 2017-04-26 Nathan Sidwell <nathan@acm.org>
27269 * tree.h (crc32_unsigned_n): Declare.
27270 (crc32_unsigned, crc32_unsigned): Make inline.
27271 * tree.c (crc32_unsigned_bits): Replace with ...
27272 (crc32_unsigned_n): ... this.
27273 (crc32_unsigned, crc32_byte): Remove.
27274 (crc32_string): Remove unnecessary braces.
27276 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27278 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27279 * ipa-inline-analysis.c (MAX_TIME): Remove.
27280 (account_size_time): Use sreal for time.
27281 (dump_inline_summary): Update.
27282 (estimate_function_body_sizes): Update.
27283 (estimate_edge_size_and_time): Update.
27284 (estimate_calls_size_and_time): Update.
27285 (estimate_node_size_and_time): Update.
27286 (inline_merge_summary): Update.
27287 (inline_update_overall_summary): Update.
27288 (estimate_time_after_inlining): Update.
27289 (inline_read_section): Update.
27290 (inline_write_summary): Update.
27291 * ipa-inline.c (compute_uninlined_call_time): Update.
27292 (compute_inlined_call_time): Update.
27293 (recursive_inlining): Update.
27294 (inline_small_functions): Update.
27295 (dump_overall_stats): Update.
27296 * ipa-inline.h: Include sreal.h.
27297 (size_time_entry): Turn time to sreal.
27298 (inline_summary): Turn self_time nad time to sreal.
27300 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27302 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27304 (sreal::stream_out, sreal::stream_in): New.
27305 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27307 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27309 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27312 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27315 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27317 (dimode_scalar_chain::compute_convert_gain): Ditto.
27318 (dimode_scalar_chain::make_vector_copies): Ditto.
27319 (dimode_scalar_chain::convert_reg): Ditto.
27320 (dimode_scalar_chain::convert_insn): Ditto.
27321 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27322 (VI248_AVX512BW_1): New mode iterator.
27323 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27324 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27327 2017-04-25 Martin Sebor <msebor@redhat.com>
27329 PR tree-optimization/80497
27330 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27331 constants are representable in HOST_WIDE_INT.
27332 (parse_directive): Ditto.
27334 2017-04-25 Martin Sebor <msebor@redhat.com>
27337 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27338 (new_zero_array): Adjust signature.
27339 (dom_info::dom_init): Used unsigned rather that size_t.
27340 (dom_info::dom_info): Same.
27342 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27343 Jakub Jelinek <jakub@redhat.com>
27346 * config/arm/arm.c: Include gimple.h.
27347 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27348 returns negative, increment ncrn only if it returned positive.
27349 (arm_needs_doubleword_align): Return int instead of bool,
27350 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27351 members, but if there is any such non-FIELD_DECL
27352 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27353 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27354 returns negative, increment nregs only if it returned positive.
27355 (arm_setup_incoming_varargs): Likewise.
27356 (arm_function_arg_boundary): Emit -Wpsabi note if
27357 arm_needs_doubleword_align returns negative, return
27358 DOUBLEWORD_ALIGNMENT only if it returned positive.
27360 2017-04-25 Marek Polacek <polacek@redhat.com>
27363 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27364 first argument to type.
27366 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27369 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27370 type checks to test for compatibility instead of equality.
27372 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27373 Jakub Jelinek <jakub@redhat.com>
27376 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27378 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27379 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27380 the alignment computation, but return their maximum in warn_alignment.
27381 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27382 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27384 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27385 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27388 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27390 * config/arc/simdext.md (dmpyh): Fix typo.
27392 2017-04-25 Richard Biener <rguenther@suse.de>
27394 PR tree-optimization/80492
27395 * alias.c (compare_base_decls): Handle registers with asm
27396 specification conservatively.
27397 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27398 compare_base_decls returning dont-know properly.
27400 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27402 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27403 (legitimate_offset_address_p): New function.
27404 (arc_legitimate_address_p): Use above function.
27406 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27408 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27410 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27412 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27413 ACCH registers whenever they are available.
27415 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27417 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27418 double regs fix when not used.
27420 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27422 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27424 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27425 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27427 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27429 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27430 when emitting short ADD instructions.
27432 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27434 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27436 (cmpsi_cc_c_insn): Likewise.
27437 (cbranchsi4_scratch): Compute proper instruction length using
27438 compact_hreg_operand.
27439 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27441 2017-04-25 Richard Biener <rguenther@suse.de>
27443 PR middle-end/80509
27444 * passes.c (pass_manager::pass_manager): Initialize
27445 m_name_to_pass_map.
27447 2017-04-25 Richard Biener <rguenther@suse.de>
27449 PR tree-optimization/79201
27450 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27452 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27455 * config/s390/vector.md: Split MEM->GPR vector moves for
27456 non-s_operand addresses.
27458 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27461 * config/s390/predicates.md (reload_const_wide_int_operand): New
27463 * config/s390/s390.md ("movti"): Remove d/P alternative.
27464 ("movti_bigconst"): New pattern definition.
27466 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27469 * s390-protos.h (s390_expand_cs_hqi): Removed.
27470 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27471 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27472 modes as well as CCZ1mode and CCZmode.
27473 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27474 signature of s390_emit_compare_and_swap.
27475 (s390_expand_cs_hqi): Likewise, make static.
27476 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27477 compare-and-swap, in some cases.
27478 (s390_expand_cs): Wrapper function.
27479 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27481 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27482 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27483 patterns for small and large integers. Forbid symref memory operands.
27484 Move expander to s390.c. Require cc register.
27485 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27486 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27487 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27488 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27489 symref memory operands. Remove CC mode and call s390_match_ccmode
27491 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27493 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27495 * config/s390/s390.md (define_peephole2): New peephole to help
27496 combining the load-and-test pattern with volatile memory.
27498 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27500 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27501 with CCZmode for TARGET_Z196.
27503 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27505 PR rtl-optimization/80501
27506 * combine.c (make_compound_operation_int): Set subreg_code to SET
27507 even for AND with mask of the sign bit of mode.
27509 PR rtl-optimization/80500
27510 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27511 sum's initial value.
27513 2017-04-25 Julian Brown <julian@codesourcery.com>
27514 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27516 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27518 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27520 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27522 2017-04-25 Julian Brown <julian@codesourcery.com>
27523 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27525 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27526 (thunderx2t99_sha): New Reservation.
27528 2017-04-25 Julian Brown <julian@codesourcery.com>
27529 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27531 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27532 type for 1-element load.
27534 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27536 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27538 2017-04-24 Martin Jambor <mjambor@suse.cz>
27540 PR tree-optimization/80293
27541 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27542 char arrays not totally scalarizable if it is false.
27543 (analyze_all_variable_accesses): Pass correct value in the new
27544 parameter. Add a statistics counter.
27546 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27548 PR middle-end/79931
27549 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27551 2017-04-24 Richard Biener <rguenther@suse.de>
27553 PR tree-optimization/80494
27554 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27555 out for complex types.
27557 2017-04-24 Richard Biener <rguenther@suse.de>
27559 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27560 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27561 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27562 (DFS): Adjust and never fail.
27563 (sccvn_dom_walker::fail): Remove.
27564 (sccvn_dom_walker::before_dom_children): Adjust.
27565 (run_scc_vn): Likewise and never fail.
27566 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27567 (pass_fre::execute): Likewise.
27569 2017-04-24 Richard Biener <rguenther@suse.de>
27571 PR tree-optimization/79725
27572 * tree-ssa-sink.c (statement_sink_location): Return whether
27573 failure reason was zero uses. Move that check later.
27574 (sink_code_in_bb): Deal with zero uses by removing the stmt
27577 2017-04-24 Richard Biener <rguenther@suse.de>
27580 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27581 pointer-based references.
27583 2017-04-24 Richard Biener <rguenther@suse.de>
27586 * pass_manager.h (pass_manager::operator new): Remove.
27587 (pass_manager::operator delete): Likewise.
27588 * passes.c (pass_manager::operator new): Remove.
27589 (pass_manager::operator delete): Likewise.
27590 (pass_manager::pass_manager): Zero individual pass members.
27592 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27595 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27596 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27597 Check "XEXP (src, 1)" operand here.
27598 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27599 Check "XEXP (src, 1)" operand here.
27600 (dimode_scalar_chain::make_vector_copies): Detect count register
27601 of a shift instruction. Zero extend count register from QImode
27602 to DImode to satisfy vector shift pattern count operand predicate.
27603 Substitute vector shift count operand with a DImode copy.
27604 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27607 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27609 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27610 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27611 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27612 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27613 (UNSPEC_NOREX_MEM): Remove definition.
27615 2017-04-21 Richard Biener <rguenther@suse.de>
27617 PR tree-optimization/79547
27618 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27619 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27620 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27621 without any constraints.
27623 2017-04-21 Richard Biener <rguenther@suse.de>
27625 PR tree-optimization/78847
27626 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27628 2017-04-21 Richard Biener <rguenther@suse.de>
27630 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27631 (build_distinct_type_copy): Likewise.
27632 (build_variant_type_copy): Likewise.
27633 * tree.c (build_qualified_type): Pass down mem-stat info.
27634 (build_distinct_type_copy): Likewise.
27635 (build_variant_type_copy): Likewise.
27637 2017-04-21 Richard Biener <rguenther@suse.de>
27639 PR tree-optimization/80237
27640 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27642 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27643 for a simplified result.
27645 2016-04-21 Richard Biener <rguenther@suse.de>
27647 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27648 sth as strict as a simple_iv but a chrec without symbols and an
27649 operand defined in the loop we are peeling (and not some subloop).
27650 (propagate_constants_for_unrolling): Propagate all constants.
27652 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27655 * config/i386/i386.c (print_reg): Remove assert for disalowed
27656 regno values, call output_operand_lossage instead.
27658 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27661 * config/i386/constraints.md (Yc): New register constraint.
27662 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27663 Use Yc constraint for alternative 2 of operand 0. Remove
27664 preferred_for_speed attribute.
27666 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27668 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27669 lastprivate clauses in SIMT case.
27671 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27673 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27675 2017-04-20 Richard Biener <rguenther@suse.de>
27677 PR tree-optimization/57796
27678 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27679 as N scalar stores.
27680 (vect_model_load_cost): Cost gathers as N scalar loads.
27682 2017-04-20 Richard Biener <rguenther@suse.de>
27684 * ggc-page.c (ggc_allocated_p): Rename to ...
27685 (safe_lookup_page_table_entry): ... this and return the lookup
27687 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27689 2017-04-20 Richard Biener <rguenther@suse.de>
27691 PR tree-optimization/80453
27692 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27693 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27694 from the conditions.
27695 (vn_phi_eq): Pass them down.
27696 (vn_phi_lookup): Record them.
27697 (vn_phi_insert): Likewise.
27699 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27701 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27702 uninitialized variable warning to avoid buffer overrun.
27704 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27707 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27708 is suppressed for '{ 0 }' in C.
27710 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27712 * BASE-VER: Set to 8.0.0.
27714 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27716 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27717 priority .init_array and .fini_array section with SECTION_NOTYPE
27720 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27722 PR middle-end/80423
27723 * tree.h (build_array_type): Add typeless_storage default argument.
27724 * tree.c (type_cache_hasher::equal): Also compare
27725 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27726 (build_array_type): Add typeless_storage argument, set
27727 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27729 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27730 (build_array_type): Likewise. Add typeless_storage argument.
27732 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27733 Jakub Jelinek <jakub@redhat.com>
27735 PR tree-optimization/80426
27736 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27737 operation on symbolic operands, also compute the overflow for the
27738 invariant part when the operation degenerates into a negation.
27740 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27743 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27744 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27747 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27749 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27752 * config/avr/avr.c (tree.h): Include it.
27753 (cgraph.h): Include it.
27754 (avr_encode_section_info): Don't warn for uninitialized progmem
27755 variable if it's just an alias.
27757 2017-04-19 Richard Biener <rguenther@suse.de>
27760 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27761 when needed by AutoPGO.
27763 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27766 * doc/lto.texi: Remove an extra 'that'.
27768 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27770 PR rtl-optimization/80429
27771 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27772 are only used in debug insns.
27774 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27775 Vladimir Makarov <vmakarov@redhat.com>
27777 * config/sparc/predicates.md (input_operand): Add comment. Return
27778 true for any memory operand when LRA is in progress.
27779 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27781 2017-04-18 Jeff Law <law@redhat.com>
27784 * mips.md ({return,simple_return}_internal): Do not overwrite
27787 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27789 PR tree-optimization/80443
27790 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27791 instead of adding 1, subtract -1 and similarly instead of subtracting
27794 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27796 PR rtl-optimization/80357
27797 * haifa-sched.c (tmp_bitmap): New variable.
27798 (model_recompute): Handle duplicate use records.
27799 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27800 (free_global_sched_pressure_data): Free it.
27802 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27805 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27806 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27807 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27808 instead of SYSTEM_HEADER_DIR.
27810 2017-04-18 Jeff Law <law@redhat.com>
27812 PR middle-end/80422
27813 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27814 predecessors after walking up the insn chain.
27816 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27819 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27820 sizetype type into debug info.
27822 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27825 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27826 unneeded test for TARGET_UPPER_REGS_SF.
27827 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27829 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27832 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27833 instead of gsi_after_labels.
27835 2017-04-18 Jeff Law <law@redhat.com>
27837 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27841 2017-04-13 Jeff Law <law@redhat.com>
27842 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27843 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27845 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27848 * config/avr/avr.c (intl.h): Include it.
27849 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27851 2017-04-18 Martin Liska <mliska@suse.cz>
27853 PR gcov-profile/78783
27854 * gcov-tool.c (gcov_output_files): Validate that destination
27855 file is either removed by the tool or by a user.
27857 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27858 Guy Benyei <guybe@mellanox.com>
27860 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27861 block, and do not negate it, the stored id is already negative.
27863 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27865 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27867 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27870 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27871 masks of options that should be turned off if the VSX vector
27872 options are turned off.
27873 (OTHER_P8_VECTOR_MASKS): Likewise.
27874 (OTHER_VSX_VECTOR_MASKS): Likewise.
27875 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27876 rs6000_disable_incompatible_switches to validate no type switches
27878 (rs6000_incompatible_switch): New function to disallow turning on
27879 other vector options if -mno-vsx, -mno-power8-vector, or
27880 -mno-power9-vector are specified.
27882 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27884 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27886 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27888 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27889 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27890 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27891 (ARG_POINTER_CFA_OFFSET): Likewise.
27893 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27895 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27896 conditions to take advantage of various optimizations.
27898 2017-04-13 Jeff Law <law@redhat.com>
27900 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27901 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27902 (zero_extendsidi2_dext): Likewise.
27904 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27907 * fold-const.c (fold_ternary_loc): Revert
27908 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27911 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27913 PR rtl-optimization/80343
27914 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27915 new scratch pseudo.
27917 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27920 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27921 to ubsan_encode_value.
27923 2017-04-13 Jeff Law <law@redhat.com>
27925 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27926 appearing in DEBUG_INSNs.
27928 2017-04-13 Martin Liska <mliska@suse.cz>
27930 PR gcov-profile/80413
27931 * gcov-io.c (gcov_write_string): Copy to buffer just when
27932 allocated size is greater than zero.
27934 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27937 * dwarf2out.c (decls_for_scope): Ignore declarations of
27938 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27940 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27943 * ipa-visibility.c (non_local_p): Fix typos.
27944 (localize_node): When localizing symbol in same comdat group,
27945 dissolve the group only when we know external symbols are going
27947 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27949 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27951 PR tree-optimization/79390
27952 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27953 order does not result in usable sequence, retry with reversed operand
27959 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27960 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27961 op0 instead of fold_convert_loc (loc, type, arg0).
27963 2017-04-12 Jeff Law <law@redhat.com>
27965 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27966 has a delay slot in the generated code.
27968 * config/cris/cris.md (cris_preferred_reload_class): Return
27969 GENNONACR_REGS rather than GENERAL_REGS.
27971 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27974 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27975 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27976 signedness of the result type.
27978 2017-04-12 Richard Biener <rguenther@suse.de>
27979 Jeff Law <law@redhat.com>
27981 PR tree-optimization/80359
27982 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27983 trim stores to TARGET_MEM_REFs.
27985 2017-04-12 Richard Biener <rguenther@suse.de>
27987 PR tree-optimization/79390
27988 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27989 threading case even more.
27991 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27994 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27995 for quad_address_p for TImode, instead of just not indexed_address.
27997 2017-04-12 Richard Biener <rguenther@suse.de>
27998 Bernd Edlinger <bernd.edlinger@hotmail.de>
28000 PR middle-end/79671
28001 * alias.c (component_uses_parent_alias_set_from): Handle
28002 TYPE_TYPELESS_STORAGE.
28003 (get_alias_set): Likewise.
28004 * tree-core.h (tree_type_common): Add typeless_storage flag.
28005 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
28006 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
28007 for types containing members with TYPE_TYPELESS_STORAGE.
28008 (place_field): Likewise.
28009 (layout_type): Likewise for ARRAY_TYPE.
28010 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
28011 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
28012 TYPE_TYPELESS_STORAGE.
28013 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
28015 2017-04-12 Jakub Jelinek <jakub@redhat.com>
28018 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
28019 first argument to type.
28021 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28025 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
28026 CONST0_RTX (mode) rather than const0_rtx where appropriate.
28027 (rs6000_expand_binop_builtin): Likewise.
28028 (rs6000_expand_ternop_builtin): Likewise; also add missing
28029 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
28030 vshasigma built-ins.
28031 * doc/extend.texi: Document that vec_xxpermdi's third argument
28032 must be a constant.
28034 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
28036 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
28037 Use shift_const cost parameter when calculating gain of STV shifts.
28039 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
28041 PR rtl-optimization/70478
28042 * lra-constraints.c (process_alt_operands): Check memory for
28043 disfavoring memory insn operand.
28045 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28047 PR middle-end/80100
28048 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
28049 left shift in unsigned HOST_WIDE_INT type.
28051 PR rtl-optimization/80385
28052 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
28053 (not (neg X)) into (plus X -1) for complex or non-integral modes.
28056 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
28057 if they have any depend clauses.
28059 2017-04-11 Martin Liska <mliska@suse.cz>
28062 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
28063 * ipa-split.c (split_function): Create a local comdat symbol
28064 if caller is in a comdat group.
28066 2017-04-11 Martin Liska <mliska@suse.cz>
28069 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
28072 2017-04-11 Martin Sebor <msebor@redhat.com>
28074 PR middle-end/80364
28075 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
28076 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
28078 (directive::set_width, directive::set_precision, format_character):
28080 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
28083 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
28086 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
28087 conflict, set target->arch_name instead of target->cpu_name.
28089 2017-04-11 Richard Biener <rguenther@suse.de>
28091 PR tree-optimization/80374
28092 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
28093 build_zero_cst, remove fold_convertible_p check again.
28095 2017-04-11 Martin Liska <mliska@suse.cz>
28098 * ubsan.c (instrument_object_size): Do not instrument register
28101 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28104 * config/i386/i386-builtin-types.def
28105 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28106 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28107 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28108 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28109 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28110 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28111 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28112 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28113 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28114 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28115 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28116 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28117 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28118 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28119 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28120 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28121 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28122 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28123 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28124 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28125 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28126 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28127 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28128 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28129 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28130 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28131 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28132 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28133 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28134 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28135 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28136 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28137 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28138 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28139 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28140 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28141 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28142 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28143 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28144 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28145 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28146 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28147 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28148 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28149 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28150 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28152 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28153 flag to second_arg_count, handle 4 argument function type _COUNT
28154 aliases, handle second_arg_count on second argument rather than last.
28156 2017-04-10 Jeff Law <law@redhat.com>
28158 PR tree-optimization/80374
28159 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28160 record anything if we can not convert integer_zero_node to the
28163 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28166 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28167 Enhance special handling given to the TARGET_P9_MINMAX option in
28168 relation to certain other options.
28170 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28172 PR tree-optimization/80153
28173 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28174 remove POINTER_PLUS_EXPR's base part directly, rather than through
28177 2017-04-10 Richard Biener <rguenther@suse.de>
28178 Bin Cheng <bin.cheng@arm.com>
28180 PR tree-optimization/80153
28181 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28182 the first element of pointer type aff_tree. Build result expr in
28184 (add_elt_to_tree): Convert to type unconditionally. Remove other
28185 fold_convert calls.
28186 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28187 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28189 2017-04-10 Richard Biener <rguenther@suse.de>
28191 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28194 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28196 PR rtl-optimization/70478
28197 * lra-constraints.c (curr_small_class_check): New.
28198 (update_and_check_small_class_inputs): New.
28199 (process_alt_operands): Update curr_small_class_check. Disfavor
28200 alternative insn memory operands. Check available regs for small
28203 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28206 * config/mips/mips.opt (-mvirt): Update description.
28207 * doc/invoke.texi (-mvirt): Likewise.
28209 2017-04-10 Richard Biener <rguenther@suse.de>
28211 PR middle-end/80362
28212 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28213 looking for NEGATE_EXPR in -A / -B to A / B folding.
28215 2017-04-10 Martin Liska <mliska@suse.cz>
28217 PR gcov-profile/80224
28218 * gcov.c (print_usage): Fix usage string.
28219 (get_gcov_intermediate_filename): Remove.
28220 (output_gcov_file): Use both for normal and intermediate format.
28221 (generate_results): Do not initialize special file for
28222 intermediate format.
28224 2017-04-10 Richard Biener <rguenther@suse.de>
28226 PR tree-optimization/80304
28227 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28230 2017-04-10 Nathan Sidwell <nathan@acm.org>
28233 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28234 (rs6000_init_builtins): Use it.
28236 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28238 * config/arm/arm.md (<mrc>): Add mode to SET source.
28239 (<mrrc>): Likewise.
28241 2017-04-10 Richard Biener <rguenther@suse.de>
28243 PR middle-end/80344
28244 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28246 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28249 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28250 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28251 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28252 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28253 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28254 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28255 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28256 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28257 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28258 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28259 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28260 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28261 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28262 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28263 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28264 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28265 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28266 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28267 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28268 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28269 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28270 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28271 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28273 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28275 PR rtl-optimization/70478
28276 * lra-constraints.c: Reverse the last patch.
28278 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28280 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28281 Add comment for WCHAR_T.
28283 2017-04-08 Martin Liska <mliska@suse.cz>
28286 2017-04-07 Martin Liska <mliska@suse.cz>
28289 * ipa-split.c (split_function): Add function part to a same comdat
28292 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28295 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28297 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28299 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28300 (vec_pairing): ...to this.
28301 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28302 (rs6000_sched_init): Adjust for name change.
28303 (struct rs6000_sched_context): Likewise.
28304 (rs6000_init_sched_context): Likewise.
28305 (rs6000_set_sched_context): Likewise.
28307 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28313 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28315 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28316 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28317 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28319 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28321 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28323 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28325 PR rtl-optimization/70703
28326 * ira-color.c (update_conflict_hard_regno_costs): Use
28327 int64_t instead of HOST_WIDE_INT.
28329 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28331 PR rtl-optimization/70478
28332 * lra-constraints.c (process_alt_operands): Disfavor alternative
28333 insn memory operands.
28335 2017-04-07 Jeff Law <law@redhat.com>
28337 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28338 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28340 2017-04-07 Martin Liska <mliska@suse.cz>
28343 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28344 Show error message instead of an ICE.
28346 2017-04-07 Martin Liska <mliska@suse.cz>
28349 * ipa-split.c (split_function): Add function part to a same comdat
28352 2017-04-07 Richard Biener <rguenther@suse.de>
28354 PR middle-end/80341
28355 * tree.c (get_unwidened): Also handle ! for_type case for
28357 * convert.c (do_narrow): Split out from ...
28358 (convert_to_integer_1): ... here. Do not pass final truncation
28359 type to get_unwidened for TRUNC_DIV_EXPR.
28361 2017-04-07 Richard Biener <rguenther@suse.de>
28363 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28365 (aff_combination_const): Adjust.
28366 (aff_combination_scale): Likewise.
28367 (aff_combination_add_elt): Likewise.
28368 (aff_combination_add_cst): Likewise.
28369 (aff_combination_convert): Likewise.
28370 (add_elt_to_tree): Likewise. Remove unused argument.
28371 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28373 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28375 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28377 * config/arm/arm.c (arm_default_short_enums): Use
28378 ARM_DEFAULT_SHORT_ENUMS.
28379 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28381 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28384 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28385 members with redundant out-of-class redeclaration.
28387 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28390 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28391 * config/i386/i386.md (*zero_extendsidi2):
28392 Add (?*x,*x) and (?*v,*v) alternatives.
28394 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28397 * config/i386/i386.c (ix86_expand_builtin)
28398 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28399 mode from insn data. Convert operands to insn operand mode.
28400 Copy operands that don't satisfy insn predicate to a register.
28402 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28404 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28407 2017-04-06 Richard Biener <rguenther@suse.de>
28409 PR tree-optimization/80334
28410 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28411 preserve alignment of accesses.
28413 2017-04-06 Richard Biener <rguenther@suse.de>
28415 PR tree-optimization/80262
28416 * tree-sra.c (build_ref_for_offset): Preserve address-space
28418 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28419 Drop useless address-space information on MEM_REF offsets.
28421 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28423 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28425 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28427 PR rtl-optimization/70703
28428 * ira-color.c (update_conflict_hard_regno_costs): Use
28429 HOST_WIDE_INT instead of long.
28431 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28434 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28435 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28437 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28438 for x86_64 target. Handle -m3dnowa option.
28440 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28442 PR rtl-optimization/70703
28443 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28444 (update_conflict_hard_regno_costs): Use long instead of unsigned
28445 arithmetic for cost calculation.
28447 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28448 Bernd Edlinger <bernd.edlinger@hotmail.de>
28451 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28454 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28457 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28458 ptr_mode with Pmode throughout.
28459 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28460 into probe_stack_range and use DImode.
28462 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28465 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28466 call_eh_return is true.
28468 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28470 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28471 Initialize last_match_fntype_index.
28473 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28476 * tree-nvr.c: Include internal-fn.h.
28477 (pass_return_slot::execute): Ignore internal calls without
28480 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28481 Richard Biener <rguenther@suse.de>
28484 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28485 captures used multiple times, except for the last use.
28486 * generic-match-head.c: Include gimplify.h.
28488 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28490 PR tree-optimization/79390
28491 * target.h (struct noce_if_info): Declare.
28492 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28493 * target.def (noce_conversion_profitable_p): New target hook.
28494 * ifcvt.h (struct noce_if_info): New type, moved from ...
28495 * ifcvt.c (struct noce_if_info): ... here.
28496 (noce_conversion_profitable_p): Renamed to ...
28497 (default_noce_conversion_profitable_p): ... this. No longer
28499 (noce_try_store_flag_constants, noce_try_addcc,
28500 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28501 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28502 instead of noce_conversion_profitable_p.
28503 * config/i386/i386.c: Include ifcvt.h.
28504 (ix86_option_override_internal): Don't override
28505 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28506 (ix86_noce_conversion_profitable_p): New function.
28507 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28508 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28509 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28510 * doc/tm.texi: Regenerated.
28512 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28514 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28517 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28520 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28521 instructions for small multiply cores.
28523 2017-04-04 Jeff Law <law@redhat.com>
28525 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28527 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28528 that are not set by the loop over the elements.
28530 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28533 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28534 int mode, convert_modes it to mode as unsigned, otherwise use
28535 lowpart_subreg to mode rather than SImode.
28536 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28537 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28538 Use DImode instead of SImode for the shift count operand.
28539 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28542 2017-04-04 Richard Biener <rguenther@suse.de>
28544 PR middle-end/80281
28545 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28546 arithmetic done for the negate or the plus. Simplify.
28547 (A - (-B) -> A + B): Likewise.
28548 * fold-const.c (split_tree): Make sure to not negate pointers.
28550 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28552 PR rtl-optimization/60818
28553 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28554 a compare of comparisons with the thing compared if this results
28555 in a different machine mode.
28557 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28559 * alias.c (base_alias_check): Fix typo in comment.
28560 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28561 * cgraphunit.c (symbol_table::compile): Likewise.
28562 * collect2.c (maybe_run_lto_and_relink): Likewise.
28563 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28564 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28565 * config/avr/avr.c (avr_map_op_t): Likewise.
28566 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28567 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28568 * config/epiphany/epiphany.md (movcc): Likewise.
28569 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28570 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28572 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28573 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28574 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28575 * config/sh/sh.c (sh_rtx_costs): Likewise.
28576 * fold-const.c (fold_truth_andor): Likewise.
28577 * genautomata.c (collapse_flag): Likewise.
28578 * gengtype.h (struct type::u::s): Likewise.
28579 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28580 * input.c (FORMAT_AMOUNT): Likewise.
28581 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28582 (known_aggs_to_agg_replacement_list): Likewise.
28583 * ipa-inline-analysis.c: Likewise.
28584 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28585 * ipa-polymorphic-call.c
28586 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28587 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28588 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28590 * modulo-sched.c (apply_reg_moves): Likewise.
28591 * omp-expand.c (build_omp_regions_1): Likewise.
28592 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28593 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28594 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28595 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28596 * value-prof.c: Likewise.
28597 * var-tracking.c (val_reset): Likewise.
28599 2017-04-03 Richard Biener <rguenther@suse.de>
28601 PR tree-optimization/80275
28602 * fold-const.c (split_address_to_core_and_offset): Handle
28605 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28607 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28608 descriptors is at least equal to that of functions.
28610 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28612 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28614 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28617 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28618 (mov<IMOD4:mode>): New expander.
28619 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28621 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28623 PR rtl-optimization/79405
28624 * fwprop.c (propagations_left): New variable.
28625 (forward_propagate_into): Decrement it.
28626 (fwprop_init): Initialize it.
28627 (fw_prop): If the variable has reached zero, stop propagating.
28628 (fwprop_addr): Ditto.
28630 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28633 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28634 a FUNCTION_DECL, pass it as decl instead of origin to
28637 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28639 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28642 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28645 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28646 TARGET_VSX_SMALL_INTEGER.
28648 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28650 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28651 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28653 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28655 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28656 extraction from odd-numbered MSA register.
28658 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28660 PR middle-end/80173
28661 * expmed.c (store_bit_field_1): Don't attempt to create
28662 a word subreg out of hard registers wider than word if they
28663 have HARD_REGNO_NREGS of 1 for their mode.
28665 PR middle-end/80163
28666 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28667 conversions to integer types wider than word and pointer.
28670 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28671 (rtx_equal_for_cselib_p): Pass 0 to it.
28672 * cselib.c (cselib_hasher::equal): Likewise.
28673 (rtx_equal_for_cselib_1): Add depth argument. If depth
28674 is 128, don't look up VALUE locs and punt. Increment
28675 depth in recursive calls when walking VALUE locs.
28677 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28679 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28680 (make_gcov_file_name): Use the canonical path name for generating
28682 (read_line): Fix handling of files with ascii null bytes.
28684 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28686 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28687 to initialise a vector register instead
28688 of using a const_int.
28690 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28692 PR translation/80189
28693 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28694 diagnostic messages.
28696 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28699 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28700 (dfp_diex_<mode>): Update mode of operand 1.
28701 * doc/extend.texi (dxex, dxexq): Document change to return type.
28702 (diex, diexq): Document change to argument type.
28704 2017-03-30 Martin Jambor <mjambor@suse.cz>
28707 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28708 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28709 it reflects the signature changes performed at the callee side.
28710 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28711 to cgraph_build_function_type_skip_args.
28712 (build_function_decl_skip_args): Adjust call to the above function.
28714 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28717 * config/i386/sse.md
28718 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28719 register as dest whenever it is a MEM not rtx_equal_p to the
28720 corresponding dup operand, and when forcing into reg move the
28721 reg into the memory afterwards.
28722 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28723 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28724 for the force_reg mode.
28725 (avx512vl_vextractf128<mode>): Use register as dest either
28726 always when a MEM, or when it is a MEM not rtx_equal_p to the
28727 corresponding dup operand, or even not when it is a CONST_VECTOR
28728 depending on the mode and lo vs. hi.
28729 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28731 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28732 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28733 Likewise. Require that operands[2] is even.
28734 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28735 Remove extraneous parens. Require that operands[2] is a multiple
28737 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28738 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28739 disallow memory then.
28741 2017-03-30 Richard Biener <rguenther@suse.de>
28743 PR tree-optimization/77498
28744 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28745 to non-constants over backedges.
28747 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28749 PR rtl-optimization/80233
28750 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28751 as last_combined_insn. Do not test for BARRIER_P separately.
28753 2017-03-29 Andreas Schwab <schwab@suse.de>
28756 * calls.c (prepare_call_address): Convert funexp to Pmode before
28757 copying to temp reg.
28759 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28761 PR tree-optimization/80158
28762 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28763 Handle possible future case of more than one alternate
28765 (replace_rhs_if_not_dup): Likewise.
28766 (replace_one_candidate): Likewise.
28768 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28770 PR rtl-optimization/80193
28771 * ira.c (ira): Do not check allocation for LRA.
28773 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28775 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28776 (nvptx_output_simt_exit): Declare.
28777 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28778 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28779 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28780 (nvptx_declare_function_name): ...here. Emit declaration of local
28781 memory space buffer for omp_simt_enter insn.
28782 (nvptx_output_unisimt_switch): New.
28783 (nvptx_output_softstack_switch): New.
28784 (nvptx_output_simt_enter): New.
28785 (nvptx_output_simt_exit): New.
28786 * config/nvptx/nvptx.h (struct machine_function): New fields
28787 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28788 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28789 (UNSPECV_SIMT_EXIT): Ditto.
28790 (omp_simt_enter_insn): New insn.
28791 (omp_simt_enter): New expansion.
28792 (omp_simt_exit): New insn.
28793 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28795 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28796 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28797 (expand_GOMP_SIMT_EXIT): New.
28798 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28799 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28800 (GOMP_SIMT_EXIT): Ditto.
28801 * target-insns.def (omp_simt_enter): New insn.
28802 (omp_simt_exit): Ditto.
28803 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28805 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28806 (lower_rec_input_clauses): Likewise.
28807 (lower_lastprivate_clauses): Handle SIMT privatization.
28809 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28810 (ompdevlow_adjust_simt_enter): New.
28811 (find_simtpriv_var_op): New.
28812 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28813 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28815 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28816 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28817 (copy_decl_for_dup_finish): Ditto.
28819 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28821 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28824 * config/i386/i386.c (ix86_option_override_internal): Always
28825 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28827 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28829 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28831 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28833 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28834 mark new edge's irreducible flag accordign to it.
28835 (vect_do_peeling): Check loop preheader edge's irreducible flag
28836 and pass it to function slpeel_add_loop_guard.
28838 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28840 PR tree-optimization/80218
28841 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28842 Update block frequencies and counts.
28844 2017-03-28 Richard Biener <rguenther@suse.de>
28846 PR tree-optimization/78644
28847 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28848 of a simplification result we may not use it at all.
28850 2017-03-28 Richard Biener <rguenther@suse.de>
28853 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28854 without arguments, generate default definition of a SSA name.
28856 2017-03-28 Richard Biener <rguenther@suse.de>
28858 PR middle-end/80222
28859 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28860 TYPE_REF_CAN_ALIAS_ALL references.
28861 * fold-const.c (fold_indirect_ref_1): Likewise.
28863 2017-03-28 Martin Liska <mliska@suse.cz>
28866 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28867 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28869 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28870 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28872 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28873 (EXTRA_SPECS): Define.
28874 (SUBTARGET_EXTRA_SPECS): Likewise.
28875 (SUBTARGET_CPP_SPEC): Likewise.
28876 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28877 SUBTARGET_EXTRA_SPECS.
28878 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28880 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28882 * config/arc/simdext.md (vst64_insn): Update pattern.
28883 (vld32wh_insn): Likewise.
28884 (vld32wl_insn): Likewise.
28885 (vld64_insn): Likewise.
28886 (vld32_insn): Likewise.
28888 2017-03-28 Marek Polacek <polacek@redhat.com>
28891 * fold-const.c (fold_comparison): Use protected_set_expr_location
28892 instead of SET_EXPR_LOCATION.
28894 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28896 * tree.c (add_expr): Avoid name lookup warning.
28898 2017-03-27 Jeff Law <law@redhat.com>
28900 PR tree-optimization/80216
28901 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28902 function name. Limit recursion depth.
28903 (record_temporary_equivalences): Corresponding changes.
28905 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28907 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28910 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28913 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28915 * cfgcleanup.c (reg_note_cfa_p): New array.
28916 (insns_have_identical_cfa_notes): New function.
28917 (old_insns_match_p): Don't cross-jump in between /f
28918 and non-/f instructions. If both i1 and i2 are frame related,
28919 verify all CFA notes, their order and content.
28921 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28924 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28925 HImode and SImode with zero extend to DImode to one insn.
28926 (bswap<mode>2_extenddi): Likewise.
28927 (bswapsi2_extenddi): Likewise.
28928 (bswaphi2_extendsi): Likewise.
28929 (bswaphi2): Combine bswap HImode and SImode into one insn.
28930 Separate memory insns from swapping register.
28931 (bswapsi2): Likewise.
28932 (bswap<mode>2): Likewise.
28933 (bswaphi2_internal): Delete, no longer used.
28934 (bswapsi2_internal): Likewise.
28935 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28936 store, and gpr<-gpr swap insns.
28937 (bswap<mode>2_store): Likewise.
28938 (bswaphi2_reg): Register only splitter, combine with the splitter.
28939 (bswaphi2 splitter): Likewise.
28940 (bswapsi2_reg): Likewise.
28941 (bswapsi2 splitter): Likewise.
28942 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28943 the insns into load, store, and register/register insns.
28944 (bswapdi2_ldbrx): Likewise.
28945 (bswapdi2_load): Likewise.
28946 (bswapdi2_store): Likewise.
28947 (bswapdi2_reg): Likewise.
28949 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28951 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28952 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28954 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28957 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28959 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28960 special handling for target option conflicts between dform
28961 options (-mpower9-dform, -mpower9-dform-vector,
28962 -mpower9-dform-scalar) and -mno-direct-move.
28964 2017-03-27 Richard Biener <rguenther@suse.de>
28966 PR tree-optimization/80181
28967 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28969 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28971 * config/arc/predicates.md (move_double_src_operand): Replace the
28972 call to move_double_src_operand with a call to address_operand.
28974 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28976 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28977 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28978 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28980 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28982 * config/arc/predicates.md (long_immediate_loadstore_operand):
28983 Consider scaled addresses cases.
28985 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28987 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28988 restored when in interrupt.
28989 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28990 doesn't have delay slot.
28992 2017-03-27 Richard Biener <rguenther@suse.de>
28995 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28996 inlined thunk clones.
28998 2017-03-27 Jakub Jelinek <jakub@redhat.com>
29001 * asan.c (instrument_derefs): Copy over last operand from
29002 original COMPONENT_REF to the new COMPONENT_REF with
29003 DECL_BIT_FIELD_REPRESENTATIVE.
29004 * ubsan.c (instrument_object_size): Likewise.
29006 2017-03-27 Richard Biener <rguenther@suse.de>
29008 PR tree-optimization/80170
29009 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
29010 sure DR/SCEV didnt fold in constants we do not see when looking
29011 at the reference base alignment.
29013 2017-03-27 Richard Biener <rguenther@suse.de>
29015 PR middle-end/80171
29016 * gimple-fold.c (fold_ctor_reference): Properly guard against
29017 NULL return value from canonicalize_constructor_val.
29019 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
29022 * config/i386/i386.c (ix86_expand_builtin)
29023 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
29024 flags reg setting and flags reg using instructions.
29025 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
29026 clobbering instructions to zero extend op2.
29028 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
29030 * doc/install.texi (Configuration) <--with-aix-soname>:
29031 Update link to AIX ld.
29033 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
29035 PR rtl-optimization/80160
29036 PR rtl-optimization/80159
29037 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
29038 reg_alternate_class into account.
29040 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
29043 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
29044 to consider in curr_insn_transform.
29046 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29048 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
29049 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
29050 and emit_mode_inner.
29052 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29054 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
29055 argument to the overloaded builtin variants. Use the new flag to
29056 deprecate certain builtin variants.
29057 * config/s390/s390-builtin-types.def: Add new builtin types.
29058 * config/s390/s390-builtins.h: Support new flags field for
29059 overloaded builtins.
29060 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
29061 (s390_macro_to_expand): Enable vector float data type.
29062 (s390_cpu_cpp_builtins_internal): Indicate support of the new
29063 builtins by incrementing the __VEC__ version number.
29064 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
29066 (s390_resolve_overloaded_builtin): Emit error messages depending
29067 on the builtin flags.
29068 * config/s390/s390.c (s390_expand_builtin): Support additional
29069 flags argument. Change error message to match the messages
29070 emitted in s390-c.c.
29071 * config/s390/s390.md: New UNSPEC_* constants.
29072 (op_type): Add new instruction types.
29073 * config/s390/vecintrin.h: Add new builtins and test data class
29075 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
29076 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
29077 (VEC_INEXACT, VEC_NOINEXACT): New constants.
29078 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
29079 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
29080 ("vec_mergel<mode>"): V_HW -> VEC_HW.
29082 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
29083 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
29084 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
29085 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
29087 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
29088 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
29089 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
29090 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
29092 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
29093 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
29094 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
29095 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
29096 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
29097 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
29098 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
29100 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29101 ("vec_scatter_element<V_HW_4:mode>_DI")
29102 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29103 ("vec_fpint<mode>", "vflls")
29104 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29105 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29106 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29107 ("*vec_cmphe<mode>_cc"): ... these.
29109 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29110 mode constant instead of magic value.
29112 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29114 * config/s390/s390.c (s390_expand_vec_compare): Support other
29115 vector floating point modes than just V2DF.
29116 (s390_expand_vcond): Likewise.
29117 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29118 (s390_cannot_change_mode_class): Prevent mode changes between TF
29119 and V1TF in vector registers.
29120 * config/s390/s390.md (DF, SF): New mode attributes.
29121 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29122 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29123 SFmode support for VRs.
29124 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29126 (VFT, VF_HW): New mode iterators.
29127 (vw, sdx): New mode attributes.
29128 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29129 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29130 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29131 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29132 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29133 also the new vector floating point modes. Renaming to ...
29135 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29136 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29137 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29138 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29139 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29140 ("vec_unordered<mode>"): ... these.
29142 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29143 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29144 ("*vec_extendv2df"): New insn definitions.
29146 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29148 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29149 ("mulditi3_2", "*muldi3_sign"): New patterns.
29150 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29151 rename the pattern definition.
29153 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29155 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29157 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29159 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29161 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29162 instruction if possible.
29163 * config/s390/vector.md (vec_halfnumelts): New mode
29165 ("*vec_vllezlf<mode>"): New pattern.
29167 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29169 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29170 ("popcountv4si2", "popcountv2di2"): Rename to ...
29171 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29172 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29174 ("popcount<mode>2_vxe"): New pattern.
29176 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29178 * common/config/s390/s390-common.c (processor_flags_table): Add
29180 * config.gcc: Add arch12.
29181 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29182 Default to arch12 for unknown CPU model numbers.
29183 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29184 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29185 PROCESSOR_max sanity check.
29186 * config/s390/s390-opts.h (enum processor_type): Add
29188 * config/s390/s390.c (processor_table): Add arch12.
29189 (s390_expand_builtin): Add check for B_VXE flag.
29190 (s390_issue_rate): Add PROCESSOR_ARCH12.
29191 (s390_get_sched_attrmask): Likewise.
29192 (s390_get_unit_mask): Likewise.
29193 (s390_sched_score): Enable z13 scheduling for arch12.
29194 (s390_sched_reorder): Likewise.
29195 (s390_sched_variable_issue): Likewise.
29196 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29198 (s390_tune_attr): Use z13 scheduling also for arch12.
29199 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29200 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29201 (TARGET_VXE_P): New macros.
29202 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29203 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29204 * config/s390/s390.opt: Add arch12 as processor_type.
29206 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29208 * config/s390/s390.md
29209 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29210 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29211 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29213 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29214 Rename expanders to ...
29216 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29217 ("fixuns_truncdddi2_emu"): ... these.
29219 ("fixuns_trunc<mode>si2_emu"): New expander.
29221 ("*fixuns_truncdfdi2_z13"): Rename to ...
29222 ("*fixuns_truncdfdi2_vx"): ... this.
29224 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29226 * config/s390/2964.md: Remove the single element vector compare
29227 instructions which are no longer used.
29228 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29230 (s390_canonicalize_comparison): Remove handling of DFmode
29232 (s390_expand_vec_compare_scalar): Remove function.
29233 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29234 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29236 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29240 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29242 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29243 will anyway by matched by mov<mode>_64dfp.
29245 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29247 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29248 vlef/vstef. Add missing operand to vleif.
29250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29252 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29253 pair for all vector types with 64 bit elements.
29254 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29255 * config/s390/vector.md (V_HW_64): ... here.
29256 (V_128_NOSINGLE): New mode iterator.
29257 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29258 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29259 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29260 ("*vec_load_pairv2di"): Change to ...
29261 ("*vec_load_pair<mode>"): ... this one.
29263 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29265 * config/s390/constraints.md: Add comments.
29266 (jKK): Reject element sizes > 8 bytes.
29267 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29269 * config/s390/s390.md: Add the s_operand checks formerly in
29270 s390_split_ok_p to various splitters where they are still
29272 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29273 for 128 bit vectors. Plus two splitters.
29275 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29277 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29280 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29283 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29284 error if the boundary argument is not constant.
29286 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29288 PR rtl-optimization/80112
29289 * loop-doloop.c (doloop_condition_get): Don't check condition
29290 if cmp isn't SET with IF_THEN_ELSE src.
29292 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29294 PR tree-optimization/80158
29295 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29296 replacing a candidate statement, also replace it for the
29297 candidate's alternate interpretation.
29298 (replace_rhs_if_not_dup): Likewise.
29299 (replace_one_candidate): Likewise.
29301 2017-03-24 Richard Biener <rguenther@suse.de>
29303 PR tree-optimization/80167
29304 * graphite-isl-ast-to-gimple.c
29305 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29307 (translate_isl_ast_to_gimple::get_rename): Likewise.
29309 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29311 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29312 handling of certain combinations of target options, including the
29313 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29314 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29316 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29319 * config/arm/arm.md (*load_multiple): Add reload_completed to
29320 matching condition.
29322 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29323 Richard Biener <rguenth@suse.de>
29325 PR tree-optimization/79908
29326 PR tree-optimization/80136
29327 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29328 been cast away, gimplify_and_add suffices.
29330 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29332 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29334 2017-03-23 Richard Biener <rguenther@suse.de>
29336 PR tree-optimization/80032
29337 * gimplify.c (gimple_push_cleanup): Forced unconditional
29338 cleanups still have to go to the conditional_cleanups
29341 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29343 PR tree-optimization/80072
29344 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29346 (next_operand_entry_id): Change type to unsigned int.
29347 (sort_by_operand_rank): Make sure to return the right return value
29348 even if unsigned fields are bigger than INT_MAX.
29349 (struct oecount): Change cnt and id type to unsigned int.
29350 (oecount_hasher::equal): Formatting fix.
29351 (oecount_cmp): Make sure to return the right return value
29352 even if unsigned fields are bigger than INT_MAX.
29353 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29356 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29357 TREE_READONLY on result if writing it more than once.
29360 * doc/invoke.texi (-fsanitize=thread): Document that with
29361 -fnon-call-exceptions atomics are not able to throw
29365 * tsan.c: Include tree-eh.h.
29366 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29367 maybe_clean_or_replace_eh_stmt where needed.
29368 (instrument_memory_accesses): Add cfg_changed argument.
29369 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29370 if it returned true.
29371 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29373 PR rtl-optimization/63191
29374 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29375 wrapper function, moved the whole old content into ...
29376 (ix86_delegitimize_address_1): ... this. New inline function.
29377 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29378 true as last argument instead of ix86_delegitimize_address.
29380 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29382 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29383 cortexa57_branch_cost.
29385 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29387 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29389 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29392 * doc/md.texi (Constraints): Document wA constraint.
29393 * config/rs6000/constraints.md (wA): New.
29394 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29395 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29396 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29397 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29399 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29402 * gimplify.c (is_oacc_declared): New function.
29403 (oacc_default_clause): Use it to set default flags for acc declared
29404 variables inside parallel regions.
29405 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29406 declared variables.
29407 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29408 declare attribute to any decl as necessary.
29410 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29413 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29414 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29415 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29416 (arm_arch_lpae): This.
29417 * config/arm/arm.c (arm_arch7ve): Rename into ...
29418 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29419 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29422 2017-03-22 Martin Liska <mliska@suse.cz>
29425 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29426 error message instead of an ICE.
29428 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29430 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29432 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29434 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29436 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29439 2017-03-21 Martin Sebor <msebor@redhat.com>
29441 * doc/extend.texi: Use "cannot" instead of "can't."
29442 * doc/hostconfig.texi: Same.
29443 * doc/install.texi: Same.
29444 * doc/invoke.texi: Same.
29445 * doc/loop.texi: Same.
29446 * doc/md.texi: Same.
29447 * doc/objc.texi: Same.
29448 * doc/rtl.texi: Same.
29449 * doc/tm.texi: Same.
29450 * doc/tm.texi.in: Same.
29451 * doc/trouble.texi: Same.
29453 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29456 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29457 (collect_checksum_attributes): Set it.
29458 (die_checksum_ordered): Use it.
29460 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29462 PR tree-optimization/79908
29463 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29464 change: For a VA_ARG whose LHS has been cast away, use
29465 force_gimple_operand to construct the side effects.
29467 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29469 PR translation/80001
29470 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29471 more amenable to translation.
29472 (oacc_loop_auto_partitions): Likewise.
29474 2017-03-21 Marek Polacek <polacek@redhat.com>
29475 Martin Sebor <msebor@redhat.com>
29477 PR tree-optimization/80109
29478 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29479 on INTEGRAL_TYPE_P.
29481 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29482 Segher Boessenkool <segher@kernel.crashing.org>
29485 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29486 check reg_used_between_p between insn and one of succ or succ2
29487 depending on if succ is artificial insn not inserted into insn
29490 2017-03-21 Martin Liska <mliska@suse.cz>
29492 PR gcov-profile/80081
29493 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29494 * doc/gcc.texi: Include gcov-dump stuff.
29495 * doc/gcov-dump.texi: New file.
29497 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29499 PR rtl-optimization/79150
29500 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29501 conditional jump, if the jump is the last insn of the loop.
29503 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29504 Richard Biener <rguenth@suse.de>
29506 PR tree-optimization/79908
29507 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29508 been cast away, use force_gimple_operand to construct the side
29511 2017-03-21 Martin Liska <mliska@suse.cz>
29513 PR libfortran/79956
29514 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29517 2017-03-21 Brad Spengler <spender@grsecurity.net>
29520 * plugin.c (htab_hash_plugin): New function.
29521 (add_new_plugin): Use it and adjust.
29522 (parse_plugin_arg_opt): Adjust.
29523 (init_one_plugin): Likewise.
29525 2017-03-21 Richard Biener <rguenther@suse.de>
29527 PR tree-optimization/80032
29528 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29529 if set force the cleanup to happen unconditionally.
29530 (gimplify_target_expr): Push inserted clobbers with force_uncond
29531 to avoid them being removed by control-dependent DCE.
29533 2017-03-21 Richard Biener <rguenther@suse.de>
29535 PR tree-optimization/80122
29536 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29537 va_arg_pack_len when the inlined call stmt requires pack
29539 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29541 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29544 * tsan.c (instrument_builtin_call): If the memory model argument
29545 is not a constant, assume it is valid.
29548 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29551 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29553 PR rtl-optimization/79910
29554 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29555 if its dest is used by an insn before I2 (other than the combined
29556 insns themselves, which are properly handled already).
29558 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29561 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29563 * combine.c (record_used_regs): New static function.
29564 (try_combine): Handle situations where there is an additional
29565 instruction between I2 and I3 which needs to have a LOG_LINK
29569 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29571 * combine.c (try_combine): Delete redundant i1 test. Call
29572 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29574 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29577 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29578 alternatives 13/14.
29580 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29582 PR tree-optimization/80054
29583 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29584 the optimization if a PHI or any of its arguments is not dominated
29585 by the candidate's basis. Use gphi* rather than gimple* as
29587 (replace_profitable_candidates): Clean up a gimple* variable that
29588 should be a gphi* variable.
29590 2017-03-20 Martin Sebor <msebor@redhat.com>
29593 * doc/extend.texi (attribute constructor): Document present limitation.
29595 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29598 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29599 __POWER9_VECTOR__ #ifdef control, change template definition to
29600 use Power9-specific built-in function.
29601 (vec_any_eq): Likewise.
29602 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29603 to control outcomes from this test.
29604 (vector_ae_<mode>p): For VEC_F modes, likewise.
29606 2017-03-20 Ian Lance Taylor <iant@google.com>
29608 * config/i386/i386.c (ix86_function_regparm): Save an extra
29609 register for -fsplit-stack with DECL_STATIC_CHAIN.
29611 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29614 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29615 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29617 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29619 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29621 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29624 2017-03-20 Marek Polacek <polacek@redhat.com>
29627 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29629 2017-03-20 Richard Biener <rguenther@suse.de>
29631 PR tree-optimization/80113
29632 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29633 allocate extra SSA name for PHI def.
29634 (add_close_phis_to_outer_loops): Likewise.
29635 (add_close_phis_to_merge_points): Likewise.
29636 (copy_loop_close_phi_args): Likewise.
29637 (copy_cond_phi_nodes): Likewise.
29639 2017-03-20 Martin Liska <mliska@suse.cz>
29641 PR middle-end/79753
29642 * tree-chkp.c (chkp_build_returned_bound): Do not build
29643 returned bounds for a LHS that's not a BOUNDED_P type.
29645 2017-03-20 Martin Liska <mliska@suse.cz>
29649 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29650 COMPLEX_CST and VECTOR_CST.
29652 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29655 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29656 target operand. A new splitter adds the clobber statement in case
29657 the target operand is dead anyway.
29659 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29661 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29662 to age-old versions of binutils and glibc.
29664 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29666 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29668 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29670 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29672 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29674 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29675 requirement for binutils 2.13.
29677 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29679 * combine.c (try_combine): Delete redundant i1 test. Call
29680 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29682 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29684 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29685 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29687 <riscv64-*-elf>: Re-arrange section
29688 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29689 <riscv32-*-linux>: Likewise.
29690 <riscv64-*-elf>: Likewise
29691 <riscv64-*-linux>: Likewise.
29693 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29696 * aarch64.opt(verbose-cost-dump): Fix typo.
29698 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29701 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29702 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29704 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29706 * reload.c (find_reloads): When reloading a nonoffsettable address,
29707 use RELOAD_OTHER for it and its address reloads.
29709 PR rtl-optimization/79910
29710 * combine.c (record_used_regs): New static function.
29711 (try_combine): Handle situations where there is an additional
29712 instruction between I2 and I3 which needs to have a LOG_LINK
29715 2017-03-17 Jeff Law <law@redhat.com>
29717 PR tree-optimization/71437
29718 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29719 conditional in the hash table first.
29720 (vrp_dom_walker::before_dom_children): Extract condition from
29721 ASSERT_EXPR. Record condition, its inverion and any implied
29722 conditions as well.
29724 2017-03-17 Marek Polacek <polacek@redhat.com>
29725 Markus Trippelsdorf <markus@trippelsdorf.de>
29727 PR tree-optimization/80079
29728 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29731 2017-03-17 Richard Biener <rguenther@suse.de>
29733 PR middle-end/80075
29734 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29735 Properly verify the LHS before the RHS possibly claims to be
29737 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29740 2017-03-17 Martin Jambor <mjambor@suse.cz>
29742 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29743 (List of -O2 options): Likewise.
29744 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29747 2017-03-17 Tom de Vries <tom@codesourcery.com>
29749 * gcov-dump.c (print_usage): Print bug_report_url.
29751 2017-03-17 Richard Biener <rguenther@suse.de>
29753 PR middle-end/80050
29754 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29755 (parser::peek): Likewise.
29757 2017-03-17 Richard Biener <rguenther@suse.de>
29759 PR tree-optimization/80048
29760 * sese.c (free_sese_info): Properly release rename_map and
29761 copied_bb_map elements.
29763 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29765 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29766 Add linked-list forward and backlinks. Insert on
29767 construction, remove on destruction.
29768 (class pass_store_merging): Add m_stores_head field.
29769 (pass_store_merging::terminate_and_process_all_chains):
29770 Iterate over m_stores_head list.
29771 (pass_store_merging::terminate_all_aliasing_chains):
29773 (pass_store_merging::execute): Check for debug stmts first.
29774 Push new chains onto the m_stores_head stack.
29776 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29779 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29780 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29781 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29783 2017-03-16 Jeff Law <law@redhat.com>
29785 PR tree-optimization/71437
29786 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29787 member function. Implementation moved into after_dom_children
29788 member function and into the threader's thread_outgoing_edges
29790 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29791 some code into new thread_outgoing_edges.
29792 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29793 definition. Simplify marker handling (do it here). Assume we always
29794 have the available expression and the const/copies tables.
29795 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29797 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29798 * tree-vrp.c (equiv_stack): No longer file scoped.
29799 (vrp_dom_walker): New class.
29800 (vrp_dom_walker::before_dom_children): New member function.
29801 (vrp_dom_walker::after_dom_children): Likewise.
29802 (identify_jump_threads): Setup domwalker. Use it rather than
29803 walking edges in a random order by hand. Simplify setup/finalization.
29804 (finalize_jump_threads): Remove.
29805 (vrp_finalize): Do not call identify_jump_threads here.
29806 (execute_vrp): Do it here instead and call thread_through_all_blocks
29809 PR tree-optimization/71437
29810 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29812 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29814 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29815 (dom_opt_dom_walker::thread_across_edge): Remove
29816 handle_dominating_asserts argument. All callers changed.
29817 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29818 changes. Remove calls to lhs_of_dominating_assert. Other
29819 uses of handle_dominating_asserts turn into unconditional code
29820 (simplify_control_stmt_condition_1): Likewise.
29821 (simplify_control_stmt_condition): Likewise.
29822 (thread_through_normal_block, thread_across_edge): Likewise.
29823 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29824 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29825 object if it is not an SSA_NAME.
29826 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29827 before calling into the VRP specific simplifiers.
29828 (identify_jump_threads): Remove handle_dominating_asserts
29831 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29834 * tree-diagnostic.c (default_tree_printer): No longer static.
29835 * tree-diagnostic.h (default_tree_printer): New prototype.
29837 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29839 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29840 Change ins into fmov.
29842 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29844 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29845 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29846 Use h_con constraint for operand 1.
29847 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29848 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29850 2017-03-15 Jeff Law <law@redhat.com>
29852 PR tree-optimization/71437
29853 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29854 (record_temporary_equivalences): Use it.
29856 PR tree-optimization/71437
29857 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29858 tree-ssa-scopedtables.
29859 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29860 (record_conditions, record_cond, vuse_eq): Likewise.
29861 (record_edge_info): Adjust to API tweak of record_conditions.
29862 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29863 (record_temporary_equivalences, optimize_stmt): Likewise.
29864 (eliminate_redundant_computations): Likewise.
29865 (record_equivalences_from_stmt): Likewise.
29866 * tree-ssa-scopedtables.c: Include options.h and params.h.
29867 (vuse_eq): New function, moved from tree-ssa-dom.c
29868 (build_and_record_new_cond): Likewise.
29869 (record_conditions): Likewise. Accept vector of conditions rather
29870 than edge_equivalence structure for first argument.
29871 for the first argument.
29872 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29873 from tree-ssa-dom.c.
29874 (avail_exprs_stack::record_cond): Likewise.
29875 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29876 from tree-ssa-dom.c.
29877 (avail_exprs_stack): Add new member functions lookup_avail_expr
29879 (record_conditions): Declare.
29881 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29884 * lra-constraints.c (process_alt_operands): Increase reject for
29885 reloading an input/output operand.
29887 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29890 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29891 insns to convert from signed/unsigned char/short to IEEE 128-bit
29893 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29895 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29898 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29899 subreg of inner mode for values already in registers.
29901 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29903 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29904 iteration reg is used after the loop.
29906 2017-03-14 Martin Sebor <msebor@redhat.com>
29908 PR tree-optimization/79800
29909 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29910 precision in negative-positive range.
29911 (format_floating): Call non-const overload with adjusted precision.
29913 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29916 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29919 2017-03-14 Martin Sebor <msebor@redhat.com>
29921 PR middle-end/80020
29922 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29923 * builtins.def (aligned_alloc): Use it.
29926 * Makefile.in (GTFILES): Add calls.c.
29927 * calls.c: Include "gt-calls.h".
29929 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29931 PR rtl-optimization/79728
29932 * regs.h (struct target_regs): New field
29933 x_contains_allocatable_regs_of_mode.
29934 (contains_allocatable_regs_of_mode): New macro.
29935 * reginfo.c (init_reg_sets_1): Initialize it, and change
29936 contains_reg_of_mode so it includes global regs as well.
29937 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29938 rather than contains_regs_of_mode.
29940 2017-03-14 Martin Liska <mliska@suse.cz>
29942 * doc/invoke.texi: Document options that can't be combined with
29943 -fcheck-pointer-bounds.
29945 2017-03-14 Martin Liska <mliska@suse.cz>
29947 PR middle-end/79831
29948 * doc/invoke.texi (-Wchkp): Document the option.
29950 2017-03-14 Martin Liska <mliska@suse.cz>
29952 * Makefile.in: Install gcov-dump.
29954 2017-03-14 Martin Liska <mliska@suse.cz>
29956 * multiple_target.c (expand_target_clones): Bail out for
29957 an invalid attribute.
29959 2017-03-14 Richard Biener <rguenther@suse.de>
29961 * alias.c (struct alias_set_entry): Pack properly.
29962 * cfgloop.h (struct loop): Likewise.
29963 * cse.c (struct set): Likewise.
29964 * ipa-utils.c (struct searchc_env): Likewise.
29965 * loop-invariant.c (struct invariant): Likewise.
29966 * lra-remat.c (struct cand): Likewise.
29967 * recog.c (struct change_t): Likewise.
29968 * rtl.h (struct address_info): Likewise.
29969 * symbol-summary.h (function_summary): Likewise.
29970 * tree-loop-distribution.c (struct partition): Likewise.
29971 * tree-object-size.c (struct object_size_info): Likewise.
29972 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29973 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29974 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29975 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29976 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29977 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29978 (struct _stmt_vec_info): Likewise.
29980 2017-03-14 Martin Liska <mliska@suse.cz>
29983 * multiple_target.c (create_dispatcher_calls): Check that
29984 a target can create a function dispatcher.
29986 2017-03-14 Martin Liska <mliska@suse.cz>
29989 * multiple_target.c (expand_target_clones): Drop local.local
29990 flag for default implementation.
29992 2017-03-14 Richard Biener <rguenther@suse.de>
29994 PR tree-optimization/80030
29995 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29997 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29999 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
30000 gcc_fallthrough() instead of __attribute__((fallthrough));
30002 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30004 * doc/gcc.texi: Remove "up" link to (DIR).
30005 * doc/gccint.texi: Ditto.
30007 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
30009 * doc/install.texi (Specific) <avr>: Remove reference to
30012 2017-03-13 Jeff Law <law@redhat.com>
30014 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
30015 attribute rather than comments.
30017 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
30018 match_scratch operand is highest.
30020 2017-03-13 Martin Liska <mliska@suse.cz>
30022 PR middle-end/78339
30023 * ipa-pure-const.c (warn_function_noreturn): If the declarations
30024 is a CHKP clone, use original declaration.
30026 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30028 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
30029 (arc_conditional_register_usage): Use a different allocation order
30030 when optimizing for size.
30031 * common/config/arc/arc-common.c (arc_option_optimization_table):
30032 Section anchors default on when optimizing for size.
30034 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30036 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
30038 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30040 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
30041 * config/arc/arc.md (cpu_facility): Add cd variant.
30042 (*movqi_insn): Add code density variant.
30043 (*movhi_insn): Likewise.
30044 (*movqi_insn): Likewise.
30045 (*addsi3_mixed): Likewise.
30046 (subsi3_insn): Likewise.
30048 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30050 * config/arc/arc.md (movsi_cond_exec): Update constraint.
30052 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
30054 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
30055 expressions with MINUS and UNARY ops.
30057 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30060 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
30062 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
30063 between vec_select and vector argument.
30064 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
30065 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
30066 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
30067 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
30068 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
30069 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
30071 2017-03-13 Richard Biener <rguenther@suse.de>
30074 * params.def (vect-max-peeling-for-alignment): Fix typo.
30076 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30078 * doc/install.texi (Specific) <mips-*-*>: Remove description of
30079 issue that only occurred with binutils below 2.18.
30081 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30083 * doc/install.texi (Specific) <cris-axis-elf>: No longer
30084 refer to binutils 2.11/2.12 minimum.
30086 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
30088 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
30089 ftp.kernel.org and simplify binutils requirement.
30091 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
30093 * doc/invoke.texi (Warning Options): Fix spelling of link-time
30095 (Optimize Options): Ditto. Also remove redundancy.
30097 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30099 PR translation/79848
30100 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30102 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30103 to G_ to avoid double translation.
30105 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30107 PR translation/79923
30108 * auto-profile.c (get_combined_location): Convert leading
30109 character of diagnostics to lower case and remove trailing period.
30110 (read_profile): Likewise for various diagnostics.
30111 * config/arm/arm.c (arm_option_override): Remove trailing period
30112 from various diagnostics.
30113 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30114 (msp430_expand_delay_cycles): Likewise.
30116 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30119 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30120 full command-line argument, rather than just "str".
30121 (aarch64_validate_march): Likewise.
30122 (aarch64_validate_mtune): Likewise.
30124 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30126 PR rtl-optimization/78911
30127 * lra-assigns.c (must_not_spill_p): New function.
30128 (spill_for): Use it.
30130 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30132 PR tree-optimization/79981
30133 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30134 ATOMIC_COMPARE_EXCHANGE ifn result.
30135 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30136 IFN_ATOMIC_COMPARE_EXCHANGE.
30138 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30141 * opts.c (parse_sanitizer_options): Add missing question mark to
30142 "did you mean" message.
30144 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30146 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30148 (VMULEUH_UNS): Likewise.
30149 (VMULOUB_UNS): Likewise.
30150 (VMULOUH_UNS): Likewise.
30151 * config/rs6000/rs6000.c (builtin_function_type): Remove
30152 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30154 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30157 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30158 x with result of extra_parsing_for_operand_code_0.
30159 (function_reader::extra_parsing_for_operand_code_0): Convert
30160 return type from void to rtx, returning x. When reading
30161 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30162 larger size containing struct block_symbol.
30164 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30167 -mfloat128-hardware without -m64.
30169 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30172 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30173 entries to the case statement that marks unsigned arguments to
30174 overloaded functions.
30176 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30178 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30179 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30181 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30184 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30185 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30187 2017-03-10 Martin Liska <mliska@suse.cz>
30191 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30192 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30195 2017-03-10 Olivier Hainque <hainque@adacore.com>
30197 * tree-switch-conversion (array_value_type): Start by resetting
30198 candidate type to it's main variant.
30200 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30202 PR rtl-optimization/79909
30203 * combine.c (try_combine): Use simplify_replace_rtx on individual
30204 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30205 of the whole CALL_INSN_FUNCTION_USAGE.
30207 PR tree-optimization/79972
30208 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30209 get_range_info on SSA_NAMEs. Formatting fixes.
30211 2017-03-10 Richard Biener <rguenther@suse.de>
30212 Jakub Jelinek <jakub@redhat.com>
30214 PR tree-optimization/77975
30215 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30216 edge to be constant.
30217 (get_val_for): For constant x return it. Formatting fix.
30218 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30219 would use the same bases as the current one.
30221 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30223 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30224 instead of vec_select for V1TImode.
30225 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30227 (VSX_LE_128): Add V1TI to this mode iterator.
30228 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30229 (*vsx_le_perm_store_<mode>): Likewise.
30230 (pre-reload splitter for VSX stores): Likewise.
30231 (post-reload splitter for VSX stores): Likewise.
30232 (*vsx_xxpermdi2_le_<mode>): Likewise.
30233 (*vsx_lxvd2x2_le_<mode>): Likewise.
30234 (*vsx_stxvd2x2_le_<mode>): Likewise.
30236 2017-03-09 Michael Eager <eager@eagercon.com>
30238 Correct failures with --enable-checking=yes,rtl.
30240 * config/microblaze/microblaze.c (microblaze_expand_shift):
30241 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30242 test for const0_rtx.
30243 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30244 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30246 2017-03-09 Richard Biener <rguenther@suse.de>
30248 PR tree-optimization/79977
30249 * graphite-scop-detection.c (scop_detection::merge_sese):
30250 Handle the case of extra exits to blocks dominating the entry.
30252 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30254 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30257 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30259 PR rtl-optimization/79949
30260 * lra-constraints.c (process_alt_operands): Check memory when
30261 trying to predict a cycle. Print about the overall increase.
30263 2017-03-09 Richard Biener <rguenther@suse.de>
30265 PR middle-end/79971
30266 * gimple-expr.c (useless_type_conversion_p): Preserve
30267 TYPE_SATURATING for fixed-point types.
30269 2017-03-09 Richard Biener <rguenther@suse.de>
30272 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30273 alignment of BLKmode params.
30275 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30278 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30279 (VALL_NO_V2Q): Likewise.
30281 * config/aarch64/aarch64-simd.md
30282 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30284 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30285 VALL_NO_V2Q mode iterator.
30286 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30288 2017-03-09 Martin Liska <mliska@suse.cz>
30290 PR tree-optimization/79631
30291 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30292 tree_int_cst_sign_bit just for INTEGER constants.
30294 2017-03-09 Martin Liska <mliska@suse.cz>
30298 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30301 2017-03-09 Marek Polacek <polacek@redhat.com>
30304 * tree.c (inchash::add_expr): Handle TREE_VEC.
30306 2017-03-09 Martin Liska <mliska@suse.cz>
30309 (chkp_narrow_size_and_offset): New function.
30310 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30311 (void chkp_parse_bit_field_ref): New function.
30312 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30313 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30315 2017-03-09 Martin Liska <mliska@suse.cz>
30318 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30319 (chkp_find_bounds_1): Remove gcc_unreachable.
30321 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30324 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30325 BUILT_IN_SYNC*, determine the access type from the size suffix and
30326 always build a MEM_REF with that type. Handle forgotten
30327 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30330 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30331 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30332 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30333 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30334 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30335 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30336 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30337 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30338 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30339 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30340 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30341 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30342 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30343 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30344 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30345 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30346 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30347 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30348 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30349 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30350 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30351 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30352 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30353 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30354 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30355 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30356 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30357 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30358 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30359 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30360 definitions outside of __OPTIMIZE__ guarded section.
30363 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30364 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30365 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30366 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30369 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30371 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30372 ("vfenez<mode>"): Add missing constraints.
30374 2017-03-08 Martin Sebor <msebor@redhat.com>
30377 * config/nds32/nds32.c (nds32_option_override):
30378 Fix misspelled diagnostic.
30380 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30383 * gimplify.c (gimplify_omp_for): Replace index var in outer
30384 taskloop statement with an artificial variable and add
30385 OMP_CLAUSE_PRIVATE clause for it.
30387 2017-03-08 Richard Biener <rguenther@suse.de>
30389 PR tree-optimization/79955
30390 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30391 for accesses that are completely outside of the variable.
30393 2017-03-08 Andrew Haley <aph@redhat.com>
30395 PR tree-optimization/79943
30396 * tree-ssa-loop-split.c (compute_new_first_bound): When
30397 calculating the new upper bound, (END-BEG) should be added, not
30400 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30402 * config/avr/avr.md (setmemhi): Make sure match_dup
30403 operand number comes before match_scratch.
30405 2017-03-08 Richard Biener <rguenther@suse.de>
30407 PR tree-optimization/79920
30408 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30409 with ncopies == 1 to ...
30410 (vect_transform_slp_perm_load): ... here. Properly compute
30411 all element loads by iterating VF times over the group. Do
30412 not handle ncopies (computed in a broken way) in
30413 vect_create_mask_and_perm.
30415 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30418 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30419 is a uniform vector, use uniform_vector_p return value instead of
30420 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30422 2017-03-07 Marek Polacek <polacek@redhat.com>
30424 PR middle-end/79809
30425 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30426 (alloca_call_type): Likewise.
30428 2017-03-07 Martin Liska <mliska@suse.cz>
30430 * gcov.c (process_args): Put comment to correct location.
30432 2017-03-07 Martin Liska <mliska@suse.cz>
30434 PR middle-end/68270
30435 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30436 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30437 (chkp_narrow_bounds_for_field): Likewise.
30438 (chkp_parse_array_and_component_ref): Pass one more argument to
30441 2017-03-07 Richard Biener <rguenther@suse.de>
30443 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30446 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30448 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30449 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30451 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30454 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30455 to end of description.
30456 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30458 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30460 PR rtl-optimization/79901
30461 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30463 (*avx512f_<code><mode>3<mask_name>): ... this.
30464 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30465 iterator instead of VI8_AVX2_AVX512BW.
30467 PR rtl-optimization/79901
30468 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30469 min/max expander, expand it using expand_vec_cond_expr.
30472 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30475 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30478 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30479 to void * for PCH reasons.
30480 * dwarf2out.c (output_loc_operands, output_die): Cast
30481 v.val_vec.array to unsigned char *.
30483 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30486 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30489 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30491 PR rtl-optimization/79571
30492 * lra-constraints.c (process_alt_operands): Calculate static
30493 reject and subtract it from overall when only addresses will be
30496 2017-03-06 Julia Koval <julia.koval@intel.com>
30499 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30500 incoming stack boundary to 128 for 64-bit targets.
30502 2017-03-06 Richard Biener <rguenther@suse.de>
30504 PR tree-optimization/79894
30505 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30506 to NULL after folding it.
30508 2017-03-06 Richard Biener <rguenther@suse.de>
30510 PR tree-optimization/79824
30511 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30512 check disabling peeling for gaps.
30514 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30516 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30517 attributes): Document gettimeofday.
30519 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30521 * config/s390/s390.c (s390_option_override_internal): Set
30522 PARAM_MIN_VECT_LOOP_BOUND
30524 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30526 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30527 * config/s390/s390.md: Likewise.
30529 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30532 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30533 (<avx2_avx512>_perm<mode>): Rename to ...
30534 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30536 (<avx512>_perm<mode>_mask): Rename to ...
30537 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30539 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30540 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30541 instead of VI8F_256_512.
30542 (avx512f_perm<mode>): New define_expand.
30543 (avx512f_perm<mode>_mask): Likewise.
30544 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30545 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30547 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30549 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30550 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30552 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30554 2017-03-06 Martin Liska <mliska@suse.cz>
30557 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30558 when having a SSA NAME w/o VAR_DECL assigned to it.
30560 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30562 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30563 msa_dpsub_<su>_d): Fix MODE for vec_select.
30565 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30567 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30569 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30571 2017-03-06 Richard Biener <rguenther@suse.de>
30573 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30574 * plugin.c (register_plugin_info): Likewise.
30575 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30577 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30579 * config/i386/sse.md (sse_storehps, sse_storelps,
30580 avx_<castmode><avxsizesuffix>_<castmode>,
30581 avx512f_<castmode><avxsizesuffix>_<castmode>,
30582 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30583 in condition that at least one operand is not a MEM.
30585 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30587 PR middle-end/79805
30588 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30589 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30591 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30592 gimple_call_nothrow_p flag based on whether original builtin can throw.
30593 If it can, emit following stmts on the fallthrough edge.
30594 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30595 don't create new bb if inserting just debug stmts on the edge, try to
30596 insert them on the fallthru bb or just reset debug stmts.
30598 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30601 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30602 restore recog_data (including the operand rtxes inside it) around
30603 the call to get_insn_template.
30605 2017-03-03 Martin Sebor <msebor@redhat.com>
30607 PR tree-optimization/79699
30608 * context.c (context::~context): Free MPFR caches to avoid
30609 a memory leak on program exit.
30611 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30613 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30614 Use wide_int::ulow () instead of .elt (0).
30616 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30618 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30619 (*pushxf): Limit oF constraint to 32bit targets and add oC
30620 constraint for 64bit targets.
30621 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30622 (*pushdf): Change rmF constraint to rmC.
30624 2017-03-03 Martin Liska <mliska@suse.cz>
30626 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30627 Remove unused variable.
30629 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30632 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30633 is a memory operand, increase num_memory.
30634 (ix86_expand_args_builtin): Likewise.
30636 2017-03-03 Jan Hubicka <jh@suse.cz>
30639 * ipa-devirt.c (maybe_record_node): Properly handle
30640 __cxa_pure_virtual visibility.
30642 2017-03-03 Martin Liska <mliska@suse.cz>
30644 PR tree-optimization/79803
30645 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30647 (pass_loop_prefetch::execute): Disabled optimization if an
30648 assumption about L1 cache size is not met.
30650 2017-03-03 Martin Liska <mliska@suse.cz>
30652 PR rtl-optimization/79574
30653 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30654 (hash_scan_set): Likewise.
30655 (dump_hash_table): Likewise.
30656 (hoist_code): Likewise.
30658 2017-03-03 Richard Biener <rguenther@suse.de>
30660 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30661 in place of uhigh ().
30662 (fixed_convert_from_real): Likewise.
30664 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30667 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30669 2017-03-03 Richard Biener <rguenther@suse.de>
30671 PR middle-end/79818
30672 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30673 TYPE_OVERFLOW_UNDEFINED check.
30675 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30677 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30679 (vector_ae_<mode>_p): Likewise.
30680 (vector_nez_<mode>_p): Likewise.
30681 (vector_ne_v2di_p): Likewise.
30682 (vector_ae_v2di_p): Likewise.
30683 (vector_ne_<mode>_p): Likewise.
30684 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30686 (vsx_tsqrt<mode>2_fe): Likewise.
30688 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30691 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30693 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30695 PR rtl-optimization/79780
30696 * cprop.c (one_cprop_pass): When second and further conditional trap
30697 in a single basic block is turned into an unconditional trap, turn it
30698 into a deleted note to avoid RTL verification failures.
30700 2017-03-02 Richard Biener <rguenther@suse.de>
30702 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30704 2017-03-02 Richard Biener <rguenther@suse.de>
30706 PR tree-optimization/79345
30708 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30709 param and abort the walk, returning -1 if it is hit.
30710 (walk_aliased_vdefs): Take a limit param and pass it on.
30711 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30712 defaulting to 0 and return a signed int.
30713 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30714 (check_defs): New helper.
30715 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30716 about uninitialized memory.
30717 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30718 bogus uninitialized warning.
30719 (fixed_convert_from_real): Likewise.
30721 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30723 PR tree-optimization/66768
30724 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30725 iv_use if base object can't be determined.
30727 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30729 PR tree-optimization/79345
30730 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30731 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30732 (get_pattern_stats): Initialize it.
30733 * genemit.c (gen_expand): Verify match_scratch numbers come after
30734 match_operand/match_dup numbers.
30735 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30736 match_scratch numbers.
30737 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30739 * config/s390/s390.md (trunctdsd2): Likewise.
30741 2017-03-02 Richard Biener <rguenther@suse.de>
30743 * wide-int.h (wide_int_storage::operator=): Implement in terms
30746 2017-03-02 Richard Biener <rguenther@suse.de>
30748 PR tree-optimization/79777
30749 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30750 the to insert expression to sth existing.
30752 2017-03-01 Martin Sebor <msebor@redhat.com>
30754 PR middle-end/79692
30755 * gimple-ssa-sprintf.c
30756 (directive::known_width_and_precision): New function.
30757 (format_integer): Use it.
30758 (get_mpfr_format_length): Consider the full range of precision
30759 when computing %g output with the # flag. Set the likely byte
30760 count to 3 rather than 1 when precision is indeterminate.
30761 (format_floating): Correct the lower bound of precision.
30763 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30765 * doc/invoke.texi: Document default code model for 64-bit Linux.
30767 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30770 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30771 udiv rather than div since input pattern is unsigned.
30773 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30775 * config/i386/i386.c (print_reg): Warn for values of
30776 unsupported size in integer register.
30778 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30781 * config/rs6000/predicates.md (current_file_function_operand): Do
30782 not allow self calls to be local if the function is replaceable.
30784 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30787 * config/rs6000/altivec.h (vec_ctz and others): Change the
30788 preprocessor macro that controls conditional compilation from
30789 _ARCH_PWR9 to __POWER9_VECTOR__.
30790 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30791 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30792 control (instead of _ARCH_PWR9 control) so that template
30793 definition uses power9-specific function.
30794 (vec_any_eq): Likewise.
30795 (vec_all_ne): Change macro definition to use a power9-specific
30796 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30797 _ARCH_PWR9 control).
30798 (vec_any_eq) Likewise.
30799 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30800 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30801 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30802 support for xvcmpnedp instruction.
30803 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30804 macro expansion so that Power9 implementation of vec_all_ne does
30805 not use the AltiVec predicate framework.
30806 (VCMPNEH_P): Likewise.
30807 (VCMPNEW_P): Likewise.
30808 (VCMPNED_P): Likewise.
30809 (VCMPNEFP_P): Likewise.
30810 (VCMPNEDP_P): Likewise.
30811 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30812 implementation of vec_any_eq to not use AltiVec predicate
30814 (VCMPAEH_P): Likewise.
30815 (VCMPAEW_P): Likewise.
30816 (VCMPAED_P): Likewise.
30817 (VCMPAEFP_P): Likewise.
30818 (VCMPAEDP_P): Likewise.
30819 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30820 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30821 not use the AltiVec predicate framework.
30822 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30823 of vec_any_eq to not use AltiVec predicate framework.
30824 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30825 support for predefined __POWER9_VECTOR__ macro to indicate that
30826 Power9 instruction selection is enabled.
30827 (altivec_overloaded_builtins): Remove extraneous
30828 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30829 function argument types RS6000_BTI_bool_V16QI and
30830 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30831 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30832 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30833 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30834 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30835 Power9 for implementations of vec_cmpne. Change the signature for
30836 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30837 (representing vec_all_ne) to remove the previously described first
30838 argument of type RS6000_BTI_INTSI, as this was an artifact of
30839 reliance on the AltiVec predicate framework, which is no longer
30840 used in the implementation of these functions. Add
30841 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30842 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30843 since, unlike the AltiVec predicate framework implementation, we
30844 do not share function descriptors between vec_alle and vec_anyeq.
30845 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30846 set of modes that receive special treatment even when
30847 TARGET_P9_VECTOR is true. The special treatment emits code that
30848 does not depend on Power9 instructions.
30849 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30850 define_expand to not rely on AltiVec predicate framework.
30851 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30853 (vector_ne_v2di_p): Change this define_expand to not rely on
30854 AltiVec predicate framework.
30855 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30857 (vector_ne_<mode>_p): Change this define_expand to not rely on
30858 AltiVec predicate framework.
30859 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30861 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30862 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30863 define_insn pattern.
30864 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30865 define_insn pattern because the xvcmpne<VSs>. instruction is not
30867 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30868 instruction is not supported.
30870 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30872 * config/nvptx/nvptx.c: Include intl.h.
30874 2017-03-01 Martin Jambor <mjambor@suse.cz>
30877 * ipa-prop.h (ipa_bits): Removed field known.
30878 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30879 to pointers. Adjusted their comments to warn about their sharing.
30880 (ipcp_transformation_summary): Change bits to a vector of pointers.
30881 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30882 (ipa_get_ipa_bits_for_value): Declare.
30883 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30884 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30885 (ipa_bits_hash_table): Likewise.
30886 (ipa_vr_ggc_hash_traits): Likewise.
30887 (ipa_vr_hash_table): Likewise.
30888 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30889 being pointers and vr_known being removed.
30890 (ipa_set_jf_unknown): Likewise.
30891 (ipa_get_ipa_bits_for_value): New function.
30892 (ipa_set_jfunc_bits): Likewise.
30893 (ipa_get_value_range): New overloaded functions.
30894 (ipa_set_jfunc_vr): Likewise.
30895 (ipa_compute_jump_functions_for_edge): Use the above functions to
30896 construct bits and vr parts of jump functions.
30897 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30898 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30900 (ipcp_grow_transformations_if_necessary): Also allocate
30901 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30903 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30904 them. Fix too long lines.
30905 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30906 vr_known being removed.
30907 (ipa_read_jump_function): Use new setter functions to construct bits
30908 and vr parts of jump functions or set them to NULL.
30909 (write_ipcp_transformation_info): Adjust for bits being pointers.
30910 (read_ipcp_transformation_info): Likewise.
30911 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30913 Include gt-ipa-prop.h.
30914 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30916 (ipcp_store_bits_results): Likewise.
30917 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30918 Do not write to existing jump functions but use a temporary instead.
30920 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30923 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30924 attempt to use its first operand as BIT_FIELD_REF base.
30926 2017-03-01 Richard Biener <rguenther@suse.de>
30928 PR middle-end/79721
30929 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30930 interpolating formula in wrapping arithmetic.
30931 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30933 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30935 PR tree-optimization/79734
30936 * tree-vect-generic.c (expand_vector_condition): Optimize
30937 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30938 Handle VEC_COND_EXPR where comparison has different inner width from
30939 type's inner width.
30941 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30943 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30944 markup, and similar issues. Remove @opindex entries for things
30945 that aren't options. Add missing -mmpy-option entries.
30947 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30949 PR tree-optimization/79737
30950 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30951 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30952 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30953 instead of byte_size. Formatting fix.
30954 (shift_bytes_in_array_right): Formatting fix.
30956 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30959 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30960 condition on optimize for the leaf function test.
30962 2017-02-28 Martin Liska <mliska@suse.cz>
30965 * read-rtl-function.c (function_reader::handle_unknown_directive):
30966 Bail out when one uses -flto.
30968 2017-02-28 Martin Liska <mliska@suse.cz>
30970 * common.opt: Replace space with tabular for options of <number>
30972 * config/i386/i386.opt: Show <number> value for
30973 -mlarge-data-threshold.
30974 * opts.c (print_filtered_help): Do not display number in hexadecimal
30977 2017-02-28 Martin Liska <mliska@suse.cz>
30979 * common.opt: Fix --help=option -Q for options which are of
30982 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30984 * config/i386/i386.c (print_reg): Error out for values
30985 of 8-bit size in invalid integer register.
30987 2017-02-28 Martin Sebor <msebor@redhat.com>
30989 PR tree-optimization/79691
30990 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30992 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30995 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30996 gcc_unreachable with output_operand_lossage.
30998 2017-02-28 Richard Biener <rguenther@suse.de>
31000 PR tree-optimization/79740
31001 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
31003 (visit_nary_op): Insert the nary into the hashtable if we
31004 pattern-matched sth.
31005 * tree-ssa-pre.c (eliminate_insert): Robustify.
31007 2017-02-28 Richard Biener <rguenther@suse.de>
31009 PR middle-end/79731
31010 * fold-const.c (decode_field_reference): Reject out-of-bound
31013 2017-02-28 Jakub Jelinek <jakub@redhat.com>
31015 * config/i386/i386.c: Include intl.h.
31016 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
31017 instead of just cond ? "..." : "...".
31018 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
31019 * coverage.c (read_counts_file): Likewise.
31020 * omp-offload.c: Include intl.h.
31021 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
31022 of just cond ? "..." : "...".
31023 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
31024 of just cond ? "..." : "...".
31026 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
31029 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
31031 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
31032 'tune for' CPU name.
31033 * config/arm/arm-cpu-data.h: Regenerated.
31035 2017-02-28 Richard Biener <rguenther@suse.de>
31037 PR tree-optimization/79732
31038 * tree-inline.c (expand_call_inline): Do not shadow var.
31040 2017-02-28 Richard Biener <rguenther@suse.de>
31042 PR tree-optimization/79723
31043 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
31044 address-space properly.
31046 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
31048 * doc/optinfo.texi (Optimization groups): Fix option used for
31050 * doc/invoke.texi (-fopt-info): Document "omp".
31051 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
31052 (OPTGROUP_ALL): Add OPTGROUP_OMP.
31053 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
31054 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
31055 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
31057 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
31059 * dumpfile.c (optgroup_options): Instead of "openmp", associate
31060 OPTGROUP_OMP with "omp".
31062 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
31065 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
31066 for arithmetic shift of unsigned V2DI.
31068 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
31070 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
31071 arc/linux.h headers.
31072 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
31073 (LINK_SPEC): Likewise.
31074 (ARC_TLS_EXTRA_START_SPEC): Likewise.
31075 (EXTRA_SPECS): Likewise.
31076 (STARTFILE_SPEC): Likewise.
31077 (ENDFILE_SPEC): Likewise.
31078 (LIB_SPEC): Likewise.
31079 (TARGET_SDATA_DEFAULT): Likewise.
31080 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
31081 (MULTILIB_DEFAULTS): Likewise.
31082 (DWARF2_UNWIND_INFO): Likewise.
31083 * config/arc/big.h: New file.
31084 * config/arc/elf.h: Likewise.
31085 * config/arc/linux.h: Likewise.
31086 * config/arc/t-uClibc: Remove.
31088 2017-02-27 Bin Cheng <bin.cheng@arm.com>
31090 PR tree-optimization/77536
31091 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
31092 (tree_transform_and_unroll_loop): Use above function to compute the
31093 estimated niter of unrolled loop and use it when scaling profile.
31094 Also use count info rather than frequency if it's non-zero.
31095 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
31096 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
31097 (vect_transform_loop): Call above function.
31099 2017-02-27 Richard Biener <rguenther@suse.de>
31101 PR tree-optimization/45397
31102 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31103 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31104 (visit_nary_op): Add pattern matching for CSEing sign-changed
31105 or truncated operations with wider ones.
31107 2017-02-27 Richard Biener <rguenther@suse.de>
31109 PR tree-optimization/79690
31110 * tree-vect-stmts.c (vectorizable_store): Use vector type
31111 built from the DR with address-space.
31113 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31115 * doc/invoke.texi (Optimize Options): Refine the description
31116 of asan-use-after-return.
31118 2017-02-25 Alan Modra <amodra@gmail.com>
31120 PR rtl-optimization/79584
31121 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31122 base, not ad->base_term, the reg within base. Remove assertion
31123 that ad->base == ad->base_term. Replace gen_int_mode using
31124 bogus mode with const0_rtx.
31126 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31128 PR middle-end/79396
31129 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31130 FMA_EXPR like tcc_binary or tcc_unary.
31132 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31135 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31137 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31138 (output_loc_operands): Handle DW_OP_call_ref and
31139 DW_OP_GNU_variable_value.
31140 (struct variable_value_struct): New type.
31141 (struct variable_value_hasher): Likewise.
31142 (variable_value_hash): New variable.
31143 (string_types): Remove.
31144 (copy_loc_descr): New function.
31145 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31146 (prepend_loc_descr_to_each): New function.
31147 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31148 instead of add_loc_descr_to_each if the first argument is single
31149 location list and the second has multiple.
31150 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31151 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31152 when looking for variable value which doesn't have other location info.
31153 (loc_list_from_tree): Formatting fix.
31154 (gen_array_type_die): Simplify DW_AT_string_length handling.
31155 (adjust_string_types): Remove.
31156 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31157 string_types. Call resolve_variable_values.
31158 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31159 (resolve_addr_in_expr): Likewise. Add A argument.
31160 (copy_deref_exprloc): Remove deref argument. Adjust for the
31161 original expression being DW_OP_GNU_variable_value with optionally
31162 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31163 optionally after it.
31164 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31165 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31166 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31167 (variable_value_hasher::hash, variable_value_hasher::equal): New
31169 (resolve_variable_value_in_expr, resolve_variable_value,
31170 resolve_variable_values, note_variable_value_in_expr,
31171 note_variable_value): New functions.
31172 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31175 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31178 * opts.h (handle_generated_option): Add GENERATED_P argument.
31179 * opts-common.c (handle_option): Adjust function comment.
31180 (handle_generated_option): Add GENERATED_P argument, pass it to
31182 (control_warning_option): Pass false to handle_generated_option
31184 * opts.c (maybe_default_option): Pass true to handle_generated_option
31186 * optc-gen.awk: Likewise.
31188 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31190 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31191 a REG, look at the REG it is a SUBREG of.
31192 (splitter for cmpeqsi_t): Ditto.
31194 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31196 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31197 the special USEs with the pattern of the insn, not the insn itself.
31199 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31202 * doc/invoke.texi: Document -mload-store-pairs.
31204 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31205 Sandra Loosemore <sandra@codesourcery.com>
31207 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31208 argument isn't a CONST_INT.
31209 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31210 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31211 (nios2_validate_compare): Bypass alternate compare logic if *op2
31212 is not a CONST_INT.
31213 (ldstwm_operation_p): Return false if first_base is not a REG or
31214 if first_offset is not a CONST_INT.
31216 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31218 * config/cris/cris.md: Use correct operand in a define_peephole2.
31220 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31222 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31224 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31226 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31227 this_insn if it is an INSN or JUMP_INSN.
31228 (force_offsettable): Look at base, not at addr.
31229 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31230 on things that aren't necessarily CONST_INTs.
31232 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31234 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31235 -mfpmath=sse is the default also for x86-32 targets with SSE2
31236 instruction set when @option{-ffast-math} is enabled
31238 2017-02-24 Jeff Law <law@redhat.com>
31240 PR rtl-optimizatoin/79286
31241 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31244 2017-02-24 Richard Biener <rguenther@suse.de>
31246 PR tree-optimization/79389
31247 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31250 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31252 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31253 function comment to reflect reality.
31254 (loop_exits_before_overflow): Fix typo in function description.
31256 2017-02-24 Richard Biener <rguenther@suse.de>
31258 PR tree-optimization/79389
31259 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31260 properly that a threading opportunity exists. Detect conditional
31261 copy/constant propagation opportunities.
31263 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31265 * config/visium/visium.md (type): Add trap.
31266 (b): New mode attribute.
31267 (*btst): Rename into...
31268 (*btst<mode>): ...this and adjust.
31269 (*cbranchsi4_btst_insn): Rename into...
31270 (*cbranch<mode>4_btst_insn): ...this and adjust.
31271 (trap): New define_insn.
31273 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31275 PR tree-optimization/79389
31276 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31277 (noce_reversed_cond_code): New function.
31278 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31279 reversed_comparison_code. Formatting fix.
31280 (noce_try_store_flag): Test rev_cond != NULL in addition to
31281 reversed_comparison_code.
31282 (noce_try_store_flag_constants): Likewise.
31283 (noce_try_store_flag_mask): Likewise.
31284 (noce_try_addcc): Use rev_cond if non-NULL instead of
31285 reversed_comparison_code.
31286 (noce_try_cmove_arith): Likewise. Formatting fixes.
31287 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31288 (noce_find_if_block): Initialize rev_cond.
31289 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31290 instead of false as last argument never attempt to reverse it
31293 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31295 PR tree-optimization/79663
31296 * tree-predcom.c (combine_chains): Process refs in reverse order
31297 only for ZERO length chains, and add explaining comment.
31299 2017-02-23 Jeff Law <law@redhat.com>
31301 PR tree-optimization/79578
31302 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31303 in call to operand_equal_p.
31305 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31308 * config/i386/cpuid.h: Fix another undefined behavior.
31310 2017-02-23 Richard Biener <rguenther@suse.de>
31312 PR tree-optimization/79683
31313 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31314 vector types for data-refs.
31316 2017-02-23 Martin Liska <mliska@suse.cz>
31318 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31320 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31322 PR middle-end/79665
31323 * internal-fn.c (get_range_pos_neg): Moved to ...
31324 * tree.c (get_range_pos_neg): ... here. No longer static.
31325 * tree.h (get_range_pos_neg): New prototype.
31326 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31327 are known to be in between 0 and signed maximum inclusive, try to
31328 expand both unsigned and signed divmod and use the cheaper one from
31331 2017-02-22 Jeff Law <law@redhat.com>
31333 PR tree-optimization/79578
31334 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31335 to compare base operands.
31337 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31340 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31341 gpc_reg_operand instead of fpr_reg_operand.
31343 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31345 * config/mips/mips.c (mips_return_in_memory): Force FP
31346 vector types to be returned in memory for o32 ABI.
31348 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31350 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31351 instead of DW_TAG_member for static data member declarations and don't
31352 set no_linkage_name for static inline data members.
31353 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31356 2017-02-22 Martin Liska <mliska@suse.cz>
31358 * doc/invoke.texi: Replace inequality signs with square brackets
31361 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31364 * lra-constraints.c (simplify_operand_subreg): Handle
31365 WORD_REGISTER_OPERATIONS targets.
31367 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31370 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31371 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31372 elimination by swapping fld*.
31374 2017-02-22 Richard Biener <rguenther@suse.de>
31376 PR tree-optimization/79673
31377 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31378 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31379 irrelevant address-space qualifiers and avoiding a
31380 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31382 2017-02-22 Richard Biener <rguenther@suse.de>
31384 PR tree-optimization/79666
31385 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31386 to not symbolically negate if that may introduce undefined
31389 2017-02-22 Martin Liska <mliska@suse.cz>
31392 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31393 * data-streamer-out.c (streamer_write_gcov_count_stream):
31395 * value-prof.c (stream_out_histogram_value): Make assert more
31396 precise based on type of counter.
31398 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31401 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31402 Use nonimmediate_operand instead of memory_operand for operand 1.
31403 (float-extend standard_x87sse_constant_load splitter): Ditto.
31405 2017-02-21 Jeff Law <law@redhat.com>
31407 PR tree-optimization/79621
31408 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31409 blocks with edges to themselves.
31411 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31414 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31415 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31416 Use gimple_call_builtin_p.
31419 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31420 on temporarily removed DEBUG_INSNs.
31422 PR tree-optimization/79649
31423 * tree-loop-distribution.c (classify_partition): Give up on
31424 non-generic address space loads/stores.
31426 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31428 * doc/loop.texi (Loop manipulation): Remove nonexistent
31429 tree_ssa_loop_version from the documentation.
31430 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31432 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31435 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31436 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31437 * config/rs6000/rs6000.c: Include except.h.
31438 (rs6000_expand_split_stack_prologue): Call
31439 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31441 2017-02-21 Martin Jambor <mjambor@suse.cz>
31444 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31445 have been analyzed.
31447 2017-02-21 Martin Jambor <mjambor@suse.cz>
31449 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31450 for backward compatibility only.
31451 * doc/invoke.texi (Option Summary): Remove all references to
31452 -fipa-cp-alignment.
31454 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31458 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31460 * lra-constraints.c (curr_insn_transform): Handle
31461 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31463 2017-02-21 Martin Liska <mliska@suse.cz>
31465 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31467 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31469 PR translation/79638
31470 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31472 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31475 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31476 (arm_function_ok_for_sibcall): Return false for an indirect call by
31477 descriptor if all the argument registers are used.
31478 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31479 alignment of the function.
31481 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31483 PR tree-optimization/61441
31484 * simplify-rtx.c (simplify_const_unary_operation): For
31485 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31486 the sNaN unmodified.
31488 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31490 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31491 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31492 instead of SYSTEM_HEADER_DIR.
31494 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31495 Martin Liška <mliska@suse.cz>
31497 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31498 Fix typos and grammar, use active voice, and clarify.
31500 2017-02-20 Marek Polacek <polacek@redhat.com>
31502 PR middle-end/79537
31503 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31506 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31508 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31511 * config/i386/i386.c (ix86_expand_builtin): Handle
31512 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31513 ix86_builtins_isa[fcode].isa as a requirement of those
31514 flags and any other flag in the bitmask.
31515 (ix86_init_mmx_sse_builtins): Use 0 instead of
31516 ~OPTION_MASK_ISA_64BIT as mask.
31517 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31518 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31519 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31520 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31522 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31525 * lra-constraints.c (split_reg): Check requested split mode
31526 is supported by the register.
31528 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31530 * lra-constraints.c (simplify_operand_subreg): Remove early
31533 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31536 * lra-constraints.c (curr_insn_transform): Tighten condition
31537 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31539 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31542 * lra-constraints.c (curr_insn_transform): Handle
31543 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31545 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31548 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31550 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31552 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31555 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31558 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31560 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31563 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31565 * final.c (last_columnnum, override_columnnum): New variables.
31566 (final_start_function): Set last_columnnum, pass it to begin_prologue
31567 hook and pass 0 to dwarf2out_begin_prologue.
31568 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31569 to source_line debug hook.
31570 (notice_source_line): Compute last_columnnum and for debug_column_info
31571 return true on column changes.
31572 * debug.h (struct gcc_debug_hooks): Add column argument to
31573 source_line and begin_prologue hooks.
31574 (debug_nothing_int_charstar_int_bool): Remove prototype.
31575 (debug_nothing_int_int_charstar,
31576 debug_nothing_int_int_charstar_int_bool): New prototypes.
31577 (dwarf2out_begin_prologue): Add column argument.
31578 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31579 begin_prologue hooks.
31580 (debug_nothing_int_charstar_int_bool): Remove.
31581 (debug_nothing_int_int_charstar,
31582 debug_nothing_int_int_charstar_int_bool): New functions.
31583 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31584 through to dwarf2out_source_line.
31585 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31586 (dwarf2out_source_line): Add column argument, emit it if requested.
31587 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31589 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31590 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31591 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31592 through to dwarf2out_begin_prologue.
31593 (vmsdbgout_source_line): Add column argument, pass it through to
31594 dwarf2out_source_line.
31595 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31596 dbxout_source_line caller.
31597 (dbxout_source_line): Add column argument.
31599 * common.opt (gno-column-info, gcolumn-info): New options.
31600 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31601 (check_die): Also test for multiple DW_AT_decl_column attributes.
31602 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31603 DW_AT_decl_column if requested.
31604 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31606 (gen_variable_die): Likewise.
31607 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31608 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31611 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31612 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31613 (ix86_handle_option): Handle OPT_m3dnowa.
31614 * doc/invoke.texi (-m3dnowa): Document.
31615 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31616 -m3dnowa instead of -m3dnow -march=athlon.
31619 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31620 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31622 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31625 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31626 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31627 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31628 generator for vsx_xxpermdi_<mode>_be.
31629 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31630 force big-endian semantics.
31631 (vsx_xxpermdi_<mode>_be): New define_expand with same
31632 implementation as previous version of vsx_xxpermdi_<mode>.
31634 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31636 PR tree-optimization/79327
31637 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31638 variable, its initialization and use.
31640 2017-02-17 Julia Koval <julia.koval@intel.com>
31642 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31643 (OPTION_MASK_ISA_PKU_UNSET): New.
31644 (ix86_handle_option): Handle -mrdpid.
31645 * config/i386/cpuid.h (bit_RDPID): New.
31646 * config/i386/driver-i386.c (host_detect_local_cpu):
31647 Detect RDPID feature.
31648 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31649 * config/i386/i386-c.c (ix86_target_macros_internal):
31651 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31652 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31653 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31654 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31655 * config/i386/i386.md (define_insn "rdpid"): New.
31656 * config/i386/i386.opt Add -mrdpid.
31657 * config/i386/immintrin.h (_rdpid_u32): New.
31659 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31661 PR rtl-optimization/79541
31662 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31663 instead of transforming it into USE.
31665 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31667 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31668 If HONOR_SNANS (SFmode) force the input to a register.
31669 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31670 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31671 an frsp or similar insn.
31673 2017-02-17 Martin Liska <mliska@suse.cz>
31675 PR rtl-optimization/79577
31676 * params.def (selsched-max-sched-times): Increase minimum to 1.
31678 2017-02-17 Martin Liska <mliska@suse.cz>
31680 PR rtl-optimization/79574
31681 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31683 2017-02-17 Martin Liska <mliska@suse.cz>
31685 PR tree-optimization/79529
31686 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31687 ssa_defined_default_def_p to handle cases which are implicitly
31689 * tree-ssa.c (ssa_defined_default_def_p): New function.
31690 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31691 which are implicitly defined.
31692 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31694 2017-02-17 Richard Biener <rguenther@suse.de>
31696 PR middle-end/79576
31697 * params.def (max-ssa-name-query-depth): Limit to 10.
31699 2017-02-17 Richard Biener <rguenther@suse.de>
31701 PR tree-optimization/79552
31702 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31705 2017-02-17 Richard Biener <rguenther@suse.de>
31708 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31710 2017-02-17 Marek Polacek <polacek@redhat.com>
31712 PR middle-end/79536
31713 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31714 (fold_negate_expr): New wrapper.
31716 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31718 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31719 Correct terminology and de-emphasize pre-standard behavior.
31721 2017-02-16 Alan Modra <amodra@gmail.com>
31723 PR rtl-optimization/79286
31724 * ira.c (def_dominates_uses): New function.
31725 (update_equiv_regs): Don't create an equivalence for insns that
31726 may trap where the register def does not dominate the use.
31728 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31730 PR rtl-optimization/78127
31731 * lra.c (lra): Call lra_eliminate before finish the loop after
31734 2017-02-16 Richard Biener <rguenther@suse.de>
31736 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31738 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31739 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31740 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31741 (isl_val_int_from_wi): New function.
31742 (extract_affine_gmp): Rename to ...
31743 (extract_affine_wi): ... this, take a widest_int.
31744 (extract_affine_int): Just wrap extract_affine_wi.
31745 (add_param_constraints): Use isl_val_int_from_wi.
31746 (add_loop_constraints): Likewise, and extract_affine_wi.
31748 2017-02-15 Jeff Law <law@redhat.com>
31750 PR middle-end/79521
31751 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31752 ira_init_register_move_cost_if_necessary.
31754 2017-02-15 Martin Sebor <msebor@redhat.com>
31756 PR middle-end/32003
31757 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31758 removed in a prior commit.
31760 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31762 PR tree-optimization/79347
31763 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31764 counters during peeling.
31766 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31768 * Makefile.in (site.exp): Remove "set ISLVER".
31770 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31773 * real.c (real_from_integer): Call real_convert even for decimal.
31775 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31778 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31780 2017-02-14 Andrew Pinski <apinski@cavium.com>
31782 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31783 cores and change the partno/implementer to be correct.
31784 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31785 the 'B" as the implementer.
31786 * config/aarch64/aarch64-tune.md: Regenerate.
31788 2017-02-14 Carl Love <cel@us.ibm.com>
31790 * config/rs6000/rs6000.c: Add case statement entry to make the
31791 xvcvuxdsp built-in argument unsigned.
31792 * config/rs6000/vsx.md: Fix the source and return operand types so they
31793 match the instruction definitions from the ISA document. Fix typo
31794 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31797 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31800 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31801 member early_clobber_alts.
31802 * lra-lives.c (reg_early_clobber_p): New.
31803 (process_bb_lives): Use it.
31804 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31805 (debug_operand_data): Initialize early_clobber_alts.
31806 (setup_operand_alternative): Set up early_clobber_alts.
31807 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31808 alternatives to new_insn_reg.
31809 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31811 (lra_update_insn_regno_info): Pass the new arg.
31813 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31815 PR middle-end/79505
31816 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31817 (new_oacc_loop_raw): Don't clear already cleared fields.
31820 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31821 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31822 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31824 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31827 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31829 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31832 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31833 the extra instruction to the right place to store 128-bit constant
31836 2017-02-14 Martin Sebor <msebor@redhat.com>
31838 PR middle-end/79448
31839 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31840 warning for strings of unknown length.
31842 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31844 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31846 2017-02-14 Jeff Law <law@redhat.com>
31849 * ira-costs.c (scan_one_insn): Initialize register move costs
31850 for pseudos seen in USE/CLOBBER insns.
31852 PR tree-optimization/79095
31853 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31854 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31855 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31856 if the operands are known to be not equal, then the resulting range
31858 (intersect_ranges): If the new range is ~[0,0] and the old range is
31859 wide, then prefer ~[0,0].
31860 * tree-vrp.c (overflow_comparison_p_1): New function.
31861 (overflow_comparison_p): New function.
31862 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31863 if NAME is used in an overflow test.
31864 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31865 overflow check that can be expressed as an equality test, then adjust
31866 ops to be that equality test.
31868 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31870 * config/s390/s390-builtin-types.def: Remove flags argument.
31871 * config/s390/s390.c (s390_init_builtins): Likewise.
31873 2017-02-14 Martin Liska <mliska@suse.cz>
31875 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31876 vector. Fix trailing white spaces.
31878 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31880 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31883 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31885 PR rtl-optimization/68664
31886 * config/arm/arm.c (arm_sched_can_speculate_insn):
31887 New function. Declare prototype.
31888 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31890 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31892 PR rtl-optimization/68664
31893 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31895 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31897 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31899 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31900 max skip bytes for function, loop and jump.
31902 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31904 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31905 ABS_EXPR for gimple dump.
31907 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31910 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31912 PR tree-optimization/79408
31913 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31914 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31915 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31916 also if rhs1 is INTEGER_CST.
31918 2017-02-14 Richard Biener <rguenther@suse.de>
31920 PR middle-end/79432
31921 * tree-into-ssa.c (insert_phi_nodes): When the function can
31922 have abnormal edges rewrite SSA names with broken use-def
31923 dominance out of SSA and register them for PHI insertion.
31925 2017-02-13 Martin Sebor <msebor@redhat.com>
31927 PR middle-end/79496
31928 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31929 clearing info.nowrite flag when snprintf size argument is a range.
31931 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31933 * cprop.c (cprop_jump): Add missing space in string literal.
31934 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31935 (get_constraint_for_component_ref): Likewise.
31936 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31937 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31938 * lra-constraints.c (process_alt_operands): Likewise.
31939 * ipa-inline.c (inline_small_functions): Likewise.
31940 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31941 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31942 * trans-mem.c (diagnose_tm_1_op): Likewise.
31943 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31944 (grid_parallel_clauses_gridifiable): Likewise.
31946 * config/nvptx/mkoffload.c (process): Add space in between
31949 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31950 "MOD4_SSE_REGS" and "ALL_REGS".
31952 * spellcheck.c (test_data): Add , in between "foo" and "food".
31954 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31957 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31958 boundary crossing check and subsequent code generation agree.
31960 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31962 * config/aarch64/aarch64.c (has_memory_op): Delete.
31963 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31966 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31968 PR rtl-optimization/79388
31969 PR rtl-optimization/79450
31970 * combine.c (distribute_notes): When removing TEM_INSN for which
31971 corresponding dest has last value recorded, invalidate that last
31974 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31976 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31977 of explicit '@'. Add missing assembly comment marker on branch costs
31980 2017-02-13 Nathan Sidwell <nathan@acm.org>
31982 * gengtype-lex.l (<in_struct>): Add '/'.
31984 2017-02-13 Martin Liska <mliska@suse.cz>
31987 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31989 2017-02-13 Richard Biener <rguenther@suse.de>
31991 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31993 * configure: Re-generate.
31994 * config.in: Likewise.
31995 * graphite-dependences.c: Simplify as if
31996 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31997 * graphite-isl-ast-to-gimple.c: Likewise.
31998 * graphite-optimize-isl.c: Likewise.
31999 * graphite-poly.c: Likewise.
32000 * graphite-sese-to-poly.c: Likewise.
32001 * graphite.h: Likewise.
32002 * toplev.c: Include isl/version.h and use isl_version () for
32003 printing the ISL version.
32004 * doc/install.texi: Update ISL requirement.
32006 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32008 * doc/standards.texi (Standards): Update reference to
32011 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
32013 * doc/extend.texi (Named Address Spaces): sourceware.org now
32015 * doc/install.texi (Binaries): Ditto.
32018 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32020 * doc/cpp.texi: Replace "stringify"/"stringification" with C
32021 standard terminology "stringize"/"stringizing" throughout.
32022 * doc/cppinternals.texi: Likewise.
32024 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
32026 * doc/extend.texi: Fix some spelling mistakes and typos.
32027 * doc/invoke.texi: Likewise.
32029 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32032 * params.def (inline-min-speedup) Change from 10 to 8.
32034 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32036 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
32039 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
32042 * ipa-inline-analysis.c (get_minimal_bb): New function.
32043 (record_modified): Use it.
32044 (remap_edge_change_prob): Handle also ancestor functions.
32046 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
32048 * doc/contrib.texi (Contributors): Remove broken link into
32049 the Mauve CVS repository.
32051 2017-02-11 Jakub Jelinek <jakub@redhat.com>
32053 PR middle-end/79454
32054 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
32055 result computation whenever lhs doesn't have vector mode, not
32056 just when it has BLKmode.
32058 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32060 * doc/makefile.texi (profiledbootstrap): Refer to the
32061 installation instructions only in textual form.
32063 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32066 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
32068 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
32070 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
32071 (Specific): Update mingw-w64 reference.
32073 (Specific): Remove broken link to Renesas RX processor.
32075 2017-02-10 Richard Biener <rguenther@suse.de>
32077 * toplev.c (process_options): Do not mention obsolete graphite
32078 options when printing sorry message about missing graphite support.
32079 Mention -floop-nest-optimize.
32081 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
32083 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
32084 (vtst_p16): Likewise.
32085 (vtstq_p8): Likewise.
32086 (vtstq_p16): Likewise.
32088 (vtstq_p64): Likewise.
32089 * config/arm/arm_neon.h (vgetq_lane_p64): New.
32090 (vset_lane_p64): New.
32091 (vsetq_lane_p64): New.
32093 2017-02-10 Jakub Jelinek <jakub@redhat.com>
32095 PR tree-optimization/79411
32096 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
32097 stmt operands are SSA_NAMEs used in abnormal phis.
32098 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32101 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32104 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32107 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32109 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32111 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32113 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32117 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32118 not arbitrary TREE_CONSTANT.
32121 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32122 "omp declare target link" attribute unless is_global_var.
32123 * omp-offload.c (find_link_var_op): Likewise.
32125 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32126 Chung-Lin Tang <cltang@codesourcery.com>
32128 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32130 (gimplify_adjust_omp_clauses): Don't delete TILE.
32131 (gimplify_omp_for): Deal with TILE.
32132 * internal-fn.c (expand_GOACC_TILE): New function.
32133 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32135 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32136 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32138 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32139 avoid DIV for outermost collapse var.
32140 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32141 Remove out of date comments, fix whitespace.
32142 * omp-general.c (omp_extract_for_data): Deal with tiling.
32143 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32144 adjust OLF_DIM_BASE value.
32145 (struct omp_for_data): Add tiling field.
32146 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32147 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32148 for auto loops. Remove default auto determining, moved to
32149 oacc_loop_fixed_partitions.
32150 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32151 stmts, add e_mask field.
32152 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32153 (oacc_thread_numbers): Use oacc_dim_call.
32154 (oacc_xform_tile): New.
32155 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32156 (finish_oacc_loop): Adjust for ifns vector.
32157 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32158 add case for GOACC_TILE fns.
32159 (oacc_loop_xform_loop): Delete.
32160 (oacc_loop_process): Iterate over call list directly, and add
32161 handling for GOACC_TILE fns.
32162 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32164 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32165 vector partitioning to outer loops. Assign 2 partitions to loops
32166 when available. Add TILE handling.
32167 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32168 (execite_oacc_device_lower): Process GOACC_TILE fns,
32169 ignore unknown specs.
32170 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32171 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32172 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32174 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32176 * configure.ac (ACX_BUGURL): Update.
32177 * configure: Regenerate.
32179 2017-02-09 Richard Biener <rguenther@suse.de>
32181 PR tree-optimization/69823
32182 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32183 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32185 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32187 * config/arc/arc-c.def: Add __NPS400__ definition.
32188 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32189 (TARGET_NPS400): Define.
32191 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32193 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32195 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32196 pointer, arch_info.
32197 (arc_cpu_types): Fill the arch_info field with a pointer into the
32198 arc_arch_types table.
32199 (arc_selected_cpu): Declare.
32200 * config/arc/arc.c (arc_selected_cpu): Make global.
32201 (arc_selected_arch): Delete.
32202 (arc_base_cpu): Delete.
32203 (arc_override_options): Remove references to deleted variables,
32204 update access to arch information.
32205 (ARC_OPT): Update access to arch information.
32206 (ARC_OPTX): Likewise.
32207 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32208 (TARGET_ARC600): Update access to arch information.
32209 (TARGET_ARC601): Likewise.
32210 (TARGET_ARC700): Likewise.
32211 (TARGET_EM): Likewise.
32212 (TARGET_HS): Likewise.
32213 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32216 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32219 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32220 condition/operands for integer GE/LE/GEU/LEU operations.
32222 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32224 PR translation/79397
32225 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32228 2017-02-08 Martin Jambor <mjambor@suse.cz>
32231 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32232 whether allocation happened.
32233 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32234 nothing was allocated.
32236 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32238 PR tree-optimization/79408
32239 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32240 constant, but SSA_NAME with a known integer range, use the minimum
32241 of that range instead of op1 to determine if modulo can be replaced
32242 with its first operand.
32244 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32246 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32248 2017-02-08 Richard Biener <rguenther@suse.de>
32250 PR tree-optimization/71824
32251 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32252 Check all loops contained in the merged region.
32254 2017-02-07 Andrew Pinski <apinski@cavium.com>
32256 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32258 2017-02-07 Andrew Pinski <apinski@cavium.com>
32260 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32261 (thunderxt88): Likewise.
32262 (thunderxt81): Disable LSE and change v8.1 to v8.
32263 (thunderxt83): Likewise.
32265 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32266 Richard Biener <rguenther@suse.de>
32268 PR middle-end/79399
32269 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32270 type from int to size_t.
32271 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32273 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32275 PR rtl-optimization/79386
32276 * cprop.c (bypass_conditional_jumps): Initialize
32277 bypass_last_basic_block already before splitting bbs after
32278 unconditional traps...
32279 (bypass_conditional_jumps): ... rather than here.
32282 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32283 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32284 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32285 fix -masm=intel patterns.
32287 2017-02-07 Richard Biener <rguenther@suse.de>
32289 PR tree-optimization/79256
32290 PR middle-end/79278
32291 * builtins.c (get_object_alignment_2): Use min_align_of_type
32292 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32293 and ADJUST_FIELD_ALIGN.
32295 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32297 * doc/tm.texi: Regenerate.
32298 * stor-layout.c (layout_decl): Adjust.
32299 (update_alignment_for_field): Likewise.
32300 (place_field): Likewise.
32301 (min_align_of_type): Likewise.
32302 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32303 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32304 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32305 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32306 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32307 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32308 * config/i386/i386.c (x86_field_alignment): Likewise.
32309 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32310 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32311 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32312 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32313 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32314 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32318 2017-01-30 Richard Biener <rguenther@suse.de>
32320 PR tree-optimization/79256
32321 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32322 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32325 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32327 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32328 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32329 builtins to SImode and emit a zero-extend, if necessary.
32331 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32333 * docs/invoke.texi (RISC-V Options): Alphabetize.
32335 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32337 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32340 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32342 * config/riscv/riscv.c: New file.
32343 * common/config/riscv/riscv-common.c: Likewise.
32344 * config.gcc: Likewise.
32345 * config/riscv/constraints.md: Likewise.
32346 * config/riscv/elf.h: Likewise.
32347 * config/riscv/generic.md: Likewise.
32348 * config/riscv/linux.h: Likewise.
32349 * config/riscv/multilib-generator: Likewise.
32350 * config/riscv/peephole.md: Likewise.
32351 * config/riscv/pic.md: Likewise.
32352 * config/riscv/predicates.md: Likewise.
32353 * config/riscv/riscv-builtins.c: Likewise.
32354 * config/riscv/riscv-c.c: Likewise.
32355 * config/riscv/riscv-ftypes.def: Likewise.
32356 * config/riscv/riscv-modes.def: Likewise.
32357 * config/riscv/riscv-opts.h: Likewise.
32358 * config/riscv/riscv-protos.h: Likewise.
32359 * config/riscv/riscv.h: Likewise.
32360 * config/riscv/riscv.md: Likewise.
32361 * config/riscv/riscv.opt: Likewise.
32362 * config/riscv/sync.md: Likewise.
32363 * config/riscv/t-elf-multilib: Likewise.
32364 * config/riscv/t-linux: Likewise.
32365 * config/riscv/t-linux-multilib: Likewise.
32366 * config/riscv/t-riscv: Likewise.
32367 * configure.ac: Likewise.
32368 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32369 Waterman as RISC-V maintainers.
32370 * doc/install.texi: Add RISC-V entries.
32371 * doc/invoke.texi: Add RISC-V options section.
32372 * doc/md.texi: Add RISC-V constraints section.
32373 * configure: Regenerated.
32375 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32378 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32379 false values to be constant vectors with all 0 or all 1 bits set.
32380 (vcondu<mode><mode>): Likewise.
32381 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32383 (fpmask_comparison_operator): Update comment.
32384 (vecint_comparison_operator): New predicate.
32385 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32386 vector conditionals when the true and false values are constant
32387 vectors with all 0 bits or all 1 bits set.
32389 2017-02-06 Martin Sebor <msebor@redhat.com>
32391 PR tree-optimization/79376
32392 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32394 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32396 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32397 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32398 to simplify split condition.
32400 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32402 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32403 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32406 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32408 PR rtl-optimization/68664
32409 * target.def (can_speculate_insn): New hook.
32410 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32411 * doc/tm.texi: Regenerate.
32412 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32413 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32414 (rs6000_sched_can_speculate_insn): New function.
32416 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32418 PR tree-optimization/79284
32419 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32420 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32421 vectorizable_mask_load_store, vectorizable_operation,
32422 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32423 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32424 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32425 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32426 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32427 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32428 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32429 is_gimple_assign (stmt). Replace another such test with
32430 is_gimple_assign (stmt).
32432 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32435 * config/avr/avr.c (rtl-iter.h): Include it.
32436 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32437 (avr_legitimate_combined_insn): ...and implementation.
32439 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32441 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32442 * config/s390/s390.c (s390_const_operand_ok)
32443 (s390_canonicalize_comparison, s390_extract_part)
32444 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32445 (s390_contiguous_bitmask_p, s390_rtx_costs)
32446 (legitimize_pic_address): Likewise.
32447 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32448 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32449 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32450 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32451 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32453 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32455 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32456 REGNO($0) == REGNO($1).
32458 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32460 * config/s390/linux.h(SIZE_TYPE): Add comment.
32462 2017-02-06 Julian Brown <julian@codesourcery.com>
32463 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32464 Virendra Pathak <virendra.pathak@broadcom.com>
32466 * config/aarch64/aarch64-cores.def: Change the scheduler
32468 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32469 * config/aarch64/thunderx2t99.md: New file.
32471 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32473 * doc/standards.texi (Go Language): Update link to language
32476 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32478 * tree-eh.c (lower_resx): Sanitize profile.
32479 (cleanup_empty_eh_move_lp): Likewise.
32481 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32484 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32486 * cfgloopmanip.h (loop_version): Update prototype.
32487 * modulo-sched.c (sms_schedule): Update call of loop_version.
32488 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32489 * tree-parloops.c (gen_parallel_loop): Likewise.
32490 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32491 * tree-ssa-loop-split.c (split_loop): Likewise.
32492 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32493 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32495 2017-02-05 Martin Liska <mliska@suse.cz>
32498 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32500 (print_operand_address): Initialize a struct to zero.
32502 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32504 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32505 garbage collector only in textual form.
32507 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32509 * doc/extend.texi (x86 specific memory model extensions for
32510 transactional memory): Simplify a phrase.
32512 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32515 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32516 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32517 (atomic_storedi_1): Likewise.
32519 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32521 PR tree-optimization/79338
32522 * tree-parloops.c (gather_scalar_reductions): Don't call
32523 vect_analyze_loop_form for loop->inner before destroying loop's
32526 2017-02-03 Martin Sebor <msebor@redhat.com>
32528 PR tree-optimization/79327
32529 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32530 when precision has resulted in leading zeros.
32531 (format_integer): Adjust the likely counter to assume an unknown
32532 argument that may be zero is non-zero.
32534 2017-02-03 Jason Merrill <jason@redhat.com>
32537 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32538 avoid copying non-taken branch.
32540 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32542 PR tree-optimization/79340
32543 * tree-vect-loop.c (vectorizable_reduction): Release
32544 vec_defs elements after safe_splicing them into other vectors.
32547 PR tree-optimization/79327
32548 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32549 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32551 (format_integer): Use wide_int_to_tree instead of build_int_cst
32552 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32553 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32554 of shortest and longest sequence.
32556 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32558 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32559 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32561 2017-02-03 Walter Lee <walt@tilera.com>
32564 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32565 after initial stackframe link reg save.
32566 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32568 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32571 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32572 wu for stxssp alternative.
32574 2017-02-03 Martin Sebor <msebor@redhat.com>
32576 PR tree-optimization/79352
32577 * gimple-fold.c (get_range_strlen): Add argument.
32578 (get_range_strlen): Change return type to bool.
32579 (get_maxval_strlen): Pass in a dummy argument.
32580 * gimple-fold.h (get_range_strlen): Change return type to bool.
32581 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32582 * tree.h (array_at_struct_end_p): Add argument.
32583 * tree.c (array_at_struct_end_p): Handle it.
32585 2017-02-03 Martin Liska <mliska@suse.cz>
32588 * multiple_target.c (create_dispatcher_calls): Redirect edge
32589 from a caller of a dispatcher.
32590 (expand_target_clones): Make the clones local.
32591 (ipa_target_clone): Do both target clones and resolvers.
32592 (ipa_dispatcher_calls): Remove the pass.
32593 (pass_dispatcher_calls::gate): Likewise.
32594 (make_pass_dispatcher_calls): Likewise.
32595 * passes.def (pass_target_clone): Put as very first IPA early
32598 2017-02-03 Martin Liska <mliska@suse.cz>
32600 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32601 in case of a function with ifunc attribute.
32603 2017-02-03 Martin Liska <mliska@suse.cz>
32605 * cgraph.c (cgraph_node::dump): Dump function version info.
32606 * symtab.c (symtab_node::dump_base): Add missing new line.
32608 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32610 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32611 (ifcombine_ifandif): Use it.
32613 2017-02-03 Martin Liska <mliska@suse.cz>
32615 * doc/invoke.texi: Document default value for
32616 use-after-scope-direct-emission-threshold.
32618 2017-02-03 Martin Liska <mliska@suse.cz>
32620 PR tree-optimization/79339
32621 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32622 (format_floating): Likewise.
32624 2017-02-03 Martin Liska <mliska@suse.cz>
32627 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32629 (ipa_node_params_t::remove): Likewise.
32630 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32631 initialization from removed ipa_node_params_t::insert.
32632 (ipa_node_params::~ipa_node_params): Move from removed
32633 ipa_node_params_t::release.
32634 * symbol-summary.h (symbol_summary::m_released): New member.
32635 Do not release a summary twice. Do not allow to call finalizer
32636 for types of a summary that live in GGC memory.
32638 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32640 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32643 2017-02-02 Martin Sebor <msebor@redhat.com>
32645 PR middle-end/79275
32646 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32647 (format_string): Tighten up the range of output for non-constant
32648 strings and correct the expected range for wide non-constant strings.
32650 2017-02-02 Martin Sebor <msebor@redhat.com>
32652 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32654 PR middle-end/32003
32655 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32657 (-fdump-tree-@var): Add to index and document how to come up
32658 with pass-specific option and dump file names.
32659 (-fdump-passes): Clarify where to look for output.
32661 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32663 PR middle-end/77445
32664 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32665 statistics of the analyzed path; allow threading for speed when
32666 any of BBs along the path are optimized for speed.
32668 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32670 PR middle-end/78468
32671 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32672 settings of the virtual registers.
32675 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32677 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32678 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32681 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32683 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32684 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32686 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32688 * config/s390/s390.md: Add missing comments with the expanded
32690 * config/s390/vector.md: Likewise.
32691 * config/s390/vx-builtins.md: Likewise.
32693 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32696 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32697 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32698 conditions on a single line.
32700 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32702 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32703 __S390_VX__ to __VX__.
32705 2017-02-01 Andrew Pinski <apinski@cavium.com>
32707 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32708 stmt_info to record_stmt_cost.
32709 (vect_get_known_peeling_cost): Pass stmt_info if known to
32711 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32712 cpu_vector_cost field into
32713 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32714 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32715 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32716 splitting of scalar_stmt_cost and vec_stmt_cost.
32717 (thunderx_vector_cost): Likewise.
32718 (cortexa57_vector_cost): LIkewise.
32719 (exynosm1_vector_cost): Likewise.
32720 (xgene1_vector_cost): Likewise.
32721 (thunderx2t99_vector_cost): Improve after the splitting of the two
32723 (aarch64_builtin_vectorization_cost): Update for the splitting of
32724 scalar_stmt_cost and vec_stmt_cost.
32726 2017-02-01 Torvald Riegel <triegel@redhat.com>
32727 Richard Henderson <rth@redhat.com>
32729 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32730 conditional on existance of a fast atomic load.
32731 * optabs-query.c (can_atomic_load_p): New function.
32732 * optabs-query.h (can_atomic_load_p): Declare it.
32733 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32734 no fast atomic load is available for the particular size of access.
32735 (expand_atomic_compare_and_swap): Likewise.
32736 (expand_atomic_load): Likewise.
32737 (expand_atomic_store): Likewise.
32738 (expand_atomic_fetch_op): Likewise.
32739 * testsuite/lib/target-supports.exp
32740 (check_effective_target_sync_int_128): Remove x86 because it provides
32741 no fast atomic load.
32742 (check_effective_target_sync_int_128_runtime): Likewise.
32744 2017-02-01 Richard Biener <rguenther@suse.de>
32746 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32747 (graphite_transform_loops): Provide opt-info for optimized nests.
32748 * tree-parloop.c (parallelize_loops): Provide opt-info for
32749 parallelized loops.
32751 2017-02-01 Richard Biener <rguenther@suse.de>
32753 PR middle-end/79315
32754 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32755 was not set before.
32757 2017-02-01 Richard Biener <rguenther@suse.de>
32759 PR tree-optimization/71824
32760 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32761 Verify the loops are valid in the merged SESE region.
32762 (scop_detection::can_represent_loop_1): Check analyzing the
32763 evolution of the number of iterations in the region succeeds.
32765 2017-01-31 Ian Lance Taylor <iant@golang.org>
32767 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32768 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32770 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32772 PR preprocessor/79210
32773 * input.c (get_substring_ranges_for_loc): Replace line_width
32774 assertion with error-handling.
32776 2017-01-31 Richard Biener <rguenther@suse.de>
32778 PR tree-optimization/77318
32779 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32780 (create_pw_aff_from_tree): Take loop parameter.
32781 (add_condition_to_pbb): Pass loop of the condition to
32782 create_pw_aff_from_tree.
32784 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32786 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32787 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32789 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32793 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32795 (convert_int_to_float128): Likewise.
32796 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32797 (convert_int_to_float128): Likewise.
32798 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32799 (UNSPEC_IEEE128_CONVERT): Likewise.
32800 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32801 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32802 Use local variables for IBM extended format.
32803 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32804 (fix_trunc<mode>si2_fprs): Likewise.
32805 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32806 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32807 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32808 to know that we can now have integers of all sizes in vector
32810 (fix<uns>_<mode>di2_hw): Likewise.
32811 (float<uns>_<mode>si2_hw): Likewise.
32812 (fix_<mode>si2_hw): Likewise.
32813 (fixuns_<mode>si2_hw): Likewise.
32814 (float<uns>_<mode>di2_hw): Likewise.
32815 (float_<mode>di2_hw): Likewise.
32816 (float_<mode>si2_hw): Likewise.
32817 (floatuns_<mode>di2_hw): Likewise.
32818 (floatuns_<mode>si2_hw): Likewise.
32819 (xscvqp<su>wz_<mode>): Delete, no longer used.
32820 (xscvqp<su>dz_<mode>): Likewise.
32821 (xscv<su>dqp_<mode>): Likewise.
32822 (ieee128_mfvsrd_64bit): Likewise.
32823 (ieee128_mfvsrd_32bit): Likewise.
32824 (ieee128_mfvsrwz): Likewise.
32825 (ieee128_mtvsrw): Likewise.
32826 (ieee128_mtvsrd_64bit): Likewise.
32827 (ieee128_mtvsrd_32bit): Likewise.
32829 2017-01-31 Martin Liska <mliska@suse.cz>
32832 * ipa-prop.c (ipa_free_all_node_params): Call release method
32833 instead of ~sumbol_summary to not to trigger double times
32836 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32838 PR tree-optimization/71691
32839 * bitmap.h (class auto_bitmap): New.
32840 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32841 is_maybe_undefined instead of ssa_undefined_value_p.
32843 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32845 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32846 __S390_ARCH_LEVEL__ to __ARCH__.
32848 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32850 PR tree-optimization/79267
32851 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32852 if should_remove_lhs_p is true.
32854 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32857 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32858 (add_alignment_attribute): New.
32859 (base_type_die): Add alignment attribute.
32860 (subrange_type_die): Likewise.
32861 (modified_type_die): Likewise.
32862 (gen_array_type_die): Likewise.
32863 (gen_descr_array_type_die: Likewise.
32864 (gen_enumeration_type_die): Likewise.
32865 (gen_subprogram_die): Likewise.
32866 (gen_variable_die): Likewise.
32867 (gen_field_die): Likewise.
32868 (gen_ptr_to_mbr_type_die): Likewise.
32869 (gen_struct_or_union_type_die): Likewise.
32870 (gen_subroutine_type_die): Likewise.
32871 (gen_typedef_die): Likewise.
32872 (base_type_cmp): Compare alignment attribute.
32874 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32877 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32878 (setb_unsigned) New pattern for setb with CCUNS.
32879 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32880 subfc./subfe sequence to avoid overflow problems. Generate a
32881 shorter sequence with cmpld/setb for power9.
32882 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32883 for generating subfc. instruction.
32884 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32885 now uses this instruction.
32887 2017-01-30 Ian Lance Taylor <iant@google.com>
32890 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32891 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32893 2017-01-30 Martin Sebor <msebor@redhat.com>
32895 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32896 Move constant to the right of a relational operator.
32897 (get_mpfr_format_length, format_character, format_string): Ditto.
32898 (should_warn_p, maybe_warn): Same.
32900 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32902 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32905 * asan.c (get_translation_unit_decl): Remove function.
32906 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32908 2017-01-30 Martin Liska <mliska@suse.cz>
32910 PR gcov-profile/79259
32911 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32912 -fprofile-generate.
32914 2017-01-30 Martin Liska <mliska@suse.cz>
32917 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32918 Initialize variables with NULL value.
32920 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32923 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32925 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32927 2017-01-30 Richard Biener <rguenther@suse.de>
32929 PR tree-optimization/79276
32930 * tree-vrp.c (process_assert_insertions): Properly adjust common
32931 when removing a duplicate.
32933 2017-01-30 Richard Biener <rguenther@suse.de>
32935 PR tree-optimization/79256
32936 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32937 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32939 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32941 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32944 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32945 ("*r<noxa>sbg_<mode>_sll_bitmask")
32946 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32947 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32948 Use contiguous_bitmask_nowrap_operand.
32950 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32953 * config/rs6000/altivec.h (vec_xl): Revise #define.
32954 (vec_xst): Likewise.
32956 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32958 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32960 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32962 PR rtl-optimization/79194
32963 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32964 traps before call to bypass_conditional_jumps.
32966 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32968 PR tree-optimization/71374
32969 * lra-constraints.c (check_conflict_input_operands): New.
32970 (match_reload): Use it.
32972 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32975 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32976 account to calculate conflict_set.
32978 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32980 PR rtl-optimization/78559
32981 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32982 other_insn in combine.
32984 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32986 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32987 uint16_type_node for BT_UINT16.
32989 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32991 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32992 "RTL Tests" to menu.
32993 (GIMPLE Tests): New node.
32994 (RTL Tests): New node.
32996 2017-01-27 Richard Biener <rguenther@suse.de>
32998 PR tree-optimization/79245
32999 * tree-loop-distribution.c (distribute_loop): Apply cost
33000 modeling also to detected patterns.
33002 2017-01-27 Richard Biener <rguenther@suse.de>
33004 PR tree-optimization/71433
33005 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
33006 (compare_assert_loc): New function.
33007 (process_assert_insertions): Sort and optimize assert locations
33008 to remove duplicates and push down identical assertions on
33009 edges to their destination block.
33011 2017-01-27 Richard Biener <rguenther@suse.de>
33013 PR tree-optimization/79244
33014 * tree-vrp.c (remove_range_assertions): Forcefully propagate
33015 out SSA names even if abnormal.
33017 2017-01-27 Jakub Jelinek <jakub@redhat.com>
33019 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
33020 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
33021 instead of MPFR_RNDN.
33023 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
33026 * arm.c (arm_option_override): Don't call build_target_option_node
33027 until after doing all option overrides.
33028 (arm_valid_target_attribute_tree): Likewise.
33030 2017-01-27 Martin Liska <mliska@suse.cz>
33032 * doc/invoke.texi (-fprofile-arcs): Document profiling support
33033 for {cd}tors and C++ {cd}tors.
33035 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
33037 * config/s390/s390.md ("*setmem_long_and")
33038 ("*setmem_long_and_31z"): Use zero_extend instead of and.
33040 2017-01-26 Martin Sebor <msebor@redhat.com>
33042 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
33045 2017-01-26 Martin Sebor <msebor@redhat.com>
33047 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
33048 HAVE_DFmode before using XFmode or DFmode.
33049 (parse_directive): Avoid using the z length modifier to avoid
33050 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
33052 PR middle-end/78703
33053 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
33054 to accept adjustment as an array.
33055 (get_int_range): New function.
33056 (struct directive): Make width and prec arrays.
33057 (directive::set_width, directive::set_precision): Call get_int_range.
33058 (format_integer, format_floating): Handle width and precision ranges.
33059 (format_string, parse_directive): Same.
33061 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33064 * dwarf2out.c (generate_skeleton_bottom_up): For children with
33065 comdat_type_p set, just clone them, but keep the children in the
33069 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
33070 which have direct callers with -fvar-tracking-assignments enabled
33072 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
33073 inside of type units.
33075 2017-01-26 Martin Sebor <msebor@redhat.com>
33077 PR middle-end/78703
33078 * gimple-ssa-sprintf.c (struct result_range): Add likely and
33080 (struct format_result): Replace number_chars, number_chars_min,
33081 and number_chars_max with a single member of struct result_range.
33083 (format_result::operator+=): Adjust.
33084 (struct fmtresult): Remove bounded. Handle likely and unlikely
33086 (fmtresult::adjust_for_width_or_precision): New function.
33087 (fmtresult:type_max_digits): New function.
33088 (bytes_remaining): Handle likely and unlikely counters.
33089 (min_bytes_remaining): Remove.
33090 (format_percent): Simplify.
33091 (format_integer, format_floating): Set likely and unlikely counters.
33092 (get_string_length, format_character, format_string): Same.
33093 (format_plain, should_warn_p): New function.
33094 (maybe_warn): Call should_warn_p. Update diagnostic messages
33095 and handle those for all directives, including plain strings.
33096 (format_directive): Handle likely and unlikely counters.
33097 Remove unnecessary quoting from diagnostics. Add an informational
33099 (add_bytes): Remove.
33100 (pass_sprintf_length::compute_format_length): Simplify.
33101 (try_substitute_return_value): Handle likely and unlikely counters.
33103 2017-01-26 Carl Love <cel@us.ibm.com>
33105 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33106 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33108 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33111 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33112 endianess for subregs into account.
33113 * lra-constraints.c (lra_constraints): Do risky transformations
33114 always on the first iteration.
33115 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33116 last_call_used_reg_set.
33117 (process_bb_lives): Define and use last_call_used_reg_set.
33118 * lra.c (lra): Always continue after lra_constraints on the first
33121 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33123 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33125 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33127 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33129 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33130 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33131 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33132 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33133 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33134 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33135 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33136 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33137 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33139 2017-01-26 Marek Polacek <polacek@redhat.com>
33142 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33143 for the third operand.
33145 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33147 PR middle-end/79236
33148 * omp-low.c (struct omp_context): Add simt_stmt field.
33149 (scan_omp_for): Return omp_context *.
33150 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33151 context to the _simt_ SIMD stmt.
33152 (lower_omp_for): For combined SIMD with sibling _simt_
33153 SIMD, make sure to use the same decls in _looptemp_
33154 clauses as in the sibling.
33156 2017-01-26 David Sherwood <david.sherwood@arm.com>
33158 PR middle-end/79212
33159 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33162 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33165 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33166 emit fld b; fld a; if possible.
33168 * brig-builtins.def: Update copyright years.
33169 * config/arm/arm_acle_builtins.def: Update copyright years.
33171 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33174 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33175 constraint instead of o for the stxsd instruction.
33177 2017-01-25 Carl Love <cel@us.ibm.com>
33179 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33180 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33182 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33184 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33186 2017-01-25 Richard Biener <rguenther@suse.de>
33188 PR tree-optimization/69264
33189 * target.def (vector_alignment_reachable): Improve documentation.
33190 * doc/tm.texi: Regenerate.
33191 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33193 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33194 earlier changes with respect to TYPE_USER_ALIGN.
33195 (vector_alignment_reachable_p): Likewise. Improve dumping.
33197 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33200 * config/arm/arm.md (xordi3): Force constant operand into a register
33203 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33205 * doc/invoke.texi (-fstore-merging): Correct default optimization
33206 levels at which it is enabled.
33207 (-O): Move -fstore-merging from list to...
33210 2017-01-25 Richard Biener <rguenther@suse.de>
33213 * omp-expand.c: Include debug.h.
33214 (expand_omp_taskreg): Make sure to generate early debug before
33215 outlining anything from a function.
33216 (expand_omp_target): Likewise.
33217 (grid_expand_target_grid_body): Likewise.
33219 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33222 * asan.c (get_translation_unit_decl): New function.
33223 (asan_add_global): Extract modules file name from globals
33224 TRANSLATION_UNIT_DECL name.
33226 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33229 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33230 for long calls with APCS frame and VFP.
33232 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33234 * cfg.c (original_copy_tables_initialized_p): New function.
33235 * cfg.h (original_copy_tables_initialized_p): New decl.
33236 * cfgrtl.c (relink_block_chain): Guard the call to
33237 free_original_copy_tables with a call to
33238 original_copy_tables_initialized_p.
33239 * cgraph.h (symtab_node::native_rtl_p): New decl.
33240 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33241 (symtab_node::needed_p): Don't assert for early assembly output
33242 for __RTL functions.
33243 (cgraph_node::finalize_function): Set "force_output" for __RTL
33245 (cgraph_node::analyze): Bail out early for __RTL functions.
33246 (analyze_functions): Update assertion to support __RTL functions.
33247 (cgraph_node::expand): Bail out early for __RTL functions.
33248 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33250 * function.h (struct function): Update comment for field
33252 * gimple-expr.c: Include "tree-pass.h".
33253 (gimple_has_body_p): Return false for __RTL functions.
33254 * Makefile.in (OBJS): Add run-rtl-passes.o.
33255 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33257 (gcc::pass_manager::get_clean_slate): New accessor.
33258 * passes.c: Include "insn-addr.h".
33259 (should_skip_pass_p): Add logging. Update logic for running
33260 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33261 property-provider override so it is only done for gimple passes.
33263 (skip_pass): New function.
33264 (execute_one_pass): Call skip_pass when skipping passes.
33265 * read-md.c (md_reader::read_char): Support filtering
33266 the input to a subset of line numbers.
33267 (md_reader::md_reader): Initialize fields
33268 m_first_line and m_last_line.
33269 (md_reader::read_file_fragment): New function.
33270 * read-md.h (md_reader::read_file_fragment): New decl.
33271 (md_reader::m_first_line): New field.
33272 (md_reader::m_last_line): New field.
33273 * read-rtl-function.c (function_reader::create_function): Only
33274 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33275 curr_properties. Set DECL_INITIAL to a dummy block.
33276 (read_rtl_function_body_from_file_range): New function.
33277 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33279 * run-rtl-passes.c: New file.
33280 * run-rtl-passes.h: New file.
33282 2017-01-24 Jeff Law <law@redhat.com>
33284 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33287 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33289 PR tree-optimization/79159
33290 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33291 (record_nonwrapping_iv): Improve boundary using above function if no
33292 value range information.
33294 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33295 Martin Jambor <mjambor@suse.cz>
33297 * brig-builtins.def: New file.
33298 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33299 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33300 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33301 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33302 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33303 * builtin-types.def (BT_INT8): New.
33304 (BT_INT16): Likewise.
33305 (BT_UINT8): Likewise.
33306 (BT_UINT16): Likewise.
33307 (BT_FN_ULONG): Likewise.
33308 (BT_FN_UINT_INT): Likewise.
33309 (BT_FN_UINT_ULONG): Likewise.
33310 (BT_FN_UINT_LONG): Likewise.
33311 (BT_FN_UINT_PTR): Likewise.
33312 (BT_FN_ULONG_PTR): Likewise.
33313 (BT_FN_INT8_FLOAT): Likewise.
33314 (BT_FN_INT16_FLOAT): Likewise.
33315 (BT_FN_UINT32_FLOAT): Likewise.
33316 (BT_FN_UINT16_FLOAT): Likewise.
33317 (BT_FN_UINT8_FLOAT): Likewise.
33318 (BT_FN_UINT64_FLOAT): Likewise.
33319 (BT_FN_UINT16_UINT32): Likewise.
33320 (BT_FN_UINT32_UINT16): Likewise.
33321 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33322 (BT_FN_INT_PTR_INT): Likewise.
33323 (BT_FN_UINT_PTR_UINT): Likewise.
33324 (BT_FN_LONG_PTR_LONG): Likewise.
33325 (BT_FN_ULONG_PTR_ULONG): Likewise.
33326 (BT_FN_VOID_UINT64_UINT64): Likewise.
33327 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33328 (BT_FN_INT8_INT8_INT8): Likewise.
33329 (BT_FN_INT16_INT16_INT16): Likewise.
33330 (BT_FN_INT_INT_INT): Likewise.
33331 (BT_FN_UINT_FLOAT_UINT): Likewise.
33332 (BT_FN_FLOAT_UINT_UINT): Likewise.
33333 (BT_FN_ULONG_UINT_UINT): Likewise.
33334 (BT_FN_ULONG_UINT_PTR): Likewise.
33335 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33336 (BT_FN_UINT_UINT_UINT): Likewise.
33337 (BT_FN_VOID_UINT_PTR): Likewise.
33338 (BT_FN_UINT_UINT_PTR: Likewise.
33339 (BT_FN_UINT32_UINT64_PTR): Likewise.
33340 (BT_FN_INT_INT_UINT_UINT): Likewise.
33341 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33342 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33343 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33344 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33345 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33346 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33347 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33348 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33349 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33350 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33351 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33352 * doc/frontends.texi: List BRIG FE.
33353 * doc/install.texi (Testing): Add BRIG tesring requirements.
33354 * doc/invoke.texi (Overall Options): Mention BRIG.
33355 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33357 2017-01-24 Richard Biener <rguenther@suse.de>
33359 PR translation/79208
33360 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33362 2017-01-24 Martin Jambor <mjambor@suse.cz>
33365 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33366 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33367 and known_contexts.
33369 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33371 PR middle-end/79123
33372 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33373 casts from signed to unsigned really don't have a range.
33375 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33377 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33378 GMP_RNDx for compatiblity.
33380 2017-01-24 Martin Liska <mliska@suse.cz>
33383 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33384 that would prevent us to call alloca with -1 as argument.
33386 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33388 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33389 Avoid -Wformat-security warning.
33391 2017-01-23 Andrew Pinski <apinski@cavium.com>
33393 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33396 2017-01-23 Martin Sebor <msebor@redhat.com>
33398 PR middle-end/78703
33399 * gimple-ssa-sprintf.c (warn_level): New global.
33400 (format_integer): Use it here and throughout the rest of the file.
33401 Use the same switch to compute sign as base.
33402 (maybe_warn): New function.
33403 (format_directive): Factor out warnings into maybe_warn.
33404 Add debugging output. Use warn_level.
33405 (add_bytes): Use warn_level.
33406 (pass_sprintf_length::compute_format_length): Add debugging output.
33407 (try_substitute_return_value): Same.
33408 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33410 PR middle-end/78703
33411 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33412 (struct fmtresult, format_integer, format_floating): Adjust.
33413 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33414 (get_string_length, format_string,format_directive): Same.
33415 (pass_sprintf_length::compute_format_length): Same.
33416 (try_substitute_return_value): Simplify slightly.
33418 PR middle-end/78703
33419 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33420 (fmtresult::operator+=): Outlined.
33421 (struct fmtresult): Add ctors.
33422 (struct conversion_spec): Rename...
33423 (struct directive): ...to this. Add and remove data members.
33424 (directive::set_width, directive::set_precision): New functions.
33425 (format_percent): Use fmtresult ctor.
33426 (get_width_and_precision): Remove.
33427 (format_integer): Make naming changes. Avoid computing width and
33429 (format_floating): Same. Adjust indentation.
33430 (format_character, format_none): New functions.
33431 (format_string): Moved character handling to format_character.
33432 (format_directive): Remove arguments, change return type.
33433 (parse_directive): New function.
33434 (pass_sprintf_length::compute_format_length): Move directive
33435 parsing to parse_directive.
33437 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33439 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33440 (assign_assembler_name_if_needed): ... this.
33441 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33442 (assign_assembler_name_if_needed): ... this.
33443 (free_lang_data_in_cgraph): Adjust callers.
33444 * cgraphunit.c (cgraph_node::analyze): Likewise.
33445 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33448 2017-01-23 Richard Biener <rguenther@suse.de>
33450 PR tree-optimization/79088
33451 PR tree-optimization/79188
33452 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33453 resetting loop bounds after last path deletion. Reset loop
33454 bounds of the target loop, make code match the comments.
33455 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33456 Make sure loops need no fixups.
33458 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33460 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33461 exponent support with double type for first argument.
33462 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33463 type returned by __builtin_vec_extract_sig,
33464 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33465 functions from "vector int" to "vector unsigned int" or from
33466 "vector long long int" to "vector unsigned long long int".
33467 Changed type returned by __builtin_vec_extract_exp,
33468 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33469 functions from "vector int" to "vector unsigned int" or from
33470 "vector long long int" to "vector unsigned long long int".
33471 Changed return type of __builtin_vec_test_data_class,
33472 __builtin_vec_test_data_class_sp, and
33473 __builtin_vec_test_data_class_dp from "vector int" to
33474 "vector bool int" or from "vector long long int" to "vector bool
33475 long long int" and changed second argument type from "unsigned
33476 int" to "int". Added new overloaded function forms "vector float
33477 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33478 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33479 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33480 double, vector unsigned long long int)" and "vector double
33481 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33482 long int)". Changed return type of
33483 __builtin_scalar_test_data_class and
33484 __builtin_scalar_test_data_class_sp and
33485 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33486 int" and changed second argument from "unsigned int" to "int".
33487 Changed type returned by __builtin_scalar_test_neg,
33488 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33489 from "int" to "bool int". Added new overloaded function form
33490 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33491 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33492 exponent double-precision with floating point first argument.
33493 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33494 documentation of scalar_test_data_class, scalar_test_neg,
33495 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33496 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33497 vec_test_data_class built-in functions to reflect refinements in
33498 their type signatures.
33500 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33502 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33504 (aarch64_elf_asm_destructor): Likewise.
33506 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33508 PR rtl-optimization/78634
33509 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33510 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33511 * ifcvt.c (noce_try_cmove): Add missing cost check.
33513 PR rtl-optimization/71724
33514 * combine.c (if_then_else_cond): Look for situations where it is
33515 beneficial to undo the work of one of the recursive calls.
33517 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33519 PR tree-optimization/70754
33520 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33521 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33522 combined stmt before it if not NULL.
33523 (combine_chains): Process refs reversely and compute dominance point
33526 2017-01-23 Martin Liska <mliska@suse.cz>
33528 PR tree-optimization/79196
33529 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33530 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33532 (strlen_optimize_stmt): Call the renamed function.
33534 2017-01-23 Michael Matz <matz@suse.de>
33536 PR tree-optimization/78384
33537 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33539 2017-01-23 Richard Biener <rguenther@suse.de>
33541 PR tree-optimization/79186
33542 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33543 both incoming edges before moving an assert.
33545 2017-01-23 Martin Jambor <mjambor@suse.cz>
33547 * ipa-prop.c (load_from_param_1): Removed.
33548 (load_from_unmodified_param): Bits from load_from_param_1 put back
33550 (load_from_param): Removed.
33551 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33552 with stmt. Reverted back to use of load_from_unmodified_param.
33554 2017-01-23 Martin Jambor <mjambor@suse.cz>
33557 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33558 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33559 field a pointer to garbage collected vector, mark lattices and
33560 ipcp_orig_node with GTY((skip)).
33561 (ipa_get_param_count): Adjust to descriptors being a pointer.
33562 (ipa_get_param): Likewise.
33563 (ipa_get_type): Likewise.
33564 (ipa_get_param_move_cost): Likewise.
33565 (ipa_set_param_used): Likewise.
33566 (ipa_get_controlled_uses): Likewise.
33567 (ipa_set_controlled_uses): Likewise.
33568 (ipa_is_param_used): Likewise.
33569 (ipa_node_params_t): Move into garbage collector. New methods insert
33571 (ipa_node_params_sum): Annotate wth GTY(()).
33572 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33574 (ipa_load_from_parm_agg): Adjust declaration.
33575 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33576 * ipa-profile.c (ipa_profile): Likewise.
33577 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33578 (ipa_populate_param_decls): Make descriptors parameter garbage
33580 (ipa_dump_param): Adjust to descriptors being a pointer.
33581 (ipa_alloc_node_params): Likewise.
33582 (ipa_initialize_node_params): Likewise.
33583 (load_from_param_1): Make descriptors parameter garbage collected.
33584 (load_from_unmodified_param): Likewise.
33585 (load_from_param): Likewise.
33586 (ipa_load_from_parm_agg): Likewise.
33587 (ipa_node_params::~ipa_node_params): Removed.
33588 (ipa_free_all_node_params): Remove call to delete operator.
33589 (ipa_node_params_t::insert): New.
33590 (ipa_node_params_t::remove): Likewise.
33591 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33592 copy known_csts and known_contexts vectors.
33593 (ipa_read_node_info): Adjust to descriptors being a pointer.
33594 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33596 (ipcp_transform_function): Make descriptors variable garbage
33599 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33601 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33602 * config/i386/avx512dqintrin.h: Ditto.
33603 * config/i386/avx512fintrin.h: Ditto.
33604 * config/i386/i386.c: Handle new builtins.
33605 * config/i386/i386-builtin.def: Add new builtins.
33606 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33607 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33609 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33610 Martin Liska <mliska@suse.cz>
33612 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33613 * asan.c (asan_expand_poison_ifn): Support stores and use
33614 appropriate ASAN report function.
33615 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33616 * internal-fn.def (ASAN_POISON_USE): Declare.
33617 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33618 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33619 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33620 ASAN_POISON calls w/o LHS.
33621 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33622 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33623 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33624 * gimplify.c (asan_poison_variables): Add attribute
33625 use_after_scope_memory to variables that really needs to live
33627 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33628 having the attribute.
33630 2017-01-23 Martin Liska <mliska@suse.cz>
33632 * asan.c (create_asan_shadow_var): New function.
33633 (asan_expand_poison_ifn): Likewise.
33634 * asan.h (asan_expand_poison_ifn): New declaration.
33635 * internal-fn.c (expand_ASAN_POISON): Likewise.
33636 * internal-fn.def (ASAN_POISON): New builtin.
33637 * sanopt.c (pass_sanopt::execute): Expand
33638 asan_expand_poison_ifn.
33639 * tree-inline.c (copy_decl_for_dup_finish): Make function
33641 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33642 * tree-ssa.c (is_asan_mark_p): New function.
33643 (execute_update_addresses_taken): Rewrite local variables
33644 (identified just by use-after-scope as addressable) into SSA.
33646 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33648 * doc/install.texi (Specific): opensource.apple.com uses https
33649 now. Remove trailing slash.
33651 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33653 * README.Portability: Remove note on an Irix compatibility issue.
33655 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33657 * gcov.c (INCLUDE_ALGORITHM): Define.
33658 (INCLUDE_VECTOR): Define.
33659 No longer include <vector> and <algorithm> directly.
33661 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33663 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33665 * doc/invoke.texi (Code Gen Options): Ditto.
33667 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33670 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33672 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33674 rtl-optimization/79125
33675 * cprop.c (local_cprop_pass): Handle cases where we make an
33676 unconditional trap.
33678 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33682 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33683 read from, for big endian.
33685 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33687 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33688 register pauth builtins for LP64 only.
33690 2017-01-20 Marek Polacek <polacek@redhat.com>
33693 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33696 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33698 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33700 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33701 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33702 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33704 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33707 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33708 in big-endian mode when they are not a single duplicated value.
33710 2017-01-20 Richard Biener <rguenther@suse.de>
33712 * BASE-VER: Bump to 7.0.1.
33714 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33716 * omp-low.c (omplow_simd_context): New struct. Use it...
33717 (lower_rec_simd_input_clauses): ...here and...
33718 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33719 references to idx, lane, max_vf, is_simt.
33721 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33723 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33726 2017-01-20 Martin Jambor <mjambor@suse.cz>
33728 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33729 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33731 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33732 (GTFILES): Rename hsa.c to hsa-common.c.
33733 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33734 * hsa-dump.c: Likewise.
33735 * hsa-gen.c: Likewise.
33736 * hsa-regalloc.c: Likewise.
33737 * ipa-hsa.c: Likewise.
33738 * omp-expand.c: Likewise.
33739 * omp-low.c: Likewise.
33740 * toplev.c: Likewise.
33742 2017-01-20 Marek Polacek <polacek@redhat.com>
33745 * doc/invoke.texi: Document -Wduplicated-branches.
33746 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33747 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33748 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33749 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33750 return 0 only when not OEP_LEXICOGRAPHIC.
33751 (fold_build_cleanup_point_expr): Use the expression
33752 location when building CLEANUP_POINT_EXPR.
33753 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33754 * tree.c (add_expr): Handle error_mark_node.
33756 2017-01-20 Martin Liska <mliska@suse.cz>
33759 * tree-profile.c (init_ic_make_global_vars): Do not call
33761 (gimple_init_gcov_profiler): Likewise.
33763 2017-01-20 Martin Liska <mliska@suse.cz>
33766 * cgraph.h (maybe_create_reference): Remove argument and
33768 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33770 * ipa-cp.c (create_specialized_node): Likewise.
33771 * symtab.c (symtab_node::maybe_create_reference): Handle
33772 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33774 2017-01-20 Martin Liska <mliska@suse.cz>
33776 * read-rtl-function.c (function_reader::create_function): Use
33777 build_decl instread of build_decl_stat.
33779 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33781 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33782 * config/i386/avx512dqintrin.h: Ditto.
33783 * config/i386/avx512fintrin.h: Ditto.
33784 * config/i386/i386-builtin-types.def: Add new types.
33785 * config/i386/i386.c: Handle new types.
33786 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33787 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33788 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33789 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33790 (__builtin_ia32_kshiftridi): New.
33791 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33793 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33797 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33798 define to rs6000_init_stack_protect_guard.
33799 (rs6000_init_stack_protect_guard): New function.
33801 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33802 Yunqiang Su <yunqiang.su@imgtec.com>
33804 * config.gcc (supported_defaults): Add madd4.
33805 (with_madd4): Add validation.
33806 (all_defaults): Add madd4.
33807 * config/mips/mips.opt (mmadd4): New option.
33808 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33810 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33812 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33813 (ISA_HAS_FUSED_MADD4): Likewise.
33814 * doc/invoke.texi (-mmadd4): Document the new option.
33815 * doc/install.texi (--with-madd4): Document the new option.
33817 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33819 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33820 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33821 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33822 (aarch64_init_pauth_hint_builtins): New.
33823 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33824 (aarch64_expand_builtin): Expand new builtins.
33826 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33828 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33829 * combine-stack-adj.c (no_unhandled_cfa): Handle
33830 REG_CFA_TOGGLE_RA_MANGLE.
33831 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33832 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33833 info for return address signing.
33834 (aarch64_expand_epilogue): Likewise.
33836 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33838 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33839 * config/aarch64/aarch64-protos.h
33840 (aarch64_return_address_signing_enabled): New declaration.
33841 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33843 (aarch64_expand_prologue): Sign return address before it's pushed onto
33845 (aarch64_expand_epilogue): Authenticate return address fetched from
33847 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33848 (aarch64_attributes): New function attributes for "sign-return-address".
33849 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33850 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33851 ("*do_return"): Generate combined instructions according to key index.
33852 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33853 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33855 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33856 * config/aarch64/aarch64.opt (msign-return-address=): New.
33857 * doc/extend.texi (AArch64 Function Attributes): Documents
33858 "sign-return-address=".
33859 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33861 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33863 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33864 overall option summary.
33866 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33868 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33869 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33870 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33871 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33873 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33875 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33876 -mpower9-minmax by default for -mcpu=power9.
33877 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33878 128-bit floating point.
33880 2017-01-20 Alan Modra <amodra@gmail.com>
33882 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33883 optimizing for size.
33885 2017-01-20 Alan Modra <amodra@gmail.com>
33888 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33889 for strcmp and strncmp from corresponding builtin decl.
33891 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33893 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33894 instead of i386/rtems-64.h.
33895 * config/i386/rtems-64.h: Remove.
33897 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33901 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33903 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33905 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33907 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33908 Change int to HOST_WIDE_INT.
33909 * config/aarch64/aarch64-protos.h
33910 (aarch64_simd_gen_const_vector_dup): Likewise.
33911 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33913 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33915 * langhooks-def.h (lhd_type_for_size): New decl.
33916 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33917 * langhooks.c (lhd_type_for_size): New function, taken from
33920 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33922 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33923 define_bypass for CR latency.
33924 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33925 (power9-alu2): Add define_bypass for CR latency.
33927 (power9-mul): Update insn latency.
33928 (power9-mul-compare): Update insn latency, bypass latency and remove
33931 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33933 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33935 * config/aarch64/aarch64.md
33936 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33937 aarch64_nopcrelative_literal_loads.
33938 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33940 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33942 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33943 TARGET_LOONGSON_3A.
33944 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33946 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33949 * config.gcc (supported_defaults): Add lxc1-sxc1.
33950 (with_lxc1_sxc1): Add validation.
33951 (all_defaults): Add lxc1-sxc1.
33952 * config/mips/mips.opt (mlxc1-sxc1): New option.
33953 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33955 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33956 __mips_no_lxc1_sxc1.
33957 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33958 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33959 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33961 2017-01-19 Richard Biener <rguenther@suse.de>
33963 PR tree-optimization/72488
33964 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33965 sure to restore SSA info.
33966 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33968 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33970 PR rtl-optimization/79121
33971 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33972 of the inner type when shifting an extended value.
33974 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33977 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33978 interposable aliases.
33980 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33983 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33984 Use the evmergelohi instruction.
33985 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33986 (mov_sitf_e500_subreg8_2_be): Likewise.
33987 (mov_sitf_e500_subreg12_2_le): Likewise.
33988 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33989 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33990 (mov_sitf_e500_subreg8_2_le): Likewise.
33991 (mov_sitf_e500_subreg12_2_be): Likewise.
33993 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33995 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33996 attribute from vecsimple to vecperm.
33997 (altivec_vbpermq2): Likewise.
33999 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34002 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
34004 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34005 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
34006 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
34007 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
34008 case where N arg is SIZE_MAX.
34009 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
34010 (cmpstrsi): Add pattern.
34012 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34014 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
34015 __builtin_vec_revb builtins.
34016 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
34017 built-in functions to support generation of the ISA 3.0 XXBR<x>
34018 vector byte reverse instructions.
34019 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
34020 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
34021 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
34022 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
34023 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
34024 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
34025 (P9V_BUILTIN_VEC_REVB): Likewise.
34026 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
34027 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
34028 (p9_xxbrq_v16qi): Likewise.
34029 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
34030 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
34031 (p9_xxbrh_v8hi): Likewise.
34032 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
34033 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
34034 vec_revb built-in functions.
34036 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
34038 PR rtl-optimization/78952
34039 * config/i386/i386.md (any_extract): New code iterator.
34040 (*insvqi_2): Use any_extract for source operand.
34041 (*insvqi_3): Use any_shiftrt for source operand.
34043 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
34045 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
34047 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
34049 2017-01-18 Matthias Klose <doko@ubuntu.com>
34051 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
34053 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34055 * config/rs6000/altivec.h (vec_bperm): Change #define.
34056 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
34057 (altivec_vbpermq2): New define_insn.
34058 (altivec_vbpermd): Likewise.
34059 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
34060 function interface.
34061 (VBPERMD): Likewise.
34062 (VBPERM): New polymorphic function interface.
34063 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
34064 Add entries for P9V_BUILTIN_VEC_VBPERM.
34065 * doc/extend.texi: Add interfaces for vec_bperm.
34067 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34069 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
34070 first letter of error messages.
34071 (s390_resolve_overloaded_builtin): Likewise.
34072 * config/s390/s390.c (s390_expand_builtin): Likewise.
34073 (s390_invalid_arg_for_unprototyped_fn): Likewise.
34074 (s390_valid_target_attribute_inner_p): Likewise.
34075 * config/s390/s390.md ("tabort"): Likewise.
34077 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
34079 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
34080 (ISA_AVOID_DIV_HILO): New macro.
34081 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
34082 (ISA_HAS_DDIV): Likewise.
34084 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34086 * doc/invoke.texi (fabi-version): Correct number of occurrences.
34088 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34090 * doc/invoke.texi (fabi-version): Spelling fix.
34092 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34095 * doc/invoke.texi (fabi-version): Mention mangling fix for
34098 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34101 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34103 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34106 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34107 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34109 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34111 (stack_protect_test): Ditto.
34112 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34113 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34115 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34116 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34117 -mstack-protector-guard-offset=.
34118 (RS/6000 and PowerPC Options): Ditto.
34120 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34122 * config/i386/i386.h (MASK_CLASS_P): New define.
34123 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34124 there are no registers from different register sets also when
34125 mask registers are used. Update function comment.
34126 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34127 to (*k/*r) and (*k/*km) alternatives.
34129 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34131 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34132 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34133 (EH_RETURN_HANDLER_RTX): New define.
34134 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34135 Force frame pointer in EH return functions.
34136 (aarch64_expand_epilogue): Add barrier for eh_return.
34137 (aarch64_final_eh_return_addr): Remove.
34138 (aarch64_eh_return_handler_rtx): New function.
34139 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34141 (aarch64_eh_return_handler_rtx): New prototype.
34143 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34145 * config/rs6000/altivec.h (vec_rlmi): New #define.
34146 (vec_vrlnm): Likewise.
34147 (vec_rlnm): Likewise.
34148 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34149 (UNSPEC_VRLNM): Likewise.
34150 (VIlong): New mode iterator.
34151 (altivec_vrl<VI_char>mi): New define_insn.
34152 (altivec_vrl<VI_char>nm): Likewise.
34153 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34155 (VRLDNM): Likewise.
34156 (RLNM): New polymorphic function entry.
34157 (VRLWMI): New monomorphic function entry.
34158 (VRLDMI): Likewise.
34159 (RLMI): New polymorphic function entry.
34160 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34161 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34162 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34165 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34168 * dwarf2out.c (field_byte_offset): Restore the
34169 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34170 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34171 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34174 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34177 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34179 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34182 * dwarf2out.c (add_data_member_location_attribute): For constant
34183 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34184 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34185 DW_AT_byte_size attributes.
34187 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34189 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34190 after forcing to constant memory when the code model is medium.
34192 2017-01-17 Julia Koval <julia.koval@intel.com>
34195 * config/i386/avx512fintrin.h
34196 (_mm512_i32gather_ps): Change __addr type to void const*.
34197 (_mm512_mask_i32gather_ps): Ditto.
34198 (_mm512_i32gather_pd): Ditto.
34199 (_mm512_mask_i32gather_pd): Ditto.
34200 (_mm512_i64gather_ps): Ditto.
34201 (_mm512_mask_i64gather_ps): Ditto.
34202 (_mm512_i64gather_pd): Ditto.
34203 (_mm512_mask_i64gather_pd): Ditto.
34204 (_mm512_i32gather_epi32): Ditto.
34205 (_mm512_mask_i32gather_epi32): Ditto.
34206 (_mm512_i32gather_epi64): Ditto.
34207 (_mm512_mask_i32gather_epi64): Ditto.
34208 (_mm512_i64gather_epi32): Ditto.
34209 (_mm512_mask_i64gather_epi32): Ditto.
34210 (_mm512_i64gather_epi64): Ditto.
34211 (_mm512_mask_i64gather_epi64): Ditto.
34212 (_mm512_i32scatter_ps): Change __addr type to void*.
34213 (_mm512_mask_i32scatter_ps): Ditto.
34214 (_mm512_i32scatter_pd): Ditto.
34215 (_mm512_mask_i32scatter_pd): Ditto.
34216 (_mm512_i64scatter_ps): Ditto.
34217 (_mm512_mask_i64scatter_ps): Ditto.
34218 (_mm512_i64scatter_pd): Ditto.
34219 (_mm512_mask_i64scatter_pd): Ditto.
34220 (_mm512_i32scatter_epi32): Ditto.
34221 (_mm512_mask_i32scatter_epi32): Ditto.
34222 (_mm512_i32scatter_epi64): Ditto.
34223 (_mm512_mask_i32scatter_epi64): Ditto.
34224 (_mm512_i64scatter_epi32): Ditto.
34225 (_mm512_mask_i64scatter_epi32): Ditto.
34226 (_mm512_i64scatter_epi64): Ditto.
34227 (_mm512_mask_i64scatter_epi64): Ditto.
34228 * config/i386/avx512pfintrin.h
34229 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34230 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34231 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34232 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34233 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34234 (_mm512_prefetch_i32scatter_ps): Ditto.
34235 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34236 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34237 (_mm512_prefetch_i64scatter_pd): Ditto.
34238 (_mm512_prefetch_i64scatter_ps): Ditto.
34239 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34240 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34241 * config/i386/avx512vlintrin.h
34242 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34243 (_mm_mmask_i32gather_ps): Ditto.
34244 (_mm256_mmask_i32gather_pd): Ditto.
34245 (_mm_mmask_i32gather_pd): Ditto.
34246 (_mm256_mmask_i64gather_ps): Ditto.
34247 (_mm_mmask_i64gather_ps): Ditto.
34248 (_mm256_mmask_i64gather_pd): Ditto.
34249 (_mm_mmask_i64gather_pd): Ditto.
34250 (_mm256_mmask_i32gather_epi32): Ditto.
34251 (_mm_mmask_i32gather_epi32): Ditto.
34252 (_mm256_mmask_i32gather_epi64): Ditto.
34253 (_mm_mmask_i32gather_epi64): Ditto.
34254 (_mm256_mmask_i64gather_epi32): Ditto.
34255 (_mm_mmask_i64gather_epi32): Ditto.
34256 (_mm256_mmask_i64gather_epi64): Ditto.
34257 (_mm_mmask_i64gather_epi64): Ditto.
34258 (_mm256_i32scatter_ps): Change __addr type to void*.
34259 (_mm256_mask_i32scatter_ps): Ditto.
34260 (_mm_i32scatter_ps): Ditto.
34261 (_mm_mask_i32scatter_ps): Ditto.
34262 (_mm256_i32scatter_pd): Ditto.
34263 (_mm256_mask_i32scatter_pd): Ditto.
34264 (_mm_i32scatter_pd): Ditto.
34265 (_mm_mask_i32scatter_pd): Ditto.
34266 (_mm256_i64scatter_ps): Ditto.
34267 (_mm256_mask_i64scatter_ps): Ditto.
34268 (_mm_i64scatter_ps): Ditto.
34269 (_mm_mask_i64scatter_ps): Ditto.
34270 (_mm256_i64scatter_pd): Ditto.
34271 (_mm256_mask_i64scatter_pd): Ditto.
34272 (_mm_i64scatter_pd): Ditto.
34273 (_mm_mask_i64scatter_pd): Ditto.
34274 (_mm256_i32scatter_epi32): Ditto.
34275 (_mm256_mask_i32scatter_epi32): Ditto.
34276 (_mm_i32scatter_epi32): Ditto.
34277 (_mm_mask_i32scatter_epi32): Ditto.
34278 (_mm256_i32scatter_epi64): Ditto.
34279 (_mm256_mask_i32scatter_epi64): Ditto.
34280 (_mm_i32scatter_epi64): Ditto.
34281 (_mm_mask_i32scatter_epi64): Ditto.
34282 (_mm256_i64scatter_epi32): Ditto.
34283 (_mm256_mask_i64scatter_epi32): Ditto.
34284 (_mm_i64scatter_epi32): Ditto.
34285 (_mm_mask_i64scatter_epi32): Ditto.
34286 (_mm256_i64scatter_epi64): Ditto.
34287 (_mm256_mask_i64scatter_epi64): Ditto.
34288 (_mm_i64scatter_epi64): Ditto.
34289 (_mm_mask_i64scatter_epi64): Ditto.
34290 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34291 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34292 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34293 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34294 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34295 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34296 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34297 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34298 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34299 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34300 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34301 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34302 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34303 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34304 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34305 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34306 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34307 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34308 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34309 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34310 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34311 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34312 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34313 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34314 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34315 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34316 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34317 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34318 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34319 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34320 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34321 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34322 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34323 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34324 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34325 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34326 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34327 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34328 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34329 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34330 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34331 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34332 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34333 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34334 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34335 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34336 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34337 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34338 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34339 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34340 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34341 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34342 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34343 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34344 definitions accordingly.
34346 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34347 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34350 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34353 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34356 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34357 hard regs for inner regno.
34359 2017-01-17 Martin Liska <mliska@suse.cz>
34362 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34363 assumption and add comment.
34365 2017-01-17 Nathan Sidwell <nathan@acm.org>
34367 * ipa-visibility.c (localize_node): New function, broken out of ...
34368 (function_and_variable_visibility): ... here. Call it.
34370 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34372 PR middle-end/77445
34373 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34374 correctly set frequency of oudgoing edge.
34375 (duplicate_thread_path): Fix profile updating.
34377 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34380 * configure.ac: Add GCC_BASE_VER.
34381 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34382 version from BASE-VER file.
34383 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34384 (gcc.o): Depend on $(BASEVER).
34385 * common.opt (dumpfullversion): New option.
34386 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34387 * doc/invoke.texi: Document -dumpfullversion.
34388 * doc/install.texi: Document --with-gcc-major-version-only.
34389 * configure: Regenerated.
34391 2017-01-17 Richard Biener <rguenther@suse.de>
34393 PR tree-optimization/71433
34394 * tree-vrp.c (register_new_assert_for): Merge same asserts
34395 on all incoming edges.
34396 (process_assert_insertions_for): Handle insertions at the
34399 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34401 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34402 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34404 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34407 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34410 2017-01-17 Alan Modra <amodra@gmail.com>
34413 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34414 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34415 symbolic stack limit when pic.
34417 2017-01-16 Martin Sebor <msebor@redhat.com>
34419 PR tree-optimization/78608
34420 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34422 2017-01-16 Jeff Law <law@redhat.com>
34425 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34426 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34427 for several include directories that may be relative to sysroot.
34428 * config/i386/x-mingw32 (gplus_includedir): Define.
34429 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34430 (native_system_includedir): Likewise.
34431 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34432 override if TARGET_SYSTEM_ROOT is defined.
34433 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34435 PR tree-optimization/79090
34436 PR tree-optimization/33562
34437 PR tree-optimization/61912
34438 PR tree-optimization/77485
34439 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34440 and computed trims into the dump file.
34442 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34444 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34446 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34449 * gimplify.c (gimplify_init_constructor): If want_value and
34450 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34454 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34455 sequence. Formatting fixes.
34456 (doloop_optimize): Formatting fixes.
34459 * gcc.c (debug_level_greater_than_spec_func): New function.
34460 (static_spec_functions): Add debug-level-gt spec function.
34461 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34463 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34464 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34465 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34466 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34467 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34468 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34470 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34472 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34473 QImode fixups to general and mask registers only.
34475 2017-01-16 Carl Love <cel@us.ibm.com>
34477 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34478 for built-in functions
34479 vector signed char vec_nabs (vector signed char)
34480 vector signed short vec_nabs (vector signed short)
34481 vector signed int vec_nabs (vector signed int)
34482 vector signed long long vec_nabs (vector signed long long)
34483 vector float vec_nabs (vector float)
34484 vector double vec_nabs (vector double)
34485 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34487 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34488 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34489 * doc/extend.texi: Update the documentation file for the new built-in
34492 2017-01-16 Martin Sebor <msebor@redhat.com>
34494 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34497 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34499 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34500 UNSPEC_VSX__XXSPLTD to require special splat handling.
34502 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34505 * system.h: Poison strndup.
34507 2017-01-16 Alan Modra <amodra@gmail.com>
34510 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34513 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34515 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34517 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34519 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34520 call recog here. Assert that INSN_CODE (insn) is non-negative.
34522 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34525 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34527 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34528 in the currently scheduled RTL fragment.
34530 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34532 PR rtl-optimization/78751
34533 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34536 2017-01-14 Jeff Law <law@redhat.com>
34538 PR tree-optimization/79090
34539 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34540 variable length stores.
34541 (compute_trims): Delete dead assignment to *trim_tail.
34542 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34545 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34547 PR rtl-optimization/78626
34548 PR rtl-optimization/78727
34549 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34550 of a block, and split such blocks after everything else is finished.
34552 2017-01-14 Alan Modra <amodra@gmail.com>
34555 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34556 target legitimate_combined_insn.
34557 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34558 (rs6000_legitimate_combined_insn): New function.
34559 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34561 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34562 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34563 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34565 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34567 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34569 2017-01-13 Jeff Law <law@redhat.com>
34571 PR tree-optimization/33562
34572 PR tree-optimization/61912
34573 PR tree-optimization/77485
34574 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34576 (delete_dead_assignment): Likewise.
34577 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34578 statement to delete_dead_call and delete_dead_assignment.
34580 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34583 * substring-locations.c (format_warning_va): Strengthen case 1 so
34584 that both endpoints of the substring must be within the format
34585 range for just the substring to be printed.
34587 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34589 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34590 * config/i386/i386.c (ix86_target_string): Add missing options
34591 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34592 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34593 flags_other and ix86_target_other to flags2_other. Display unknown
34595 (ix86_valid_target_attribute_inner_p): Add missing options and
34596 reorder options by implied ISAs, as in ix86_target_string.
34598 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34600 * hash-table.h (hash_table::too_empty_p): New function.
34601 (hash_table::expand): Use it.
34602 (hash_table::traverse): Likewise.
34603 (hash_table::empty_slot): Use sizeof (value_type) instead of
34604 sizeof (PTR) to convert bytes to elements. Shrink the table
34605 if the current size is excessive for the current number of
34608 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34610 * ira-costs.c (record_reg_classes): Break from the inner loop
34611 early once alt_fail is known to be true. Update outer loop
34612 handling accordingly.
34614 2017-01-13 Jeff Law <law@redhat.com>
34616 * tree-ssa-dse.c (decrement_count): New function.
34617 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34618 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34619 when we know the partially dead statement is a mem* function.
34621 PR tree-optimization/61912
34622 PR tree-optimization/77485
34623 * tree-ssa-dse.c: Include expr.h.
34624 (maybe_trim_constructor_store): New function.
34625 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34627 PR tree-optimization/33562
34628 PR tree-optimization/61912
34629 PR tree-optimization/77485
34630 * doc/invoke.texi: Document new dse-max-object-size param.
34631 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34632 * tree-ssa-dse.c: Include params.h.
34633 (dse_store_status): New enum.
34634 (initialize_ao_ref_for_dse): New, partially extracted from
34636 (valid_ao_ref_for_dse, normalize_ref): New.
34637 (setup_live_bytes_from_ref, compute_trims): Likewise.
34638 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34639 (maybe_trim_partially_dead_store): Likewise.
34640 (maybe_trim_complex_store): Likewise.
34641 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34642 Track what bytes live from the original store. Return tri-state
34643 for dead, partially dead or live.
34644 (dse_dom_walker): Add constructor, destructor and new private members.
34645 (delete_dead_call, delete_dead_assignment): New extracted from
34647 (dse_optimize_stmt): Make a member of dse_dom_walker.
34648 Use initialize_ao_ref_for_dse.
34650 PR tree-optimization/33562
34651 PR tree-optimization/61912
34652 PR tree-optimization/77485
34653 * sbitmap.h (bitmap_count_bits): Prototype.
34654 (bitmap_clear_range, bitmap_set_range): Likewise.
34655 * sbitmap.c (bitmap_clear_range): New function.
34656 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34658 2017-01-13 Martin Liska <mliska@suse.cz>
34661 * function.c (set_cfun): Add new argument force.
34662 * function.h (set_cfun): Likewise.
34663 * ipa-inline-transform.c (inline_call): Use the function when
34664 strict alising from is dropped for function we inline to.
34666 2017-01-13 Richard Biener <rguenther@suse.de>
34668 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34669 for dumping GIMPLE INTEGER_CSTs.
34671 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34673 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34674 to 201112L since C++17.
34676 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34679 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34680 if -fsanitize=kernel-address is present.
34682 2017-01-13 Richard Biener <rguenther@suse.de>
34684 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34685 as _Literal ( type ) number in case usual suffixes do not
34686 preserve all information.
34688 2017-01-13 Richard Biener <rguenther@suse.de>
34690 PR tree-optimization/77283
34691 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34692 and ssa-iterators.h.
34693 (is_feasible_trace): Implement a cost model based on joiner
34696 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34699 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34700 char or short to __float128/_Float128 directly.
34702 2017-01-12 Martin Sebor <msebor@redhat.com>
34704 to -Wformat-overflow.
34705 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34706 (min_bytes_remaining): Same.
34707 (get_string_length): Same.
34708 (format_string): Same.
34709 (format_directive): Same.
34711 (pass_sprintf_length::handle_gimple_call): Same.
34713 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34715 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34716 info.nowrite calls with no lhs that can't throw. Return bool
34717 whether gsi_remove has been called or not.
34718 (pass_sprintf_length::handle_gimple_call): Return bool whether
34719 try_substitute_return_value called gsi_remove. Formatting fix.
34720 (pass_sprintf_length::execute): Don't use gsi_remove if
34721 handle_gimple_call returned true.
34724 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34725 be removed due to side-effects, don't remove following barrier nor
34726 turn the successor edge into fallthru edge.
34728 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34731 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34732 element-reversing loads and stores as not swappable.
34734 2017-01-12 Nathan Sidwell <nathan@acm.org>
34735 Nicolai Stange <nicstange@gmail.com>
34737 * combine.c (try_combine): Don't ignore result of overlap checking
34738 loop. Combine overlap & asm check into single loop.
34740 2017-01-12 Richard Biener <rguenther@suse.de>
34742 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34743 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34745 2017-01-12 Richard Biener <rguenther@suse.de>
34747 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34748 and TS_TARGET_OPTION directly derive from TS_BASE.
34749 * tree-core.h (tree_optimization_option): Derive from tree_base.
34750 (tree_target_option): Likewise.
34752 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34754 * config/i386/i386.c (memory_address_length): Increase len
34755 only when rip_relative_addr_p returns false.
34757 2017-01-11 Julia Koval <julia.koval@intel.com>
34759 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34760 (OPTION_MASK_ISA_SGX_SET): New.
34761 (ix86_handle_option): Handle OPT_msgx.
34762 * config.gcc: Added sgxintrin.h.
34763 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34764 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34765 * config/i386/i386.c (ix86_target_string): Add -msgx.
34767 (ix86_option_override_internal): Handle new options.
34768 (ix86_valid_target_attribute_inner_p): Add sgx.
34769 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34770 * config/i386/i386.opt: Add msgx.
34771 * config/i386/sgxintrin.h: New file.
34772 * config/i386/x86intrin.h: Add sgxintrin.h.
34774 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34777 * fold-const.c (maybe_nonzero_address): Return 1 for function
34779 (tree_single_nonzero_warnv_p): Don't handle function local objects
34783 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34786 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34789 * opts.c: Include "spellcheck.h"
34790 (struct string_fragment): New struct.
34791 (struct edit_distance_traits<const string_fragment &>): New
34793 (get_closest_sanitizer_option): New function.
34794 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34796 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34798 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34800 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34801 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34802 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34803 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34804 for initial die_offset if dwarf_split_debug_info.
34805 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34806 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34808 (output_skeleton_debug_sections): Formatting fix. Use
34809 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34810 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34812 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34814 * config/arm/cortex-a53.md: Add bypasses for
34815 cortex_a53_r2f_cvt.
34816 (cortex_a53_r2f): Only use for transfers.
34817 (cortex_a53_f2r): Likewise.
34818 (cortex_a53_r2f_cvt): Add reservation for conversions.
34819 (cortex_a53_f2r_cvt): Likewise.
34821 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34823 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34824 to all inlined functions, change static to extern.
34826 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34829 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34831 (arm_assemble_integer): Likewise.
34833 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34835 * config.gcc: Use new awk script to check CPU, FPU and architecture
34836 parameters for --with-... options.
34837 * config/arm/parsecpu.awk: New file
34838 * config/arm/arm-cpus.in: New file.
34839 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34841 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34843 * config/arm/t-arm: Update dependency rules.
34844 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34845 of processing .def files.
34846 * config/arm/genopt.sh: Deleted.
34847 * config/arm/gentune.sh: Deleted.
34848 * config/arm/arm-cores.def: Deleted.
34849 * config/arm/arm-arches.def: Deleted.
34850 * config/arm/arm-fpus.def: Deleted.
34851 * config/arm/arm-tune.md: Regenerated.
34852 * config/arm/arm-tables.opt: Regenerated.
34853 * config/arm/arm-cpu.h: New generated file.
34854 * config/arm/arm-cpu-data.h: New generated file.
34855 * config/arm/arm-cpu-cdata.h: New generated file.
34857 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34860 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34862 (input_varpool_node): Unpack dynamically_initialized bit.
34864 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34866 PR rtl-optimization/79032
34867 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34868 the alignment of the adjusted memory reference against that of MODE,
34869 instead of the alignment of the original memory reference.
34871 2017-01-11 Martin Jambor <mjambor@suse.cz>
34873 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34875 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34876 decorated functions.
34878 2017-01-11 Richard Biener <rguenther@suse.de>
34880 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34881 set range/nonnull info for PHI results. Do not set it on
34882 stmts marked for removal.
34884 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34886 * expr.c (store_field): In the bitfield case, fetch the return value
34887 from the registers before applying a single big-endian adjustment.
34888 Always do a final load for a BLKmode value not larger than a word.
34890 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34893 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34894 that we correctly handle column numbers greater than
34895 LINE_MAP_MAX_COLUMN_NUMBER.
34897 2017-01-10 Martin Sebor <msebor@redhat.com>
34899 PR middle-end/78245
34900 * gimple-ssa-sprintf.c (get_destination_size): Call
34901 {init,fini}object_sizes.
34902 * tree-object-size.c (addr_object_size): Adjust.
34903 (pass_through_call): Adjust.
34904 (pass_object_sizes::execute): Adjust.
34905 * tree-object-size.h (fini_object_sizes): Declare.
34907 2017-01-10 Martin Sebor <msebor@redhat.com>
34909 PR tree-optimization/78775
34910 * builtins.c (get_size_range): Move...
34911 * calls.c: ...to here.
34912 (alloc_max_size): Accept zero argument.
34913 (operand_signed_p): Remove.
34914 (maybe_warn_alloc_args_overflow): Call get_size_range.
34915 * calls.h (get_size_range): Declare.
34917 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34919 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34920 from TI's devices.csv file as of September 2016.
34921 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34923 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34925 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34926 * doc/invoke.texi: Likewise.
34927 * doc/md.texi: Likewise.
34928 * doc/objc.texi: Likewise.
34930 2017-01-10 Joshua Conner <joshconner@google.com>
34932 * config/arm/fuchsia-elf.h: New file.
34933 * config/fuchsia.h: New file.
34934 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34935 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34937 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34939 2016-01-10 Richard Biener <rguenther@suse.de>
34941 PR tree-optimization/79034
34942 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34943 Propagate out degenerate PHIs in the joiner.
34945 2017-01-10 Martin Liska <mliska@suse.cz>
34947 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34948 (sort_congruence_classes_by_decl_uid): Likewise.
34949 (sort_congruence_class_groups_by_decl_uid): Likewise.
34950 (sem_item_optimizer::merge_classes): Sort class, groups in these
34951 classes and members in the groups by DECL_UID of declarations.
34952 This would make merge operations stable.
34954 2017-01-10 Martin Liska <mliska@suse.cz>
34956 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34957 usage of m_classes_vec.
34958 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34959 (sem_item_optimizer::get_group_by_hash): Likewise.
34960 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34961 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34962 (sem_item_optimizer::verify_classes): Likewise.
34963 (sem_item_optimizer::process_cong_reduction): Likewise.
34964 (sem_item_optimizer::dump_cong_classes): Likewise.
34965 (sem_item_optimizer::merge_classes): Likewise.
34966 * ipa-icf.h (congruence_class_hash): Rename from
34967 congruence_class_group_hash. Remove declaration of m_classes_vec.
34969 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34971 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34972 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34973 * config.gcc: Add avx512vpopcntdqintrin.h.
34974 * config/i386/avx512vpopcntdqintrin.h: New.
34975 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34976 * config/i386/i386-builtin-types.def: Add new types.
34977 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34978 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34979 __builtin_ia32_vpopcountq_v8di_mask): New.
34980 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34981 __AVX512VPOPCNTDQ__.
34982 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34983 (PTA_AVX512VPOPCNTDQ): Define.
34984 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34985 TARGET_AVX512VPOPCNTDQ_P): Define.
34986 * config/i386/i386.opt: Add mavx512vpopcntdq.
34987 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34988 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34990 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34992 PR middle-end/77484
34993 * predict.def (PRED_CALL): Set to 67.
34995 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34997 * expr.c (store_field): In the bitfield case, if the value comes from
34998 a function call and is of an aggregate type returned in registers, do
34999 not modify the field mode; extract the value in all cases if the mode
35000 is BLKmode and the size is not larger than a word.
35002 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
35005 * config/i386/cpuid.h: Fix undefined behavior.
35007 2017-01-04 Jeff Law <law@redhat.com>
35009 PR tree-optimization/79007
35010 PR tree-optimization/67955
35011 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
35012 conservative for pt.null when flag_non_call_exceptions is on.
35014 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35016 PR translation/79019
35017 PR translation/79020
35018 * params.def (PARAM_INLINE_MIN_SPEEDUP,
35019 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
35020 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
35022 * config/avr/avr.opt (maccumulate-args): Likewise.
35023 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
35024 * common.opt (freport-bug): Likewise.
35025 * cif-code.def (CIF_FINAL_ERROR): Likewise.
35026 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
35027 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
35028 translatable string.
35029 * config/i386/i386.c (function_value_32): Likewise.
35030 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
35031 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
35033 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
35034 * common/config/msp430/msp430-common.c (msp430_handle_option):
35036 * symtab.c (symtab_node::verify_base): Likewise.
35037 * opts.c (set_debug_level): Likewise.
35038 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
35039 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
35040 missing whitespace to translatable strings.
35041 * config/avr/avr.md (bswapsi2): Fix typo in comment.
35042 * config/sh/superh.h: Likewise.
35043 * config/i386/xopintrin.h: Likewise.
35044 * config/i386/znver1.md: Likewise.
35045 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
35046 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
35047 * double-int.h (struct double_int): Likewise.
35048 * double-int.c (div_and_round_double): Likewise.
35049 * wide-int.cc: Likewise.
35050 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
35051 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
35052 * cfgcleanup.c (crossjumps_occured): Renamed to ...
35053 (crossjumps_occurred): ... this.
35054 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
35057 PR tree-optimization/78899
35058 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
35059 returning bool return struct loop *, NULL for failure and the new
35061 (versionable_outer_loop_p): Don't version outer loop if it has
35062 dont_vectorized bit set.
35063 (tree_if_conversion): When versioning outer loop, ensure
35064 tree_if_conversion is performed also on the inner loop of the
35065 non-vectorizable outer loop copy.
35066 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
35067 LOOP_VECTORIZED in inner loop of the scalar outer loop and
35068 prevent vectorization of it.
35069 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
35070 the outer loop vectorization of the non-scalar version is attempted
35071 before vectorization of the inner loop in scalar version. If
35072 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
35073 vectorization of its inner loop.
35074 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
35075 has 2 inner loops, rename also on edges from bb whose single pred
35076 is outer_loop->header. Fix typo in function comment.
35078 2017-01-09 Martin Sebor <msebor@redhat.com>
35081 * asan.c (asan_emit_stack_protection): Increase local buffer size
35082 to avoid snprintf truncation warning.
35084 2017-01-09 Andrew Pinski <apinski@cavium.com>
35086 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
35087 to reference thunderx2t99 for the tuning structure
35088 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
35090 (thunderx2t99_extra_costs): This.
35091 * config/aarch64/aarch64-tune.md: Regenerate.
35092 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
35093 (vulcan_addrcost_table): This.
35094 (vulcan_regmove_cost): Rename to ...
35095 (thunderx2t99_regmove_cost): This.
35096 (vulcan_vector_cost): Rename to ...
35097 (thunderx2t99_vector_cost): this.
35098 (vulcan_branch_cost): Rename to ...
35099 (thunderx2t99_branch_cost): This.
35100 (vulcan_tunings): Rename to ...
35101 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35102 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35104 2017-01-09 Martin Jambor <mjambor@suse.cz>
35108 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35109 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35110 (propagate_vr_accross_jump_function): Use the above function for all
35111 value range computations for pass-through jump functions and type
35112 converasion from explicit value range values.
35113 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35114 parameters from TYPE_ARG_TYPES.
35115 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35116 (ipa_write_node_info): Stream type of the actual argument.
35117 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35119 2017-01-09 Martin Liska <mliska@suse.cz>
35122 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35123 (lookup_compiler): Do not show error message with have_E.
35125 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35127 PR tree-optimization/78938
35128 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35129 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35130 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35131 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35134 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35136 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35139 2017-01-09 Richard Biener <rguenther@suse.de>
35141 PR tree-optimization/78997
35142 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35143 name condition properly.
35145 2017-01-09 Richard Biener <rguenther@suse.de>
35148 * dwarf2out.c (is_cxx): New overload with context.
35149 (is_naming_typedef_decl): Use it.
35151 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35153 * invoke.texi (Option Summary): Correct spacing in option lists
35154 and add line breaks to fix over-long lines.
35156 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35158 PR middle-end/17660
35160 * extend.texi (Common Variable Attributes): Add xref to GCC
35161 Internals manual to explain mode attribute keywords.
35163 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35166 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35167 and Preprocessor Options.
35168 (Options for Linking): Document -pthread here....
35169 (RS/6000 and PowerPC Options): ...not here.
35170 (Solaris 2 Options): ...or here.
35171 * doc/cppopts.texi: Document -pthread.
35173 2017-01-08 Martin Sebor <msebor@redhat.com>
35175 PR middle-end/77708
35176 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35177 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35178 New member functions.
35179 (format_directive): Used them.
35181 (pass_sprintf_length::handle_gimple_call): Same.
35182 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35183 to avoid truncation for any argument.
35184 (extract_affine_mul): Same.
35185 * tree.c (get_file_function_name): Same.
35187 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35189 PR middle-end/77484
35190 * predict.def (PRED_INDIR_CALL): Set to 86.
35192 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35194 PR preprocessor/54124
35195 * doc/cppopts.texi: Reformat -d subtable to list the full name
35196 of the options. Add cross-reference to the docs for the general
35197 compiler -d options.
35198 * doc/invoke.texi (Developer Options): Add cross-reference to the
35199 preprocessor-specific -d option documentation.
35201 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35203 PR preprocessor/13498
35204 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35205 redudant material, and reflect new command-line options.
35206 (System Headers): Likewise.
35208 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35210 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35211 -isystem, and -idirafter. Copy-edit.
35212 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35213 default for -ftrack-macro-expansion. Delete obsolete and
35214 badly-formatted implementation details about -fdebug-cpp output.
35215 * doc/cppwarnopts.texi: Copy-edit.
35217 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35220 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35221 that the transition from a max line width >= 1<<10 to narrower
35222 lines works correctly.
35224 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35226 * doc/options.texi (PerFunction): New.
35227 * opt-functions.awk (switch_flags): Map both Optimization and
35228 PerFunction to CL_OPTIMIZATION.
35229 * opth-gen.awk: Test for PerFunction flag along with
35231 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35232 it only when the latter is present. Skip those that don't in
35233 the hash function generator.
35234 * common.opt (fvar-tracking): Mark as PerFunction instead of
35236 (fvar-tracking-assignments): Likewise.
35237 (fvar-tracking-assignments-toggle): Likewise.
35238 (fvar-tracking-uninit): Likewise.
35240 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35242 PR translation/79018
35243 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35246 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35249 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35250 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35251 TARGET_LONG_JUMP_TABLE_OFFSETS.
35252 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35253 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35254 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35255 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35256 * config/m68k/m68k.md (tablejump expander): Likewise.
35257 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35258 TARGET_LONG_JUMP_TABLE_OFFSETS.
35259 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35260 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35262 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35263 David Holsgrove <david.holsgrove@xilinx.com>
35265 * common/config/microblaze/microblaze-common.c
35266 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35267 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35269 * config/microblaze/microblaze.c (microblaze_must_save_register)
35270 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35272 (microblaze_eh_return): New function.
35273 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35274 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35275 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35276 * config/microblaze/microblaze.md (eh_return): New pattern.
35278 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35280 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35281 GCC_DIAGNOSTIC_STRINGIFY): Define.
35283 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35285 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35287 * config/arm/arm.md (<mcrr>): New.
35289 * config/arm/arm.c (arm_arch5te): New.
35290 (arm_option_override): Set arm_arch5te.
35291 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35293 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35294 (arm_mcrr_qualifiers): ... this. New.
35295 (MRRC_QUALIFIERS): Define to...
35296 (arm_mrrc_qualifiers): ... this. New.
35297 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35299 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35300 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35301 (MRRCI, mrrc, MRRC): New.
35302 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35303 VUNSPEC_MRRC2): New.
35305 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35307 * config/arm/arm.md (<mcr>): New.
35309 * config/arm/arm.c (arm_coproc_builtin_available): Add
35310 support for mcr, mrc, mcr2 and mrc2.
35311 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35312 (arm_mcr_qualifiers): ... this. New.
35313 (MRC_QUALIFIERS): Define to ...
35314 (arm_mrc_qualifiers): ... this. New.
35315 (MCR_QUALIFIERS): Define to ...
35316 (arm_mcr_qualifiers): ... this. New.
35317 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35319 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35320 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35321 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35322 VUNSPEC_MRC2): New.
35324 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35326 * config/arm/arm.md (*ldc): New.
35330 * config/arm/arm.c (arm_coproc_builtin_available): Add
35331 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35332 (arm_coproc_ldc_stc_legitimate_address): New.
35333 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35334 'qualifier_const_pointer'.
35335 (LDC_QUALIFIERS): Define to...
35336 (arm_ldc_qualifiers): ... this. New.
35337 (STC_QUALIFIERS): Define to...
35338 (arm_stc_qualifiers): ... this. New.
35339 * config/arm/arm-protos.h
35340 (arm_coproc_ldc_stc_legitimate_address): New.
35341 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35342 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35343 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35344 stc2, stcl, stc2l): New.
35345 * config/arm/constraints.md (Uz): New.
35346 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35347 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35348 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35349 VUNSPEC_STC2L): New.
35351 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35353 * config/arm/arm.md (<cdp>): New.
35354 * config/arm/arm.c (neon_const_bounds): Rename this ...
35355 (arm_const_bounds): ... this.
35356 (arm_coproc_builtin_available): New.
35357 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35358 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35359 (CDP_QUALIFIERS): Define to...
35360 (arm_cdp_qualifiers): ... this. New.
35362 (arm_expand_builtin_args): Add case for 6 arguments.
35363 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35364 (arm_const_bounds): ... this.
35365 (arm_coproc_builtin_available): New.
35366 * config/arm/arm_acle.h (__arm_cdp): New.
35368 * config/arm/arm_acle_builtins.def (cdp): New.
35370 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35371 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35372 'arm_const_bounds'.
35373 * config/arm/types.md (coproc): New.
35374 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35375 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35376 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35377 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35379 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35381 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35382 (UBINOP_QUALIFIERS): New.
35384 (acle_builtin_data): New. Change comment.
35385 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35386 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35387 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35388 arm_acle_builtins.def.
35389 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35390 (arm_init_acle_builtins): New.
35391 (CRC32_BUILTIN): Remove.
35392 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35393 crc32cb, crc32ch and crc32cw.
35394 (arm_init_crc32_builtins): Remove.
35395 (arm_init_builtins): Use arm_init_acle_builtins rather
35396 than arm_init_crc32_builtins.
35397 (arm_expand_acle_builtin): New.
35398 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35399 * config/arm/arm_acle_builtins.def: New.
35401 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35403 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35404 (arm_builtin_datum): ... this.
35405 (arm_init_neon_builtin): Rename to ...
35406 (arm_init_builtin): ... this. Add a new parameters PREFIX
35407 and USE_SIG_IN_NAME.
35408 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35409 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35410 'arm_builtin_datum'.
35411 (arm_init_vfp_builtins): Likewise.
35412 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35413 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35414 (arm_expand_neon_args): Rename to ...
35415 (arm_expand_builtin_args): ... this. Rename builtin_arg
35416 enum values and differentiate between ARG_BUILTIN_MEMORY
35417 and ARG_BUILTIN_NEON_MEMORY.
35418 (arm_expand_neon_builtin_1): Rename to ...
35419 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35420 values, arm_expand_builtin_args and add bool parameter NEON.
35421 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35422 (arm_expand_vfp_builtin): Likewise.
35423 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35425 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35427 PR middle-end/77484
35428 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35429 * predict.c (tree_estimate_probability_bb): Reverse direction of
35430 polymorphic call predictor.
35432 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35434 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35435 (determine_pass_name_match): ...this new function and...
35436 (should_skip_pass_p): ...this new function.
35438 2017-01-06 Nathan Sidwell <nathan@acm.org>
35440 * ipa-visibility.c (function_and_variable_visibility): Reformat
35441 comments and long lines. Remove extrneous if.
35442 * symtab.c (symtab_node::make_decl_local): Fix code format.
35443 (symtab_node::set_section_for_node): Fix comment typo.
35445 2017-01-06 Martin Liska <mliska@suse.cz>
35448 * lra-constraints.c: Rename invariant to lra_invariant.
35449 * predict.c (set_even_probabilities): Initialize e to NULL.
35451 2017-01-05 Martin Sebor <msebor@redhat.com>
35453 PR tree-optimization/78910
35454 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35455 (format_integer): Correct off-by-one error in the handling
35456 of precision with negative numbers in signed conversions..
35458 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35460 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35462 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35464 PR tree-optimization/71016
35465 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35466 factor_out_conditional_conversion. Formatting fix.
35467 (factor_out_conditional_conversion): Add cond_stmt argument.
35468 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35469 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35472 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35474 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35475 read-rtl-function.o, and selftest-rtl.o.
35476 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35477 (selftest::aarch64_test_loading_full_dump): New function.
35478 (selftest::aarch64_run_selftests): New function.
35479 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35480 selftest::aarch64_run_selftests.
35481 * config/i386/i386.c
35482 (selftest::ix86_test_loading_dump_fragment_1): New function.
35483 (selftest::ix86_test_loading_call_insn): New function.
35484 (selftest::ix86_test_loading_full_dump): New function.
35485 (selftest::ix86_test_loading_unspec): New function.
35486 (selftest::ix86_run_selftests): Call the new functions.
35487 * emit-rtl.c (maybe_set_max_label_num): New function.
35488 * emit-rtl.h (maybe_set_max_label_num): New decl.
35489 * function.c (instantiate_decls): Guard call to
35490 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35491 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35493 * gensupport.c (gen_reader::gen_reader): Pass "false"
35494 for new "compact" param of rtx_reader.
35495 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35496 rather than an empty string for NULL strings.
35497 * read-md.c: Potentially include config.h rather than bconfig.h.
35498 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35499 (have_error): New global, copied from errors.c.
35500 (md_reader::read_name): Rename to...
35501 (md_reader::read_name_1): ...this, adding "out_loc" param,
35502 and converting "missing name or number" to returning false, rather
35504 (md_reader::read_name): Reimplement in terms of read_name_1.
35505 (md_reader::read_name_or_nil): New function.
35506 (md_reader::read_string): Handle "(nil)" by returning NULL.
35507 (md_reader::md_reader): Add new param "compact".
35508 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35509 (md_reader::read_file): New method.
35510 * read-md.h (md_reader::md_reader): Add new param "compact".
35511 (md_reader::read_file): New method.
35512 (md_reader::is_compact): New accessor.
35513 (md_reader::read_name): Convert return type from void to file_location.
35514 (md_reader::read_name_or_nil): New decl.
35515 (md_reader::read_name_1): New decl.
35516 (md_reader::m_compact): New field.
35517 (noop_reader::noop_reader): Pass "false" for new "compact" param
35519 (rtx_reader::rtx_reader): Add new "compact" param.
35520 (rtx_reader::read_rtx_operand): Make virtual and convert return
35521 type from void to rtx.
35522 (rtx_reader::read_until): New decl.
35523 (rtx_reader::handle_any_trailing_information): New virtual function.
35524 (rtx_reader::postprocess): New virtual function.
35525 (rtx_reader::finalize_string): New virtual function.
35526 (rtx_reader::m_in_call_function_usage): New field.
35527 (rtx_reader::m_reuse_rtx_by_id): New field.
35528 * read-rtl-function.c: New file.
35529 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35530 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35531 (selftest::verify_three_block_rtl_cfg): New decl.
35532 * read-rtl-function.h: New file.
35533 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35534 For host, include function.h, memmodel.h, and emit-rtl.h.
35535 (one_time_initialization): New function.
35536 (struct compact_insn_name): New struct.
35537 (compact_insn_names): New array.
35538 (find_code): Handle insn codes in compact dumps.
35539 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35540 (bind_subst_iter_and_attr): Likewise.
35541 (add_condition_to_string): Likewise.
35542 (add_condition_to_rtx): Likewise.
35543 (apply_attribute_uses): Likewise.
35544 (add_current_iterators): Likewise.
35545 (apply_iterators): Likewise.
35546 (initialize_iterators): Guard usage of apply_subst_iterator with
35547 #ifdef GENERATOR_FILE.
35548 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35549 (md_reader::read_mapping): Likewise.
35550 (add_define_attr_for_define_subst): Likewise.
35551 (add_define_subst_attr): Likewise.
35552 (read_subst_mapping): Likewise.
35553 (check_code_iterator): Likewise.
35554 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35556 (one_time_initialization): New function.
35557 (rtx_reader::read_until): New method.
35558 (read_flags): New function.
35559 (parse_reg_note_name): New function.
35560 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35561 Handle reuse_rtx ids.
35562 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35563 Add parsing support for RTL dumps, mirroring the special-cases in
35564 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35565 values, and calling handle_any_trailing_information.
35566 (rtx_reader::read_rtx_operand): Convert return type from void
35567 to rtx, returning return_rtx. Handle case 'e'. Call
35568 finalize_string on XSTR and XTMPL fields.
35569 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35570 "(nil)" values were omitted. Call the postprocess vfunc on the
35572 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35573 class ctor. Initialize m_in_call_function_usage. Call
35574 one_time_initialization.
35575 * rtl-tests.c (selftest::test_uncond_jump): Call
35576 set_new_first_and_last_insn.
35577 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35578 * selftest-rtl.c: New file.
35579 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35580 (selftest::get_insn_by_uid): New decl.
35581 * selftest-run-tests.c (selftest::run_tests): Call
35582 read_rtl_function_c_tests.
35583 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35584 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35587 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35589 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35590 operands in a special way. Assert that pos+len <= mode precision.
35592 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35594 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35595 3 argument Alias with unlimited for the negative form.
35596 (fno-vect-cost-model): Removed.
35598 2017-01-05 Martin Liska <mliska@suse.cz>
35600 * hsa-gen.c (gen_hsa_divmod): New function.
35601 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35603 2017-01-05 Martin Liska <mliska@suse.cz>
35606 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35609 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35611 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35612 small constant length operands.
35614 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35616 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35617 between loop iterations.
35619 2017-01-05 Martin Liska <mliska@suse.cz>
35622 * gimplify.c (gimplify_decl_expr): Compare to
35623 asan_poisoned_variables instread of checking flags.
35624 (gimplify_target_expr): Likewise.
35625 (gimplify_expr): Likewise.
35626 (gimplify_function_tree): Conditionally initialize
35627 asan_poisoned_variables.
35629 2017-01-04 Jeff Law <law@redhat.com>
35631 PR tree-optimizatin/78812
35632 * rtl.h (contains_mem_rtx_p): Prototype.
35633 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35634 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35635 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35636 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35637 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35639 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35641 * input.c (assert_char_at_range): Default-initialize actual_range.
35643 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35645 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35646 to match the caller.
35648 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35650 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35651 insns after final jump in test to emit dummy move.
35653 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35655 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35656 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35658 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35660 * multiple_target.c (create_dispatcher_calls): Init e_next.
35661 * tree-ssa-loop-split.c (split_loop): Init border.
35662 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35665 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35670 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35671 (altivec_register_operand): Do not return true if the operand
35672 contains a SUBREG mixing SImode and SFmode.
35673 (vsx_register_operand): Likewise.
35674 (vsx_reg_sfsubreg_ok): New predicate.
35675 (vfloat_operand): Do not return true if the operand contains a
35676 SUBREG mixing SImode and SFmode.
35677 (vint_operand): Likewise.
35678 (vlogical_operand): Likewise.
35679 (gpc_reg_operand): Likewise.
35680 (int_reg_operand): Likewise.
35681 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35682 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35683 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35685 (rs6000_emit_move_si_sf_subreg): New helper function.
35686 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35687 fixup SUBREGs involving SImode and SFmode.
35688 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35689 numbers for the new peephole2 optimization.
35690 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35691 the GLIBC math library that do AND/IOR/XOR operations on single
35692 precision floating point.
35693 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35694 target macros to say whether we need to avoid SUBREGs mixing
35696 (TARGET_ALLOW_SF_SUBREG): Likewise.
35697 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35698 (UNSPEC_SI_FROM_SF): Likewise.
35699 (iorxor): Change spacing.
35700 (and_ior_xor): New iterator for AND, IOR, and XOR.
35701 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35702 (movdi_from_sf_zero_ext): Likewise.
35703 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35704 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35705 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35706 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35707 (fms<mode>4): Likewise.
35708 (fnma<mode>4): Likewise.
35709 (fnms<mode>4): Likewise.
35710 (nfma<mode>4): Likewise.
35711 (nfms<mode>4): Likewise.
35713 2017-01-04 Marek Polacek <polacek@redhat.com>
35716 * doc/invoke.texi: Document -Wpointer-compare.
35718 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35720 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35723 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35724 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35727 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35730 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35731 documentation of the powerpc_popcntb_ok attribute.
35732 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35733 code to issue warning messages if a requested CPU configuration is
35734 not supported by the binary (assembler and loader) toolchain.
35735 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35736 made to define a built-in function that has been disabled.
35737 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35738 made to define a built-in function that has been disabled.
35739 (altivec_init_builtins): Add comment explaining why definition
35740 of the DST built-in functions is not preceded by an assertion
35741 check. Add assertions to prevent ICE if attempts are made to
35742 define an altivec predicate or an abs* built-in function that has
35744 (htm_init_builtins): Add comment explaining why definition of the
35745 htm built-in functions is not preceded by an assertion check.
35747 2017-01-04 Jeff Law <law@redhat.com>
35749 PR tree-optimizatin/67955
35750 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35751 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35752 the points-to solution does not include pt_null. Use DECL_PT_UID
35755 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35757 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35758 Use gen_int_mode instead of gen_lopwart for const_int operands.
35760 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35762 PR tree-optimization/71563
35763 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35764 out of range value - has low bits known to be zero.
35766 2017-01-04 Alan Modra <amodra@gmail.com>
35768 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35769 * configure: Regenerate.
35770 * config.in: Regenerate.
35772 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35775 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35776 a substring of the message, but strcmp with the whole message. Ifdef
35777 ENABLE_NLS, translate the message first using dgettext.
35779 2017-01-03 Jeff Law <law@redhat.com>
35781 PR tree-optimizatin/78856
35782 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35783 (mark_threaded_blocks): Remove code to truncate thread paths that
35784 cross multiple loop headers. Instead invalidate the cached loop
35785 iteration information and handle case of a thread path walking
35786 into an irreducible region.
35788 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35791 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35792 assertions. Add support for doing the signbit if the IEEE 128-bit
35793 floating point value is in a GPR.
35794 * config/rs6000/rs6000.md (Fsignbit): Delete.
35795 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35796 Update the length attribute if the value is in a GPR.
35797 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35798 the sign or zero extension instruction, since the value is always 0/1.
35799 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35802 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35803 extracting SImode to a GPR register so that we can generate a
35804 store, limit the vector to be in a traditional Altivec register
35805 for the vextuwrx instruction.
35807 2017-01-03 Ian Lance Taylor <iant@google.com>
35809 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35811 2017-01-03 Martin Sebor <msebor@redhat.com>
35813 PR tree-optimization/78696
35814 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35815 precision. Use MPFR for %f for greater fidelity. Correct handling
35817 (pass_sprintf_length::compute_format_length): Set width and precision
35818 specified by asrerisk to void_node for vararg functions.
35819 (try_substitute_return_value): Adjust dump output.
35821 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35823 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35825 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35827 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35828 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35829 -mlra/-mno-lra was passed to the compiler.
35831 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35833 PR rtl-optimization/65618
35834 * emit-rtl.c (try_split): Move initialization of "before" and
35835 "after" to just before the call to emit_insn_after_setloc.
35837 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35839 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35841 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35843 * dwarf2out.c (gen_enumeration_type_die): When
35844 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35846 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35848 PR tree-optimization/78965
35849 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35850 Change first argument from const call_info & to call_info &. For %n
35851 set info.nowrite to false.
35853 PR middle-end/78901
35854 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35855 possibly throwing calls.
35857 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35858 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35859 and fns handling, rather than in a separate case SSA_NAME.
35861 2017-01-02 Jeff Law <law@redhat.com>
35863 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35864 fixes for first_period and second_period variables.
35866 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35869 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35870 (*insvqi_1): New insn pattern.
35871 (*insvqi_1_mem_rex64): Ditto.
35872 (*insvqi_2): Ditto.
35873 (*insvqi_3): Rename from *insvqi.
35875 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35877 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35879 * doc/cfg.texi (Edges): Remove reference to Java.
35880 (Maintaining the CFG): Ditto.
35882 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35884 PR middle-end/77674
35885 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35886 transparent aliases.
35888 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35890 PR middle-end/77484
35891 * predict.def (PRED_CALL): Update hitrate.
35892 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35893 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35894 into direct/indirect/polymorphic variants.
35896 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35898 Update copyright years.
35900 * gcc.c (process_command): Update copyright notice dates.
35901 * gcov-dump.c (print_version): Ditto.
35902 * gcov.c (print_version): Ditto.
35903 * gcov-tool.c (print_version): Ditto.
35904 * gengtype.c (create_file): Ditto.
35905 * doc/cpp.texi: Bump @copying's copyright year.
35906 * doc/cppinternals.texi: Ditto.
35907 * doc/gcc.texi: Ditto.
35908 * doc/gccint.texi: Ditto.
35909 * doc/gcov.texi: Ditto.
35910 * doc/install.texi: Ditto.
35911 * doc/invoke.texi: Ditto.
35913 Copyright (C) 2017 Free Software Foundation, Inc.
35915 Copying and distribution of this file, with or without modification,
35916 are permitted in any medium without royalty provided the copyright
35917 notice and this notice are preserved.